Air Quality and Ecological Impacts: Relating Sources to Effects 978-0-08-095201-7

Citation preview

Air Quality and Ecological Impacts: Relating Sources to Effects Edited by Allan H. Legge Biosphere Solutions, Calgary, Alberta, Canada

Amsterdam – Boston – Heidelberg – London – New York – Oxford Paris – San Diego – San Francisco – Singapore – Sydney – Tokyo

Elsevier 30 Corporate Drive, Suite 400, Burlington, MA 01803, USA Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands Linacre House, Jordan Hill, Oxford OX28DP, UK First edition 2009 Copyright r 2009 Elsevier Ltd. All rights reserved No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher. Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone (þ44) (0) 1865 843830; fax (þ44) (0) 1865 853333; email: [email protected]. Alternatively you can submit your request online by visiting the Elsevier web site at http://www.elsevier.com/locate/permissions, and selecting Obtaining permission to use Elsevier material Notice No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress ISBN: 978-0-08-095201-7 ISSN: 1474-8177 For information on all Elsevier publications visit our website at books.elsevier.com

Printed and bound in Hungary 09 10 11 12 13 10 9 8 7 6 5 4 3 2 1

Legge Cover Information From Left to Right Far left: Navajo Coal-Fired Power Generating Station near Page, Arizona. Middle top left: Acute foliar injury to the leaves of Wild Rose (Rosa acicularis Lindl.) exposed to elevated concentrations of SO2 when leaves were wet with water droplets is seen as reddish-brown interveinal necrotic areas with smooth and regular margins. When the leaves of Wild Rose are dry when exposed to elevated concentrations of SO2, the margins of the reddish-brown necrotic areas of acute foliar injury are very irregular rather than smooth. Middle bottom left: Acute foliar injury to alsike clover (Trifolium hybridum L.) from exposure to elevated concentrations of SO2 with injury seen as an ivory-colored marginal and interveinal necrosis extending towards the midrib of each of the three leaflets. Middle top right: Acute foliar injury to the needles of Lodgepole Pine (Pinus contorta Dougl. ex Loud.). Current year needles have only slight reddish-brown tip necrosis while second, third and fourth years needles have more injury. Third and fourth year needles are being prematurely dropped. Middle bottom right: Acute foliar injury to Balsam Poplar (Populus balsamifera L.) from chronic exposure to hydrogen fluoride (HF) with injury seen as a dark brown tip and marginal necrosis with an inner yellow band of chlorotic leaf tissue. Far right: Air quality monitoring station (background) and tower (foreground) measuring ambient air quality for selected air pollutants downwind of a sour-gas processing plant near Calgary, Alberta.

Dedication to Dr David F. Karnosky (1949–2008)

This book is dedicated to our friend and colleague Dr David F. Karnosky, who passed away suddenly on October 24, 2008. Dr Karnosky had an immense love of science, best manifested in his prodigious publication record spanning over 30 years. He established a truly global reputation for himself in forest science through sustained contributions in several disciplines including tree breeding, genetics,

physiology, and ecosystem processes. However, it was certainly David’s total commitment to the study of air pollution effects on forests that was his real passion. David exerted many significant leadership roles, most notably as Director of the Aspen Free Air Carbon Dioxide Enrichment (FACE) experiment in Rhinelander, Wisconsin. In 2008, Aspen FACE included some 100 scientists from 9 countries, and continues to serve as a platform for multidisciplinary research and student training. This world-class experiment on the impact of elevated levels of CO2 (carbon dioxide) and O3 (ozone) on forest ecosystems would neither have been initiated, nor continued over a decade of productive research without David’s vision, and his selfless contributions of time and effort. Dr Karnosky received his BS, MS, and PhD (1975) from the University of Wisconsin-Madison. In 2005, he received an IUFRO Scientific Achievement Award, and in 2006, an honorary doctorate from the University of Tartu (Estonia). David also held several patents, was Director of the Ecosystem Science Center at Michigan Technological University, in Houghton, Michigan and most recently, the Robbins Chair in Sustainable Management of the Environment. Despite his outstanding scientific achievements, David always managed to remain a humble and very generous individual, whose contribution to air pollution effects science and professional development of others will be clearly missed.

ix

List of Contributors Jean-Christophe Aznar

INRS-ETE, 490 de la Couronne, Quebec City, Quebec G1K 9A9, Canada; E-mail: [email protected]

Christian Be´gin

Natural Resources Canada, Geological Survey of Canada, 490 de la Couronne, Quebec City, Quebec G1K 9A9, Canada; E-mail: [email protected]

U. Brand

Department of Earth Sciences, Brock University, St. Catherines, Ontario L2S 3A1, Canada; E-mail: [email protected]

A. Bytnerowicz

USDA Forest Service, Pacific Southwest Research Station, 4955 Canyon Crest Drive, Riverside, CA 92507-6099, USA; E-mail: [email protected]

Vicent Calatayud

Fundacio´n CEAM, Parque Tecnolo´gico, Charles R. Darwin 14, Paterna, 46980 Valencia, Spain; E-mail: [email protected]

D.H. Campbell

US Geological Survey, MS 415, Water Resources Division, Denver Federal Center, Denver, CO 80225, USA; E-mail: [email protected]

J. Neil Cape

Centre for Ecology and Hydrology, Bush Estate, Penicuik, Midlothian EH26 0QB, UK; E-mail: [email protected]

D.W. Clow

US Geological Survey, MS 415, Water Resources Division, Denver Federal Center, Denver, CO 80225, USA; E-mail: [email protected]

x

List of Contributors

B. Craig

Ecological Monitoring and Assessment Network (EMAN), Environment Canada, Canada Centre for Inland Waters, 867 Lakeshore Road, Burlington, Ontario L7R 4A6, Canada; E-mail: [email protected]

Tom Dann

Environment Canada, Environmental Protection Service, Air Toxics Section, 355 River Road, Gloucester, Ontario K1A 0H3, Canada; E-mail: [email protected]

Luit J. De Kok

Laboratory of Plant Physiology, University of Groningen, P.O. Box 14, 9750 AA Haren, The Netherlands; E-mail: [email protected]

M.E. Fenn

USDA Forest Service, Pacific Southwest Research Station, 4955 Canyon Crest Drive, Riverside, CA 92507-6099, USA; E-mail: [email protected]

Warren Heilman

USDA Forest Service, Northern Research Station, 1407 South Harrison Road, East Lansing, MI 48823, USA; E-mail: [email protected]

Philip K. Hopke

Center for Air Resources Engineering and Science, and Department of Chemical Engineering, Clarkson University, Box 5708, Potsdam, NY 13699-5708, USA; E-mail: [email protected]

Hans-Ju¨rgen Ja¨ger

Institute for Plant Ecology, Heinrich-Buff-Ring 26-32, Justus-Liebig-University, D-35392, Giessen, Germany; E-mail: [email protected]. uni-giessen.de

David F. Karnosky

School of Forest Resources and Environmental Science, Michigan Technological University, 1400 Townsend Drive, Houghton, MI 49931-1295, USA; E-mail: [email protected]

Sagar V. Krupa

Department of Plant Pathology, University of Minnesota, 495 Borlaug Hall, 1991 Upper Buford Circle, St. Paul, MN 55108, USA; E-mail: [email protected]

xi

List of Contributors

Allan H. Legge

Biosphere Solutions, 1601, 11th Avenue NW, Calgary, Alberta T2N 1H1, Canada; E-mail: [email protected]

Joe¨lle Marion

INRS-ETE, 490 de la Couronne, Quebec City, Quebec G1K 9A9, Canada; E-mail: [email protected]

D.P. McCarthy

Department of Earth Sciences, Brock University, St. Catherines, Ontario L2S 3A1, Canada; E-mail: [email protected]

S.B. McLaughlin

Department of Ecology and Evolutionary Biology, University of Tennessee, Knoxville, TN 37996-0211, USA Current Address: 76 Briar Patch Lane, Raphine, VA 24472, USA; E-mail: [email protected]

Milla´n Milla´n

Fundacio´n CEAM, Parque Tecnolo´gico, Charles R. Darwin 14, Paterna, 46980 Valencia, Spain; E-mail: [email protected]

N.P. Molotch

Department of Civil and Environmental Engineering, 5732 Boelter Hall, University of California, Los Angeles, Ca 90095-1593, USA; E-mail: [email protected]

Miloslav Nosal

Department of Mathematics and Statistics, University of Calgary, Calgary, Alberta T2N 1N4, Canada; E-mail: [email protected]

P.E. Padgett

USDA Forest Service, Pacific Southwest Research Station, 4955 Canyon Crest Drive, Riverside, CA 92507-6099, USA; E-mail: [email protected]

Jose´ Luis Palau

Fundacio´n CEAM, Parque Tecnolo´gico, Charles R. Darwin 14, Paterna, 46980 Valencia, Spain; E-mail: [email protected]

Kevin E. Percy

Natural Resources Canada, Canadian Forest Service-Atlantic Forestry Centre, 1350 Regent Street, Fredericton, New Brunswick E3B 5P7, Canada; E-mail: [email protected]

xii

List of Contributors

J.E. Pleim

Atmospheric Science Modeling Division, Air Resources Laboratory, National Oceanic and Atmospheric Administration (NOAA), Research Triangle Park, NC 27711, USA; E-mail: [email protected]

Richard L. Poirot

Air Pollution Control Division, Vermont Department of Environmental Conservation, Building 3 South, 103 South Main Street, Waterbury, VT 05671-0402, USA; E-mail: [email protected]

Maria Sanz

Fundacio´n CEAM, Parque Tecnolo´gico, Charles R. Darwin 14, Paterna, 46980 Valencia, Spain; E-mail: [email protected]

Martine M. Savard

Natural Resources Canada, Geological Survey of Canada, 490 de la Couronne, Quebec City, Quebec G1K 9A9, Canada; E-mail: [email protected]

J.O. Sickman

Department of Environmental Sciences, University of California, Riverside, CA 92521, USA; E-mail: [email protected]

Jaak Sober

School of Forest Resources and Environmental Science, Michigan Technological University, 1400 Townsend Drive, Houghton, Mi 49931-1295, USA; E-mail: [email protected]

Anna Smirnoff

Natural Resources Canada, Geological Survey of Canada, 490 de la Couronne, Quebec City, Quebec G1K 9A9, Canada; E-mail: [email protected]

C. Elisabeth E. Stuiver

Laboratory of Plant Physiology, University of Groningen, P.O. Box 14, 9750 AA Haren, The Netherlands; E-mail: [email protected]

Ineke Stulen

Laboratory of Plant Physiology, University of Groningen, P.O. Box 14, 9750 AA Haren, The Netherlands; E-mail: [email protected]

xiii

List of Contributors

G.S. Tonnesen

Center for Environmental Research and Technology, Bourns College of Engineering, University of California, Riverside, CA 92521, USA; E-mail: [email protected]

K.C. Weathers

The Cary Institute of Ecosystem Studies, 65 Sharon Turnpike, P.O. Box AB, Millbrook NY 12545-0129, USA; E-mail: [email protected]

Liping Yang

National Laboratory of Soil Testing and Fertilizer Recommendation, Institute of Agricultural Resources and Regional Planning, Chinese Academy of Agricultural Sciences, No. 12 South Zhongguancun Street, Beijing 100081, China; E-mail: [email protected]

xv

Acknowledgments The contents of this book arose from an International Symposium entitled ‘‘Relating Atmospheric Source Apportionment to Vegetation Effects: Establishing Cause and Effect Relationships’’ held at the Banff Park Lodge in Banff, Alberta, Canada on April 25, 2005 which preceded the 37th Air Pollution Workshop held at the same location April 26–28, 2005. Financial assistance for the International Symposium and the 37th Air Pollution Workshop is gratefully acknowledged and was provided by the following: Environment Canada, Toronto, Ontario; Alberta Environment, Edmonton, Alberta; Biosphere Solutions, Calgary, Alberta; Canadian Natural Resources Limited, Calgary, Alberta; Syncrude Canada Limited, Fort McMurray, Alberta; Suncor Energy Inc., Fort McMurray, Alberta; Alberta Energy and Utilities Board, Calgary, Alberta; Albian Sands Energy Inc., Fort McMurray, Alberta; Duke Energy Gas Transmission, Fort St. John, British Columbia; Anadarko Canada Corporation, Calgary, Alberta; OPTI Canada Inc., Calgary, Alberta; Jaques Whitford, Calgary, Alberta; and Enersul Limited Partnership, Calgary, Alberta. Appreciation is extended to the following peer reviewers whose thoughtful comments and constructive suggestions greatly enhanced the quality and clarity of the final chapters: Christian P. Andersen, Simon A. Bareham, J. Neil Cape, Boris Chevone, Alan W. Davison, Andreas Fangmeir, Linda Geiser, Malcolm Hawkesford, Andreas Klumpp, Sagar V. Krupa, Steven W. Leavitt, Samuel L. McLaughlin, Kevin E. Percy, Richard Poirot, David H. S. Richardson, Bret Schichtel, and David A. Weinstein Many thanks are extended to Caroline Simpson whose meticulous formatting and thoughtful editing skills for completeness and consistency of the text in each of the chapters made an enormous contribution to the final product. Marilyn Croot’s computer graphics skills greatly helped in improving the quality and consistency of a number of illustrations in this book, for which I am grateful. Enormous appreciation and thanks are extend to Sagar V. Krupa, Chief Editor of the Elsevier Book Series ‘‘Developments in Environmental Science’’ for his continual support, assistance, patience, and understanding during the preparation and completion of this book.

xvi

Acknowledgments

These acknowledgments would not be complete without thanking Dr Lisa Welikovitch and Dr Alexander Bayes, my cardiologist and heart surgeon, respectively, of the Foothills Medical Centre, Calgary, Alberta, whose knowledge and incredible skills during very challenging times for me made it possible for me to be able to complete this book. Allan H. Legge Book Editor

xvii

Preface Environmental pollution has played a critical role in human lives since the early history of the nomadic tribes. During the last millennium, industrial revolution, increased population growth and urbanization have been the major determinants in shaping our environmental quality. Initially primary air pollutants such as sulfur dioxide and particulate matter were of concern. For example, the killer fog of London in 1952 resulted in significant numbers of human fatality leading to major air pollution control measures. During the 1950s, scientists also began to understand the cause and atmospheric mechanisms for the formation of the Los Angeles photochemical smog. We now know that surface level ozone and photochemical smog are a worldwide problem at regional and continental scales, with specific geographic areas of agriculture, forestry and natural resources, including their biological diversity at risk. As studies continue on the atmospheric photochemical processes, air pollutant transport, their atmospheric transformation and removal mechanisms, so is the effort to control the emissions of primary pollutants (sulfur dioxide, oxides of nitrogen, hydrocarbons and carbon monoxide), mainly produced by fossil fuel combustion. During mid-1970s environmental concerns regarding the occurrence of ‘‘acidic precipitation’’ began to emerge to the forefront. Since then, our knowledge of the adverse effects of air pollutants on human health and welfare (terrestrial and aquatic ecosystems and materials) has begun to rise substantially. Similarly, studies have been directed to improve our understanding of the accumulation of persistent inorganic (heavy metals) and organic (polyaromatic hydrocarbons, polychlorinated biphenyls) chemicals in the environment and their impacts on sensitive receptors, including human beings. Use of fertilizers (excess nutrient loading) and herbicides and pesticides in both agriculture and forestry and the related aspects of their atmospheric transport, fate and deposition; their direct runoff through the soil and impacts on ground and surface water quality and environmental toxicology have become issues of much concern. In the recent times, environmental literacy has become an increasingly important factor in our lives, particularly in the so-called developed

xviii

Preface

nations. Currently the scientific, public and political communities are much concerned with the increasing global scale air pollution and the consequent global climate change. There are efforts being made to totally ban the use of chlorofluorocarbon and organo-bromine compounds at the global scale. However, during this millennium many developing nations will become major forces governing environmental health as their populations and industrialization grow at a rapid pace. There is an on-going international debate regarding policies and the mitigation strategies to be adopted to address the critical issue of climate change. Human health and environmental impacts and risk assessment and the associated cost-benefit analyses, including global economy are germane to this controversy. An approach to understanding environmental issues in general and in most cases, mitigation of the related problems requires a systems analysis and a multi- and inter-disciplinary philosophy. There is an increasing scientific awareness to integrate environmental processes and their products in evaluating the overall impacts on various receptors. As momentum is gained, this approach constitutes a challenging future direction for our scientific and technical efforts. The objective of the book series Developments in Environmental Science is to facilitate the publication of scholarly works that address any of the described topics, as well as those that are related. In addition to edited or single and multi-authored books, the series also considers conference proceedings and paperback computer-software packages for publication. The emphasis of the series is on the importance of the subject topic, the scientific and technical quality of the content and timeliness of the work. Sagar V. Krupa Chief Editor, Book Series

xix

Introduction Air quality regulatory policies limit the emissions of primary pollutants such as particulate matter, oxides of sulphur and nitrogen and volatile hydrocarbons. However, these primary pollutants in general are of local concern, whereas secondary pollutants formed in the atmosphere such as ozone, fine and nano-particle aerosols including excess nitrogen loading in the environment and acidic deposition are of regional, continental and even global scale concern. These concerns are also inherently coupled with the issue of global climate change. Both primary and secondary pollutants have direct and indirect (through soils, food chain, etc.) effects on ecosystems and the environment. Such effects can include visible injury on vegetation and adverse impacts on growth, productivity, nutritive quality and community structure and biological diversity, with or without symptoms of injury to the plant foliage. Although numerous case studies have been conducted on the impacts of emissions from specific or isolated point sources on vegetation, such efforts become highly complex on a regional scale when a number of sources emit common pollutants (e.g., oxides of sulphur) that have adverse effects on vegetation at a given location. Identification and apportionment of the contributions of an individual or a specific source to the observed or predicted impacts is extremely important in emission reductions relevant to that source and in environmental management. From the perspective of air quality management, atmospheric science and computer-based methods exist for the apportionment of the contributions of individual sources to the composition of the atmosphere at a given geographic location. However, such methods have not been coupled with vegetation and ecosystem impact assessment. That need and the importance for reviewing the state of the knowledge in the subject matter to promote and encourage future efforts form the basis for this volume. The discussion by Philip Hopke provides the background, theory and application of atmospheric source apportionment and the methods and models used for the estimation of the contributions to the airborne

xx

Introduction

concentrations that originate from the emissions from both natural and anthropogenic sources to environmental receptors. Receptor models such as Chemical Mass Balance (CMB), UNIMIX, Positive Matrix Factorization (PMF) and Potential Source Contribution Function (PSCF) are discussed. Neil Cape describes the various mechanisms by which plants accumulate atmospheric sulphur and nitrogen compounds as well as trace metals and organic compounds and provides examples of their natural background concentrations and their elevated levels due to uptake from the atmosphere. The use of plants as biological indicators of air quality is also discussed. Richard Poirot discusses how trace metals are used in source apportionment, including finger print libraries for different source categories. Such a discussion is also directed to how the influences of more than one source with confounding fingerprints are resolved regarding their individual contributions. Palau et al. provide an example of a unique study using the characteristics of complex terrain meteorology coupled with trace metal accumulation in plant receptors exposed to emissions from the Andorra Power Plant at Teruel (Aragon, Spain) in a vegetation impact assessment. An analogous treatment focusing on the response of selected tree species in an urban airshed near Montreal (Canada) using dendroisotopic analyses is presented by Savard et al., which suggests that the values of the stable elemental isotopes of oxygen and carbon found in the cellulose of trees rings can be used for evaluating both air quality and climate. McCarthy et al. show how differences in urban air quality can be characterized using a standardized lichen tissue in mess bags hung on trees distributed across 156 sites in Hamilton (Ontario, Canada). The presentation by Fenn et al. presents an informative discussion of the very unique challenges faced when trying to measure atmospheric deposition in arid as well as snow dominated regions in western North America. McLaughlin and Nosal show how high-resolution dendrometer systems have been used to evaluate the effects of ambient ozone concentrations on mature loblolly pine trees (Pinus taeda L.) in the field. They show that by quantifying and modeling the specific effects of ambient ozone exposure in the presence of selected key climatic variables that episodic exposure to ozone may limit the growth of mature trees more than previously assumed from controlled studies with seedlings and saplings. Percy et al. present another approach utilizing plant response data from a series of aspen clones (Populus tremuloides Michx.) and white birch (Betula papyrifera Marsh.) along with selected climatic variables from the Aspen

Introduction

xxi

Free Air Carbon Dioxide Enrichment (FACE) O3 exposure experiment in Wisconsin to develop ozone exposure-based growth response models. The discussions by Ja¨ger and Krupa and De Kok et al. bring forth the very important aspects of frequently ignored components of exposure– response relationships. As with essential elements such as sulphur, even non-essential chemical constituents such as ozone have stimulatory or positive effects at low exposure levels. The recognition and application of the Hormesis concept is a very important and critical consideration in modern toxicology and unless such stimulatory or positive effects are accounted for, current numerical definitions of dose–response relationships in vegetation effects literature must be considered as being subject to artifacts since they do not describe the complete response surface. Overall the topics presented in this book represent an attempt to link some very important aspects of source-oriented air quality characterization to the assessment of vegetation effects and in the preparation of Environmental Impact Statements (EISs) or Environmental Impact Assessments (EIAs). Hopefully the reader will find the multi-disciplinary, inter-facial scientific contents of this book to be innovative and educational. Allan H. Legge

Developments in Environmental Science, Volume 9 Allan H. Legge (Editor) Copyright r 2009 Elsevier Ltd. All rights reserved. ISSN: 1474-8177/DOI:10.1016/S1474-8177(08)00201-5

1

Chapter 1 Theory and Application of Atmospheric Source Apportionment Philip K. Hopke Abstract Source apportionment is the estimation of the contributions to the airborne concentrations that arise from the emissions of natural and anthropogenic sources. To obtain a source apportionment, data analysis tools called receptor models are applied to elicit information on the sources of air pollutants from the measured constituent concentrations. Typically, they use the chemical composition data for airborne particulate matter samples. In such cases, the outcome is the identification of the pollution source types and estimates of the contribution of each source type to the observed concentrations. It can also involve efforts to identify the locations of the sources through the use of ensembles of air parcel back trajectories. In recent years, there have been improvements in the factor analysis methods that are applied in receptor modeling, as well as easier application of trajectory methods. These developments are reviewed and typical applications to data from national parks, wilderness, and other Class 1 visibility locations in the United States are presented in this chapter.

1.1. Introduction

The management of air quality is a difficult but important problem. In general, it involves the identification of the sources of materials emitted into the air, the quantitative estimation of the emission rates of the pollutants, the understanding of the transport of the substances from the sources to downwind locations, and the knowledge of the physical and chemical transformation processes that can occur during that transport.

Corresponding author: E-mail: [email protected]

2

Philip K. Hopke

All of those elements can then be put together into a mathematical model that can be used to estimate the changes in observable airborne concentrations that might be expected to occur if various actions are taken. Such actions could include limitations placed on source emissions as new industries are built and begin to function and the imposition of emission controls on existing facilities in order to reduce the pollutant concentrations. However, the atmosphere is a very complex system and it is necessary to greatly simplify the descriptions of reality in order to produce a mathematical model capable of being calculated even on the largest and fastest computers. Significant improvements have been made over the past 30 years in the mathematical modeling of dispersion of pollutants in the atmosphere. However, there are still many instances when the models are insufficient to permit the full development of effective and efficient air quality management strategies particularly for airborne particulate matter. Thus, it is necessary to have other methods available to assist in the identification of sources and the apportionment of the observed pollutant concentrations to those sources. Such methods are called receptor-oriented or receptor models since they are focused on the behavior of the ambient environment at the point of impact as opposed to the source-oriented dispersion models that focus on the emissions, transport, dilution, and transformations that occur beginning at the source and following the pollutants to the sampling or receptor site. These methods have been applied primarily to airborne particulate matter. A comprehensive view of the field can be found in Hopke (1985, 1991). As part of a larger report, Seigneur et al. (1997) provided a comprehensive review of both dispersion and receptor models through 1996.

1.2. Background

The fundamental principle of receptor modeling is that mass conservation can be assumed and a mass balance analysis can be used to identify and apportion sources of airborne particulate matter in the atmosphere. This methodology has generally been referred to within the air pollution research community as receptor modeling (Hopke, 1985, 1991). The approach to obtaining a data set for receptor modeling is to determine a large number of chemical constituents such as elemental concentrations in a number of samples. It is assumed that the measured concentrations are the result of a summation of the mass contributions of a number of independent sources or source types. Thus, a mass balance equation can be written to account for all m chemical species in the n samples as

Theory and Application of Atmospheric Source Apportionment

3

contributions from p-independent sources. xij ¼

p X

gip fpj þ eij

(1.1)

p¼1

where xij is the measured concentration of the jth species in the ith sample, fpj the concentration of the jth species in material emitted by source p, gip the contribution of the pth source to the ith sample, and eij the portion of the measurement that cannot be fit by the model. There exist a set of natural physical constraints on the system that must be considered in developing any model for identifying and apportioning the sources of airborne particle mass (Henry, 1991). The fundamental, natural physical constraints that must be obeyed are: (1) The model must reproduce the original data; the model must explain the observations. (2) The predicted source compositions must be non-negative; a source cannot have a negative elemental concentration. (3) The predicted source contributions to the aerosol must all be nonnegative; a source cannot emit negative mass. (4) The sum of the predicted elemental mass contributions for each source must be less than or equal to total measured mass for each element; the whole is greater than or equal to the sum of its parts. Although developing and applying these models, it is necessary to keep these constraints in mind in order to be certain of obtaining physically realistic solutions. There are some additional complicating aspects to the solution of the problem outlined in Eq. (1.1) since emission source profiles do not remain constant over time and environmental data tend to have relatively high noise in the measurements. These complications limit the resolutions that can be obtained using receptor models. 1.3. Receptor models 1.3.1. Sources known

There are a variety of ways to solve Eq. (1.1) depending on the information that is available. If the number and nature of the sources in the region are known (i.e., p and fik), then the only unknown is the mass contribution of each source to each sample, gkj. Winchester and Nifong (1971) and Miller et al. (1972) were the first to independently suggest this approach. The problem is typically solved using an effective-variance

4

Philip K. Hopke

least-squares approach (Cooper et al., 1984) that is now generally referred to as the Chemical Mass Balance (CMB) model. Software (Watson et al., 1990) is available from the U.S. Environmental Protection Agency at www.epa.gov/ttn/SCRAM (last accessed on July 15, 2008). Solution methods using multivariate calibration methods have also been proposed, and are summarized in the earlier review (Seigneur et al., 1997). There have not been any new method developments in this area although updated versions of the software are available. There have not been many new source profiles developed that are readily available. The U.S. Environmental Protection Agency’s library is contained in SPECIATE, which is available from www.epa.gov/ttn/CHIEF (last accessed on July 15, 2008). There are a number of new profiles particularly for sparkignition and diesel vehicles that have been measured, but they have not yet been added to the database although an effort is in progress that should produce a new library in the near future. Chow and Watson (2002) reviewed that work and other recent CMB studies. CMB is most useful for primary emissions where the chemical characteristics of the particles permit their apportionment. Secondary particles are difficult since they represent the product of atmospheric transformations of gaseous emissions into particles and are generally treated as specific chemical species such as sulfate, nitrate, and ammonium or ammonium sulfate and ammonium nitrate. 1.3.2. Sources unknown

The area of active method development has been in the methods to be used when the source profiles are not known. These are forms of factor analysis, but quite different from traditional Principal Components Analysis and related techniques. In factor analysis, the problem is expanded to the solution of the source profiles and contributions over a set of samples. Thus, the basic equation in matrix form is X ¼ GF 0

(1.2)

The two new approaches are Unmix (Henry & Kim, 1989; Kim & Henry, 1999, 2000; Lewis et al., 2003) and Positive Matrix Factorization (PMF) (Paatero, 1997, 1999). 1.3.2.1. Unmix

Unmix is based on an eigenvalue analysis. The Unmix model is a new type of multivariate receptor model based on principal component analysis (PCA). The model uses a new transformation method based on

Theory and Application of Atmospheric Source Apportionment

5

the Self-Modeling Curve Resolution (SMCR) techniques. Since a unique solution is not possible (Henry, 1987), the SMCR technique restricts the feasible region of the real solution into a small region with explicit physical constraints. Such as source compositions must be greater than or equal to zero. Explicit physical constraints form linear inequality constraints in the space spanned by the eigenvectors, and these constraints form the feasible region in eigenvectors’ space. Unmix is designed to resolve the most important sources contributing to the measured mass concentrations. The model was applied to particulate matter composition data from Phoenix (Lewis et al., 2003). The analysis generated source profiles and overall average percentage source contribution estimates for five source categories: gasoline engines (3374%), diesel engines (1672%), secondary sulfate (1972%), crustal/soil (2272%), and vegetative burning (1072%). In this study, the authors were able to separate motor vehicle contributions into diesel and spark-ignition sources. Diesel emissions were identified by high elemental carbon relative to the organic carbon whereas spark-ignition vehicles had a profile with more organic than elemental carbon. They found a substantial difference in the contribution of diesel emissions between weekend and weekday samples. The U.S. Environmental Protection Agency has developed this model as a standalone program that is being beta tested and is expected to be available in the near future. 1.3.2.2. PMF

PMF takes a very different approach to the factor analysis problem from the prior forms of factor analysis. All of the other methods use an eigenvector analysis based on a singular value decomposition (SVD). The X matrix can also be defined: 0 (1.3) X ¼ USV 0 ¼ U S V þ E 0  where Uand V are the first p columns of the U and V matrices. The U and V matrices are calculated from eigenvalue–eigenvector analyses of the XXu and XuX matrices, respectively. It can be shown (Lawson & Hanson, 1974; Malinowski, 1991) that the second term on the right side of Eq. (1.3) estimates X in the least-squares sense that it gives the lowest possible value for " #2 p m X m X n n X X X 2 ev ¼ xv€  gip f pj (1.4) i¼1 j¼1

i¼1 j¼1

p¼1

Thus, an eigenvector analysis is an implicit least-squares analysis in that it is minimizing the sum of squared residuals for the model. Paatero and

6

Philip K. Hopke

Tapper (1993) show that in effect in PCA, there is scaling of the data by column or by row and that this scaling will lead to distortions in the analysis. They further show that optimum scaling of the data would be to scale each data point individually so as to have the more precise data having more influence on the solution than points that have higher uncertainties. However, they show that point-by-point scaling results in a scaled data matrix that cannot be reproduced by a conventional factor analysis based on the singular value decomposition. Thus, PMF takes the approach of an explicit least-squares approach in which the method minimizes the object function: 0 12 p P g f x  ip pj C n X m B v€ X p¼1 B C (1.5) Q¼ B C @ A s ij j¼1 i¼1

where sij is an estimate of the ‘‘uncertainty’’ in the jth variable measured in the ith sample. The factor analysis problem is then to minimize Q(E) with respect to G and F with the constraint that each of the elements of G and F is to be non-negative. Over the past several years, several approaches to solving the PMF problem have been developed. Initially, a program called PMF2 utilizes a unique algorithm (Paatero, 1997) for solving the factor analytic task. For small- and medium-sized problems, this algorithm was found to be more efficient than Alternating Least Squares (ALS) methods (Hopke et al., 1998). Subsequently, an alternative approach that provides a flexible modeling system has been developed for solving the various PMF factor analyses least squares problems (Paatero, 1999). This approach, called the multilinear engine (ME), has been applied to environmental problems that involve the solution of more complex models (Begum et al., 2005a; Chueinta et al., 2004; Hopke et al., 2003; Paatero et al., 2003; Xie et al., 1999a; Zhao et al., 2004), but has not yet been widely used. A preliminary user-friendly version of the program implementing Eq. (1.1) is available at www.epa.gov/ttn/scram (last accessed on July 15, 2008). PMF as implemented in PMF2 has been widely applied to a variety of data sets. It was initially used to analyze data sets of major ion compositions of daily precipitation samples collected over a number of sites in Finland (Juntto & Paatero, 1994) and samples of bulk precipitation (Anttila et al., 1995) in which they are able to obtain considerable information on the sources of these ions. Polissar et al. (1996) applied the PMF2 program to Arctic data from seven National

Theory and Application of Atmospheric Source Apportionment

7

Park Service (NPS) sites in Alaska as a method to more quantitatively resolve the major source contributions. Over the past decade, there have been many applications of PMF to various source-receptor modeling problems. Polissar et al. (1998) reanalyzed an augmented set of Alaskan NPS data and resolved additional sources. Xie et al. (1999a, 1999b) have made several analyses of data from an 11-year series of particulate matter samples taken at Alert, N.W.T. Polissar et al. (1999) have examined the semi-continuous aerosol data collected by NOAA at their atmospheric observatory at Barrow, Alaska. Lee et al. (1999) have applied PMF to urban aerosol compositions in Hong Kong. They were able to identify up to nine sources that provided a good apportionment of the airborne particulate matter. Paterson et al. (1999) applied PMF to air quality and temperature data collected at a series of sites around the southern end of Lake Michigan in 1997 and used three factors to reproduce 75% of the variation in the data. Huang et al. (1999) analyzed elemental composition data for particulate matter samples collected at Narragansett, RI, using both PMF and conventional factor analysis. They were able to resolve more components with more physically realistic compositions with PMF. Thus, the approach is gaining interest because it does have some inherent advantages particularly through its ability to individually weight each data point. PMF is somewhat more complex and harder to use, but it appears to provide improved resolution of sources and better quantification of impacts of those sources than PCA (Huang et al., 1999). Yakovleva et al. (1999) applied PMF to particle composition data taken from personal samplers as well as indoor and outdoor samplers around the home in which the person with the sampler lived. These data were analyzed as both 2- and 3-way problems. In the 3-way analysis, it is possible to ascertain the extent of indoor particle concentration that is of ambient origins. Chueinta et al. (2000) have analyzed PM10 composition data. In this chapter, they introduce a source contribution rose analogous to a wind rose to help provide information on the direction of the source relative to the receptor site. Ramadan et al. (2000) applied PMF to a set of daily data from Phoenix, AR. In this analysis, separate profiles were resolved for diesel and spark-ignition vehicles. Lewis et al. (2003) analyzed the same data using Unmix and find relatively similar results for those sources that contribute the largest amounts to the ambient mass concentrations. Aerosol chemical composition data for PM2.5 samples collected from 1988 to 1995 at Underhill, Vermont were analyzed by Polissar et al. (2001a). An 11-factor solution was obtained. Sources representing wood burning, coal and oil combustion, the coal combustion emissions plus

8

Philip K. Hopke

photochemical sulfate production, metal production plus municipal waste incineration, and the emissions from motor vehicles were identified. Emissions from smelting of non-ferrous metal ores, arsenic smelting, as well as soil particles and particles with high concentrations of Na were also identified by PMF. Song et al. (2001) have analyzed similar data from Brigantine, NJ and Washington, DC and compare these results with those from Underhill. In these data, similar two sulfur factors (winter and summer) were observed. Additional studies have been performed on the data from Brigantine, NJ (Lee et al., 2002); Spokane, WA (Kim et al., 2003a); and Atlanta (Kim et al., 2003b). Also, PMF has been applied to particle size distribution data measured in Seattle (Kim et al., 2004a) and Pittsburgh (Zhou et al., 2004a). To improve source identification as well as to separate primary sources of carbonaceous particles into traffic-related carbonaceous particle sources, and residual oil combustion, the analyses have recently begun using the Interagency Monitoring of Protected Visual Environments (IMPROVE) temperature resolved carbon fractions in PMF analysis for PM data from Crater Lake and Lassen National Parks (Liu et al., 2003), Atlanta, GA (Kim et al., 2004b), Washington, DC (Kim & Hopke, 2004b), San Gorgonio, CA (Zhao & Hopke, 2004), Brigantine, NJ (Kim & Hopke, 2004b), Seattle, WA (Kim et al., 2004b; Maykut et al., 2003), Mammoth Cave National Park (Zhao & Hopke, 2006), and Great Smoky National Park (Kim & Hopke, 2006). To illustrate the nature of the results obtained from such analyses, the analysis of the data from San Gorgonio National Wilderness will be presented. 1.3.3. Illustrative example of PMF

The IMPROVE monitoring program (Malm et al., 1994; http://vista. cira.colostate.edu/improve, last accessed on July 15, 2008) was established in 1985 to aid the creation of Federal and State implementation plans for the protection of visibility in 156 national parks and wilderness areas. One of the objectives of the IMPROVE program is to identify chemical species and emission sources responsible for existing man-made visibility impairment. The IMPROVE program has also been a key participant in visibility-related research, including the advancement of monitoring instrumentation, analysis techniques, visibility modeling, policy formulation, and source attribution field studies. Aerosol samples in this study were collected beginning in March 1988 at the San Gorgonio National Wilderness IMPROVE site (latitude: 34.19241N, longitude: 116.90131W, altitude 1705 m). This site is near

Theory and Application of Atmospheric Source Apportionment

9

San Bernardino, CA and downwind of the greater Los Angeles area, which is an area dominated by mobile source emissions. Initially, the elements were measured by Proton-Induced X-ray Emission (PIXE) for Na-Mn and Proton Elastic Scattering Analysis (PESA) for H. Ion chromatography was used to measure anions extracted from the Nylon filter. There were several significant changes over time in the sample analysis methods. Quartz filters were measured using a Thermal Optical Reflectance (TOR) method for OC and EC (Chow et al., 1993). In the summer, 1992, X-ray fluorescence (XRF) was added to decrease the minimum detection limits (MDL) for the elements between Fe and Pb. There were changes in the denuders in 1996 with the addition of glycerin and a change in manufacturer of the Nylon filters in 1998. At the end of 2001, the elemental analysis for elements lighter than Fe was switched from PIXE to Cu-anode XRF. This change lead to a decrease in MDL for most elements, a small increase in MDL for Al and a substantial increase in the MDL for sodium. PESA was retained for the measurement of the H concentrations. The robustness of PMF analysis to such variations through the application of individual data point weights, permit the use of all of the samples from March 2, 1988 to May 30, 2003. After 2000, the IMPROVE program changed the sampling schedule from 24-h samples collected twice per week (on Wednesday and Saturday) to a 24-h sample every third day. Data pretreatment can be briefly summarized as follows. In the original measurement data set, some species have more than 50% missing or below detection limit (BDL) values. It is not a good method to exclude an element just based on its percentage of missing values and BDLs. Recently, a practical statistical method based on signal/noise (S/N) ratio was proposed to confirm the variable (element) bad or not (Paatero & Hopke, 2003). The element with the S/N larger than 2 can be considered as a normal element, the element with the S/N between 0.2 and 2 can be considered as a weak element, and the element with the S/N less than 0.2 can be considered as bad variable. The bad element should be excluded from analysis, unless it is a must considered element. The uncertainties can be multiplied by a factor of 2 or 3 to down-weight the weak elements. In addition, it is recommended that the element with more than 90–95% missing values and BDLs be excluded no matter how its S/N is. There is a strong relationship between XRF sulfur and IC sulfate, so in accordance with the IMPROVE recommendation (http://vista.cira.colostate.edu/ improve, last accessed on July 15, 2008), only XRF sulfur should typically be used. Recent studies (Kim et al., 2004a; Kim & Hopke, 2004a, 2004b; Zhao & Hopke, 2004) have found it advantageous to use OP and subtract its value from that of EC1 to provide two of the variables in the data.

10

Philip K. Hopke

Figure 1.1. Profiles of the resolved sources at the San Gorgonio National Wilderness IMPROVE site.

In the study of the San Gorgonio data, seven sources were resolved for the ambient PM2.5 (Zhao & Hopke, 2004). They were (1) local soil, (2) diesel engine emissions, (3) secondary nitrate, (4) secondary sulfate with secondary organic, (5) aged sea salt, (6) gasoline engine emissions, and (7) Asian dust. The corresponding source profiles and contributions are shown in Figs. 1.1 and 1.2, respectively. An interesting feature of this solution is the existence of two sources with different OC/EC fractions that have been tentatively identified as

Theory and Application of Atmospheric Source Apportionment

11

Figure 1.2. Time series of the contributions of the identified sources at the San Gorgonio National Wilderness IMPROVE site.

diesel and spark ignition vehicle. Diesel and gasoline are composed of many compounds, and their emissions are also a complex mixture of compounds. Watson et al. (1994) provided characterizations of gasoline and diesel vehicle emissions obtained in dynamometer tests using the same IMPROVE TOR method as was used to produce the OC/EC data at San Gorgonio. These profiles can be compared to those developed by Watson et al. (1994) based on 1990 measurements in Phoenix (Fig. 1.3). In the literature profiles (Watson et al., 1994), the diesel OC1 abundance

12

Philip K. Hopke

Figure 1.3. Comparison of the diesel (top) and gasoline (bottom) carbon thermal fractions extracted from the San Gorgonio site data and those reported by Watson et al. (1994).

is higher than the gasoline OC1 value. The EC1 fraction in gasoline emissions is generally more abundant than that in diesel emissions. The EC1 fraction in this figure includes the OP. The EC2 fraction in the diesel emissions is significantly higher than that in the gasoline emissions. There is very little EC3 in either diesel emissions or gasoline emissions. Figure 1.3 shows the source profiles attributed to diesel emissions and the gasoline emissions obtained in this study. OP was separated from EC1, so if OP was restored into EC1, source 2 in Fig. 1.1 could be assigned to be diesel emission with high concentrations of OC1, EC1, and EC2. Source 6 with high OC and EC1 represents gasoline emission. To highlight these source profiles in comparison with direct measurements, the profiles of diesel and gasoline obtained in this study with relative concentrations of OC and EC are shown in Fig. 1.3 along with the respective profiles measured in dynamometer studies by Watson et al. (1994). EC1 in Fig. 1.3 included the OP in order to be comparable with the definition of EC1 as used in Watson et al. (1994). The diesel profile of San Gorgoio agrees with that of Watson et al. (1994) with respect to the

Theory and Application of Atmospheric Source Apportionment

13

Figure 1.4. Relative contributions of the identified sources to the PM2.5 mass.

key fractions (i.e., OC1, EC2 for diesel). The gasoline profile differs from that of Watson et al. (1994), but it is similar to the profiles reported by Kim and Hopke (2004a, 2004b) and Kim et al. (2004a). Moreover, the significant amount of EC in the diesel profile and the large amount of OC in the gasoline profile support these assignments. However, Shah et al. (2004) have shown that slow moving and stop-and-go diesel engine emissions include roughly equal amounts of OC and EC using a different measurement protocol for the carbon so there may be some slow moving diesel included with the gasoline vehicle contributions. The diesel profile does not show the Fe and Mn that has been observed in other studies as diesel fuel addictive (Kim et al., 2004a; Kim & Hopke, 2004a, 2004b; Maykut et al., 2003). Some Si in the diesel profile may be due to the mixture of soil during the transportation. The Fe and Ca in the gasoline profile may be mainly from the catalyst-equipped gasoline vehicles (Schauer et al., 2002) and also possibly from the soil. Figure 1.4 shows that the average contributions of diesel and gasoline emissions to total PM2.5 mass are 9.2% and 7%, respectively. In addition to diesel and gasoline emissions, five other sources were identified in this study. They are soil, secondary nitrate, secondary sulfate with secondary organics, aged sea salt, and Asian desert dust, respectively. Source 1 represents soil with high concentrations of Al, Si, Ca, and Fe, and contributes 6.7% to the total PM2.5 mass at this site. The ratio of Al

14

Philip K. Hopke

to Si in this source is 0.91 much higher than the typical ratio observed in soils, 0.293 (Mason, 1966). This source appears to include some Al that should have been assigned to the other sources. Source 3 shows secondary nitrate with high concentration of nitrate and contributes 37.9% to the total PM2.5 mass. Nitrate is formed in the atmosphere predominantly through oxidation of NOx and the suspected origin of this source is the mobile emissions. This region has abundant ammonia available that favors the formation of NH4NO3. This source shows a seasonal trend with high contributions in winter because low temperature shifts the equilibrium system of NO3 and HNO3 toward the particle phase, increasing the mass of NH4NO3 (Seinfeld & Pandis, 1998). However, the temporal contribution plot for this source also shows peaks in the summer of 1992. This result reflects the temporal variation of measured NO3 concentration as has been observed in prior summer studies in nearby Riverside, CA (Liu et al., 2000). Source 4 with high concentrations of sulfur and OCs represents secondary sulfate with secondary organics, and contributes 24.6% to the total PM2.5 mass. Combustion may be a major contribution to SO2, which will be further changed to SO4 through the photooxidation reaction, so H and K shown in this source might be the co-products of combustion. The inclusion of potassium suggests some admixture of wood smoke and/or meat cooking (Schauer et al., 1999, 2001). OC in this profile could be present because of the condensation of the co-products of combustion and secondary organic compounds formed in the atmosphere by the same processes that converted the SO2 to sulfate. The temporal contribution plot of this source shows a seasonal variation with high peaks in the summer when photooxidation (Harrison & Perry, 1986) of gaseous precursors is at a high rate. Source 5 appears to be aged sea salt with high concentrations of Na, S, and NO3. This source contributes 6.7% to the total PM2.5 mass. The low concentration of chloride in this source may be due to the conversion from NaCl to Na2SO4 or NaNO3 during transportation (Liu et al., 2003). Source 7 shows high concentration of Si and contributes 8% to the total PM2.5 mass. The Al concentration is very low in this source and represents the rotational problem noted for source 1. This source is assigned as Asian desert dust. It can be seen from the temporal contribution plot of this source that almost every spring (from March to May) shows a high peak, especially the spring of 2001. Dust storms occur almost every spring in the deserts of Western China, such as Taklamakan, Gobi, and Ordos Deserts. Husar et al. (2001) reported that in April 1998, several intense dust storms occurred over the Gobi Desert in Western China and Mongolia. In particular, the storm on April 19, 1998 produced a dust

Theory and Application of Atmospheric Source Apportionment

15

cloud that crossed the Pacific and reached much of the west coast of North America. Thus, the peak of April 29, 1998 in the contribution plot of this source may correspond to this sand storm. Recently, the studies of the aerosol of the Crater Lake and Lassen Volcanic National Parks have reported the similar result and the corresponding date was also April 29, 1998 (Liu et al., 2003). They suggested that such events might be observed more frequently at high altitude sites such as San Gorgonio. The peak on April 13, 2001 in the contribution plot is likely to be due to a known major sand storm occurred in China. In order to explore this hypothesis, air parcel back trajectories with three different starting heights (10, 100, and 1000 m above model ground level) and two different end times (16:00 pm and 23:00 pm UTC) on April 13, 2001 were calculated using the NOAA HYSPLIT (Hybrid Single-Particle Lagrangian Integrated Trajectory) model (http://www.arl.noaa.gov/ready/hysplit4.html, last accessed on July 15, 2008, also Draxler & Hess, 1998). All of the trajectories went across the northern Pacific Ocean suggesting eastern Asia as the source area (Fig. 1.5). A comparison between this source profiles and the reported composition of Asian mineral dust (Nishikawa et al., 2000) is shown in the Table 1.1 with respect to the key crustal elements, Al, Si, Ti, Fe, and Ca. The composition of all the elements in the two studies, with the exception of Al, were in reasonable agreement and support the assignment of this source as Asian dust. This source profile also shows some S and OC, probably because the dust was mixed with anthropogenic air pollutants during the transport across China. Sources 1 and 7 both show high concentration of Si, but it can be seen from Fig. 1.2 that their temporal contribution plots are different, suggesting two different sources called ‘‘local soil’’ and ‘‘Asian dust’’. The temporal contribution plots of these two sources both showed a high peak on November 25, 2002. The Si concentration of that date was extremely high (12 times) compared to the average Si concentration. As for the low concentration of Al in source 7, the possible reasons are (1) Al might be incorporated into other sources that represent stronger sources of Al and (2) more than 1/3 of the Al concentration measurements were missing or below the detection limit. Another test of the effectiveness of PMF analysis is the comparison of the predicted PM mass versus the measured PM mass. The predicted PM mass of each sample was obtained from the sum of scaled source contribution values. The predicted PM2.5 mass is compared to the measured mass in Fig. 1.6. The slope of the line is 0.87870.008. The intercept is 0.9270.07, and the squared correlation coefficient, R2, is 0.90.

16 Philip K. Hopke

Figure 1.5.

Calculated backward air parcel trajectory plots for three different heights and two different times on April 13, 2001.

17

Theory and Application of Atmospheric Source Apportionment Table 1.1. The composition comparison between the Asian dusts of this study and the reference Element a

Ratio a

Al

Ca

Fe

Si

Ti

1.0

0.533

1.085

0.444

1.035

The ratio of the concentration derived by this study to that of Nishikawa et al. (2000).

Figure 1.6. Time series of Si concentrations measured at the Washington, DC IMPROVE network site.

1.4. Methods incorporating back trajectories

Dispersion models discussed elsewhere describe the transport of the particles from a source to the sampling location. However, using an analogous model of atmospheric transport, a model calculates the position of the air being sampled backward in time from the receptor site from various starting times throughout the sampling interval. There are a number of such models that have been developed to study atmospheric transport. This report will concentrate on a particular model, HYSPLIT model (Draxler & Hess, 1998). This model is accessible on-line and is quite easy to use (http://www.arl.noaa.gov/ss/models/hysplit.html, last accessed on August 14, 2008). The trajectories can then be used in Residence Time Analysis (RTA) (Poirot & Wishinski, 1986; Poirot et al., 2001), Areas of Influence Analysis (AIA) (Malm et al., 1990), Quantitative Bias Trajectory Analysis (QTBA) (Keeler & Samson, 1989), Potential Source Contribution Function (PSCF) (Ashbaugh et al., 1985), and Residence Time Weighted Concentrations (RTWC) (Stohl, 1996). RTA, AIA, QTBA, and RTWC each have been used only in a few publications and those results were reviewed by Seigneur et al. (1997). Recently a simplified version of QTBA has been described by Zhou et al. (2004b) and Brook et al. (2004) and applied to transport in northeastern U.S. and southeastern Canada. Hopke et al. (2005) have shown that a two-site RTA analysis gave good correspondence with the known emissions from coal- and oil-fired power plants in the northeastern U.S.

18

Philip K. Hopke

1.5. Using the trajectories

One of the uses for the trajectories is to examine the origins of observed peaks in the time series of observed species. For example, Fig. 1.6 shows the time series of silicon measured in Washington, DC as part of the IMPROVE network (http://vista.cira.colostate.edu/views/, last accessed on July 15, 2008). It can be seen that there are sharp spikes in concentrations and they occur typically in July and August although there are occasional one in April as well. Figure 1.7 shows the back trajectory plot for July 7, 1993 whereas Fig. 1.8 shows the plot for April 22, 2001. It can be seen that the July 1993 event is associated with intercontinental transport of dust from the Saharan region of Africa while the April 2001 is associated with the transport of Asian dust from China. The trajectories for the March to May crustal element events typically are associated with Asian dust transport whereas the more common July to August events come from the transport of Saharan dust. Similar analyses can be done in examining

Figure 1.7. 1993.

Back trajectory associated with the Si event seen in Washington, DC on July 7,

Theory and Application of Atmospheric Source Apportionment

19

Figure 1.8. Back trajectory associated with the Si event seen in Washington, DC on April 22, 2001.

events observed in the resolved source apportionments produced by PMF analyses. 1.5.1. Potential source contribution function

The Potential Source Contribution Function (PSCF) receptor model is the most widely used of the trajectory ensemble methods. It was originally developed by Ashbaugh et al. (1985) and Malm et al. (1986). It has been applied in a series of studies over a variety of geographical scales (Biegalski & Hopke, 2004; Cheng & Lin, 2001; Cheng et al., 1993a, 1993b, 1996; Fan et al., 1995; Gao et al., 1993, 1994, 1996; Hoh & Hites, 2004; Hsu et al., 2003a, 2003b; Lin et al., 2001; Plaisance et al., 1996, 1997; Poissant, 1999; Polissar et al., 2001a, 2001b; Wang et al., 2004; Xie et al., 1999c; Yli-Tuomi et al., 2003; Zeng & Hopke, 1989). Air parcel back trajectories ending at a receptor site are represented by segment endpoints. Each endpoint has two coordinates (e.g., latitude, longitude) representing the central location of an air parcel at a particular time. To calculate the PSCF, the whole geographic region covered by the trajectories is divided into an array of grid cells whose size is dependent

20

Philip K. Hopke

on the geographical scale of the problem so that the PSCF will be a function of locations as defined by the cell indices i and j. Let N be the total number of trajectory segment endpoints during the whole study period, T. If n segment trajectory endpoints fall into the ijth cell (represented by nij), the probability of this event, A, is given by nij P½Aij  ¼ (1.6) N where P[Aij] is a measure of the residence time of a randomly selected air parcel in the ijth cell relative to the time period T. Suppose in the same ijth cell there is a subset of mij segment endpoints for which the corresponding trajectories arrive at a receptor site at the time when the measured concentrations are higher than a prespecified criterion value. In the study described in the next section, the criteria values were the calculated mean values for each species at each site. The probability of this high concentration event, Bij, is given by P[Bij], mij (1.7) P½Bij  ¼ N Like P[Aij] this subset probability is related to the residence time of air parcel in the ijth cell but the probability B is for the contaminated air parcels. The PSCF is defined as Pij ¼

P½Bij  mij ¼ P½Aij  nij

(1.8)

Pij is the conditional probability that an air parcel which passed through the ijth cell had a high concentration upon arrival at the trajectory endpoint. Although the trajectory segment endpoints are subject to uncertainty, a sufficient number of endpoints should provide accurate estimates of the source locations if the location errors are random and not systematic. Cells containing emission sources would be identified with conditional probabilities close to one if trajectories that have crossed the cells effectively transport the emitted contaminant to the receptor site. The PSCF model thus provides a means to map the source potentials of geographical areas. It does not apportion the contribution of the identified source area to the measured receptor data. Xie et al. (1999c) used PSCF to examine the locations of the sources identified by the PMF analysis of the data from Alert. The results of these analyses were in agreement with earlier efforts that examined the PSCF maps for the individual chemical constituents in the particle samples. Poissant (1999) used PSCF to examine the likely source locations for total gaseous mercury observed in the St. Lawrence River valley. During

Theory and Application of Atmospheric Source Apportionment

21

the winter, fall, and spring period the distribution of potential sources reasonably reproduces the North American Hg emission inventory. However, because a single fixed criterion was over the entire year and transport from many of the strong source areas was weak during the summer months, few source areas are observed during the summer data where the concentrations were the lowest. Polissar et al. (2001a) applied PSCF to the PMF results for the data from Underhill, VT. The results helped to clarify the nature of the sources. For example, the high degree of overlap in the source regions for the winter and summer ‘‘coal-fired power plant’’ source type suggested that the observed sulfate was coming from the same emission sources. Polissar et al. (2001b) examined the particle data (black carbon, light scattering, and condensation nuclei counts) collected at Point Barrow, AK. They found that they could distinguish between biogenic sources of the small particles seen only with the condensation nuclei counter from anthropogenic larger particles that scatter and absorb light. The biogenic particles came primarily from the open areas of the North Pacific Ocean whereas most of the anthropogenic particles came from known industrialized areas of Russia. 1.5.2. Application of PSCF

There have been several efforts to test the ability of the PSCF approach to identify specific sources using natural wildfire events. Cheng and Lin (2001) have shown that PSCF can successfully locate emission source locations by examining data related to the 1998 Central American smoke events that had impacts on sampling sites in the Southern Great Plains. Recently, Begum et al. (2005b) had the opportunity to further test the method. During early July 2002, there was a major forest fire in central Quebec whose plume penetrated down the coast of the eastern United States to well south of Philadelphia, where semi-continuous measurements were being made of a variety of airborne particle properties (Jeong et al., 2004a, 2004b). This event permitted testing the ability of PSCF analysis to identify the known source location of this fire. The main objective of this study was to identify possible source locations of atmospheric aerosol by combining air parcel back trajectories with the semi-continuous particle data and ascertain if the PSCF analysis clearly defines the fire location. Particulate matter mass and composition were measured in Philadelphia from July 1 to August 3, 2002. Mass concentrations of PM2.5, organic carbon, elemental carbon, black carbon, and sulfate were measured at a site northeast of Philadelphia (latitude 40.041N, longitude 75.001W).

22

Philip K. Hopke

The monitoring site was located 400 m away from a light traffic road and 1 km away from a heavy traffic highway. It is approximately 15 km from downtown Philadelphia. Continuous PM2.5 concentrations every 30 min were obtained by using a 30C Tapered Element Oscillating Microbalance (TEOM) (Rupprecht and Patashnick, Albany, NY) with a sample equilibration system (SES) dryer. Light scattering (bscat) was measured using a Radiance Research Model 908 integrating nephelometer (lower detection limit and accuracy are about 106 m1 and 20%, respectively) with a Nafion drier on the inlet to reduce the influence of particle-bound water. Carbonaceous material in PM2.5 was measured by a semicontinuous OC/EC field instrument (Sunset Laboratory, Forest Grove, Oregon, USA). Sulfate concentrations were measured with high time resolution using a continuous sulfate analyzer originally designed by the Harvard School of Public Health (HSPH) (George Allen, pvt. comm.) and built at Clarkson University. The continuous sulfate analyzer included three main parts, sample pretreatment system to segregate PM2.5 using a Sharp Cut Cyclone (SCC), and removal of positive and negative interferences such as SO2 and NOx using a sodium carbonate-coated denuder and a carbon monolith. The heated converter (Thermo Environmental Model 350, Thermo Environmental Instruments Inc, Franklin, Massachusetts, USA) was operated at around 7001C to reduce particulate sulfate to SO2 and the SO2 gas was introduced into an SO2 analyzer (Thermo Environmental 43C-TLS). To estimate the background concentration, a zero particle purge system periodically supplied filtered air to the furnace that served as the blank values. These blank values were then subtracted from the measured concentrations. There were problems with the sulfate analyzer at the beginning of the sampling period and no data were collected until July 6. Sulfate values were not corrected for the converter efficiency in the semicontinuous analyzer since the PSCF analysis only requires good relative values. The sulfate, PM2.5, bscat, OC, and EC data are described in detail by Jeong et al. (2004a, 2004b). A 2-h time interval was chosen for sampling, analysis, and instrumental cool down period. The PSCF results are displayed in the form of maps of the area of interest on which the PSCF values ranging from 0 to 1 are displayed in terms of a color scale. The maps for OC and EC are shown in Figs. 1.9 and 1.10. Both maps show large areas in the region to the east of the southern end of Hudson Bay. Beginning on July 2, there was a large boreal forest fire in this area. A cold front carried this material down into the mid-Atlantic region beginning July 6. Its influence lasted from July 6 to 10 and the PSCF analysis clearly identified the fires’

Theory and Application of Atmospheric Source Apportionment

23

Figure 1.9. Potential source contribution function plot for OC measured in Philadelphia, PA, July 2002.

Figure 1.10. Potential source contribution function plot for EC measured in Philadelphia, PA, July 2002.

24

Philip K. Hopke

Figure 1.11. Map of fire locations in Quebec in 2002. The map was obtained from http://atlas.gc.ca/site/english/maps/environment/forestfires/dailyhotspots2002 (last accessed on August 16, 2008).

locations. Figure 1.11 shows the location of the fires (http://atlas.gc.ca/ site/english/maps/environment/forestfires/, last accessed on July 15, 2008) as identified in satellite images. There is clearly good correspondence between the hotspot locations identified from satellite images and the high PSCF valued grid cells. In addition to the fire location, there are other small high potential areas such as Pittsburgh, Toledo to Michigan City, central Illinois, St. Louis and northeastern Missouri to southwestern Iowa. Urban areas are expected to be OC and EC source areas. The nature of the areas in central Illinois, Missouri, and Iowa are uncertain. There were additional areas in western Quebec and in Ontario near Lake Huron that could be additional fire zones. The PM2.5 map (Fig. 1.12) also shows a small area of high potential to the east of the southern end of Hudson Bay. This area is similar in location

Theory and Application of Atmospheric Source Apportionment

25

Figure 1.12. Potential source contribution function plot for PM2.5 measured in Philadelphia, PA, July 2002.

Figure 1.13. Potential source contribution function plot for bscat measured in Philadelphia, PA, July 2002.

to those discussed for the OC and EC maps. The fire aerosol produced a peak 2-h mass concentration value of 160 mg m3 (Jeong et al., 2004b). Figure 1.13 presents the PSCF map for bscat. No area in the vicinity of Hudson Bay corresponds to the wildfire area appearing in this map.

26

Philip K. Hopke

Figure 1.14. Potential source contribution function plot for sulfate measured in Philadelphia, PA, July 2002.

The nephelometer failed on July 6 and thus, there were no light scattering data for the period of the fire event. Thus, the location of the fires could not be observed in the PSCF map for bscat because there were no data to guide the analysis. In comparison, the PSCF map for sulfate (Fig. 1.14) does not show an influence of the wildfire since it is not a significant SO2 source. Instead, the map shows the major source region located in the western end of the river valley in southern Indiana and Illinois and northern Kentucky. There is also an area of potential influence in the area around southern Lake Michigan and trailing across northern Illinois. The Gary-Hammond area of Indiana has major steel mills that may be the source of the SO2. There is also an area in east Texas where there are a number of ligniteburning power plants. Similar patterns have been seen for sulfate measured in Washington, DC (Kim & Hopke, 2004a) and Brigantine, NJ (Kim & Hopke, 2004b). These results support the utility of the PSCF method. In this application, the location of the Quebec fire area was clearly identified correctly similarly to the results reported by Cheng and Lin (2001). The PSCF probability maps were readily interpretable in terms of known source regions. These results lend support to the utility of the methods for the identification of likely major source areas for transported species.

Theory and Application of Atmospheric Source Apportionment

27

1.6. Future directions

The U.S. Environmental Protection Agency has deployed an extensive network of samplers to collect particle samples in urban areas for particle composition measurements. This Chemical Speciation Network has 54 sites for which samples are collected every third day and more than 170 other sites where data are collected either every third day or every sixth day. Thus, an extensive database is becoming available for receptor modeling of urban areas (Kim et al., 2005). There have been new developments in semi-continuous sampling/ analysis systems for major components of the ambient aerosol. Commercial instruments are now available for sulfate, nitrate, organic and elemental carbon. These systems typically provide hourly data. Research instruments have been developed to measure ions or trace elements on a time scale of every 30 min. These instruments permit the observation of rapidly changing concentrations that occur when the sampling site is affected by the plume of particular point sources. Such data will make factor analysis models much more effective because the wind data can be used to determine that certain downwind source contributions can be forced to zero. With greater variability in the data, more sources will be more reliably extracted from the data. There is growing interest in the use of resolved source contributions in epidemiological studies of the relationships between airborne particle and adverse human health effects. It is thought unlikely that all particles have equal toxicity and thus, the problem then exists as to how to organize data characterizing particle samples to enter appropriate statistical models. There are too many components typically measured and there is often high correlation among them because they do come from a limited number of common sources. Thus, it is anticipated that there will be an increased demand for easy to use software that will permit even complex receptor models to be applied to a wider variety of available data. For ecosystem problems, it is certainly possible to identify which sources are producing the most particulate mass and contributing the potentially toxic constituents. It is also possible to identify the likely locations of distant sources and the potential of transboundary pollution that is leading to the degradation of remote areas. Such information should be useful in developing effective strategies for mitigating the key sources and producing economically efficient improvements in air quality. Finally, now that the EPA has declared areas of the United States to be in non-attainment of the PM2.5 ambient air quality standard, there will need to be application of these new receptor model methods to data like those from the Speciation Network to provide information for state and

28

Philip K. Hopke

local air quality management strategy development. A similar problem will arise in Europe as the new European PM10 standards start to be enforced and areas are identified that have problems that require identification and quantitative apportionment of particle sources. Thus, receptor models continue to be developed and improved and there appears to be a substantial need for the application in the near future.

ACKNOWLEDGMENTS

The author would like to acknowledge all of the past members of his research group who have contributed to the development and application of the models described in this chapter. The author gratefully acknowledges the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model and READY website (http://www.arl.noaa.gov/ready.html) used in this publication. The work in our group has been supported by a number of agencies including the U.S. Environmental Protection Agency, the California Air Resources Board, the New York State Energy Research and Development Authority, the New Jersey Department of Environmental Protection, the Delaware Department of Natural Resources and Environmental Control, Lake Michigan Air Directors Consortium (LADCO), and the International Atomic Energy Agency.

REFERENCES Anttila, P., Paatero, P., Tapper, U., and Ja¨rvinen, O. 1995. Application of positive matrix factorization to source apportionment: Results of a study of bulk deposition chemistry in Finland. Atmos. Environ. 29, 1705–1718. Ashbaugh, L.L., Malm, W.C., and Sadeh, W.Z. 1985. A residence time probability analysis of sulfur concentrations at Grand Canyon National Park. Atmos. Environ. 19, 1263–1270. Begum, B., Hopke, P.K., and Zhao, W. 2005a. Source identification of fine particles in Washington DC by expanded factor analysis modeling. Environ. Sci. Technol. 55, 227–240. Begum, B.A., Kim, E., Jeong, C.H., Lee, D.H., and Hopke, P.K. 2005b. Evaluation of the potential source contribution function using the 2002 Quebec forest fire episode. Atmos. Environ. 39(20), 3719–3724. Biegalski, S.R., and Hopke, P.K. 2004. Total potential source contribution function analysis of trace elements determined in aerosol samples collected near Lake Huron. Environ. Sci. Technol. 38, 4276–4284. Brook, J.R., Johnson, D., and Mamedov, A. 2004. Determination of the source areas contributing to regionally high warm season PM2.5 in eastern North America. J. Air Waste Manag. Assoc. 54, 1162–1169.

Theory and Application of Atmospheric Source Apportionment

29

Cheng, M.D., and Lin, C.J. 2001. Receptor modeling for smoke of 1998 biomass burning in Central America. J. Geophys. Res. 106, 22871–22886. Cheng, M.-D., Hopke, P.K., Barrie, L., Rippe, A., Olson, M., and Landsberger, S. 1993a. Qualitative determination of source regions of long-range transported aerosol using data collected at Canadian high Arctic. Environ. Sci. Tech. 27, 2063–2071. Cheng, M.D., Hopke, P.K., and Zeng, Y. 1993b. A receptor-oriented methodology for determining source regions of particle sulfate composition observed at Dorset, Ontario. J. Geophys. Res. 98, 16839–16849. Cheng, M.-D., Gao, N., and Hopke, P.K. 1996. Source apportionment study of nitrogen species measured in southern California in 1987. J. Environ. Eng. 122, 183–190. Chow, J., and Watson, J. 2002. Review of PM2.5 and PM10 apportionment for fossil fuel combustion and other sources by the chemical mass balance receptor model. Energy Fuels 16, 222–260. Chow, J.W., Waston, J.G., Pritchett, L.C., Pierson, W.R., Fraizer, C.A., and Purcell, R.G. 1993. The DRI thermal/optical reflectance carbon analysis system, description, evaluation and applications in U.S. air quality studies. Atmos. Environ. 27A, 1185–1201. Chueinta, W., Hopke, P.K., and Paatero, P. 2000. Investigation of sources of atmospheric aerosol urban and suburban residential areas in Thailand by positive matrix factorization. Atmos. Environ. 34, 3319–3329. Chueinta, W., Hopke, P.K., and Paatero, P. 2004. A multilinear model for spatial pattern analysis of the measurement of haze and visual effects project. Environ. Sci. Technol. 38, 544–554. Cooper, J.A., Watson, J.G., and Huntzicker, J.J. 1984. The effective variance weighting for least squares calculations applied to the mass balance receptor model. Atmos. Environ. 18, 1347–1355. Draxler, R.R., and Hess, G.D. 1998. An overview of the Hysplit_4 modeling system for trajectories, dispersion, and deposition. Aust. Meteorol. Mag. 47, 295–308. Fan, A., Hopke, P.K., Raunemaa, T., Oblad, M., and Pacyna, J.M. 1995. A study on the potential sources of air pollutants observed at Tjorn, Sweden. Environ. Sci. Pollut. Res. 2, 107–115. Gao, N., Cheng, M.D., and Hopke, P.K. 1993. Potential source contribution function analysis and source apportionment of sulfur species measured at Rubidoux, CA during the Southern California Air Quality Study, 1987. Anal. Chim. Acta 277, 369–380. Gao, N., Cheng, M.D., and Hopke, P.K. 1994. Receptor modeling of airborne ionic species collected in SCAQS, 1987. Atmos. Environ. 28, 1447–1470. Gao, N., Hopke, P.K., and Reid, N.W. 1996. Possible sources of some trace elements found in airborne particles and precipitation in Dorset, Ontario. J. Air Waste Manag. Assoc. 46, 1035–1047. Harrison, R.M., and Perry, R. 1986. Handbook of air pollution analysis (second ed.). Chapman and Hall, London. Henry, R.C. 1987. Current factor analysis models are ill-posed. Atmos. Environ. 21, 1815–1820. Henry, R.C. 1991. Multivariate receptor models. In: Hopke, P.K., ed. Receptor modeling for air quality management. Elsevier Science Publishers, Amsterdam, pp. 117–147. Henry, R.C., and Kim, B.M. 1989. Extension of self-modeling curve resolution to mixtures of more than three components. Part 1. Finding the basic feasible region. Chemometr. Intell. Lab. Syst. 8, 205–216. Hoh, E., and Hites, R.A. 2004. Sources of toxaphene and other organochlorine pesticides in North America as determined by air measurements and potential source contribution function analyses. Environ. Sci. Technol. 38, 4187–4194.

30

Philip K. Hopke

Hopke, P.K. 1985. Receptor modeling in environmental chemistry. Wiley, New York. Hopke, P.K., ed. 1991. Receptor modeling for air quality management. Elsevier Science, Amsterdam. Hopke, P.K., Paatero, P., Jia, H., Ross, R.T., and Harshman, R.A. 1998. Three-way (PARAFAC) factor analysis: Examination and comparison of alternative computational methods as applied to ill-conditioned data. Chemometr. Intell. Lab. Syst. 43, 25–42. Hopke, P.K., Ramadan, Z., Paatero, P., Norris, G., Landis, M., Williams, R., and Lewis, C.W. 2003. Receptor modeling of ambient and personal exposure samples: 1998 Baltimore particulate matter epidemiology-exposure study. Atmos. Environ. 37, 3289–3302. Hopke, P., Zhou, L., and Poirot, R.L. 2005. Reconciling trajectory ensemble receptor model results with emissions. Environ. Sci. Technol. 39, 7980–7983. Hsu, Y.-K., Holsen, T.M., and Hopke, P.K. 2003a. Locating and quantifying PCB sources in Chicago: Receptor modeling and field sampling. Environ. Sci. Technol. 37, 681–690. Hsu, Y.-K., Holsen, T.M., and Hopke, P.K. 2003b. Comparison of hybrid receptor models to locate PCB sources in Chicago. Atmos. Environ. 37, 545–562. Huang, S., Rahn, K.A., and Arimoto, R. 1999. Testing and optimizing two factor-analysis techniques on aerosol at Narragansett, Rhode Island. Atmos. Environ. 33, 2169–2185. Husar, R.B., Tratt, D.M., Schichtel, B.A., Falke, S.R., Li, F., Jaffe, D., Gasso, S., Gill, T., Laulainen, N.S., Lu, F., Reheis, M.C., Chun, Y., Westphal, D., Holben, B.N., Gueymard, C., McKendry, I., Kuring, N., Feldman, G.C., McClain, C., Frouin, R.J., Merrill, J., DuBois, D., Vignola, F., Murayama, T., Nickovic, S., Wilson, W.E., Sassen, K., Sugimoto, N., and Malm, W.C. 2001. Asian dust events of April 1998. J. Geophys Res. 106(D16), 18317–18330. Jeong, C.H., Hopke, P.K., Kim, E., and Lee, D.W. 2004a. Measurement of real-time PM2.5 mass, sulfate, and carbonaceous aerosols at the multiple monitoring sites. Atmos. Environ. 38, 5247–5256. Jeong, C.H., Hopke, P.K., Kim, E., and Lee, D.W. 2004b. The comparison between thermal-optical transmittance elemental carbon and aethalometer black carbon measured at multiple monitoring sites. Atmos. Environ. 38, 5193–5204. Juntto, S., and Paatero, P. 1994. Analysis of daily precipitation data by positive matrix factorization. Environmetrics 5, 127–144. Keeler, G.J., and Samson, P.J. 1989. Spatial representativeness of trace element ratios. Environ. Sci. Technol. 23, 1358–1384. Kim, B.M., and Henry, R.C. 1999. Extension of self-modeling curve resolution to mixtures of more than three components. Part 2. Finding the complete solution. Chemometr. Intell. Lab. Syst. 49, 67–77. Kim, B.M., and Henry, R.C. 2000. Extension of self-modeling curve resolution to mixtures of more than three components. Part 3. Atmospheric aerosol data simulation studies. Chemometr. Intell. Lab. Syst. 52, 145–154. Kim, E., and Hopke, P.K. 2004a. Source apportionment of fine particles at Washington, DC utilizing temperature resolved carbon fractions. J. Air Waste Manag. Assoc. 54, 773–785. Kim, E., and Hopke, P.K. 2004b. Improving source identification of fine particles in a rural northeastern U.S. area utilizing temperature resolved carbon fractions. J. Geophys. Res. 109, D09204. Kim, E., and Hopke, P.K. 2006. Characterization of fine particle sources in the Great Smoky Mountains Area. Sci. Total Environ. 368, 781–794. Kim, E., Larson, T.V., Hopke, P.K., Slaughter, J.C., Sheppard, L.E., and Claiborn, C. 2003a. Source apportionment of PM2.5 in an arid northwest U.S. city by positive matrix factorization. Atmos. Res. 66, 291–305.

Theory and Application of Atmospheric Source Apportionment

31

Kim, E., Hopke, P.K., and Edgerton, E.S. 2003b. Source identification of Atlanta aerosol by positive matrix factorization. J. Air Waste Manag. Assoc. 53, 731–739. Kim, E., Hopke, P.K., Larson, T.V., and Covert, D.S. 2004a. Analysis of ambient particle size distributions using positive matrix factorization and Unmix. Environ. Sci. Technol. 38, 202–209. Kim, E., Hopke, P.K., and Edgerton, E.S. 2004b. Improving source identification of Atlanta aerosol using temperature resolved carbon fractions in positive matrix factorization. Atmos. Environ. 38, 3349–3362. Kim, E., Hopke, P.K., and Qin, Y. 2005. Estimation of organic carbon blank values and error structures of the speciation trends network data for source apportionment. J. Air Waste Manag. Assoc. 55(8), 1190–1199. Lawson, C.L., and Hanson, R.J. 1974. Solving least-squares problems. Prentice-Hall, Englewood Cliffs, NJ. Lee, E., Chan, C.K., and Paatero, P. 1999. Application of positive matrix factorization in source apportionment of particulate pollutants in Hong Kong. Atmos. Environ. 33, 3201–3212. Lee, J.H., Yoshida, Y., Turpin, B.J., Hopke, P.K., Poirot, R.L., Lioy, P.J., and Oxley, J.C. 2002. Identification of sources contributing to mid-Atlantic regional aerosol. J. Air Waste Manag. Assoc. 52, 1186–1205. Lewis, C.W., Norris, G.A., Henry, R.C., and Conner, T.L. 2003. Source apportionment of Phoenix PM2.5 aerosol with the Unmix receptor model. J. Air Waste Manag. Assoc. 53, 325–338. Lin, C., Cheng, M., and Schroeder, W. 2001. Transport patterns and potential sources of total gaseous mercury measured in Canadian high Arctic in 1995. Atmos. Environ. 35, 1141–1154. Liu, D.Y., Prather, K.A., and Hering, S. 2000. Variations in the size and chemical composition of nitrate-containing particles in Riverside, CA. Aerosol Sci. Technol. 33, 71–86. Liu, W., Hopke, P.K., and Vancuren, R.A. 2003. Origins of fine aerosol mass in the western United States using positive matrix factorization. J. Geophys. Res. 108(D23), Paper No. 4716, doi:10.1029/2003JD003678. Malinowski, E.R. 1991. Factor analysis in chemistry (second ed.). Wiley, New York. Malm, W.C., Johnson, C.E., and Bresch, J.F. 1986. Application of principal component analysis for purposes of identifying source-receptor relationships. In: Pace, T.G., ed. Receptor methods for source apportionment. Air Pollution Control Association, Pittsburgh, PA, pp. 127–148. Malm, W.C., Gebhart, K.A., and Henry, R.C. 1990. An investigation of the dominant source regions of fine sulfur in the western United States and their areas of influence. Atmos. Environ. 24A, 3047–3060. Malm, W.C., Sisler, J.F., Huffman, D., Eldred, R.A., and Cahill, T.A. 1994. Spatial and seasonal trends in particle concentration and optical extinction in the United States. J. Geophys. Res. 99, 1347–1370. Mason, B. 1966. Principles of geochemistry (third ed.). Wiley, New York. Maykut, N.N., Lewtas, J., Kim, E., and Larson, T.V. 2003. Source apportionment of PM2.5 at an urban IMPROVE site in Seattle, WA. Environ. Sci. Technol. 37, 5135–5142. Miller, M.S., Friedlander, S.K., and Hidy, G.M. 1972. A chemical element balance for the Pasadena aerosol. J. Colloid Interface Sci. 39, 65–176. Nishikawa, M., Hao, Q., and Morita, M. 2000. Preparation and evaluation of certified reference materials for Asian mineral dust. Global Environ. Res. 4(1), 103–113. Paatero, P. 1997. Least squares formulation of robust, non-negative factor analysis. Chemometr. Intell. Lab. Syst. 37, 23–35.

32

Philip K. Hopke

Paatero, P. 1999. The multilinear engine—a table-driven least squares program for solving multilinear problems, including the n-way parallel factor analysis model. J. Comput. Graph. Stat. 8, 854–888. Paatero, P., and Tapper, U. 1993. Analysis of different modes of factor analysis as least squares fit problems. Chemometr. Intell. Lab. Syst. 18, 183–194. Paatero, P., and Hopke, P.K. 2003. Discarding or downweighting high noise variables in factor analytic models. Anal. Chim. Acta 490, 277–289. Paatero, P., Hopke, P.K., Hoppenstock, J., and Eberly, S. 2003. Advanced factor analysis of spatial distributions of PM2.5 in the eastern U.S.. Environ. Sci. Technol. 37, 2460–2476. Paterson, K.G., Sagady, J.L., Hooper, D.L., Bertman, S.B., Carroll, M.A., and Shepson, P.B. 1999. Analysis of air quality data using positive matrix factorization. Environ. Sci. Technol. 33, 635–641. Plaisance, H., Coddeville, P., Roussel, L., and Guillermo, R.A. 1996. Qualitative determination of the source locations of precipitation constituents in Morvan, France. Environ. Technol. 17, 977–986. Plaisance, H., Galloo, J.C., and Guillermo, R. 1997. Source identification and variation in the chemical composition of precipitation at two rural sites in France. Sci. Total Environ. 206, 79–93. Poirot, R.L., and Wishinski, P.R. 1986. Visibility, sulfate, and air mass history associated with the summertime aerosol in northern Vermont. Atmos. Environ. 20, 1457–1469. Poirot, R.L., Wishinski, P.R., Hopke, P.K., and Polissar, A.V. 2001. Comparative application of multiple receptor methods to identify aerosol sources in northern Vermont. Environ. Sci. Technol. 35, 4622–4636. Poissant, L. 1999. Potential sources of atmospheric total gaseous mercury in the St. Lawrence River valley. Atmos. Environ. 33, 2537–2547. Polissar, A.V., Hopke, P.K., Malm, W.C., and Sisler, J.F. 1996. The ratio of aerosol optical absorption coefficients to sulfur concentrations, as an indicator of smoke from forest fires when sampling in polar regions. Atmos. Environ. 30, 1147–1157. Polissar, A.V., Hopke, P.K., Malm, W.C., and Sisler, J.F. 1998. Atmospheric aerosol over Alaska: 2. Elemental composition and sources. J. Geophys. Res. 103, 19045–19057. Polissar, A.V., Hopke, P.K., Paatero, P., Kaufman, Y.J., Hall, D.K., Bodhaine, B.A., Dutton, E.G., and Harris, J.M. 1999. The aerosol at Barrow, Alaska: Long-term trends and source locations. Atmos. Environ. 33, 2441–2458. Polissar, A.V., Hopke, P.K., and Poirot, R.L. 2001a. Atmospheric aerosol over Vermont: Chemical composition and sources. Environ. Sci. Technol. 35, 4604–4621. Polissar, A.V., Hopke, P.K., and Harris, J.M. 2001b. Source regions for atmospheric aerosol measured at Barrow, Alaska. Environ. Sci. Technol. 35, 4214–4226. Ramadan, Z., Song, X.-H., and Hopke, P.K. 2000. Identification of sources of Phoenix aerosol by positive matrix factorization. J. Air Waste Manag. Assoc. 50, 1308–1320. Schauer, J.J., Kleeman, M.J., Cass, G.R., and Simoneit, B.R.T. 1999. Measurement of emissions from air pollution sources. 1. C1 through C29 organic compounds from meat charbroiling. Environ. Sci. Technol. 33, 1566–1577. Schauer, J.J., Kleeman, M.J., Cass, G.R., and Simoneit, B.R.T. 2001. Measurement of emissions from air pollution sources. 3. C1-C29 organic compounds from fireplace combustion of wood. Environ. Sci. Technol. 35, 1716–1728. Schauer, J.J., Kleeman, M.J., Cass, G.R., and Simoneit, B.R.T. 2002. Measurement of emissions from air pollution sources. 5. C1-C32 organic compounds from gasolinepowered motor vehicles. Environ. Sci. Technol. 36, 1169–1180. Seigneur, C., Pai, P., Louis, J.F., Hopke, P.K., and Grosjean, D. 1997. Review of air quality models for particulate matter, Report 4669. American Petroleum Institute, Washington, DC, 311 pp.

Theory and Application of Atmospheric Source Apportionment

33

Seinfeld, J.H., and Pandis, S.N. 1998. Atmospheric chemistry and physics, from air pollution to climate change. Wiley, New York. Shah, S.D., Cocker III, D.R., Wayne, J.M., and Norbeck, J.M. 2004. Emission rates of particulate matter and elemental and organic carbon from in-use diesel engines. Environ. Sci. Technol. 38, 2544–2550. Song, X.H., Polissar, A.V., and Hopke, P.K. 2001. Source of fine particle composition in the northeastern US. Atmos. Environ. 35, 5277–5286. Stohl, A. 1996. Trajectory statistics—a new method to establish source-receptor relationships of air pollutants and its application to the transport of particulate sulfate in Europe. Atmos. Environ. 30, 579–587. Wang, Y.Q., Zhang, X.Y., Arimoto, R., Cao, J.J., and Shen, Z.X. 2004. The transport pathways and sources of PM10 pollution in Beijing during spring 2001, 2002 and 2003. Geophys. Res. Lett. 31, Art. No. L14110. Watson, J.G., Robinson, N.F., Chow, J.C., Henry, R.C., Kim, B.M., Pace, T.G., Meyer, E.L., and Nguyen, Q. 1990. The USEPA/DRI chemical mass balance receptor model, CMB 7.0. Environ. Software 5, 38–49. Watson, J.G., Chow, J.C., Lowenthal, D.H., Pritchett, L.C., and Frazier, C.A. 1994. Differences in the carbon composition of source profiles for diesel and gasoline powered vehicles. Atmos. Environ. 28, 2493–2505. Winchester, J.W., and Nifong, G.D. 1971. Water pollution in Lake Michigan by trace elements from pollution aerosol fallout. Water Air Soil Pollut. 1, 50–64. Xie, Y.-L., Hopke, P.K., Paatero, P., Barrie, L.A., and Li, S.-M. 1999a. Identification of source nature and seasonal variations of Arctic aerosol by the multilinear engine. Atmos. Environ. 33, 2549–2562. Xie, Y.L., Hopke, P., Paatero, P., Barrie, L.A., and Li, S.M. 1999b. Identification of source nature and seasonal variations of Arctic aerosol by positive matrix factorization. J. Atmos. Sci. 56, 249–260. Xie, Y.-L., Hopke, P., Paatero, P., Barrie, L.A., and Li, S.M. 1999c. Locations and preferred pathways of possible sources of Arctic aerosol. Atmos. Environ. 33, 2229–2239. Yakovleva, E., Hopke, P.K., and Wallace, L. 1999. Receptor modeling assessment of PTEAM data. Environ. Sci. Technol. 33, 3645–3652. Yli-Tuomi, T., Hopke, P.K., Paatero, P., Basunia, M.S., Landsberger, S., Viisanen, Y., and Paatero, J. 2003. Atmospheric aerosol over Finnish Arctic: Source analysis by the multilinear engine and the potential source contributions function. Atmos. Environ. 37, 4381–4392. Zeng, Y., and Hopke, P.K. 1989. A study on the sources of acid precipitation in Ontario, Canada. Atmos. Environ. 23, 1499–1509. Zhao, W., and Hopke, P.K. 2004. Source apportionment for ambient particles in the San Gorgonio Wilderness. Atmos. Environ. 38, 5901–5910. Zhao, W., and Hopke, P.K. 2006. Source identification for fine aerosols in Mammoth Cave National Park. Atmos. Res. 80, 309–322. Zhao, W., Hopke, P.K., and Karl, T. 2004. Source identification of volatile organic compounds in Houston. Environ. Sci. Technol 38, 1338–1347. Zhou, L., Kim, E., Hopke, P.K., Stanier, C., and Pandis, S. 2004a. Advanced factor analysis on Pittsburgh particle size distribution data. Aerosol Sci. Technol. 38(S1), 118–132. Zhou, L., Hopke, P.K., and Liu, W. 2004b. Comparison of two trajectory-based models for locating particle sources for two rural New York sites. Atmos. Environ. 38, 1955–1963.

Developments in Environmental Science, Volume 9 Allan H. Legge (Editor) Copyright r 2009 Elsevier Ltd. All rights reserved. ISSN: 1474-8177/DOI:10.1016/S1474-8177(08)00202-7

35

Chapter 2 Use of Trace Metals as Source Fingerprints Richard L. Poirot Abstract This chapter introduces and updates the Fast Aerosol-Sensing Tools for Natural Event Tracking (FASTNET) and Combined Aerosol Trajectory (CATT) tools, and gives examples of applications for analyzing aerosol pollution events and space/time patterns and trends in selected aerosol ionic, carbonaceous, and trace element species. The goal is to encourage use of and feedback on these tools from a broad, international community of air pollution researchers and analysts, so that a growing herd of ‘‘fast cats’’ may soon enhance the rate at which our collective knowledge of the causes and effects of air pollution evolves.

2.1. Introduction

In an ideal world, the pace at which we are able to find, acquire, understand, merge, and analyze air pollution-related data would be limited only by the speed of our mouse clicks, as we sit sipping latte in a wireless Starbucks Cafe´, and all the information we needed would appear magically on our laptop computer screens. In the real world, such information rarely lies at our fingertips, we face obstacles such as those articulated in the 1989 National Academy of Science (NAS, 1989) report on ‘‘Information technology and the conduct of research: The users view.’’ The researcher is not aware of all the relevant data; if he is aware of the data, he cannot get access to them; if he can access the data, he cannot read them; if he can read the data, he does not know how good they are; and if he finds the data to be good, he cannot merge them with other data.

Corresponding author: E-mail: [email protected]

36

Richard L. Poirot

There have been considerable advances in the field of information technology in the 15 years since this rather gloomy NAS assessment, and there is currently a multitude of air quality-related data accessible in various online data archival and retrieval systems. At the same time, the air pollution researcher’s ‘‘data problem’’ has taken on new dimensions, as the sheer number, size, and complexity of the ‘‘relevant’’ data have expanded exponentially. This ‘‘data deluge’’ problem is especially acute for those with an interest in aerosol pollution, as aerosols are so inherently complex and as there are so many different kinds of relevant data—from extensive, new, surface-based monitoring networks, meteorological and aerosol forecast models, satellite imagery and associated data products, etc. FASTNET (Fast Aerosol-Sensing Tools for Natural Event Tracking; Poirot et al., 2004) and CATT (Combined Aerosol Trajectory Tools; Husar et al., 2004) are two recently developed, online data acquisition and analysis tools that will help improve our efficiency as air quality analysts. Both tools were developed by the Center for Air Pollution Impact and Trends Analysis (CAPITA) at Washington University, with funding support from the five U.S. Regional Planning Organizations (RPOs), and are maintained by CAPITA under the larger domain and infrastructure of the DataFed Federation (http://datafed.net/, last accessed on July 15, 2008), developed with initial funding support from the National Science Foundation (Husar, 2001) and the National Aeronautics and Space Administration (NASA) (Falke & Husar, 2003). This chapter introduces and updates the FASTNET and CATT tools, and gives examples of applications for analyzing aerosol pollution events and space/time patterns and trends in selected aerosol ionic, carbonaceous, and trace element species. The goal is to encourage use of and feedback on these tools from a broad, international community of air pollution researchers and analysts, so that a growing herd of ‘‘fast cats’’ may soon enhance the rate at which our collective knowledge of the causes and effects of air pollution evolves.

2.2. Methods

FASTNET and CATT are two of several ‘‘projects’’ developed and maintained within—and dependent on the fundamental data and architecture developed for—the DataFed Federation (http://datafed.net/; see Fig. 2.1, last accessed on August 19, 2008). Both projects use a common data viewer (Fig. 2.2) that allows users to access, explore, screen, aggregate, layer, display, and acquire air quality-related data

Figure 2.1.

Datafed Federation (http://Datafed.net, last accessed on July 17, 2008).

Use of Trace Metals as Source Fingerprints 37

38

Richard L. Poirot

Figure 2.2. Data Viewer (http://webapps.datafed.net/datafed.aspx, last accessed on July 17, 2008).

from roughly 100 different sources. These various data can be selected from an extensive data catalog (‘‘data catalog’’) through which some data are directly accessed from their native data repositories, maintained by the data generators, whereas others are automatically cached by DataFed.net to ensure archiving, facilitate navigation and merging of relational space/time features, or generate ‘‘value-added’’ data products. For the FASTNET project, 14 specific datasets, summarized in Fig. 2.3, are emphasized in the DataFed catalog. These include various surface-based aerosol, meteorology, and visibility data; aerosol forecast model results, and satellite data and images. Many of these data sets are available in near-real time, but have only become available (or archived) quite recently, while others, such as the filter-based aerosol chemistry data from the IMPROVE network (Interagency Monitoring of Protected

NOAA/NCDC

DOD/NRL

USDA/USFS

NASA/USFS

GOES

NASA/GSFC

NASA/CAPITA

NPS/CIRA

NPS/CIRA

MWH

NPS

NEXTRAD

NAAPS

HMS FIRE

MODIS FIRE

GASP

TOMS AI

SEAWiFS

VIEWS CHEM

ATAD

MWH webcam

NPS webcam

FASTNET datasets and characteristics.

5/1/2003

NOAA/STI

ASOS STI

Figure 2.3.

7/30/1998

NOAA/PSWC

4/5/2004

7/19/2004

1/1/1988

3/1/1988

1/1/2000

7/25/1996

7/16/2004

8/1/2003

10/1/2003

1/1/2001

4/19/1997

7/1/2002

91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 Start Date

SURF MET

90

EPA/STI

89

AIRNOW

88

Provider

Dataset/ Time

aerosol aerosol

hour 6 hour day day .5 hour day day 3 hour

now now now now now now

aerosol archived archived

day day

now now

aerosol

meteorol

aerosol

fire

fire

aerosol

meteorol

aerosol

6 hour

cache

archived

cache

5/31/2002

1/1/2003

12/31/2002

aerosol

hour

meteorol

now

cache

hour

aerosol

now

cache

Type

hour

Time Resolution

now

End Date

Use of Trace Metals as Source Fingerprints 39

40

Richard L. Poirot

Visual Environments) are available with a time lag of approximately 1 year, but have a longer historical record. The CATT project currently uses ‘‘only’’ the IMPROVE aerosol data and 5-day backward trajectories calculated for IMPROVE sites and sample dates by the National Park Service, using the National Oceanic and Atmospheric Administration (NOAA) Atmospheric Transport and Dispersion (ATAD; Heffter, 1980) model. CATT links the aerosol chemistry and trajectory data, allowing detailed exploration of space, time, and composition patterns. Additional details on FASTNET data tools, techniques, and sample applications are available in Poirot et al. (2004). The FASTNET homepage, illustrated in Fig. 2.4, can be accessed at http://datafed.net/ projects/FASTNET/FASTNET_Links.htm (last accessed on July 15, 2008). Additional details on CATT data, tools, techniques, and sample applications are available in Husar et al. (2004). The CATT homepage, illustrated in Fig. 2.5, can be accessed at http://datafed.net/projects/catt/ CATT_Links.htm (last accessed on July 15, 2008).

2.3. Sample applications

‘‘SURF_MET’’ is one of the many useful FASTNET datasets, and includes hourly global surface meteorological observations from more than 10,000 sites, with especially dense coverage over North America. These data are provided initially and in near-real time by the NOAA World Data Center (WDC) for Meteorology (http://www.ncdc.noaa.gov/ oa/wmo/wdcamet.html, last accessed on July 15, 2008), in Asheville, NC, but the historical archive is not accessible online. Starting in mid-1998, the Plymouth State Weather Center (PSWC; http://vortex.plymouth.edu/, last accessed on July 15, 2008) at Plymouth State College, New Hampshire began routinely downloading and archiving these data, and has made them accessible to CAPITA. At CAPITA, the WDC/PSWC data are routinely downloaded and reformatted in a relational database to facilitate exploration of both space and time patterns in the data. CAPITA also uses the visibility (visual range), temperature, and dew point data to calculate several ‘‘processed data variables,’’ including light extinction (Bext—in inverse megameters, calculated in this case as 3000/ Visual Range in kilometer), humidity-screened light extinction (FBext— with hours with relative humidity W90% filtered out), and humidityscreened and adjusted light extinction (RHBext, which first filters out RH W90%, and then reduces the remaining data values by an inverse hygroscopic growth function). This RHBext variable relates very closely to concentrations of fine particle PM2.5 mass, but includes much more

Figure 2.4.

FASTNET website (http://datafedwiki.wustl.edu/index.php/FASTNET, last accessed on July 17, 2008).

Use of Trace Metals as Source Fingerprints 41

42

Richard L. Poirot

Figure 2.5. CATT website (http://datafed.net/projects/catt/CATT_Links.htm, last accessed on July 17, 2008).

dense spatial and temporal coverage than current PM2.5 networks. CAPITA also adds a wind vector display option (SURF_MET_WIND) based on the wind speed and direction data. These processed data variables, along with the raw data are then redistributed via FASTNET, providing universal access to highly ‘‘usable’’ forms of the data— generated initially by 10,000 individual weather observation stations, collected and posted by the WDC, downloaded and archived by the PSWC, accessed, processed, and redistributed by CAPITA, and subsequently employed by a broader FASTNET ‘‘user community’’ for analysis of air pollution events. The raw data were not collected for this purpose, but take on added value as they pass through many helping hands. Figure 2.6 shows both the RHBext and wind vector spatial displays over the U.S. for February 19, 2004 at 17:00 UTC. Note the two separate

-125

-120

-115

Figure 2.6.

-110

-105

-100

-95

RHBext and wind vectors on February 19, 2004, 17:00 UTC.

Data Provider: Plymouth State Weather Center

24

26

28

30

32

34

36

38

40

42

44

46

48

50

52

SURF_MET.RHBext: Surface Extinction Coefficient & Surface Wind Vectors

-90

-85

-80

-75

Delivery: DataFed.Net

-70

800 450 100

RHBext

KPTT 2004-02-19T17:00:00 190

Use of Trace Metals as Source Fingerprints 43

44

Richard L. Poirot

areas of high (dried) extinction—one along the eastern half of the U.S.– Canada border, and another near the Texas–Mexico border. The northeastern event, associated with very low wind speeds, was subsequently identified by the FASTNET user community as being caused primarily by high ammonium nitrate concentrations under low mixing heights (Poirot & Husar, 2004b). The southwestern event, associated with very high wind speeds, was subsequently identified by the FASTNET user community as being caused primarily by high windblown dust concentrations (Husar & Poirot, 2004). Figure 2.7 shows other FASTNET options and data, illustrating the smoke impacts from the July 2002 Quebec forest fires with a spectral reflectance (350 nm) image from the SEAWIFS satellite, underlying the concurrent 1-h SURF_MET RHBext data on the left, and the 24-h IMPROVE organic carbon (OC; red) and sulfate (SO4; yellow) on the right side of the figure. The satellite view integrates across the vertical column, and the surface data indicate where the highest smoke concentrations mixed down to the surface. Surface impacts were highest in the eastern Mid-Atlantic region, with generally lower concentrations closer to the source area, as much of the smoke passed by aloft (Poirot & Husar, 2004a; Taubman et al., 2004). Zooming out (FASTNET zoom tool) to cover a slightly larger area and adding the IMPROVE chemical data on the right, it can also be observed that there were also concurrently moderately high SO4 concentrations, but ‘‘displaced’’ to the west and south of the Ohio River Valley region, which most typically experiences the highest summer SO4 concentrations. For readers who may have an interest in further exploration of the combined SEAWIFS images and surface meteorology data, the view (and the underlying data that can be browsed) used for the left side of Fig. 2.7 can be found by opening the FASTNET Viewer and then selecting ‘‘File,’’ ‘‘Open Page,’’ ‘‘RichP,’’ and ‘‘Qfire1.page.’’ From here, you can explore other time periods of interest, zoom or pan to other regions, add or delete other datasets, change data scales and rendering options, download (some) data in ASCII csv format, and save your page by ‘‘File,’’ ‘‘Save Page As,’’ and then create your own directory and file name. One of the various and most ‘‘simple’’ options provided by the CATT tool is to select a single sample day for the IMPROVE network and plot the back trajectories (4 per day, 5 days back) for all the sites with data on that date. Relative concentrations of different chemical species of interest can also be plotted, and the trajectories can be ‘‘color weighted’’ to emphasize the flows associated with highest (and lowest) species concentrations. Note that the scales in Fig. 2.8 are different from each other for the color-weighted trajectories, ranging from W20 mg/m3 for OC, to 10 mg/m3

-86

-84

-82

-80

-78

-76

-74

-72

-68

-66

Delivery: DataFed.Net

-70

RP11 2002-07-07 116.00.00 NaN

Delivery: DataFed.Net

-100 -98 -96 -94 -92 -90 -88 -86 -84 -82 -80 -78 -76 -74 -72 -70 -68 Data Provider: Coop Inst for Research Atmosphere, Colo State

24

26

28

30

32

34

36

38

40

42

44

46

48

50

Figure 2.7. July 7, 2002, SEAWIFS Spectral Reflectance (350 nm) and Surface RHBext (left panel) with fine organic carbon (red) and sulfate (yellow) from IMPROVE sites (right).

Data Provider: Plymouth State Weather Center

-88

32

34

36

38

40

42

44

46

48

50

SURF_MEI.FBext: Surface Extiction Coefficient

Use of Trace Metals as Source Fingerprints 45

Figure 2.8. CATT color-weighted back trajectories for IMPROVE sites on July 7, 2002, unweighted (top left) and weighted by high (red) OC (top right), SO4 (bottom right), and chloride Cl (bottom left).

46 Richard L. Poirot

Use of Trace Metals as Source Fingerprints

47

for SO4 and 1 mg/m3 for chloride (Cl). The trajectories for highest OC concentrations on July 7, 2002 clearly implicate the central Quebec origin of the smoke, whereas those weighted by highest SO4 reside primarily over the Ohio River Valley—even as the highest observed SO4 concentrations were displaced to the south and west of that high emission region. On the same day, the highest (but not very high) Cl concentrations were observed at far western sites and associated with marine flows. A similar set of color-weighted trajectories are displayed in Fig. 2.9 for August 12, 2002, when a major SO4 event was impacting the Northeast (upper left) and northwestern forest fires were contributing to high OC at several western sites (upper right). A major carbon study was being conducted at Yosemite National Park during this time period (McMeeking et al., 2004) and carbon-14 dating, soluble potassium, and a number of molecular organic wood smoke tracers, including levoglucosan, retene, vanillian, dehydroabietic acid, and 7-oxo-dehydroabietic acid, all confirmed the strong influence of wood smoke—traced, in turn, primarily to fires in Oregon, based on OC concentrations during this time period (Engling et al., 2004). The origins of smoke plumes from large, isolated forest fires, such as those impacting the Northeast on July 7, 2002 and the Northwest on August 12, 2002, are relatively easy to trace, in part because the emissions modulations are so extreme in space and time. The origins of large SO4 events, however, are much more difficult to trace to individual sources, because there are so many contributing sources, and because SO4 is predominantly a secondary pollutant, resulting from highly variable atmospheric transformation of gaseous SO2. The trajectories in the lower two panels of Fig. 2.9 are color weighted by highest concentrations of selenium (Se; left) and nickel (Ni; right). For this event, the highest selenium trajectories correspond to the highest SO4 trajectories, passing over the Ohio River Valley, whereas the highest nickel trajectories are much more constrained to the Northeast. Expanding from individual events to multi-year patterns, several of the CATT tools allow analyses of long-term patterns based on various gridded ensemble trajectory metrics (see bottom row of CATT Links page in Fig. 2.5). One of these optional gridded trajectory metrics is referred to as ‘‘incremental probability,’’ which is described in detail in the CATT Illustrated Instruction Manual. Briefly, a grid, for which domain and grid-square size can be user defined, is employed to aggregate trajectory endpoints from large numbers of trajectories for each grid square. For a given site and time domain (also user defined), an ‘‘all day probability field’’ is calculated for all IMPROVE sample days. A grid square’s probability is expressed as the fraction of trajectory endpoints in that

Figure 2.9. CATT color-weighted back trajectories for IMPROVE sites on August 12, 2004 emphasizing highest concentrations of SO4 (top left), OC (top right), Se (bottom left), and Ni (bottom right).

48 Richard L. Poirot

Use of Trace Metals as Source Fingerprints

49

square divided by the total in all squares. For a selected pollutant species, a ‘‘high day probability field’’ is calculated in the same way but limited to a high subset of pollutant concentration days, where ‘‘high’’ (or any other part of the distribution) can be defined by the user either as an absolute value or as a percentile range of the distribution. The incremental probability is the high day probability minus the all day probability. How much more likely is a location to be upwind if the pollutant was higher than it is on an everyday basis? Other optional ensemble trajectory metrics include the potential source contribution function, upwind average concentration, and concentrationweighted probability. Each can be calculated for individual sites, or aggregated for groups of sites. The gridded results can be displayed in CATT, with user-defined rendering options, and/or exported in ASCII csv format, which includes the latitude and longitude of grid-square locations and the calculated metric. These csv files can then be entered directly into spatial analysis programs such as ArcViews for additional analysis and more refined plotting. Figure 2.10 shows the CATT incremental probability fields for highest selenium and nickel aggregated for all IMPROVE sites from 2000 through 2003. If selenium is high, the air has most likely resided over the ‘‘Midwest,’’ and if nickel is high, the air has most likely resided over the ‘‘East Coast,’’ especially the Northeast and Southeast coastal areas. In Figure 2.11, these gridded incremental probability data have been exported to ArcView, interpolated, and are compared with (1998 EPA EGRID) U.S. SO2 emissions from coal (left) and oil (right) utility sources. Selenium appears to be a good tracer for influence from coalburning emissions, whereas nickel (like vanadium) is a good tracer for residual oil combustion. These observations are consistent with recent modeling results from the Positive Matrix Factorization (PMF) and Unmix receptor models for a number of northeastern IMPROVE sites (Coutant et al., 2002; Lee et al., 2003; Poirot et al., 2001; Polissar et al., 2001; Song et al. 2001), which identified coal and oil combustion sources with strong selenium and nickel contents, respectively. For those receptor-modeling studies that also included trajectory interpretation of the sources (Poirot et al., 2001; Polissar et al., 2001; Lee et al., 2003), the coal and oil sources were associated with flows from the Midwest and Northeast urban corridors, respectively. The models, which can only identify sources with fixed chemical compositions, also tended to divide the coal source influence into two separate ‘‘source components,’’ one with a high sulfur:selenium ratio and one with a lower sulfur:selenium ratio, interpreted as representing the maximal and

50

Richard L. Poirot VIEWS_OL:SEf LYBR1 2002-07-31 NaN ug/m3

Combined Aerosol Trajectory Tool (CATT) 54 52 50 48 46 44 42 40 38 36 34 32 30 28 SEf ug/m3 26 0.001 0.0008 24 0.0006 0.0004 22 0.0002 0 20 -125 -120

Monitoring Network IMPROVE

Se -115

-110

-105

-100

-95

-90

-85

-80

-70

VIEWS_OL:NIf LYBR1 2002-07-31 NaN ug/m3

Combined Aerosol Trajectory Tool (CATT) 54 52 50 48 46 44 42 40 38 36 34 32 30 28 NIf ug/m3 26 0.001 0.0008 24 0.0006 0.0004 22 0.0002 0 20 -125 -120

-75

Delivery: DataFed.Net

Data Provider: Coop Inst for Research Atmosphere, Colo State

Monitoring Network IMPROVE

Ni -115

-110

-105

-100

Data Provider: Coop Inst for Research Atmosphere, Colo State

-95

-90

-85

-80

-75

-70

Delivery: DataFed.Net

Figure 2.10. Incremental probability fields for top 10% Se (top) and Ni (bottom), all IMPROVE sites 2000–2004.

minimal degrees of secondary aerosol formation—from coal sources— encountered at the receptor sites, respectively. This concept of varying sulfur:selenium ratios can be further explored over a larger domain by applying the CATT ‘‘upwind average concentration’’ calculation. For each grid square, an average pollutant

Use of Trace Metals as Source Fingerprints

51

Figure 2.11. Incremental probabilities for highest 10% Se (top) and Ni (bottom) from all IMPROVE sites 2000–2002, Compared with 1998, SO2 emissions from coal (top) and oil (bottom) utility boilers.

52

Richard L. Poirot

concentration is calculated for a receptor site for all trajectories passing through that grid square en route to the receptor. Like the incremental probability metric, this calculation can also be aggregated across multiple receptor locations, following the approach of Kenski (2004). Figure 2.12 displays the average upwind selenium (top) and sulfate (bottom) for two different 4-year time periods, 1992–1995 (left) and 2000–2003 (right). This analysis was constrained to 42 IMPROVE sites that commenced operation before 1992, so that any apparent changes over time would not be influenced by the addition of new monitoring sites closer to or further from emission source areas. The spatial patterns for both pollutants and time periods are generally quite similar, although the magnitude of the average upwind sulfate has declined substantially in recent years, whereas the magnitude of the average upwind selenium has remained relatively constant. For whatever reason(s), the effective control strategies for SO2—including both stack scrubbers and switching to lower sulfur coals—have had relatively minimal influence on selenium emissions. Figure 2.13 compares sulfate and selenium as measured at individual sites across the IMPROVE network for the 2000–2003 period on the left with the upwind average sulfate and selenium as calculated for individual grid cells during this same time period on the right. The measured species are not well correlated (R2 ¼ 0.36) due to variable secondary transformation rates that result in high sulfate:selenium ratios during periods of rapid transformation (typically during summer) and much lower ratios during colder, less humid periods when transformation rates are much slower. However, both species strongly tend to come from the same upwind locations. Figures 2.14 and 2.15 provide two perspectives on the shift in sulfate:selenium ratios over time and space. Figure 2.14 compares the upwind sulfate:selenium ratios over the 1992–1995 and 2000–2003 time periods, and shows that although the correlations are strong in both time periods, the slope has decreased by approximately 12% in recent years. Per unit of selenium, we are seeing less sulfate. Figure 2.15 shows the shifting spatial pattern of sulfate:selenium ratios over time, with the darker shading emphasizing the highest selenium:sulfate ratios. During the earlier time period, the highest selenium:sulfate ratios occurred primarily in the Northwest, where the sulfate concentrations are lowest. But in recent years, as sulfate has declined, the area of higher selenium:sulfate ratio has shifted further east. Whether this reflects unique features of emissions control equipment or a shift to lower sulfur (Western) coal—which may have slightly higher selenium relative to sulfate—is a question for future study. Selenium remains as a good tracer

Figure 2.12. Average upwind Se (top) and sulfate (bottom) for 42 long-term IMPROVE sites, over two 4-year periods, 1992–1995 (left) and 2000–2003 (right).

Use of Trace Metals as Source Fingerprints 53

0

5

10

15

20

25

0

0.002

0.006

Selenium (ug/m3)

0.004

0.008

0.01

Average Upwind Sulfate (ug/m3) 0

2

4

6

8

0

0.0008

0.0012

0.0016 Average Upwind Selenium (ug/m3)

0.0004

y = 4365x R2 = 0.85

0.002

Figure 2.13. Sulfate vs. selenium at 42 IMPROVE sites, 2000–2002, as measured at the sites (left) and as calculated for upwind average grid locations (right).

Sulfate (ug/m3)

30

54 Richard L. Poirot

Use of Trace Metals as Source Fingerprints

55

Figure 2.14. Shift in upwind SO4:Se ratios over time.

for coal sulfate, but the quantitative relationship is not constant over space and time. The overall change in average upwind sulfate (1992–1995 minus 2000–2003) is plotted in Fig. 2.16. Changes have generally been greatest over areas in the eastern U.S. and Canada, where total SO2 emissions reductions have been greatest over this time period. The area of greatest reduction, over the southern Ohio River Valley, is displaced slightly to the southwest of locations where SO2 reductions have been greatest. Possibly this may reflect more efficient sulfate formation in moist, warm airflows moving northeast off the Gulf of Mexico around the backside of stagnating summer high-pressure systems. No change is evident in western areas, and (slight) increases are indicated over areas in Alberta (and Montana) and Mexico. These apparent increases are very small and well within the uncertainty of the analysis method. It should also be cautioned that these far northern and southern areas are near the edges of

56

Richard L. Poirot

Figure 2.15. Spatial pattern of changing upwind sulfate:selenium ratios, 1992–1995 (top) and 2000–2002 (bottom).

the ATAD trajectory domain and may be further influenced by the relative scarcity of rawindsonde data used to drive the model. Figure 2.17 displays average upwind aerosol nitrate concentrations over the two time periods, and shows a slight decrease over southern

Use of Trace Metals as Source Fingerprints

57

Figure 2.16. Change in upwind average SO4 (mg/m3) between 1992–1995 and 2000–2003.

California, with a slight increase (about 0.1 mg/m3) over the central ‘‘corn belt’’ region, where relatively high agricultural ammonia emissions (rather than NOx) appear to be an important factor for aerosol nitrate formation. This area of elevated upwind nitrate appears to extend into the Canadian plains, although this may also be related to flows of colder and relatively sulfate-free air most conducive to aerosol nitrate formation. The extent to which this increase in nitrate may be due to increases in agricultural ammonium and/or decreases in acidic sulfate concentrations is not clear. This apparent change may have also been influenced by network-wide changes in IMPROVE nitrate sampling methods, although those changes were estimated to result in lower (not higher) post-96 concentrations (Schichtel, 2006). Several ‘‘unusual,’’ large-scale springtime nitrate aerosol events have been observed in the past few years over a region extending east of the earlier-indicated area of highest upwind nitrate, along the U.S.–Canada border, including several in 2004 (Poirot & Husar, 2004b; Husar & Poirot, 2004) and one in 2005 (Kenski et al., 2005). In both cases, relatively large groups of data analysts in this large transboundary region were quickly attracted by common interest to exchange data and work on a collaborative group analysis. FASTNET tools and users made substantial contributions to these analyses, and the CATT can be employed for additional retrospective analyses as soon as the slower

58

Richard L. Poirot

Figure 2.17. Average upwind nitrate for 42 IMPROVE sites, 1992–1995 (top) and 2000– 2002 (bottom).

filter chemistry and trajectory data become available. Quite possibly such events are not so much unusual as they are typical, but have only recently become readily detectable with the recent proliferation of new continuous and speciation measurements, and—equally important—with

Use of Trace Metals as Source Fingerprints

59

the development and more widespread use of fast and powerful new data acquisition and analysis tools. 2.4. Conclusions

The intent here was simply to introduce the new FASTNET and CATT data acquisition and analysis tools and encourage others to try them out (http://datafed.net, last accessed on August 19, 2008). A number of specific sample applications were presented without intent to draw specific conclusions, but rather to provoke future exploration. Like many of the featured data that improve in value by passing through many hands, the analysis tools will also be improved with use and feedback from data analysts.

ACKNOWLEDGMENTS

The many datasets presented here were graciously provided by data generators too numerous to mention individually, but each are referenced in the datafed.net catalog. Special thanks to Serpil Kayin, (MARAMA) and Gary Kleiman (NESCAUM), who provided the management for the CATT and FASTNET projects, and of course to the toolmakers: Rudy Husar, Steffan Falke, and Kari Hoijarvi at CAPITA.

REFERENCES Coutant, B., Kelly, T., Ma, J., Scott, B., Wood, B., and Main, H. 2002. Source apportionment analysis of air quality monitoring data: Phase I final report. Prepared for the Mid-Atlantic/Northeast Visibility Union and Midwest Regional Planning Organization by Battelle Memorial Institute, Columbus, OH and Sonoma Technology, Inc., Petaluma, CA, USA. Engling, G., Herkes, P., Carrillo, J., Kreidenweis, S.M., and Collett, J.L. 2004.Organic aerosol composition in yosemite national park during the 2002 yosemite aerosol characterization study. Paper #66, Air &Waste Management Association International Specialty Conference on: Regional and Global Perspectives on Haze: Causes, Consequences and Controversies, Asheville, NC, USA. Falke, S., and Husar, R. 2003. Application of NASA ESE data and tools to particulate air quality management. Proposal to NASA Earth Science REASoN Solicitation CAN-02OES-01. Heffter J.L. 1980. Air resources laboratories atmospheric transport dispersion model (ARLATAD). Technical Memo ERL ARL-81, NOAA, Rockville, MD, USA. Husar, R. 2001. Monitoring and analysis of large-scale aerosol events: Information technology research on collaboration in virtual workgroups. Proposal to NSF on Information Technology Research, NSF 00-126 ITR/AP-IM(GEO).

60

Richard L. Poirot

Husar, R., and Poirot, R. 2004. Texas-Mexico Dust Event, 19 February 2004. http://capita. wustl.edu/capita/capitareports/040219TexasDust/040219TexMexDust.ppt (last accessed on July 15, 2008). Husar, R., Poirot, R., Gebhart, K., Schichtel, B., and Malm, W. 2004. Combined aerosol trajectory tool (CATT) for the IMPROVE Chemical Dataset. Paper #97, A&WMA international specialty conference on regional and global perspectives on haze: Causes, consequences and controversies, Asheville, NC. USA. Kenski, D. 2004. Quantifying transboundary transport of PM2.5: a GIS analysis. Paper #247, 97th Annual A&WMA Conference, Indianapolis, IN, USA. Kenski, D.M., Husar, R., Harrell, M., Conaster, N., Swinford, B., and Turner, J. 2005. The great midwestern PM2.5 event of February 2005. http://www.inawma.org/files/2005-1011_The_Great_Midwestern_PM2.pdf (last accessed on August 16, 2008). Lee, J., Yoshida, Y., Turpin, B., Hopke, P., Poirot, R., Lioy, P., and Oxley, J. 2003. Identification of sources contributing to the mid-Atlantic regional aerosol. J. Air Waste Manag. Assoc. 52, 1186–1205. McMeeking, G.R., Kreidenweis, S.M., Carrico, C., Lee, T., Carrillo, J., Day, D.E., Hand, J.L., and Malm, W.C. 2004. Dry aerosol size distributions and derived optical properties during the yosemite aerosol characterization study. Paper #27, Air & Waste Management Association International Specialty Conference on Regional and Global Perspectives on Haze: Causes, consequences and controversies, Asheville, NC, USA. National Academy of Sciences (NAS). 1989. Information technology and the conduct of research: The user’s view. Report of the Panel on Information Technology and the Conduct of Research, National Academy Press, Washington, DC, USA. Poirot, R., and Husar, R. 2004a. Chemical and physical characteristics of wood smoke in the northeastern US during July 2002: Impacts from Quebec forest fires. Paper #93, Air & Waste Management Association International Specialty Conference on Regional and Global Perspectives on Haze: Causes, Consequences and Controversies, Asheville, NC, USA. Poirot, R., and Husar, R. 2004b. Winter PM Event over the Northeast and Quebec. http:// capita.wustl.edu/capita/ (last accessed on August 16, 2008). Poirot, R.L., Wishinski, P.R., Hopke, P.K., and Polissar, A.V. 2001. Comparative application of multiple receptor methods to identify aerosol sources in northern Vermont. Environ. Sci. Technol. 35, 4622–4636. Poirot, R., Husar, R., Funk, T., Raffuse, S., Dye, T., Kleiman, G., and Kenski, D. 2004. Aerosol and haze observations with the FASTNET distributed monitoring system. Paper #93, Air & Waste Management Association International Specialty Conference on Regional and Global Perspectives on Haze: Causes, Consequences and Controversies, Asheville, NC, USA. Polissar, A.V., Hopke, P.K., and Poirot, R.L. 2001. Atmospheric aerosol over vermont: Chemical composition and sources. Environ. Sci. Technol. 35, 4604–4621. Schichtel, B. 2006. A Discontinuity in the nitrate ion time series at June 1996. Data Advisory: http://vista.cira.colostate.edu/improve/Data/QA_QC/qa_qc_Branch.htm (last accessed on August 16, 2008). Song, X.H., Polissar, A.V., and Hopke, P.K. 2001. Sources of fine particle composition in the northeastern U.S. Atmos. Environ. 35, 5277–5286. Taubman, B.F., Marufu, L.T., Vant-Hull, B.-L., Piety, C.A., Doddridge, B.G., Dickerson, R.R., and Li, Z. 2004. Smoke over haze: Aircraft observations of chemical and optical properties and the effects on heating rates and stability. J. Geophys. Res. 109, D02206, doi:10.1029/2003JD003898.

Developments in Environmental Science, Volume 9 Allan H. Legge (Editor) Copyright r 2009 Elsevier Ltd. All rights reserved. ISSN: 1474-8177/DOI:10.1016/S1474-8177(08)00203-9

61

Chapter 3 Plants as Accumulators of Atmospheric Emissions J. Neil Cape* Abstract The uptake of air pollutants by plants occurs through different pathways and processes, including direct uptake of gases through stomata, deposition to the soil and subsequent root uptake, and deposition to external surfaces of bark and foliage. After uptake, pollutants may diffuse or be actively transported to different plant parts. The accumulation rate depends on the balance between the rates of uptake and removal, either by direct excretion or elimination after metabolic processing. Avoidance of toxic effects may occur through sequestering material within the plant in specialized structures, or in a way that minimizes damage. The effective use of plants as bioindicating accumulators of air pollution requires an understanding of the factors that control the concentrations of accumulated material in and on plants: the original deposition pathway; the biological, physical, and chemical processes that occur; and the effects of climate, substrate, and species on the plant itself.

3.1. Introduction

The effects of air pollutants on plants have been recognized since the early days of the Industrial Revolution, but it was only in the later part of the 20th century that concerted measures were introduced to limit pollutant emissions—and then in response to harmful local effects of emissions on human health (e.g., the Clean Air Acts in Britain in the late 1950s), rather than effects on the natural environment. The issue of long-range transport, and the debate on ‘‘acid rain,’’ which dominated the late 1970s and 1980s, introduced the need for international agreements to

Corresponding author: E-mail: [email protected]

62

J. Neil Cape

tackle air pollution on a regional scale. In Europe, and to a lesser extent in North America, such concerns have led to a very large reduction in the amounts of sulfur and nitrogen oxides emitted into the environment (Cape et al., 2003a). Plans are also in place to restrict emissions of heavy metals and persistent organic pollutants (POPs) into the atmosphere (United Nations Economic Commission for Europe (UNECE), 2005). The widespread removal or modification of clearly identifiable point sources (chimney stacks) and the increasing importance of diffuse sources, such as vehicles and farming practices, have led to problems in identifying and attributing the potential effects of air pollution to particular sources. Whereas it was relatively straightforward to link the increased sulfur content of vegetation growing close to a large coalburning power station or smelter to the emissions from the chimneys, the diffuse and complex nature of pollutant emissions today makes the use of vegetation as a marker more difficult. This review looks at the routes by which different types of air pollutants are taken up by vegetation, and their subsequent fate, and examines the factors that determine whether or not measurements on plants can be used to indicate the presence of harmful emissions in the field. It is not intended as a comprehensive review of the literature, which has been well covered by recent publications (Jeran et al., 2004; Klumpp et al., 2004; Wolterbeek et al., 2003; Zechmeister et al., 2003a), and especially a review of much of the German and central European literature on the accumulation of air pollutants by higher plants (Weiss et al., 2003).

3.2. Mechanisms of uptake for different air pollutants 3.2.1. Gases

Conceptually, the simplest route for a pollutant gas to enter a plant is through stomata. The rate of uptake through stomata is partly controlled by the degree of stomatal opening, and partly by the rate at which the gas molecules can be transported to the leaf surface, which depends on the molecular size and the degree of air turbulence. Although high concentrations of some pollutant gases, typical of areas close to point sources, have been shown to influence the degree of stomatal opening (Mansfield, 1998), the exchange of gases is passive and in proportion to the flux of water out of the leaf. The diffusion flux is determined by the relative molecular diffusion constant in air, which varies inversely with the square root of the molecular weight of the molecule concerned. It is generally assumed for most gases that the effective concentration within the leaf is

Plants as Accumulators of Atmospheric Emissions

63

zero, so that the driving potential for uptake is simply the air concentration of the gas at the leaf surface. However, this rather simple picture may conceal a range of different behavior, depending on the flux into the sub-stomatal cavity (determined by the air concentration and the stomatal conductance), the solubility of the gas in water (or in the aqueous film in the sub-stomatal cavity), and the rate of removal by transport within the apoplast or reaction with components of the apoplast or cell walls. For many gases, the limiting factor in uptake may be the stomatal conductance, which depends on the physiological state of the leaf, and external factors such as temperature, water vapor pressure deficit, soil water availability, and sunlight intensity. Consequently, the air concentration (driving potential) may not be a good indicator of the actual flux of gas into the leaf, and any subsequent effects. The ‘‘pollutant absorbed dose’’ (Fowler & Cape, 1982) has been recognized as particularly important when trying to assess the relationship between ozone concentrations and effects on plants (Emberson et al., 2000). Some scenarios are shown schematically in Fig. 3.1. In case A, a sparingly soluble unreactive gas comes to equilibrium with the leaf tissue; although only the aqueous compartments are shown in the diagram, the gas might also be lipid soluble, and so partition into the cuticle and cell walls also. The gas may be totally inert, or may be metabolized slowly, giving rise to non-detectable by-products or possibly to the accumulation of a reaction product. An example of this type of behavior would be tetrachloroethylene (C2Cl4), which comes to equilibrium with leaf tissue within a few hours of exposure (Binnie et al., 2002; Brown et al., 1998), and leads to the long-term accumulation of trichloroacetate as a metabolic product (Franzaring et al., 2000). Where the rate of removal, by transport or metabolic uptake, exceeds the transfer rate from the atmosphere, a quasi-equilibrium may be established, with a dynamic balance between the concentrations measured in the leaf and the air concentrations (as shown in Fig. 3.1, B) exemplified by benzene (C6H6) or toluene (methyl benzene) (Binnie et al., 2002). For this to happen, the aqueous solubility of the gas has to be low enough that a significant air concentration is maintained in equilibrium, and subsequent transport or metabolism of the compound within the leaf is sufficiently slow that the aqueous-phase concentration of the molecule can be sustained. For sparingly soluble gases such as ozone (O3), reaction or transport away from the sub-stomatal cavity may be sufficiently fast that concentrations in the sub-stomatal cavity are maintained close to zero, so that uptake rates are proportional to the air concentration. The main sink for O3 is reaction with ascorbate in the apoplast (Chameides, 1989; Plo¨chl et al., 2000), which is sufficiently fast to maintain uptake rates through stomata

64

J. Neil Cape air concentration

aqueous film

leaf interior

reaction, removal or accumulation A

B air concentration

aqueous film

leaf interior reaction, removal or accumulation C

D

Figure 3.1. Schematic representation of the different types of behavior involved in the uptake (and emission) of pollutant gases through stomata. The shading in the boxes, representing the aqueous film lining the sub-stomatal cavity and the apoplast, shows the extent to which the reservoir approaches equilibrium with the adjoining reservoirs. When the aqueous film is fully at equilibrium (shaded), uptake rates are controlled by internal leaf processes; when not at equilibrium (only partly shaded), uptake rates are controlled by stomatal opening or external gas transfer rates. A, slightly soluble gas that is removed slowly; air–plant equilibrium may permit slow accumulation of metabolic product (e.g., C2Cl4 to TCA). B, slightly soluble gas that is removed at rate similar to uptake rate; air–plant equilibrium not achieved, but substance present in leaf at quasi-equilibrium (e.g., benzene (C6H6)). C, slightly soluble gas that is removed rapidly; air–plant equilibrium not achieved, so gas does not accumulate in leaf, but reaction products may do so (e.g., O3 or NO2). D, slightly soluble gas that is in equilibrium with leaf interior; internal concentration greater than air concentration, leading to emission (e.g., isoprene C5H9). E, soluble gas that is removed rapidly; air–plant equilibrium not achieved, so gas does not accumulate in leaf, but reaction products may do so (e.g., SO2). F, very soluble gas that is removed rapidly; air–plant equilibrium not achieved, so gas does not accumulate in leaf, but reaction products may do so (e.g., HNO3). G, very soluble gas that is in equilibrium with leaf interior; internal concentration less than air concentration, leading to net uptake (e.g., NH3). H, very soluble gas that is in equilibrium with leaf interior; internal concentration greater than air concentration, leading to net emission (e.g., NH3).

65

Plants as Accumulators of Atmospheric Emissions air concentration

aqueous film

leaf interior reaction, removal or accumulation E

F air concentration

aqueous film

leaf interior reaction, removal or accumulation G

H

Figure 3.1. (Continued )

under field conditions (Fig. 3.1, C). Nitrogen dioxide (NO2) behaves similarly, with no evidence of an internal resistance to uptake (Rondon & Granat, 1994), suggesting that transport, or reaction is faster than uptake rates, at least at low air concentrations. Some sparingly soluble gases that have both biogenic and anthropogenic sources are likely to have higher internal concentrations than in the surrounding air, leading to emission rather than uptake under most conditions (Fig. 3.1, D). The situation is more complex for gases that have a higher solubility in water. The solubility of sulfur dioxide (SO2) depends on the pH of the aqueous medium; where it is well buffered, as in the apoplast, uptake from the gas phase can continue, provided there is a sink for the dissolved gas, through oxidation to sulfate ðSO42 Þ or other (bio) chemical reactions

66

J. Neil Cape

(Fig. 3.1, E). The high aqueous solubility of nitric acid (HNO3) and hydrogen chloride (HCl), and lack of a pH dependence on solubility (at least under field conditions) implies an effective sink within the leaf, where even modest reaction or removal rates will permit continued uptake from the atmosphere (Fig. 3.1, F). The same situation also appears to apply to some organic molecules, e.g., formaldehyde (Kondo et al., 1996). For other highly soluble gases this is not the case; for example, ammonia (NH3) concentrations inside the leaf are determined by the equilibrium between the pH, ammonium ion ðNH4þ Þ concentration, and temperature of the sub-stomatal cavity (Sutton et al., 1993). When external air concentrations are high, uptake occurs if the air concentration exceeds the equivalent equilibrium concentration within the leaf (Fig. 3.1, G). When the internal NH3 concentration exceeds that outside the leaf, emission rather than uptake occurs (Fig. 3.1, H). For some gases that are lipid soluble rather than water soluble, uptake may be faster through the exposed leaf surface than through stomata. The relative importance of lipid and aqueous pathways has been modeled for a wide range of compounds (Riederer, 1995). Absolute uptake rates are likely to be much slower than for water-soluble gases because of the slow rates of diffusion through the waxy cuticle. In principle, uptake rates by plants without a cuticle (e.g., bryophytes) could be rapid, except that the outer cell walls may not be very effective sinks for lipid-soluble gases, being composed of celluloses and structural carbohydrates. Examples of lipid-soluble molecules for which surface uptake would be relatively important are benzene (C6H6), perchloroethylene (C2Cl4) or even nitric oxide (NO). These might be expected to come to relatively rapid equilibrium with the surrounding air if removal processes inside the plant are slow. For example, concentrations of C2Cl4 in grass came to equilibrium with air concentrations after a few hours, at values suggesting significant concentration enhancement in the non-aqueous structures of the leaf. Conversely, C6H6 appeared to be broken down or transported out of the leaf faster than the uptake rate could be maintained, because the ‘‘equilibrium’’ leaf concentrations measured were less than that would have been predicted simply on the basis of its solubility in the aqueous phase (Binnie et al., 2002). The fate of the absorbed gas then dictates whether or not it will accumulate and provide a signal that could be used to assess exposure over a period of time. Much depends on whether the gas contains elements that can be used by the plant. Inside the leaf, the molecule or its immediate reaction product(s) are metabolized, transported, or sequestered. In order to be of use as a bioindicator, the original substance or its metabolic product must remain within the leaf. For elements that are key

Plants as Accumulators of Atmospheric Emissions

67

nutrients, such as N or S, incorporation into the plant structure may not lead to an increased concentration, but an increase in growth. If the supply of photosynthate or other nutrient is limited, then enhanced accumulation of ‘‘storage’’ molecules may occur. This can be seen as increases in sulfate content in leaves exposed to SO2 (e.g., Manninen & Huttunen, 2000), increases in amino acid content in response to NH3 (Pitcairn et al., 2003), the accumulation of fluoride (F) particularly in leaf margins (Weinstein & Davison, 2004), or the accumulation of metabolic by-products (e.g., trichloroacetate in response to uptake of C2Cl4; Franzaring et al., 2000).

3.2.2. Particles

Deposition of particles to vegetation is normally thought of as occurring on the external surfaces of vegetation, but some particles may be deposited directly to the sub-stomatal cavity (Fig. 3.2). As for gases, the rate of deposition depends not only on the presence of particles in the atmosphere, but on the degree of turbulence above the surface, which is related to the surface roughness (Gallagher et al., 2002). However, the deposition rate is controlled by the final stage of the

Figure 3.2. Scanning electron micrograph of the surface of a Scots pine needle showing the accumulation of spherical fuel ash particles inside the sub-stomatal cavity and on the leaf surface.

68

J. Neil Cape

deposition process, which relies both on the size of the particle and on the precise nature of the surface. Gases diffuse through the surface boundary layer close to a leaf surface by molecular diffusion. Particles, if small enough, may cross this boundary layer by diffusion, but as particles increase in size up to 0.1 mm, diffusion becomes slower, and deposition rates decrease markedly. The dependence of deposition rate on particle size is relatively simple for short vegetation (Nemitz et al., 2002), but becomes more complex over forests, where deposition rates for submicron particles are much greater than for short vegetation (Gallagher et al., 1997). As particles grow further, beyond about 1 mm in diameter, deposition rates increase because particles possess sufficient momentum to cross the leaf boundary layer and reach the surface. There, they may bounce and be resuspended, or may stick on the surface. The nature of the surface will dictate the outcome, depending on the chemical affinity of the particle for the surface (e.g., for lipophilic particles such as polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), etc.), the physical roughness (e.g., presence of leaf trichomes or epicuticular wax structures; Burkhardt et al., 1995), or whether or not the surface is wet. In general, the behavior of the stomata appears to be of little relevance, although phoretic effects may come into play for smaller particles, leading to enhanced deposition around stomata. There may also be electrostatic effects (induced dipoles) near sharp points and edges that enhance deposition locally on the surface. Local wetness (e.g., around stomata) may also lead to greater retention of particles, particularly if dry particles can deliquesce at the higher humidity close to the leaf surface (Burkhardt et al., 1999). Retention by wet surfaces is likely to be greater than by dry surfaces if impaction is the transport process involved (i.e., for large particles) because of the effects of surface tension, and the possibility of dissipating collisional energy in a liquid film. The lack of biological control (through stomatal opening) of particle deposition has been put to good use for particle capture in the use of moss bags for sampling airborne pollutants, where the large surface area of mosses acts as an efficient scavenger of airborne particulate material that mimics deposition on real vegetation (see review by Zechmeister et al., 2003a and citations therein). Once on the surface, particles are not easily removed, either physically by abrasion or chemically by dissolution in rain or dew. Several studies have demonstrated the poor efficiency of washing with water, even for mineral dusts on a waxy surface (e.g., Fortmann & Johnson, 1984). There is some debate over the efficacy of washing plant material before chemical analysis (see later). The fact that particles accumulate on plant surfaces over time illustrates the poor efficiency of rainfall for removing airborne

Plants as Accumulators of Atmospheric Emissions

69

particles from some surfaces—with some exceptions, where a combination of cuticle structure and chemistry provides very efficient cleansing of particles by rain: the ‘‘lotus effect’’ (Neinhuis & Barthlott, 1997, 1998). Water-soluble particles are gradually removed by wetting, but this process is not necessarily rapid (McCune & Lauver, 1986). Insoluble particles may become embedded in epicuticular waxes, particularly if the wax formation processes continue after leaf expansion, and may undergo chemical changes in situ after deposition (e.g., further oxidation of organic material in particles, or reaction of carbonates to sulfates). As surfaces are repeatedly wetted and dried, there may be release of metal ions that can exchange with cation exchange sites in the cuticle and migrate into the leaf, but this is likely to be a relatively slow process. 3.2.3. Wet deposition

Exchange (uptake or leaching) of solutes between leaves and surfaces wetted by rain, cloud, or dew is controlled by the nature of the plant surfaces. For lower plants, with poorly developed or non-existent waxy cuticles, the process may be relatively rapid, with the cell walls acting as a cation exchange membrane. Exchange of neutral or anionic species is likely also, but probably at slower rates. A waxy cuticle presents a significant barrier to ion exchange processes for the leaves of higher plants, partly because a waxy surface may reduce the residence time of droplets on the surface, and partly because the wax forms a barrier between the aqueous solution and the underlying hydrophilic ion exchange matrices of the cuticle or outer cell walls. However, few leaves even of higher plants have complete waxy coatings. There may be areas (e.g., within the sheath of pine needles; Leyton & Juniper, 1963) where the epicuticular waxes are sparse (Katz et al., 1989), or there may be damage caused by wind abrasion that provides hydrophilic pathways for ion exchange. Examples include uptake of radiolabeled sulfate by spruce needles (Percy & Baker, 1989), and by beech leaves after wind abrasion (Hoad et al., 1992). Moreover, stem and bark surfaces may provide sites for solute exchange, for both cations and anions (Wilson & Tiley, 1998). Evidence of anion uptake by stem surfaces has been shown in studies with trichloroacetic acid (TCA), where significant uptake of TCA occurred through a spruce canopy, despite negligible uptake by needles alone, and stem concentrations of TCA after exposure were considerably greater than for needles (Cape et al., 2003b; Dickey et al., 2004). Leaching of cations from conifer shoots occurs predominantly through stem, rather than needle surfaces (Mitterhuber et al., 1989). The very large surface area of the bark of some plants may lead to the retention of large amounts of

70

J. Neil Cape

material, possibly on ion exchange sites that act as a long-term ‘‘buffer’’ for wet-deposited ions (Hauck et al., 2001). The bark has also been used as a passive sampler for organic pollutants (Douce et al., 1997). Cloud water may deposit large quantities of water-insoluble particulate material to leaf surfaces. Measured concentrations of particles in cloud water have exceeded 100 mg L1 (Crossley et al., 1992; Hadi et al., 1995), leading to rapid accumulation of surface deposits. Wet-deposited material may be retained on the surface as water evaporates, exchanged with ions inside the leaf, or washed off as throughfall. Sparingly soluble material, e.g., large organic molecules such as pesticide residues, may have greater affinity with the leaf surface than with aqueous solution, and may partition onto the leaf, leading to surface accumulation (Riederer, 1995). 3.2.4. Deposition to soil

Apart from direct uptake from the atmosphere, root uptake of material previously deposited on the soil surface can be an important pathway for the accumulation of pollutants. For nitrogen, there appears to be selectivity in the root uptake process, with ammonium favored over nitrate (Buchmann et al., 1995; Lumme, 1994), but for many ions there appears to be little selectivity. Anions such as sulfate and trichloroacetate can be transported from roots to shoots in the transpiration stream (Cape, 1993; Heal et al., 2003; Veltkamp & Wyers, 1997). The same applies for neutral or less polar molecules such as chlorinated hydrocarbons, with the transpiration stream acting as a passive transportation mechanism to the stomata, where pollutants can be released back to the atmosphere by evaporation (Newman et al., 1997). Uptake of material solubilized in soil, for example, some metals, may represent the mobilization of many decades of earlier atmospheric deposition. In this case, the accumulation of material in the aboveground parts of a plant indicates the legacy of past pollution, rather than the present-day situation, and is perhaps more akin to the accumulation seen in plants grown on contaminated soil. Nevertheless, the indirect uptake of air pollutants via the soil and roots represents an important pathway because the material is deposited to the inside of the leaf, rather than the outside, and is more likely to interfere with normal plant function. 3.3. The use of plant accumulation as a bioindicator

As with all types of bioindicators, the accumulation of an air pollutant in plant tissue (usually above ground) can be only recognized when there is a

Plants as Accumulators of Atmospheric Emissions

71

good understanding of what is ‘‘normal’’ in an unpolluted area. The range of normal variation in an unpolluted plant species will depend on many factors, both intrinsic and external. There are likely to be genotypic variations in ‘‘normal’’ concentrations of major nutrients (e.g., N content of different spruce and pine clones; Sheppard & Cannell, 1985), as well as variations caused by differences in climate, soil, and competition from neighboring plants. Bryophytes and lichens have been widely used as accumulating bioindicators because often there is little direct input from the soil or soil water—particularly for bryophytes in ombrotrophic bogs or epiphytic species on tree trunks. For higher plants, uptake by roots includes both soil-derived material and substances that may have come from the atmosphere; separating the two sources may be difficult, particularly if the intention is to study current, rather than historical, pollution levels. However, if plants are to be used as accumulating bioindicators, the practicalities of sampling and analysis will dictate whether or not such an approach will be successful. Decisions have to be made on what to sample, where, when, and how often. The decisions will vary depending on the pollutant(s) under consideration, the plant species available, and the reasons for monitoring pollutant exposure. The ideal pollutant for such a study would be one that had uniform deposition rates over long time scales, entered leaves only through stomata, had no detrimental effect on the plant, was stored in plant tissue irreversibly without being metabolized or causing damage, and was chemically distinct from substances found in healthy plant tissue. The nearest example of this case would be hydrogen fluoride at low concentrations. In reality, however, uptake and elimination rates are not constant in time, and even if the pollutant has no physiological effect on the plant, its rate of accumulation will depend on the growth and development of the plant with time. Consequently, the concentrations measured in plant tissue will not remain constant through the year, but will exhibit temporal fluctuations, depending on the relative rates of growth and accumulation. 3.3.1. What to measure?

Usually, the accumulation of pollutant material is measured by chemical analysis of leaves, or aboveground material. This is partly because this represents the usual uptake pathway of the pollutant, and partly for convenience. However, because different pollutants are deposited in different ways, and may accumulate either inside or outside the leaf, clarity is required as to the nature of the deposition pathway. For example, a simple analysis of the total sulfur content of a leaf may

72

J. Neil Cape

provide little evidence as to whether the accumulation was caused by exposure to SO2, which might be expected to enter the leaf and cause perturbations to the biochemistry leading to the accumulation of SO42 inside the leaf, or to SO42 particles, which would accumulate on the outside of the leaf and have little if any direct effect on the plant. Researchers have recognized the problem of whether or not to wash plant material before chemical analysis, but have not always been aware of effective methods of separating internal and external material (see later). 3.3.2. Where to sample?

A decision on where to sample is linked to subsequent questions of ‘‘when’’ and ‘‘how often,’’ because the type of plant or part of a plant to be sampled for analysis will be determined by the duration of the period of accumulation required. A short-term study might only sample the new growth of a perennial plant, or the whole plant, if it has developed and grown during the sampling period. For longer-lived species, longer accumulation times are possible, and additional information may be gleaned by comparing different ages of plant material, for example, different age classes of conifer needles, to provide an indication of the rate of accumulation with time (Cape et al., 1990; Hellstrom et al., 2004; Strachan et al., 1994). For small plants, it may be impracticable to separate different plant parts before analysis, but for larger plants, the location of sampling (e.g., which whorl of a coniferous tree?) becomes an issue relating to spatial variability, to be discussed later. Furthermore, the location of a plant within its community may have a bearing on the rate of accumulation of an air pollutant—not only in terms of exposure to air pollution (e.g., the edge of a forest stand compared with the center; Beier, 1991; Lindberg & Owens, 1993; Neal et al., 1994; Ould-Dada et al., 2002), but also in terms of relative growth rates, and competition with other individuals or other species for light or nutrients. In general, the effects of such factors on accumulation rates are not well known in advance, so standardized protocols have to be adapted for sampling (Stefan et al., 2000). A thorough discussion of factors influencing the sampling design has been provided by Weiss et al. (2003). 3.3.3. When to sample?

Because rates of pollutant deposition are not uniform throughout the year, but tend to occur in ‘‘episodes’’ or only at certain times of year, the interaction with the growth and development of the plant becomes important in deciding when to sample. Given that both absolute and

73

Plants as Accumulators of Atmospheric Emissions

relative plant growth rates vary during the year, care should be taken to sample at times that minimize year-to-year variation in accumulated concentrations, particularly if the intention is to study long-term trends. The dependence of the accumulated concentration on the balance between uptake rates and plant growth, and the ensuing complexity of the temporal variation in concentration, can be illustrated by a simple model. Even without including the stochastic nature of pollutant exposure, or removal by intermittent processes such as rainfall, significant variation can be observed. Figure 3.3 shows the concentration of pollutant in a hypothetical leaf throughout a growing season. The increase in biomass is

biomass

% biomass

constant uptake, no removal constant uptake, constant removal constant uptake, sine removal sine uptake, no removal sine uptake, constant removal sine uptake, sine removal

time Figure 3.3. Simple model of the variation in the accumulation of a pollutant in a leaf throughout the growing season (the change in biomass shown by the dotted line) under different scenarios: (1) rate of pollutant uptake is proportional to the surface area only (simulating constant exposure to the pollutant), and no removal processes operate; (2) as (1), but the removal rate within the leaf is proportional to the product of leaf biomass (simulating metabolizing or export capacity) and the amount of pollutant present; (3) as (2), but the removal rate is weighted by a sine function that peaks midway through the growth period, simulating varying metabolic or export capacity throughout the leaf ’s life; (4) as (1), but the rate of pollutant uptake is weighted throughout the period by a sine function, simulating varying pollutant exposure; (5) as (4), but with removal rate calculated as (2); (6) as (4), but with removal rate also weighted by a sine function, as (3).

74

J. Neil Cape

assumed to follow a sigmoidal pattern with time (dotted line). Six scenarios are illustrated: (1) Pollutant exposure is constant over the life of the leaf, with uptake at a rate proportional to the leaf area (biomass2/3), and no removal processes operating: constant uptake, no removal. (2) Pollutant uptake is as in (1), but the pollutant is removed at a rate proportional to the biomass of the leaf and the accumulated pollutant concentration, simulating changes in the availability of detoxification processes or transport from the leaf, chosen so that once the leaf is fully expanded, uptake and elimination rates are balanced: constant uptake, constant removal. (3) Pollutant uptake is as in (1), but with the removal rate weighted by a sine function, increasing from 0 initially to 1 at the mid-point of the leaf ’s life and returning to 0 at the end of the period. This is a crude simulation of potential variability in elimination processes during a leaf ’s development, expansion, and senescence: constant uptake, sine removal. (4) Pollutant exposure is weighted using a sine function, to be 0 at t ¼ 0, rising to 1 at the mid-point and falling to 0 at the end of the period. This is a crude simulation of seasonal variation in pollutant concentrations—perhaps relevant to exposure to semi-volatile organic compounds that have higher air concentrations at higher temperatures in summer. No removal processes occur: sine uptake, no removal. (5) Pollutant exposure as in (4), but with constant removal at a rate proportional to leaf biomass and pollutant concentration, as in (2): sine uptake, constant removal. (6) Pollutant exposure as in (4), but with removal rates as in (3): sine uptake, sine removal. The patterns of concentration throughout the growing season fall into two groups: in the first, the lack of a significant removal process leads to a gradual (but not linear) increase in pollutant concentration; in the second, concentrations fluctuate, depending on the balance between uptake and removal, and with little correlation with stage of growth or age of leaf. Recent data on the accumulation of metals in mosses and lichens illustrate the complexity in the real world:  Samples of the moss Scleropodium purum (Hedw.) Limpr. growing in situ were taken every 2 weeks throughout a year, and separated into apical and basal portions (Leblond et al., 2004). Analysis of metals showed different temporal patterns, with larger variation in apical than basal portions. Two distinct patterns were observed: for barium (Ba),

Plants as Accumulators of Atmospheric Emissions

75

aluminum (Al), iron (Fe), vanadium (V), and mercury (Hg), apical concentrations were smaller than in basal (older) tissue, with a minimum in winter and a maximum in summer, differing by a factor of up to 2, in what appeared to be a consistent seasonal pattern rather than a continuing accumulation; for manganese (Mn) and calcium (Ca), concentrations in apical tissue were greater than for basal tissue but showed a similar temporal pattern of a similar magnitude and an overall increase with time, whereas zinc (Zn) concentrations showed a relatively sharp winter peak and no accumulation through the year.  Samples of the moss Abietinella abietina (Hedw.) Fleisch. growing in situ, taken over 4 months, showed a large increase in biomass, which may have been partly responsible for an overall decrease in heavy metal concentrations at the end of the sampling period. However, heavy rainfall after a long dry period may also have contributed—with high exposure rates early in the season (dry weather) and higher removal and biomass growth rates later (Zechmeister et al., 2003a).  Transplants of the lichen Evernia prunastri (L.) Ach. were exposed over a year at three sites in Italy, with sampling on 5 days and analysis for metal content. The temporal trends were similar for all metals at most sites, showing a gradual increase from January to September, and then a decrease in December. There was clear evidence of lead (Pb) emissions at the industrial site, compared with the urban and rural sites, and of enhanced chromium (Cr) at both urban and industrial sites compared with the rural site (Conti et al., 2004).  The cobalt (Co) content of transplants of the lichen Parmelia sulcata Taylor, taken monthly over a year at six sites in Portugal, showed gradual accumulation at most sites, but considerable variability. There was a strong dependence not only on the bulk deposition of Co (measured separately) but also on the timing of wet and dry periods (Freitas & Pacheco, 2004). Average data for each sampling period (Fig. 3.4) illustrate the importance of temporal variation; if samples had been taken only after 4 months, the TAP (Tapada do Outeiro) site would have shown the highest signal; at 6 months, all the sites were similar, and at 12 months, two sites (CAR, Carregado and FAR, Faralhao) were clearly accumulating much more Co than the others.

3.3.4. How often?

The frequency of sampling will be determined by the perceived objective of the monitoring strategy and the availability of resources. The identification of long-term trends over many years might require annual samples,

76

J. Neil Cape 50 CAR FAR PAL SIN TAP

Cobalt content: ng cm-2

40

30

20

10

0 1

2

3

4

5

6 7 months

8

9

10

11

12

Figure 3.4. Illustration of the temporal variation in plant accumulation of a pollutant, and the consequences for the interpretation of sampling at different times of year. Data are plotted from Freitas & Pacheco (2004) and show the accumulation of Cobalt in the lichen Parmelia sulcata at five sites in Portugal over the course of a year. The letter codes indicate the site names.

taken at the same time each year, whereas an attempt to measure uptake rates might require monthly sampling throughout a growing season. The difficulties in determining a consistent timing for annual sampling have been illustrated above, but resources may not permit frequent sampling throughout a year or growing season. As a surrogate for repeated sampling, the use of different ages of material from the same plant (e.g., different year classes of conifer needles) may give an indication of accumulation rates, but care has to be taken in interpreting such data. For example, there may be internal transfer of accumulated material from older to younger tissue, or vice versa. Moreover, the accumulation of high concentrations of a pollutant may cause early senescence and loss of older foliage, so that the older needles remaining on the tree may have pollutant concentrations that are biased toward lower values.

3.3.5. Spatial variability

The issue of spatial variability covers a range of scales, from sampling plot to geographical region, and is central to gaining an understanding of

Plants as Accumulators of Atmospheric Emissions

77

what is a ‘‘normal’’ concentration for a given analyte in a particular plant species. In some cases, the normal concentration is effectively zero, or below the detection limit of the analytical method. However, most pollutants that are of concern in terms of accumulation and possible harm to ecosystems are characterized by persistence and the ability to be transported over long distances in the atmosphere. Consequently, some level of accumulation is to be expected even in areas remote from sources. An example might be the ubiquitous occurrence of measurable amounts of pesticide residues across Europe (Hellstrom et al., 2004). In this case, higher concentrations in some parts of the continent were ascribed to greater pesticide use, but the spatial scale of the distribution pattern precluded detailed identification of sources. Where a potential source has been identified, confirmation can be obtained by using elemental accumulation in plant tissue (Di Guardo et al., 2003). However, it is important for such an application that the degree of variability, and the factors controlling it, are recognized (Kylin et al., 2003). Information is required on small spatial scales, of the order of 0.1–1 km, to investigate the influence on the surrounding environment of identifiable individual sources in a background that contains a mixture of pollutants from many different sources. Examples might be the investigation of the spatial extent of influence of an agricultural point source of ammonia (e.g., Pitcairn et al., 1998) or of a line source such as a major road (e.g., Naszradi et al., 2004). In such instances, comparisons can be made between samples taken along transects away from or across the supposed source or between samples taken upwind and downwind from the source. The assumption is usually made that differences caused by different climatic conditions or soil types are likely to be small compared with the signal caused by the presumed source. Sampling at several points along a linear transect provides confirmation of the source location, if the accumulation of the pollutant decreases monotonically away from the source. However, such an approach requires the presence of the same plant species growing under similar conditions across the whole area under study, unless cultivated indicators or transplants are used (see later). Even on small spatial scales, the variability of the plant material and growing conditions may be sufficiently large that it masks any accumulation of a pollutant. This is particularly relevant where the pollutant is a normal constituent of plant tissue, such as N. Natural plant communities, unlike agricultural crops, are generally very variable at spatial scales of a few meters, so sufficient samples must be taken to ensure that the local variability can be described, or by pooling sample material, reduced. Figure 3.5a shows the natural variability of heather

78

J. Neil Cape 10 sd/mean=0.10 8 6 4 2 0

1

1

a

6

7

1.

1.

1.

1.

1.

5

4

3

2

1.

1.

e

or

M

12 sd/mean=0.13 10 8 6 4 2 0

5

0.

b

55

0.

6

0.

65

0.

7

0.

75

0.

8

0.

85

0.

9

0.

95

0.

e

or

M

20 sd/mean=0.12 16 12 8 4 0 c

6

0.

65

0.

7

0.

75

0.

8

0.

85

0.

9

0.

95

0.

1

05

1.

1

1.

Figure 3.5. Variation in N content of (a) heather (Calluna vulgaris), (b) a lichen (Cladonia portentosa), and (c) a moss, Sphagnum capillifolium at Whim moss, Scotland (L. J. Sheppard, pers. comm.). Vertical axis shows frequency of occurrence. Horizontal axes show N content as % dry weight. Fifty-four samples of each species were taken on the same day over an area of 0.5 ha.

Plants as Accumulators of Atmospheric Emissions

79

(Calluna vulgaris) growing in an undisturbed ombrotrophic bog in southern Scotland (Sheppard et al., 2004), and reflects the difficulty in sampling (from 54 plots) over a range of plant age, in addition to variability caused by differences in surrounding or underlying species. Considerable variation may exist within a given plant species at a given site, but this is likely to be less than the variation between species. Figure 3.5b, c shows the small-scale spatial variability in the lichen Cladonia portentosa, and a moss, Sphagnum capillifolium, at the same site. Some of the variability may also be caused by canopy structure—plants growing in the understory may be exposed to different pollution regimes than those in the overstory. Even botanically similar species may take up and use the same pollutant in different ways, leading to different accumulation rates. 3.3.6. Chemical analysis of sampled material

Cleaning plant material before chemical analysis can be the most costly (in terms of time) operation in the whole sequence from sampling to chemical analysis. Different plant species may need to be separated from each other, different ages of material or live and dead tissue may need to be selected, and adhering soil removed. Once the initial separation process is complete, the next question is ‘‘to wash, or not?’’ Washing or rinsing sampled plant material with water may be an effective method for extracting or removing water-soluble material that is easily accessible, i.e., not embedded in surface waxes or overlain by later hydrophobic deposits. However, even soluble material may take more than a few seconds of wetting to be effectively removed—with the risk that wetting the surface (usually with deionized water) will cause leaching of internal solutes. An alternative technique is to remove the whole surface layer by stripping off the epicuticular wax using brief (few seconds) immersion in a non-polar solvent such as dichloromethane or chloroform. Surface material can then be analyzed directly, after extraction into water, or after acid digestion to liberate metal ions from insoluble salts. The dewaxed leaf can then be analyzed ‘‘complete,’’ or extracted with water or a buffer solution to determine the internal content of water-soluble material. Rinsing with an organic solvent effectively kills the leaf and may disrupt cells, so that the resultant extract of a dewaxed leaf may not accurately reflect the apoplastic composition of the whole leaf. However, there is a clear distinction between external and internal material. Both may provide useful indicators of accumulation from the atmosphere, and may permit greater sensitivity in identifying airborne, as opposed to soil-derived, sources. Weiss et al. (2003) have proposed a method for deciding whether or not to wash sampled material, and whether to use water or an organic solvent.

80

J. Neil Cape

Whether plant material is analyzed before or after washing, the results have to be expressed in terms of the biomass involved. For analyses that have to be made on samples of fresh tissue (e.g., thermally labile or semivolatile substances), results may be expressed relative to the fresh weight of the sample. Care in sampling and storage of semi-volatile analytes is particularly important, to avoid artifacts caused by desorption or translocation of semi-volatile material between the time of sampling and analysis. If the fresh weight is to be used as the basis for reporting data, the potential variability in hydration state at the time of sampling must be accounted for. This may be achieved by sampling before dawn, or by equilibrating material at constant humidity before analysis. Measurements may also be expressed relative to surface area, rather than sample mass, particularly if they relate to accumulation of pollutants on an external leaf surface. Automated procedures exist for measuring the area of broad-leaved plants, but are more time consuming for irregularly shaped samples or conifer needles. It should be made clear whether leaf surface areas refer to one side of the leaf only (projected leaf area) or both sides (total leaf area). More usually, results are expressed relative to the sample dry weight, but even this is not without problems if there are changes, for example, in the amount of non-structural carbohydrates in a leaf throughout a growing season (up to 30% dry weight; Linder, 1995) or throughout a plant (Niinemets, 1997), that influence the leaf area to dry weight ratio. The dynamics of the final reported value for the accumulation of a pollutant in plant tissue may, consequently, depend on the dynamics of the reference frame (dry weight or leaf area) as well as the dynamics of pollutant uptake and storage mechanisms. Some of the problems associated with defining a standard reference point can be overcome if the accumulation of a pollutant is expressed relative to something other than the biomass analyzed. This may be particularly important for elements that play a vital role in normal plant processes, such as N. Accumulation as a result of pollutant uptake may be more clearly apparent by consideration of element ratios (e.g., N:K, N:P, N:S) than by simple measurements of N as a fraction of dry weight. For metals, enrichment factors are often calculated relative to average crustal abundance, to separate anthropogenic inputs from soil-derived material. The use of such ratios may also compensate, to some extent, for the influences of different climatic conditions or growth rates in space and/or time. 3.3.7. The use of transplants and cultivated indicators

There are two basic types of transplant: the native, which involves species growing at the site of interest, and cultivated indicators, which are raised

Plants as Accumulators of Atmospheric Emissions

81

under standard conditions in a standardized soil. The use of native transplants requires appropriate matching of species and site conditions with a site that is known to be free from exposure to the pollutant of interest. Reciprocal transplants, coupled with the use of in situ transplants (to control for the effect of the disruption of transplanting), confirm that changes observed are caused by the pollutant exposure. Reciprocal transplants can also be used to demonstrate the response of the accumulating plants to reductions, as well as increases, in pollutant exposure, and the dynamics of recovery. The following two examples demonstrate their efficacy:  Reciprocal transplants of epiphytic bryophytes between sites with high and low N deposition showed that uptake of N (moving from a low N site to a high N site) was more rapid than removal of N, but that N content did decrease over a period of months when pollutant exposure was reduced (Mitchell et al., 2004).  Reciprocal transplants of a moss between a site polluted by heavy metals and a background site showed increased uptake over 3 years by plants moved to the polluted site, but retention of prior pollution by V and Pb after plants were moved from the polluted to the background site (Tabors et al., 2004). These examples illustrate the very different dynamics between elements that are part of the normal functioning of the plant (N) and those that are retained passively, probably on ion-exchange sites (Pb, V). Standardized transplanting involves the introduction of a test plant into the site under investigation to act as a receptor. Responses of test plants are generally well described, improving the interpretation of the signal being measured. Cultivated indicators are particularly suited to the evaluation of impacts from a known or suspected source. Because they are grown in standard culture media, they also eliminate the contribution from earlier deposition to soils that could be incorporated into native plant species, and so are of benefit for studying current pollution levels at locations with a prior history of pollution impacts. The use of cultivated indicators can potentially provide information on the spatial and temporal scale of air pollution, especially in the absence of local sources and ‘‘control’’ sites. The value of using cultivated indicators for this purpose over using transplants is that their growth is relatively independent of climatic and microclimatic conditions. Either native transplants or cultivated indicators may be used in successive years, to demonstrate trends with time, but the results may be compromised by climatic differences between years. Environmental variables should mimic those from where the transplanted material was

82

J. Neil Cape

grown (particularly important where epiphytes and cryptogams are being used). 3.4. The value of plants as bioindicators

Having established that plants accumulate air pollutants by a variety of different mechanisms, under what conditions does the measurement of the pollutant content of a plant provide useful information? The reasons for measuring the accumulation of air pollutants by plants fall into the following five main categories: (1) Estimating the spatial influence of a known pollutant source. (2) Attempting to derive cause–effect relationships where damage to vegetation is observed. (3) Using regional spatial patterns to infer the relative importance of different sources. (4) Estimating temporal trends, and the efficacy of pollution control measures. (5) Estimating pollutant deposition rates. 3.4.1. Spatial patterns around known sources

There are many examples in the literature, going back to the earliest measurements of foliar chemistry. In many cases, the use of foliar analysis has been related to the second reason listed, an attempt to establish cause–effect relationships. More often, the spatial pattern is described even in the absence of obvious damage from the point source, extending into areas where pollution can be detected, even though it is impossible to ascribe a cause–effect relationship to plant health. Such studies may be localized (Poykio & Torvela, 2001), or regional (Rautio et al., 1998a), depending on the size of the identified source. The spatial limits of the influence of a particular point source to some extent depend on the degree of variability in the ‘‘background’’ signal on which it is superimposed. 3.4.2. Cause–effect relationships

Where visible damage to vegetation occurs, and air pollution is thought to be the cause, the identification of increased pollutant concentrations in the foliage of affected plants, particularly if spatially correlated with the visible injury, can be used as evidence of a causal relationship between

Plants as Accumulators of Atmospheric Emissions

83

pollutant emissions and injury. However, there are problems associated with such an approach. Where plants are showing visible injury from whatever cause, or if growth rates have been suppressed, the concentrations of many elements in foliage may be affected—an increase in N or S concentrations may reflect reduced growth rates from some other cause (e.g., ozone exposure or drought) and be unrelated to the immediate cause of the plant’s response to its environment. In order to minimize such effects, samples for analysis are sometimes taken only from visibly undamaged plants or leaves (Cape et al., 1989), although this practice may bias the results if damaged foliage has been preferentially exposed to pollution. Also, the pattern of element concentration not only has to be associated with the pattern of damage (i.e., increasing as damage increases), but must also be related to the patterns of healthy growth in the surrounding region (i.e., decreasing in healthy plants). The first condition is easily satisfied along a transect away from a point source, for example, but the second condition is seldom tested explicitly—i.e., measuring the pattern of element concentration in healthy vegetation along perpendicular transects, or in the surrounding area. It is possible that gradients in some other factor such as soil condition or climate vary along the transect in a way that contributes to the damage, but which just happens to be correlated with the pattern of element concentrations in the foliage. A simplistic example is shown in Fig. 3.6, where a change in soil property and susceptibility to drought occurs in the same direction as the prevailing wind. A simple transect based on the pollution source would provide data that could be wrongly interpreted as evidence of a causal relationship. Another example might be where the pattern of uptake of one pollutant (not in itself at damaging levels) is affected by the presence of another pollutant from the same, or a different source (Jokinen, 1993; Rautio et al., 1998a). 3.4.3. Regional patterns to identify sources and transport

The use of plant accumulators over large areas permits a straightforward mapping of the distribution of trace elements in vegetation that can be interpreted in terms of air pollution or other sources of variation, such as soil type. Sampling using strict protocols and a restricted range of plant species provides an integrated measure of exposure that cannot be realistically achieved over wide spatial areas using instrumental monitoring. Such surveys can be conducted at a national scale (Dmuchowski & Bytnerowicz, 1995; Mankovska, 1997) or internationally (http://www.icp forests.org/, last accessed on July 15, 2008). The results of such surveys

84

J. Neil Cape

Damage severity

Source

Pollutant concn. Sampling transect

Gradient in soil moisture, and prevailing wind direction

Figure 3.6. Schematic representation of the danger of interpreting the results of a transect study in terms of cause–effect relationships. The variation in visible damage (diamonds) is caused by an underlying change in soil properties that makes plants more sensitive to drought toward the top right of the diagram. This variation is correlated with the dispersion of pollutants from the source as measured by foliar accumulation (circles). Measurements along the transect, in the absence of other information, could be used (wrongly) to attribute the cause of the damage to the presence of the pollutant.

have been used to identify particular ‘‘hot spots’’ that might form the basis for subsequent, more detailed, investigations, but are also used to evaluate the state of pollutant impacts in relation to neighboring regions. The results of such surveys should be interpreted only qualitatively, because the interaction of pollutant exposure with regional-scale variations in climate, soil type, plant species, or genotype, etc. makes it very difficult to associate quantitatively the concentrations measured in foliage with air or precipitation concentrations of the pollutant(s) concerned. 3.4.4. Temporal trends

National and international surveys have been used to evaluate the effects of large-scale emissions reductions, for example, across Europe (http:// www.icp-forests.org/). However, the use of biomonitoring more locally is also important as a means of evaluating the response of the environment to changes in emissions. The effect of emission reductions may not be seen clearly in average concentrations, but may be more obvious in a

Plants as Accumulators of Atmospheric Emissions

85

reduction in variability and in the frequency of occurrence of very large values close to sources (Fu¨rst et al., 2003). Again, there is the temptation to link measured changes in foliar concentrations of trace elements with changes in plant health and the incidence of visible injury, but year-to-year variations in climate may frustrate any attempts to do so reliably. 3.4.5. Uses for estimating rates of deposition

The possibility of establishing a quantitative link between plant concentrations and rates of pollutant deposition is attractive, because of the difficulty and expense of monitoring air and precipitation concentrations, let alone the rates of dry and wet deposition. However, there is still a great degree of uncertainty involved—both in our understanding of the rates and mechanisms of uptake, and in our estimates of wet and dry deposition. This is particularly the case where the same pollutant may be deposited in more than one form (e.g., sulfur as SO2 gas, as SO42 particles or as SO42 in precipitation) or by more than one pathway (e.g., via stomata, leaf surfaces, or roots). Examples of attempts to estimate deposition rates from foliar measurements are given in Sections 3.5.2 and 3.5.3. The lack of quantitative understanding of the processes involved means that empirical relationships have to be derived on the basis of ‘‘calibration’’ at a few locations, or under a restricted set of conditions. Application of such relationships across wide spatial regions, or over time, may be unreliable. Often, the calibration of foliar concentration measurements is against modeled rather than measured inputs, and may be affected by the inherent uncertainty of the model used in estimating the pollutant input at the sampling site. 3.5. Examples of the application of plant accumulation

The following examples are used to illustrate some of the principles described earlier, and are not meant to be a comprehensive review of the literature, which is beyond the scope of this chapter. The examples have been chosen from the more recent literature, although they rely on many decades of the use of plant bioaccumulators as indicators of air pollution and its effects. Most of the examples have been taken from field measurements, rather than from manipulative experiments, because of the difficulties in relating plant uptake of pollutants under field conditions (continuous, with occasional episodes, or seasonal patterns) to experiments where pollutant exposure may be acute for short periods, e.g., in

86

J. Neil Cape

biomass

% biomass

constant uptake, no removal constant uptake, constant removal pulse uptake, no removal pulse uptake, constant removal late pulse, no removal late pulse, constant removal

time Figure 3.7. A variation on the model of Fig. 3.3, demonstrating the effects on plant concentrations of a single large exposure to a pollutant, in this case simulating the experimental treatment of soil as in a fertilizer application. The total ‘‘exposure’’ is matched to the reference case of constant exposure, with a single pulse, decaying exponentially over a period of 30 days (simulating removal/dispersion processes in soil), applied at the start of the growing season or toward the end of leaf expansion. The dashed lines show the effect of including a removal process (similar to case (2) in Fig. 3.3).

fertilization experiments. Such different patterns of pollutant exposure can produce very different patterns of plant accumulation of the pollutant concerned, as illustrated schematically in Fig. 3.7. 3.5.1. Sulfur

The main issue concerning the accumulation of S by plants is the form in which the S is taken up, as the gas (SO2) or as SO42 ions from solution. The latter pathway is most likely through the soil rather than directly through the canopy; the direct uptake of particles into the plant is likely to be a minor pathway—wash-off and subsequent soil uptake is the more likely route. However, insoluble sulfate particles (e.g., gypsum, CaSO42 ) may accumulate on leaf surfaces and be included as ‘‘leaf content’’ if leaves are not washed before chemical analysis (see above). Not all the

Plants as Accumulators of Atmospheric Emissions

87

surface salts, however, are necessarily of external origin: the formation of gypsum crystals on leaf surfaces has been ascribed to the reaction of SO2 or wet-deposited SO42 with Ca2þ ions leached from leaves (Trimbacher & Weiss, 1999; Turunen et al., 1995). The direct uptake of SO2 appears to be more effective in raising leaf concentrations of S than root uptake of SO42 , although examples of the latter route have been reported. At a local scale, increased foliar S concentrations have been observed either as a result of experimental fumigation in the field (Shaw & McLeod, 1995), or in proportion to ambient air concentrations of SO2 (Manninen & Huttunen, 1995), especially where there are episodes of high gas concentration (Manninen & Huttunen, 1997). Different tree species growing at the same location show different rates of accumulation (Manninen & Huttunen, 2000)—any ‘‘calibration’’ of leaf concentrations against air concentrations must be species specific. The absorbed SO2 is mostly stored as SO42 rather than organic S in the leaf, at least in conifers (Kaiser et al., 1993), and the accumulation rate in different age classes of conifer needles usually increases with air concentration. The accumulation of SO42 may also depend on the nitrogen status of the plant (Manninen & Huttunen, 2000), and although a model of the uptake of SO2 by spruce trees was consistent with the S accumulation in foliage (Slovik et al., 1995), other means of disposing of excess S, by emission of hydrogen sulfide (H2S), have been reported (Kindermann et al., 1995). Where soil and climatic gradients are weak, regional surveys may help identify the sources of S accumulated in foliage. Even under more complex conditions, correlation analysis may point to the uptake pathway; in the UK, high S levels in tree foliage were associated more with SO2 concentrations than with wet deposition of SO42 (Innes, 1995). The uptake of SO42 from soil leading to foliar accumulation of S has been demonstrated as a result of soil additions, whether as magnesium sulfate (MgSO4) to beech trees (Ende & Huttl, 1992) or ammonium sulfate ((NH4)2SO4) to Norway spruce (Majdi & Rosengrenbrinck, 1994). However, soil additions of S may not lead to net accumulation in the plant; addition of isotopically labeled 34 SO42 to soil demonstrated the uptake of the labeled S but without any net increase in foliar S content (Giesemann et al., 1995). There can also be marked seasonal variations in S content, which may be a function of differing seasonal patterns of exposure, or of uptake mechanisms (Rautio et al., 1998b). In relation to forest health, the visible damage to trees close to point sources of SO2 has long been recognized, but on more regional scales, it has been shown that although SO2 may be an important factor, other factors can be important in determining the health status of trees exposed to similar concentrations of SO2 (Slovik, 1996).

88

J. Neil Cape

3.5.2. Nitrogen

Large point sources of N-containing pollutants, especially NH3, which is rapidly taken up by plants because of its aqueous solubility, provide very strong signals in terms of the accumulation of total N, water soluble N, NH4þ ions, and amino acids in surrounding vegetation (e.g., Pitcairn et al., 2003). In such cases, the source of the additional N is not in doubt, and N accumulation may be one of several different bioindicators available to map the extent of influence of the source. More generally, transplants and cultivated indicators have been used to assess exposure to N-containing emissions at both local and regional scales. They can provide information relatively quickly (4 weeks) on the spatial scale of eutrophication close to a point source (Leith et al., 2004). The availability of N, for example, is strongly coupled to growth potential so that the product of aboveground dry weight and foliar N concentration may be more meaningful than foliar N concentrations alone, which may be affected by growth dilution. In the cited example, the foliar N content of the cultivated indicators (Lolium perenne) was inversely proportional to the rate of plant growth (i.e., shaded, slower-growing plants had higher leaf %N) so that the total N content of the cultivated indicators was more closely related to exposure (NH3 concentrations) than plant growth rate or leaf %N alone. Where eutrophication arises from diffuse sources at low concentrations or from wet deposition, slower-growing plants such as epiphytic mosses may be used as N accumulators. In such areas, for example, the western parts of the UK, transplants may need to be left in situ for up to 1 year (Mitchell et al., 2004). Neither method can provide subsidiary information on factors, for example, P availability, that may modify the eutrophication impact. Likewise, the impact on the transplant/ cultivated indicator may not be representative of the community it has been imported into, but this may not be important if the accumulation of N is simply used as a bioindicator of exposure. Where a pollutant, such as the different forms of N, is intimately involved in the growth and vitality of plant tissue, seasonal variations may become important in establishing the relationship between exposure and tissue N content. A long-term study of tissue N in moorland heather (C. vulgaris) after 7–8 years of regular, repeated N addition (0–120 kg N ha1yr1) found a statistically significant increase in N content in June, but no treatment effects in September, January, or March (Cawley, 2001). Temporal variation may not just be seasonal—a recent study showed large (up to 30%) inter-annual variation in leaf %N for different plant species growing at the same site and sampled at the same time of year (Emmett et al., 2004).

Plants as Accumulators of Atmospheric Emissions

89

Ericaceous shrubs, including heather, have been widely used as indicators of N deposition in remote areas, and across different spatial scales. One of the largest such studies, from the UK to northern Scandinavia, demonstrated a linear increase in the foliar %N of Calluna and other ericaceous shrubs with the (modeled) wet and dry deposition of N (Pitcairn et al., 2001). However, although such strong links between foliar N and N deposition are attractive as an alternative to deposition monitoring, the relationship between deposition and foliar N is likely to be strongly species dependent, and only applicable where one form of N input (e.g., wet deposition) dominates. Differences in climate may also play a role (Hicks et al., 2000). Close to sources of NH3, uptake of NH3 may dominate and provide a different relationship between N deposition and %N in foliage. In a controlled experiment in open-top chambers, the rate of N accumulation by heather was three times faster in response to NH3 exposure than to wet-deposited NH4þ , for the same rate of N deposition (Leith et al., 2001). Recent studies with pleurocarpous mosses have indicated the potential for using such species as indicators of N deposition on a regional scale (Solga et al., 2005). This innovative approach used 15N abundance as a marker of the relative contribution of NH4þ and NO3 to the N uptake. Foliar N accumulation has also been used to demonstrate the changes over time in response to changes in atmospheric N deposition, based on chemical analysis of herbarium samples of heather (Pitcairn et al., 1995). Similar studies on mosses show the interaction between temporal and spatial trends in N deposition and N content (Baddeley et al., 1994). 3.5.3. Heavy metals

Mosses have been widely used to map heavy metal deposition. Close to primary sources of heavy metals, e.g., smelters, much of the deposition may be in the form of particles, but in areas remote from sources, both wet deposition and particle deposition contribute. On a regional scale, the use of plants as bioaccumulators is not primarily concerned with direct toxicity of metals to plants, because concentrations are generally too small. However, measurements are used to establish both spatial and temporal patterns, and recent European surveys under the aegis of the United Nations Economic Commission for Europe International Cooperative Programme (UNECE-ICP) on effects of Air Pollution on Natural Vegetation and Crops (Buse et al., 2003) using standardized methods and protocols have stimulated research into the factors that determine the observed patterns of metal deposition. This has included a detailed study on the effect of sampling design and its interaction with

90

J. Neil Cape

spatial patterns (Fernandez et al., 2005), and the effects of sampling time and small-scale spatial variability (Zechmeister et al., 2003b). Networks for moss sampling have been established since 1990, with sampling every 5 years, thereby permitting temporal changes as a result of changes in industrial emissions to be recorded (Herpin et al., 2004). However, one of the main aims of heavy metal sampling using mosses is to estimate the deposition of heavy metals on regional scales for comparison with emission estimates. Direct measurement of the wet deposition of heavy metals is possible, but requires rigorous protocols to avoid sampling artifacts and contamination. Air sampling for particles is expensive and requires access to electrical power, so an effective surrogate measurement by using a bioaccumulator has several attractions. In the absence of direct measurements on a sufficiently fine spatial scale to permit comparison with measurements on mosses, predictive methods have been developed based on known (or estimated) emissions and atmospheric transport models. Data are available from some surveys that permit calibration of moss concentrations against measured wet deposition (Berg & Steinnes, 1997), but such relationships may not hold for different climatic zones or moss species. Moreover, the estimation of dry deposition rates of particulate material is complicated by the dependence of dry deposition rates on particle size and on vegetation roughness (see above) and the variations in particle size observed, which are different for different metals. Some metals are found predominantly in the sub-micron size range, others only in the supermicron size range, whereas others have bimodal or more complex distribution patterns (Allen et al., 2001). Combination of available data on particle concentrations, from measurements or emission/transport models, permits calculation of total (wet þ dry) deposition for comparison with measurements (Fowler et al., 2006). Differences between measured concentrations in mosses and predicted deposition rates indicate areas for further measurements, and may identify unrecognized sources of heavy metal pollution.

3.6. Conclusions

The accumulation of airborne pollutants by plants is a consequence of uptake, usually passive, onto leaf surfaces, through stomata, or indirectly through the soil and root systems. Whether or not a pollutant accumulates depends on whether it can be metabolized, transported from the leaf to the root and excreted, used by the plant’s normal biochemical processes, or stored inside the leaf. All the processes involved depend on the

Plants as Accumulators of Atmospheric Emissions

91

physiological state of the plant to some extent, as well as the plant’s architecture. Consequently, accumulation rates vary greatly across plant species, even when exposed to similar pollutant concentrations. However, the use of carefully designed protocols, which minimize the effects of phenology and differential exposure and which avoid potential artifacts caused by the way samples are processed after sampling and before analysis, has provided reliable indications of the relative exposure of vegetation to different pollutants. If one is interested in potential harmful effects of pollutants on vegetation, then bioaccumulation is perhaps a more meaningful indicator of areas under pollution stress than a simple measure of air concentration or wet deposition, because plant accumulation integrates not just exposure, but the rate of uptake, which may depend on other factors such as climate. Measurements of the accumulation of pollutants in plants have provided spatial patterns that identify local and regional sources in terms of a measurable footprint on the environment, whether or not damage to vegetation is immediately obvious. They have also provided independent evidence of long-term temporal changes where air pollution measurement data did not exist. Their strength is in the relative simplicity of the processing required—and consequent cost saving—compared with instrumental methods of air quality monitoring. Although progress has been made in linking concentrations of nutrients such as N and S in plants to exposure or deposition rates, more data are needed before the relative importance of the different forms of pollutant exposure, and the responses of different species, are understood well enough to provide reliable estimates of deposition rates from foliar measurements. However, for substances that are less biologically active and more difficult to measure directly, such as heavy metals or POPs, the use of accumulation in plants will continue to be a cost-effective method for estimating spatial and temporal patterns of exposure, and for the identification of emission sources.

REFERENCES Allen, A.G., Nemitz, E., Shi, J.P., Harrison, J.P., and Greenwood, J.C. 2001. Size distributions of trace metals in atmospheric aerosols in the United Kingdom. Atmos. Environ. 35(27), 4581–4591. Baddeley, J.A., Thompson, D.B.A., and Lee, J.A. 1994. Regional and historical variation in the nitrogen content of Racomitrium lanuginosum in Britain in relation to atmospheric nitrogen deposition. Environ. Pollut. 84(2), 189–196. Beier, C. 1991. Separation of gaseous and particulate dry deposition of sulfur at a forest edge in Denmark. J. Environ. Qual. 20, 460–466.

92

J. Neil Cape

Berg, T., and Steinnes, E. 1997. Use of mosses (Hylocomium splendens and Pleurozium schreberi) as biomonitors of heavy metal deposition: From relative to absolute deposition values. Environ. Pollut. 98(1), 61–71. http://www.sciencedirect.com/science/article/ B6VB5-3SWR618-R/2/fdc3c10191a3bbd9e8350ee9669c0850 (purchase or subscription required). Binnie, J., Cape, J.N., Mackie, N., and Leith, I.D. 2002. Exchange of organic solvents between the atmosphere and grass – the use of open top chambers. Sci. Total Environ. 285(1–3), 53–67. Brown, R.H.A., Cape, J.N., and Farmer, J.G. 1998. Partitioning of chlorinated solvents between pine needles and air. Chemosphere 36(8), 1799–1810. Buchmann, N., Schulze, E.D., and Gebauer, G. 1995. N-15-ammonium and N-15-nitrate uptake of a 15 year old Picea abies plantation. Oecologia 102(3), 361–370. Burkhardt, J., Peters, K., and Crossley, A. 1995. The presence of structural surface waxes on coniferous needles affects the pattern of dry deposition of fine particles. J. Exp. Bot. 46(288), 823–831. Burkhardt, J., Kaiser, H., Goldbach, H., and Kappen, L. 1999. Measurements of electrical leaf surface conductance reveal recondensation of transpired water vapour on leaf surfaces. Plant Cell Environ. 22(2), 189–196. Buse, A., Norris, D., Harmens, H., Buker, P., Ashenden, T.W., and Mills, G. 2003. Heavy metals in European mosses: 2000/2001 survey. Report to UNECE ICP Vegetation, Centre for Ecology and Hydrology. http://icpvegetation.ceh.ac.uk Cape, J.N. 1993. The use of S-35 to study sulphur cycling in forests. Environ. Geochem. Health 15(2), 113–118. Cape, J.N., Paterson, I.S., and Wolfenden, J. 1989. Regional variation in surface properties of Norway spruce and Scots pine needles in relation to forest decline. Environ. Pollut. 58(4), 325–342. Cape, J.N., Freer-Smith, P.H., Paterson, I.S., Parkinson, J.A., and Wolfenden, J. 1990. The nutritional status of Picea abies (L) Karst. across Europe, and implications for forest decline. Trees 4(4), 211–224. Cape, J.N., Fowler, D., and Davison, A. 2003a. Ecological effects of sulfur dioxide, fluorides, and minor air pollutants: Recent trends and research needs. Environ. Int. 29, 201–211. Cape, J.N., Reeves, N.M., Schro¨der, P., and Heal, M.R. 2003b. Long-term exposure of Sitka spruce seedlings to trichloroacetic acid. Environ. Sci. Technol. 37(13), 2953–2957. Cawley, L.E. 2001. Pollutant nitrogen deposition and drought tolerance in heathland plants. PhD dissertation. Manchester Metropolitan University, Manchester, UK. Chameides, W.L. 1989. The chemistry of ozone deposition to plant leaves: Role of ascorbic acid. Environ. Sci. Technol. 23(5), 595–600. Conti, M.E., Tudino, M., Stripeikis, J., and Cecchetti, G. 2004. Heavy metal accumulation in the lichen Evernia prunastri transplanted at urban, rural and industrial sites in central Italy. J. Atmos. Chem. 49(1–3), 83–94. Crossley, A., Wilson, D.B., and Milne, R. 1992. Pollution in the upland environment. Environ. Pollut. 75, 81–87. Dickey, C.A., Heal, K.V., Stidson, R.T., Koren, R., Schroder, P., Cape, J.N., and Heal, M.R. 2004. Trichloroacetic acid cycling in Sitka spruce saplings and effects on sapling health following long term exposure. Environ. Pollut. 130(2), 165–176. http://www.sciencedirect. com/science/article/B6VB5-4BRSFJP-3/2/6f5f 7580ccc40cb04c6fe9738417e522 (purchase or subscription required).

Plants as Accumulators of Atmospheric Emissions

93

Di Guardo, A., Zaccara, S., Cerabolini, B., Acciarri, M., Terzaghi, G., and Calamari, D. 2003. Conifer needles as passive biomonitors of the spatial and temporal distribution of DDT from a point source. Chemosphere 52(5), 789–797. http://www.sciencedirect.com/ science/article/B6V74-48CWWNJ-4/2/841578b2cb80da2811dbda02969e736b (purchase or subscription required). Dmuchowski, W., and Bytnerowicz, A. 1995. Monitoring environmental pollution in Poland by chemical analysis of Scots pine (Pinus sylvestris L) needles. Environ. Pollut. 87(1), 87–104. Douce, D.S., Clench, M.R., Cooke, M., and Wang, J. 1997. Evidence for the adsorption of nitrated polycyclic aromatic hydrocarbons by tree bark. J. Chromatogr. A 786(2), 275–283. Emberson, L.D., Ashmore, M.R., Cambridge, H.M., Simpson, D., and Tuovinen, J.P. 2000. Modelling stomatal ozone flux across Europe. Environ. Pollut. 109(3), 403–413. Emmett, B.A., Jones, M.L.M., Jones, H., Wildig, J., Williams B., Davey, M., Carroll, Z., and Healey, M. 2004. Grazing/nitrogen deposition interactions in upland acid moorland. CEH report to The National Assembly for Wales and the Countryside Council for Wales, Centre for Ecology and Hydrology, 96 pp. Ende, H.P., and Hu¨ttl, R.F. 1992. Insufficient sulfur nutrition of yellowed spruce and beech in the southern Black Forest. Phyton (Horn, Austria) 32(3), 47–50. Fernandez, J.A., Real, C., Couto, J.A., Aboal, J.-R., and Carballeira, A. 2005. The effect of sampling design on extensive bryomonitoring surveys of air pollution. Sci. Total Environ. 337, 11–21. Fortmann, R.C., and Johnson, D.L. 1984. Effect of simulated rainfall on removal and in situ formation of particles on Taxus needles. Environ. Pollut. 7, 297–316. Fowler, D., and Cape, J.N. 1982. Air pollutants in agriculture and horticulture. In: Unsworth, M.H., and Ormrod, D.P., eds. Effects of gaseous air pollution in agriculture and horticulture. Butterworths, Sevenoaks, UK, pp. 3–26. Fowler, D., McDonald, A.G., Crossley, A., Nemitz, E., Leaver, D., Cape, J.N., Smith, R.I., Anderson, D., Rowland, P., Ainsworth, G., Lawlor, A.J., Guyatt, H., and Harmens, H. 2006. UK Heavy Metal Monitoring Network. Report to Defra. 120 pp. http:// www.nmw.ac.uk/cehstaff/Library/cr/C02164/Final_report_2006_finalversion.pdf Franzaring, J., van der Eerden, L.J.M., van Dijk, C.J., Helsper, J.P.F.G., Pikaar, P.J.J., Nijhuis, E.H., and Tonneijck, A.E.G. 2000. Assessing the chronic effects of atmospheric tetrachloroethylene (PER) on plants. Plant Research International, Wageningen, Note 48. Freitas, M.C., and Pacheco, A.M.G. 2004. Bioaccumulation of cobalt in Parmelia sulcata. J. Atmos. Chem. 49(1–3), 67–82. Fu¨rst, A., Smidt, S., and Herman, F. 2003. Monitoring the impact of sulphur with the Austrian Bioindicator Grid. Environ. Pollut. 125(1), 13–19. http://www.sciencedirect.com/ science/article/B6VB5-48SHS7T-3/2/d4d60c514ddceeed881a81a0ad53a617 (purchase or subscription required). Gallagher, M.W., Beswick, K.M., Duyzer, J., Westrate, H., Choularton, T.W., and Hummelshoj, P. 1997. Measurements of aerosol fluxes to Speulder forest using a micrometeorological technique. Atmos. Environ. 31(3), 359–373. Gallagher, M.W., Nemitz, E., Dorsey, J.R., Fowler, D., Sutton, M.A., Flynn, M., and Duyzer, J. 2002. Measurements and parameterizations of small aerosol deposition velocities to grassland, arable crops, and forest: Influence of surface roughness length on deposition. J. Geophys. Res. Atmos. 107(D12), Art. no. 4154, doi:10.1029/2001JD000817. Giesemann, A., Jager, H.J., and Feger, K.H. 1995. Evaluation of sulphur cycling in managed forest stands by means of stable S-isotope analysis. Plant Soil 169, 399–404. Hadi, D.A., Crossley, A., and Cape, J.N. 1995. Particulate and dissolved organic carbon in cloud water in southern Scotland. Environ. Pollut. 88(3), 299–306.

94

J. Neil Cape

Hauck, M., Jung, R., and Runge, M. 2001. Relevance of element content of bark for the distribution of epiphytic lichens in a montane spruce forest affected by forest dieback. Environ. Pollut. 112(2), 221–227. Heal, M.R., Dickey, C.A., Cape, J.N., and Heal, K.V. 2003. The routes and kinetics of trichloroacetic acid uptake and elimination in Sitka spruce (Picea sitchensis) saplings via atmospheric deposition pathways. Atmos. Environ. 37(31), 4447–4452. Hellstrom, A., Kylin, H., Strachan, W.M.J., and Jensen, S. 2004. Distribution of some organochlorine compounds in pine needles from Central and Northern Europe. Environ. Pollut. 128(1–2), 29–48. Herpin, U., Siewers, U., Markert, B., Rosolen, V., Breulmann, G., and Bernoux, M. 2004. Second German heavy-metal survey by means of mosses, and comparison of the first and second approach in Germany and other European countries. Environ. Sci. Pollut. Res. Int. 11, 57–66. Hicks, W.K., Leith, I.D., Woodin, S.J., and Fowler, D. 2000. Can the foliar nitrogen concentration of upland vegetation be used for predicting atmospheric nitrogen deposition? Evidence from field surveys. Environ. Pollut. 107(3), 367–376. Hoad, S.P., Jeffree, C.E., and Grace, J. 1992. Effects of wind and abrasion on cuticular integrity in Fagus sylvatica L. and consequences for transfer of pollutants through leaf surfaces. Agric. Ecosyst. Environ. 42(3–4), 275–289. Innes, J.L. 1995. Influence of air pollution on the foliar nutrition of conifers in Great Britain. Environ. Pollut. 88(2), 183–192. Jeran, Z., Freitas, M.C., and Wolterbeek, H.T. 2004. Assessment of levels and sources of trace element air pollution: The biomonitoring approach. J. Atmos. Chem. 49, 1–3. Jokinen, J. 1993. Estimating of 2 different kinds of forest damage mechanisms in southeastern Finland .1. The Nitrogen–Sulphur hypothesis. Forstwissenschaftliches Centralblatt 112(1–2), 27–35. Kaiser, W., Dittrich, A., and Heber, U. 1993. Sulfate concentrations in Norway spruce needles in relation to atmospheric SO2—A comparison of trees from various forests in Germany with trees fumigated with SO2 in growth chambers. Tree Physiol. 12(1), 1–13. Katz, C., Oren, R., Schulze, E.-D., and Milburn, J.A. 1989. Uptake of water and solutes through twigs of Picea abies (L.) Karst. Trees 3, 33–37. Kindermann, G., Huve, K., Slovik, S., Lux, H., and Rennenberg, H. 1995. Emission of hydrogen sulfide by twigs of conifers—A comparison of Norway spruce (Picea abies (L) Karst.), Scotch pine (Pinus sylvestris L) and blue spruce (Picea pungens Engelm). Plant Soil 169, 421–423. Klumpp, A., Ansel, W., and Klumpp, G. 2004. Urban air pollution, bioindication and environmental awareness. Cuvillier Verlag, Goettingen. Kondo, T., Hasegawa, K., Uchida, R., Onishi, M., Mizukami, A., and Omasa, K. 1996. Absorption of atmospheric formaldehyde by deciduous broad-leaved, evergreen broadleaved, and coniferous tree species. Bull. Chem. Soc. Jpn. 69(12), 3673–3679. Kylin, H., Hellstrom, A., Norstrand, E., and Zaid, A. 2003. Organochlorine pollutants in Scots pine needles—Biological and site related variation within a forest stand. Chemosphere 51(8), 669–675. Leblond, S., Gombert, S., Colin, J.L., Losno, R., and De Traubenberg, C.R. 2004. Biological and temporal variations of trace element concentrations in the moss species Scleropodium purum (Hedw.) Limpr. J. Atmos. Chem. 49(1–3), 95–110. Leith, I.D., Sheppard, L.J., Pitcairn, C.E.R., Cape, J.N., Hill, P.W., Kennedy, V.H., Tang, Y.S., Smith, R.I., and Fowler, D. 2001. Comparison of the effects of wet N deposition

Plants as Accumulators of Atmospheric Emissions

95

(NH4Cl) and dry N deposition (NH3) on UK moorland species. Water Air Soil Pollut. 130, 1043–1048. Leith, I.D., van Dijk, N., Pitcairn, C.E.R., Sheppard, L.J., and Sutton, M.A. 2004. Bioindicator methods for nitrogen based on transplantation: Standardised model plants. In: Sutton, M.A., Pitcairn, C.E.R., and Whitfield, C.P., eds. Bioindicator and biomonitoring methods for assessing the effects of atmospheric nitrogen on statutory nature conservation sites. Joint Nature Conservation Committee, JNCC Report 356, ISSN 0963-8091, pp. 100–104. http://www.jncc.gov.uk/page-3236 Leyton, L., and Juniper, B.E. 1963. Cuticle structure and water relations of pine needles. Nature 196, 770–771. Lindberg, S.E., and Owens, J.G. 1993. Throughfall studies of deposition to forest edges and gaps in montane ecosystems. Biogeochemistry 19(3), 173–194. Linder, S. 1995. Foliar analysis for detecting and correcting nutrient imbalances in Norway spruce. Ecol. Bull. 44, 178–190. Lumme, I. 1994. Nitrogen uptake of Norway spruce (Picea abies Karst.) seedlings from simulated wet deposition. For. Ecol. Manage. 63(2–3), 87–96. Majdi, H., and Rosengrenbrinck, U. 1994. Effects of ammonium sulphate application on the rhizosphere, fine-root and needle chemistry in a Picea abies (L) Karst. stand. Plant Soil 162, 71–80. Mankovska, B. 1997. Variations in sulphur and nitrogen foliar concentration of deciduous and coniferous vegetation in Slovakia. Water Air Soil Pollut. 96(1–4), 329–345. Manninen, S., and Huttunen, S. 1995. Scots pine needles as bioindicators of sulphur deposition. Can. J. For. Res. 25(10), 1559–1569. Manninen, S., and Huttunen, S. 1997. Accumulation of sulphur in and on Scots pine needles in the subarctic. Water Air Soil Pollut. 95(1–4), 147–164. Manninen, S., and Huttunen, S. 2000. Response of needle sulphur and nitrogen concentrations of Scots pine versus Norway spruce to SO2 and NO2. Environ. Pollut. 107(3), 421–436. Mansfield, T. 1998. Stomata and plant water relations: Does air pollution create problems? Environ. Pollut. 101(1), 1–11. http://www.sciencedirect.com/science/article/ B6VB53V1CG7N-1/2/15c8bd84fce1da73b7df0bb25dffb58f (purchase or subscription required) McCune, D., and Lauver, T.L. 1986. Experimental modeling of the interaction of wet and dry deposition on conifers. In: Lee, S.D., Schneider, T., Grant, L.D., and Verkerk, P.J., eds. Aerosols. Lewis Publishing, Chelsea, MI, pp. 1715–1721. Mitchell, R.J., Sutton, M.A., Truscott, A.-M., Leith, I.D., Cape, J.N., Pitcairn, C.E.R., and van Dijk, N. 2004. Growth and tissue nitrogen of epiphytic Atlantic bryophytes: Effects of increased and decreased N deposition. Funct. Ecol. 18, 322–329. Mitterhuber, E., Pfanz, H., and Kaiser, W.M. 1989. Leaching of solutes by the action of acidic rain: A comparison of efflux from twigs and single needles of Picea abies (L. Karst.). Plant Cell Environ. 12, 93–100. Naszradi, T., Badacsonyi, A., Nemeth, N., Tuba, Z., and Batic, F. 2004. Zinc, lead and cadmium content in meadow plants and mosses along the M3 Motorway (Hungary). J. Atmos. Chem. 49(1–3), 593–603. Neal, C., Ryland, G.P., Conway, T., Jeffery, H.A., Neal, M., Robson, A.J., Smith, C.J., Walls, J., and Bhardwaj, C.L. 1994. Interception of chemicals at a forest edge for a rural low-lying site at Black Wood, Hampshire, southern England. Sci. Total Environ. 142(3), 127–141. Neinhuis, C., and Barthlott, W. 1997. Characterization and distribution of water-repellent, self-cleaning plant surfaces. Ann. Bot. 79(6), 667–677.

96

J. Neil Cape

Neinhuis, C., and Barthlott, W. 1998. Seasonal changes of leaf surface contamination in beech, oak, and ginkgo in relation to leaf micromorphology and wettability. New Phytol. 138(1), 91–98. Nemitz, E., Gallagher, M.W., Duyzer, J.H., and Fowler, D. 2002. Micrometeorological measurements of particle deposition velocities to moorland vegetation. Quart. J. Roy. Meteor. Soc. 128(585), 2281–2300. Newman, L.A., Strand, S.E., Choe, N., Duffy, J., Ekuan, G., Ruszaj, M., Shurtleff, B.B., Wilmoth, J., Heilman, P., and Gordon, M.P. 1997. Uptake and biotransformation of trichloroethylene by hybrid poplars. Environ. Sci. Technol. 31(4), 1062–1067. Niinemets, U. 1997. Distribution patterns of foliar carbon and nitrogen as affected by tree dimensions and relative light conditions in the canopy of Picea abies. Trees Struct. Func. 11(3), 144–154. Ould-Dada, Z., Copplestone, D., Toal, M., and Shaw, G. 2002. Effect of forest edges on deposition of radioactive aerosols. Atmos. Environ. 36(36–37), 5595–5606. http:// www.sciencedirect.com/science/article/B6VH3-474936J-5/2/3833fc3dbe28ee4ec5bfddbed 8afa072 Percy, K.E., and Baker, E.A. 1989. Effect of simulated acid rain on foliar uptake of Rbþ and SO42- by two clones of Sitka spruce (Picea sitchensis (Bong.) Carr.). In: Bucher, J.B., and Bucher-Wallin, I., eds. Air pollution and forest decline. EAFV Birmensdorf, Interlaken, Switzerland, pp. 493–495. Proceedings of 14th International Meeting for Specialists in Air Pollution Effects on Forest Ecosystems IUFRO P2.05, October 2–8, 1988. Pitcairn, C.E.R., Fowler, D., and Grace, J. 1995. Deposition of fixed atmospheric nitrogen and foliar nitrogen content of bryophytes and Calluna vulgaris (L) Hull. Environ. Pollut. 88(2), 193–205. Pitcairn, C.E.R., Leith, I.D., Sheppard, L.J., Sutton, M.A., Fowler, D., Munro, R.C., Tang, S., and Wilson, D. 1998. The relationship between nitrogen deposition, species composition and foliar nitrogen concentrations in woodland flora in the vicinity of livestock farms. Environ. Pollut. 102(S1), 41–48. Pitcairn, C.E.R., Leith, I.D., Fowler, D., Hargreaves, K.J., Moghaddam, M., Kennedy, G.H., and Granat, L. 2001. Foliar nitrogen as an indicator of nitrogen deposition and critical loads exceedance on a European scale. Water Air Soil Pollut. 130(1–4), 1037–1042. Pitcairn, C.E.R., Fowler, D., Leith, I.D., Sheppard, L.J., Sutton, M.A., Kennedy, V., and Okello, E. 2003. Bioindicators of enhanced nitrogen deposition. Environ. Pollut. 126(3), 353–361. http://www.sciencedirect.com/science/article/B6VB5-49F84S9-4/2/01bbbb8e8 e7b871e3b56d42aeb6d930f (purchase or subscription required). Plo¨chl, M., Lyons, T., Ollerenshaw, J., and Barnes, J. 2000. Simulating ozone detoxification in the leaf apoplast through the direct reaction with ascorbate. Planta 210(3), 454–467. Poykio, R., and Torvela, H. 2001. Pine needles (Pinus sylvestris) as a bioindicator of sulphur and heavy metal deposition in the area around a pulp and paper mill complex at Kemi, northern Finland. Int. J. Environ. Anal. Chem. 79(2), 143–154. Rautio, P., Huttunen, S., and Lamppu, J. 1998a. Element concentrations in Scots pine needles on radial transects across a subarctic area. Water Air Soil Pollut. 102(3–4), 389–405. Rautio, P., Huttunen, S., and Lamppu, J. 1998b. Seasonal foliar chemistry of northern Scots pines under sulphur and heavy metal pollution. Chemosphere 37(2), 271–287. Riederer, M. 1995. Partitioning and transport of organic chemicals between the atmospheric environment and leaves. In: Trapp, S., and McFarlane, J.C., eds. Plant contamination: Modeling and simulation of organic chemical processes. CRC Press, Boca Raton, pp. 153–190. Rondon, A., and Granat, L. 1994. Studies on the dry deposition of NO2 to coniferous species at low NO2 concentrations. Tellus Series B 46(5), 339–352.

Plants as Accumulators of Atmospheric Emissions

97

Shaw, P.J.A., and McLeod, A.R. 1995. The effects of SO2 and O3 on the foliar nutrition of Scots pine, Norway spruce and Sitka spruce in the Liphook open-air fumigation experiment. Plant Cell Environ. 18(3), 237–245. Sheppard, L.J., and Cannell, M.G.R. 1985. Nutrient use efficiency of clones of Picea sitchensis and Pinus contorta. Silvae Genet. 34(4–5), 126–132. Sheppard, L.J., Crossley, A., Leith, I.D., Hargreaves, K.J., Carfrae, J.A., van Dijk, N., Cape, J.N., Sleep, D., Fowler, D., and Raven, J.A. 2004. An automated wet deposition system to compare the effects of reduced and oxidised N on ombrotrophic bog species: Practical considerations. Water Air Soil Pollut. Focus 4(6), 197–205. Slovik, S. 1996. Early needle senescence and thinning of the crown structure of Picea abies as induced by chronic SO2 pollution – 2. Field data basis, model results and tolerance limits. Glob. Chang. Biol. 2(5), 459–477. Slovik, S., Siegmund, A., Kindermann, G., Riebeling, R., and Balazs, A. 1995. Stomatal SO2 uptake and sulfate accumulation in needles of Norway spruce stands (Picea abies) in Central Europe. Plant Soil 169, 405–419. Solga, A., Burkhardt, J., Zechmeister, H.G., and Frahm, J.-P. 2005. Nitrogen content, 15N natural abundance and biomass of the two pleurocarpous mosses Pleurozium schreberi (Brid.) Mitt. and Scleropodium purum (Hedw.) Limpr. in relation to atmospheric nitrogen deposition. Environ. Pollut. 134(3), 465–473. http://www.sciencedirect.com/ science/article/B6VB5-4DXC2V8-3/2/94e6c3e9199050f48267df149c295810 (purchase or subscription required). Stefan, K., Raitio, H., Bartels, U., and Fo¨rst, A. 2000. Manual on methods and criteria for harmonized sampling, assessment, monitoring and analysis of the effects of air pollution on forests. Part IV: Sampling and analysis of needles and leaves. UNECE ICP Forests, Geneva. Strachan, W.M.J., Eriksson, G., Kylin, H., and Jensen, S. 1994. Organochlorine compounds in pine needles—Methods and trends. Environ. Toxicol. Chem. 13(3), 443–451. Sutton, M.A., Pitcairn, C.E.R., and Fowler, D. 1993. The exchange of ammonia between the atmosphere and plant communities. Adv. Ecol. Res. 24, 301–394. Tabors, G., Brumelis, G., Lapina, L., Pospelova, G., and Nikodemus, O. 2004. Changes in element concentrations in moss segments after cross—Transplanting between a polluted and non-polluted site. J. Atmos. Chem. 49(1–3), 191–197. Trimbacher, C., and Weiss, P. 1999. Needle surface characteristics and element contents of Norway spruce in relation to the distance of emission sources. Environ. Pollut. 105(1), 111–119. Turunen, M., Huttunen, S., Back, J., and Lamppu, J. 1995. Acid-rain-induced changes in cuticles and Ca distribution in Scots pine and Norway spruce seedlings. Can. J. For. Res. 25(8), 1313–1325. United Nations Economic Commission for Europe (UNECE). 2005. http://www.unece.org/ env/lrtap/status/lrtap_s.htm (last accessed on July 15, 2008). Veltkamp, A.C., and Wyers, G.P. 1997. The contribution of root-derived sulphur to sulphate in throughfall in a Douglas fir forest. Atmos. Environ. 31(10), 1385–1391. Weinstein, L.H., and Davison, A. 2004. Fluorides in the environment: Effects on plants and animals. CABI Publishing, Wallingford, UK. Weiss, P., Offenthaler, I., O¨hlinger, R., and Wimmer, J. 2003. Higher plants as accumulative bioindicators. In: Markert, B.A., Breure, A.M., and Zechmeister, H.G., eds. Bioindicators and biomarkers: Principles, concepts, and applications. Elsevier, Amsterdam, pp. 465–500. Wilson, E.J., and Tiley, C. 1998. Foliar uptake of wet-deposited nitrogen by Norway spruce: An experiment using N-15. Atmos. Environ. 32(3), 513–518.

98

J. Neil Cape

Wolterbeek, H.G., Garty, J., Reis, M.A., and Freitas, M.C. 2003. Biomonitors in use: Lichens and metal pollution. In: Markert, B.A., Breure, A.M., and Zechmeister, H.G., eds. Bioindicators and biomarkers: Principles, concepts, and applications. Elsevier, Amsterdam, pp. 377–419. Zechmeister, H.G., Grodzinska, K., and Szarek-Lukaszewska, G. 2003a. Bryophytes. In: Markert, B.A., Breure, A.M., and Zechmeister, H.G., eds. Bioindicators and biomarkers: Principles, concepts, and applications. Elsevier, Amsterdam, pp. 329–375. Zechmeister, H.G., Hohenwallner, D., Riss, A., and Hanus-Illnar, A. 2003b. Variations in heavy metal concentrations in the moss species Abietinella abietina (Hedw.) Fleisch. according to sampling time, within site variability and increase in biomass. Sci. Total Environ. 301(1–3), 55–65. http://www.sciencedirect.com/science/article/B6V78-470V 02T-3/2/d44b4c1229f4a23cd5aee4f6367c6cd9 (purchase or subscription required).

Developments in Environmental Science, Volume 9 Allan H. Legge (Editor) Copyright r 2009 Elsevier Ltd. All rights reserved. ISSN: 1474-8177/DOI:10.1016/S1474-8177(08)00204-0

99

Chapter 4 Relating Source-Specific Atmospheric Sulfur Dioxide Inputs to Ecological Effects Assessment in a Complex Terrain Jose´ Luis Palau, Sagar V. Krupa*, Vicent Calatayud, Maria Sanz and Milla´n Milla´n Abstract The Andorra Power Plant (APP) at Teruel, Aragon, Spain (401 54u 0v N, 11 8u 0uu W, 605 m ASL), with a stack height of 343 m has three units, each with a 350 MW power-generating capacity. It uses 12,000–15,000 tons of low-grade lignite daily, with high sulfur content (5–6% and estimated sulfur dioxide (SO2) emissions of 11.2 g/m3 – S). A SO2 control (scrubber) system was installed in 1999. In an effort to understand the ecological impacts of APP SO2 emissions, the path of plume transport through a highly complex terrain of mountains and valleys was documented, separately during 1995–1999 and 1999–2003, using a mobile unit equipped with a Correlation Spectrometer (COSPEC, for the SO2 levels in the plume aloft) and a rapid response pulse fluorescence detector (for the simultaneous measurements of SO2 concentrations at ground level). In 2003, 13 ecological monitoring plots consisting of Austrian pine (Pinus nigra) were established for long-term effects studies. These plots were located within the most frequent direction of plume transport. To separate the contributions of the atmosphere from those of the soil to P. nigra foliar concentrations of total S and other elements, an Elemental Enrichment Analysis (EEA) method was used. Based on the least amount of variance between several elemental concentrations in P. nigra needles and the corresponding soils in the plots, aluminum was chosen as the normalization element for computing elemental enrichment factors. Those results identifying sites with various levels of impacts were in close agreement with the results of measured plume transport and deposition in the complex Corresponding author: E-mail: [email protected]

100

Jose´ Luis Palau et al.

terrain, before and after the installation of the SO2 control system. Consequently, additional studies are underway to determine the long-term impacts of SO2 exposure on growth and productivity of P. nigra.

4.1. Introduction

Although tropospheric ozone (O3) is recognized as the most important phytotoxic air pollutant globally (Krupa et al., 2001), primary pollutants from stationary point sources continue to be of concern at the local scale regarding their direct impact on vegetation, and at the regional scale, regarding their conversion to nano and fine particles (www.epa.gov/ncea, last accessed on July 17, 2008) and excess elemental (e.g., nitrogen (N)) loadings at the surface, leading to adverse ecosystem effects (van der Hoek et al., 1998). Traditionally, the emphasis of studies on point-source emissions and vegetation effects has been on sulfur dioxide (SO2), hydrogen fluoride (HF), and trace metals (TMs) (Bell & Treshow, 2002; Legge & Krupa, 2002). In general, as specific environmental concerns arose, such studies, by their nature, were performed postmortem in the vicinity of single or isolated point sources (e.g., the West Whitecourt case study; see Krupa & Legge, 1998). Frequently, these investigations relied on spatial mapping of visible foliar injury. However, chronic physiological, growth, and biomass responses can occur without visible foliar injury. To optimize the chances of detection and quantification of such effects, source plume dispersion and meteorological characteristics should, ideally, be coupled to the selection of the proper study sites with suitable ecological characteristics (Legge et al., 1981). In addition to the use of the source plume dispersion patterns, in suitable cases, elemental tracers of the plume and their accumulation within the foliage and/or the soil can be used. Examples include the use of trace metals such as arsenic (As), beryllium (Be), chromium (Cr), cadmium (Cd), scandium (Sc), strontium (Sr), and vanadium (V) (Sloof, 1995a, 1995b) and stable elemental isotope ratios such as 34S:32S (Krouse et al., 1984; Prietzel et al., 2004). In these cases, it is important that the tracer and the phytotoxic pollutant behave similarly (regarding their trend or co-linearity) in their dispersion, deposition, and accumulation in the receptor. Establishing specific cause–effect relationships becomes problematic when clusters of point sources and their emissions impact a particular geographic location, and more than one source emits the same pollutant of interest or concern. In that context, receptor models are

Relating Source-Specific Atmospheric Sulfur Dioxide Inputs

101

applied to apportion the contribution of individual sources to the pollutant load (see Chapter 1 by Hopke in this volume). Application of receptor models requires chemical constituent-speciated profiles of potentially contributing sources and the corresponding ambient data from analyzed particulate matter samples collected at a single receptor site (Reff et al., 2007; also http://www.epa.gov/ttn/chief/ software/speciate/index.html, last accessed on July 17, 2008). Receptor models can also be applied using data from foliar accumulation of various chemical species, in lieu of ambient air measurements. However, this has not yet been accomplished in vegetation effects studies. Such efforts are further compounded by source–impact relationships that are influenced by meteorological interactions in complex terrains (mountains and valleys, or land mass and large water bodies, e.g., the Great Lakes or oceans). A number of primary pollutants, such as SO2, will accumulate in the foliage (as sulfur (S); Legge et al., 1988) (also see Section 4.3 later and Chapter 3 by Cape in this volume). However, S is an essential element and, in general, soil is the major source for its supply to plants. Thus, where the atmosphere also serves as an important source of S, its contribution should be separated from that of the soil in relating adverse plant responses to direct SO2 exposures. Thus, in this chapter, we report the results of a case study of the interrelationships between the plume dispersion of emissions from a large coal-fired power plant through a complex mountainous terrain, SO2 concentrations aloft and at ground level, and the geographic extent of the accumulation of atmospheric S and other elements in pine (Pinus spp.) foliage, in defining impacted areas for future long-term studies.

4.2. Els Ports–Maestrat regional-scale case study of atmospheric SO2 4.2.1. Emission source

The ENDESA Electricity Conglomerate’s Andorra Power Plant (APP) at Teruel, Aragon, Spain (401 54u 0uu N, 11 8u 0uu W, 605 m ASL), with a stack height of 343 m, has three units, each with a 350 MW power-generating capacity (Fig. 4.1). It uses 12,000–15,000 tons of low-grade lignite daily, with high sulfur content (5–6% and estimated SO2 emissions of 11.2 g/m3 – S in 1987) (ENDESA, 1994). At the beginning of the 1980s, forest damage began to be detected in the Els Ports–Baix Maestrat counties in the neighboring province of

102

Jose´ Luis Palau et al.

Figure 4.1. Andorra Power Plant (APP) at Teruel, Aragon, Spain (401 54u 0uu N, 11 8u 0uu W, 605 m asl), with a stack height of 343 m, has three units, each with a 350 MW powergenerating capacity.

Castello´n, and it was attributed exclusively, but without adequate scientific support, to the SO2 emissions from the APP (Table 4.1). Consequently, various organizations independently initiated studies to evaluate the damage and try to establish cause-and-effect relationships between the power plant emissions and the observed damage. Coincident with the start of the APP, some Maestrat forest species, especially the conifers, began to show evidence of a slow degradation, characterized by

Relating Source-Specific Atmospheric Sulfur Dioxide Inputs

103

Table 4.1. Andorra Power Plant (APP) case study: Chronology Year

Feature

1974 1979

Government authorizes the construction of power plant Unit no. 1 started (350 MW). Stack height ¼ 343 m; S content of lignite in the area ¼ 6% Units 2 and 3 started (700 MW) Several forests in the mountain slopes in the region start to decline Problem Assessment Committee is created Installation of pilot coal-washing facilities and electrostatic precipitators (scrubber) begins Legal actions and several scientific studies take place The CEAM air quality, meteorology, and plume dispersion modeling and vegetation effects monitoring studies begin SO2 scrubber (control system) in place Current CEAM vegetation effects studies begin

1980 1980–1981 1982–1984 1985 1991–1994 1994–1995 1999–2000 2003

needle defoliation and discoloration, combined with an excessively high rate of mortality of Pinus nigra (Austrian or European black pine). Initial exploratory studies in the area also detected a number of pathogens and pests. However, none of these studies provided conclusive proof of whether APP emissions were the cause of the observed damage. Consequently, in 1994, a systematic measurement program was started by the Centro de Estudios Ambientales del Mediterra´neo, Valencia (CEAM) that included a new air quality-monitoring network coupled with power plant (APP) plume tracking to determine the causes of the environmental problems. 4.2.2. Study area

The APP is located near the city of Andorra (Teruel), 87 km from the Spanish Mediterranean coast, at the southwestern border of the Ebro River basin. The study area consists of the Mediterranean coast (east of the power plant), the Ebro Valley (north of the power plant, running from NNW to ESE), and the northeastern ridges of the Iberian Range (south and southeast of the power plant). The area includes the semi-arid plain of Calanda (100 km inland from the coast, with a mean altitude of 600 m ASL), some mountain ranges on the northwestern side of the Iberian Range (Mediterranean forest with a mean altitude of 1000–1300 m asl), and the coastal plain of Castello´n (where the vegetation is characterized by irrigated crops). This coastal plain is delimited to the north, 7 km from the coast, by a mountain range (780 m ASL, with a very steep slope toward the coast).

104

Jose´ Luis Palau et al.

4.2.3. Field study

The Els Ports–El Maestrat field study was started in the southwestern border of the Ebro River basin in November 1994 (Fig. 4.2). The objectives were to: (a) monitor the atmospheric SO2 plume dispersion from the power plant, aloft and at ground level over a complex, mountainous terrain, under different meteorological conditions and

41.4 41.2 1,900 1,700 1,500 1,300 1,100 900 700 500 300 100

C.T. Andorra

41

Monagrega

40.8 Villores Morella

40.6

Coratxar

Vallibona Vilafranca

40.4

Benicarló

40.2 Onda GRAU

40

Almassora

39.8 (a)

-1.2

-1 -0.8 -0.6 -0.4 -0.2

0

0.2

0.4

0.6

4570 4560

1,900

4550

1,700

4540

1,500

4530

1,300

4520

1,100 900

4510

700

4500

Colors: Pine species Pinus halepensis Pinus nigra Pinus sylvestris *Pinus nigra + Pinus sylvestris

500

4490

300

4480

100

4470 (b)

690 700 710 720 730 740 750 760 770 780 790 800

Figure 4.2. (a) The northern section of the Valencia region, showing parts of the three provinces of Teruel, Tarragona (with the Ebro delta), and Castello´n. The dashed line indicates the study area, southeast of the Andorra Power Plant (APP or C.T. Andorra). The sites providing meteorological data are shown in red (60 m tall tower at the power plant, Monagrega, Villores, Morella, Grau de Castello´n, and Almassora). Topographic heights in meters and (b) Topographic locations of the Els Ports-Maestrat vegetation monitoring plot network in relation to the APP.

105

Relating Source-Specific Atmospheric Sulfur Dioxide Inputs

(b) quantify the recurrence or frequency of each dispersion scenario. A complementary dataset, independent of the plume monitoring, was generated from a parallel evaluation of the state of the vegetation health in a network of selected sites within the study area. However, only the results from the receptor (Austrian pine foliar tissue) elemental accumulation portion of a subsequent study are presented in this chapter. 4.2.4. Experimental approach

Plume transport aloft and its fumigation at ground level were respectively quantified, the former by systematic remote sensing using a mobile unit equipped with a correlation spectrometer (COSPEC) and the latter by a conventional, pulse fluorescence SO2 analyzer (Fig. 4.3). The plumetracking strategy consisted of making transects as cross-sectional as possible to the mean plume transport direction at different distances from the stack. Measurements were taken throughout the day to record any diurnal changes that might occur in the plume transport direction or in the dispersion conditions. To obtain the dispersion parameters implicitly contained in the experimental data, pseudo-Lagrangian averages were

Ultraviolet Radiation

Ro

ad

SO2 Absortion

Profiles

d

nte me tru cle Ins Vehi

EC zer ly SP CO Ana SO

ted gra n nte tio y I entra l l a r tic Conc evel n Ve L tio SO ound entra Gr onc C SO

Figure 4.3. Upper left: General view of the mobile unit used to track and monitor the Andorra Power Plant (APP) plume and its deposition; Bottom left: Close-up of the equipment used to monitor the plume (correlation spectrometer, COSPEC) and ground-level SO2 concentrations (rapid response pulse florescence analyzer); Right: Measurements were carried out from either side and across the plume. Note the differences in SO2 concentrations within the plume and simultaneously at ground level.

106

Jose´ Luis Palau et al.

computed (Milla´n, 1978). These averages were derived from the coordinates related to the center of each profile; thus, meandering effects were not taken into account. The profiles, averaged over time but not space, showed the relative diffusion of the plume by maintaining its morphologic features (Milla´n et al., 1976). 4.2.5. Statistics of wind fields at the surface

A significant aspect of the annual and seasonal wind roses calculated from the fixed meteorological tower data in the study area was the existence of ‘‘wings’’ in markedly different directions (see, e.g., the roses corresponding to the Corachar station, Figs. 4.4 and 4.5). That characteristic indicated that the dominant winds can be affected by: (a) channeling, if the wings were in opposite directions; (b) other orographic effects, if the wings were in different but not opposite directions; (c) seasonal or monthly cycles, if the wings changed direction during the year; (d) diurnal cycles, if the (monthly) rose was reduced to one wing with the diurnal data and the other wing with the nocturnal data; and (e) combinations of the above. Another important aspect was that the wind roses for each month of the year presented a marked seasonal variation in their dominant directions. In general, this seasonal cycle showed predominantly northwesterly winds during the autumn and winter, with maxima in November and February. The southeasterly component increased from March onward, and was dominant from June to August. 4.2.6. Statistics of atmospheric wind fields aloft

Diurnal advective regimes in the mid-troposphere can be identified by using the plume transport direction aloft. In the present case, the measurements aloft were used to verify the patterns of advection and turbulent diffusion governing atmospheric pollutant dynamics as an initial step in analyzing the cause-and-effect relationships between the emission source and ground-level SO2 concentrations. The measurements aloft, obtained using a vehicle equipped with a COSPEC, represent a clear advantage over the information provided by the fixed ground-level monitoring stations. A statistical analysis of the plume transport directions, coupled with the available surface meteorological information, was used to obtain an average (statistical) representation of the main wind field advection features (Fig. 4.6, Table 4.2). Plume transport aloft was predominantly from the southeast (45–1251 sector, clockwise from the north) during the autumn–spring period. Statistically similar frequencies of transport from the southeast and southwest were observed only during the

107

Relating Source-Specific Atmospheric Sulfur Dioxide Inputs Corachar; Annual N NNW 30 NW 20 WNW W

10

NNE NE ENE

0

WSW

E ESE SE SSE

SW SSW S Corachar; Spring N NNW 30 NW 20 WNW W

Corachar; Summer

NNE NE

10

ENE

0

E

WSW

ESE

SW SSW S

N NNW 20 NW 15 10 WNW 5 W 0

Corachar; Autumn N NNW 30 NW 20 WNW W

SW SSW

NNE NE ENE

0

E ESE

S

E ESE SE

S

SSE

Corachar; Winter

10

WSW

ENE

WSW SW SSW

SE SSE

NNE NE

SE SSE

N NNW 40 30 NW 20 WNW 10 0 W

ENE E ESE

WSW SW SSW

NNE NE

S

SE SSE

Figure 4.4. Frequency of wind direction (wind roses) at Corachar (a site close to Fredes Boixar biological effects monitoring plot. Refer to no. 7 in Table 4.3 and also see Fig. 4.8). The wind roses were computed as annual and seasonal averages for 1995–2002.

108

Jose´ Luis Palau et al. Corachar; Annual N NNW 30 NW 20 WNW 10 W

NNE NE ENE E

0

WSW

ESE

SW SSW

SE S

SSE

Corachar; Spring N NNW 30 NW 20 WNW 10

Corachar; Summer

NNE NE ENE

0

W

E

WSW

ESE

SW SSW

SE S

N NNW 30 NW 20 WNW 10

W

ENE E ESE SE

S

SSE

SE SSE

Corachar; Winter

0

SW SSW

ESE

S

NNE NE

WSW

E

WSW

Corachar; Autumn N 30 NNW NW 20 WNW 10

ENE

0

W

SW SSW

SSE

NNE NE

N NNW 30 NW 20 WNW 10 W

NNE NE ENE

0

E

WSW

ESE SE

SW SSW S

SSE

Figure 4.5. Frequency of wind direction (wind roses) during the day (red) and during the night (black) at Corachar (a site close to Fredes Boixar biological effects monitoring plot. Refer to #7 in Table 4.3 and also see Fig. 4.8). The wind roses were computed as annual and seasonal averages for the period 1995–2002.

109

Relating Source-Specific Atmospheric Sulfur Dioxide Inputs Plume Transport Directions (%) Total (1995 - 03) 45 NW

N NE

30 15

W

0

E

SE

SW S Plume Transport Directions (%) Spring (1995 - 03) 45 NW

Plume Transport Directions (%) Summer (1995 - 03)

N

45 NW

NE

30

NE

30 15

15 W

N

0

E

W

SE

SW

0

E

SE

SW

S

S

Plume Transport Directions (%) Autumn (1995 - 03)

Plume Transport Directions (%) Winter (1995 - 03)

45 NW

N

45 NW

NE

30 15

W

N NE

30 15

0

E

SE

SW S

W

0

E

SE

SW S

Figure 4.6. 1995–2003: Total and cumulative seasonal plume transport by direction (%) from the Andorra Power Plant (APP).

110

Jose´ Luis Palau et al.

Table 4.2. Annual and seasonal frequency statistics of the APP plume transport by direction Frequency of plume transport (%), 1995–2003

Spring Summer Autumn Winter Annual

N

NE

E

SE

S

SW

W

NW

5 10 13 7 9

8 6 17 9 10

24 12 31 27 23

34 23 29 45 31

7 8 3 3 5

3 7 0 0 3

10 19 1 3 8

11 17 6 7 10

30 51 28 21 34

70 49 72 79 66

Note: The last two columns to the right represent the average plume transport frequency statistic within the SE-to-NW and NW-to-SE directional sectors, representing easterlies and westerlies, respectively.

summer. During winter, 79% of the plume transport directions were toward the southeast quadrant (Fig. 4.6 and Table 4.2) because of recurring northwest advective conditions. This seasonal dispersion scenario is a natural consequence of the gradual migration (during late summer and autumn) of northern lows toward midlatitudes and the corresponding retreat of the Azores high-pressure system from the latitudes of the Iberian Peninsula. Under this scenario, the plume aloft is advected by the general wind circulation driven by Atlantic lows flowing northwest to southeast across the Iberian Peninsula, and/or under nocturnal drainage conditions channeled by the Ebro Basin toward the Mediterranean Sea. Only 21% of the measurements tracked the plume aloft as flowing inland (toward the northwest). During summertime, the plume transport roses aloft clearly showed two ‘‘wings’’ (Fig. 4.6), indicating a systematic daily wind-direction cycle. In the spring and summer, when anti-cyclonic conditions predominated over the Iberian Peninsula, with low winds, strong insolation, and frequent formation of the Iberian Thermal Low (ITL), wind circulation being governed by meson-scale processes. During that period, daily breeze cycles, coupled with up-slope winds, carried the air masses over the entire region and frequently influenced the dynamic behavior of the plume aloft. As a result, strong turns (1801) around the stack were documented (Pe´rez-Landa et al., 2002) throughout the summer. During the summer, plume transport toward the southeast occurred mainly during the morning (before approximately 11:00 h UTC), whereas plume transport toward the northwest or inland occurred later in the day. In comparison, plume transport toward the north or the south was mainly associated with the turns, coinciding with the activation of mesoscale circulation in the region. Some 8% of the turns were clockwise

Relating Source-Specific Atmospheric Sulfur Dioxide Inputs

111

(toward the south) and 10% were counter-clockwise (toward the north). These plume turns were very often associated with strong SO2 fumigations (high concentrations) at ground level. 4.2.7. Plume dispersion

When comparing spatial SO2 averages from the power plant for both time periods (four years before (1995–1998) and four years after (1999–2002) the implementation of the new sulfur-removal or scrubber system), the second period showed a better-defined mean plume maximum aloft toward the north when compared with the period before the installation of the control system (January 1995–June 1999). There was no significant change in the maximum values during both periods, although there was a clear decrease in the total area of the mean profile obtained after the installation of the new sulfur-removal system. In contrast, any differences in spatial averages south and west of the power plant were less conclusive as the number of plume profiles measured was significantly lower. There was a clear decrease in ground-level SO2 after June 1999, although maximum values were observed at the same geographic locations as before the installation of the control system. Ground-level SO2 concentrations also decreased within the sectors northwest and south–southeast of the power plant, whereas within the southwest sector, values remained relatively the same before and after the installation of the control system. Another relevant aspect extracted from the analysis of these spatial SO2 averages was the occurrence of vertical, directional wind shear between the mean plume aloft and the mean impacts at ground level. This behavior is typical of elevated plumes (Milla´n et al., 1986, 1987). Under ‘‘classical’’ conditions, vertical, directional wind shear is caused by vertical change in wind direction, which for flat terrains is cyclonic (downwind from the emission source, maximum impacts at ground level will be to the left of the maximum of the plume aloft). But, in the present case, the direction of the shear was anti-cyclonic. This persistent feature was observed in both four-year datasets (before and after the implementation of the new scrubber system) of the spatial averages of SO2. This feature suggests that the channeling of ground-level SO2 is not driven by synoptic effects but by local or regional effects. Thus, (1) before the new sulfur-removal system was installed, more frequent plume advection aloft toward the Els Ports–Maestrat area was detected, within a specific arc in the east–northeast and southeast directions; (2) after the new control system was installed, more frequent plume advection aloft was directed toward the northeast; (3) during both periods, there was a persistent anti-cyclonic vertical wind shear (toward the south);

112

Jose´ Luis Palau et al.

(4) mean ground-level SO2 concentrations measured along a network of roads east of the power plant decreased after June 1999 (after SO2 control), although that effect was not related to the spatial position of the maximum concentration values (which remained essentially the same before and after the installation of the scrubber system); and (5) vertical wind shear between the plume aloft and ground-level SO2 concentrations suggests that atmospheric layers near to the surface are dynamically driven by local and/or regional-scale processes associated with southeastern to southwestern topographical boundaries of the complex-terrain study area. The first two conclusions earlier suggest a significant change in the plume dynamics aloft caused by a change in the mean height of SO2 transport following the installation of the scrubber. In summary, the overall narrative in the previous sections of the case study underscores the difficulty in experimentally characterizing the dynamics of source plume dispersion and deposition in complex terrains and thus, identifying the approach to the spatial distribution and deployment of ecological study sites and relating the corresponding SO2 exposures to vegetation effects.

4.3. Atmosphere as a source of S in plant tissues 4.3.1. Assimilation vs. accumulation of S in plants

As noted previously, S is an essential nutrient for normal plant growth and development. In general, soil is the primary source of that S (as sulfate, SO42 ) for plants. However, the atmosphere can also serve as a source, particularly in S-deficient soils. Under such conditions, exposure to low levels of SO2 and uptake can result in positive effects on plant growth and productivity (Legge & Krupa, 2002). Nevertheless, all plants have a normal range of S requirements and when that is satisfied, largely by uptake from the soil, additional input from the atmosphere will result in adverse effects (Krupa & Legge, 1999, 2001). A number of investigators have used the accumulation of total S in the foliage as an indicator of plant response to atmospheric SO2 exposures (Guderian, 1977; Huttunen et al., 1985; Manninen & Huttunen, 1995). In contrast to the uptake of SO42 from the soil, once SO2 enters the foliage, it is initially converted to HSO3 and SO3 and then to SO42 , the end product representing the accumulated S species. In contrast, as an essential element, the products of metabolic assimilation of S would be found in the organic fraction (cysteine, cystine, methionine, biotin, lipoic acid, thiols, sulfoquinovose, etc.).

Relating Source-Specific Atmospheric Sulfur Dioxide Inputs

113

Legge et al. (1988) concluded that the ratios of foliar SI (inorganic S or the accumulated fraction) to So (organic S or the assimilated fraction) could be used as an indicator in plants under atmospheric S pollution. More recently, Krupa and Legge (1999, 2001) found that ST (total S or assimilated, So þ accumulated, SI) to SI ratio could also be used in understanding foliar injury caused by atmospheric SO2. Most likely, the differences between the specific results of Legge et al. (1988) and Krupa and Legge (1999, 2001) were due to the physiological differences between the plant species studied (a conifer vs. a deciduous shrub), growth conditions, SO2 exposure dynamics, metabolic rates, etc. Nevertheless, overall these approaches to partitioning assimilated vs. accumulated foliar sulfur content provide a very valuable tool for identifying the role of atmospheric S in foliar responses. However, where multiple sources of S impact a common receptor site and if the objective is to apportion the relative contributions of individual sources to the measured foliar S content, then an alternative approach, such as receptor modeling, will have to be used (see Chapter 1 by Hopke in this volume; also Sloof, 1995a, 1995b). 4.3.2. Experimental approach

As opposed to other postmortem studies, Legge et al. (1981) were one of the first with much scientific success, to actually couple a priori the results of a Gaussian plume dispersion model of a single point-source emission to an optimal experimental design in ecological effects studies. Inasmuch as ecological homogeneity and uniform biological comparability are highly desired when examining the spatial effects of an external variable such as the impacts of atmospheric SO2 exposures on vegetation, it is not always possible to implement an ideal experimental design as in the present case involving a highly complex terrain of mountains and valleys and spatial heterogeneity. Fig. 4.2 (B) shows the initial dispersion of vegetation monitoring sites relative to the location of the APP to the northwest. The main tree species in the region are: Aleppo pine (Pinus halepensis), Austrian pine, and Scots pine (Pinus sylvestris). Fig. 4.7 shows the general distribution of conifer species in the study area and the locations of the vegetation-effects study sites. Among those plots, Austrian pine was present at all locations and, therefore, was the species of choice. Furthermore, coincident with the start of the APP, it was also the species that exhibited the highest rates of mortality. The study sites (Fig. 4.8) were chosen based on the known frequency characteristics of plume transport and geographic patterns of deposition from 1995 to 2003 (Fig. 4.8 and Tables 4.2 and 4.3). Additional support

709965

719965

729965

739965

749965

759965

769965

779965

789965

799965

789965

799965

4549972

4549972

4559972

4559972

699965

114

689965

4539972 4529972

4529972

4539972

Power Plant

4519972

4519972

La Cerdler

Lionrovo Linar Pla

4509972

4509972

Mitla Vila

Mitla Vila

Frestes Pionix

Pereroles

4499972

4499972

Herbeset

T. Gros

4489972

Campello

4479972

4479972

Ares Vilafranca Pla Measoro

689965

699965

709965

719965

729965

739965

749965

759965

769965

779965

Figure 4.7. Geographic distribution of major pine species (dark areas in the map) and the specific locations of the study plots (d) (Austrian pine, Pinus nigra) relative to the Andorra Power Plant (APP), located northwest of the study sites.

Jose´ Luis Palau et al.

4489972

Creu Gelat

115

Relating Source-Specific Atmospheric Sulfur Dioxide Inputs

41.158

1,900 1,700 Alcaniz

1,500

Power Plant

40.958

1,300

Calanda

1,100

1 2

40.758

900 4

5 6

8

40.558

700

3 7 Prodes Corachar

500

Forcall Monella10 Vallibona

300

9 100

11 Ares 12 Milafrance 13

San_Mateo

Vinaroz meter a.s.l.

40.358 -0.763

-0.563

-0.363

-0.163

0.037

0.237

0.437

Figure 4.8. A detailed map showing the spatial distribution of the Austrian pine study sites, relative to the location of the Andorra Power Plant (APP). Additional information is provided in Table 4.3.

Table 4.3. Locations of Austrian pine study plots relative to the APP Plot La Cerollera Monroyo Mitja Vila Pinar Pla Pereroles Herbeset Fredes Boixar Creu Gelat Campello T. Gros Ares Vilafanca Pla Masorro Power plant

Altitude (m asl)

Distance to the power plant (km)

927 825 1367 1263 1106 1165 1194 1294 1037 1250 1255 1384 1215

31.276 33.540 47.074 52.653 40.801 45.716 54.001 48.072 54.007 53.790 61.215 62.021 65.968

580

0.000

116

Jose´ Luis Palau et al.

was provided by the computation of the wind rose presented in other figures (Figs. 4.4–4.6). In principle, the final deployment of the effectsmonitoring sites was based on the results of a Lagrangian plume tracing superimposed on a Eularian grid consisting of the occurrence of a common receptor species (Austrian pine) at all sites. 4.3.3. Establishing the spatial relationship between the point-source emissions and total S content in plants

As noted previously, appropriate analysis of the foliar S content can be used as an indicator of atmospheric SO2 exposures. However, in this case, no effort was made to separate the foliar fractions of assimilated vs. accumulated S (Krupa & Legge, 1999; Legge et al., 1988). As the main objective was to identify geographic locations in a very complex terrain where vegetation may be adversely impacted by atmospheric SO2, much reliance was placed on the results of extensive tracking of the APP plume (rather than on dispersion modeling) and ground-level SO2 measurements. Consequently, the effects study sites were not necessarily similar, except for the presence of Austrian pine in the plot. Thus, in order to separate the contribution of atmospheric S from that of soil S, an elemental enrichment method was used. Computation of elemental enrichment: EF ¼ Cx=CnðplantÞ : Cx=CnðsoilÞ

(1)

where, EF is the Enrichment factor; Cx, the Concentration of element ‘‘x’’, whose enrichment is to be determined; and Cn, the Concentration of the normalizing element ‘‘n.’’ The selection of Cn, or the normalizing element ‘‘n’’ in this case, was determined by the extent of variance in the ratios of the concentrations of a number of elements in the plant tissue and their corresponding soil concentrations in the plot. Low variance in plant to soil elemental concentration ratios would indicate soil as the main source of the element in the plant at a given site. In contrast, high ratios would indicate the contribution of an extraneous source (in this case, the atmosphere) to the specific elemental concentration in the plant. Based on such analysis, aluminum (Al) was the element of choice in the present study. The results of the Elemental Enrichment Analysis (EEA) are presented in Table 4.4. The rankings of the individual study sites, based on the total of the elemental enrichment scores, are in close agreement with the results presented in Fig. 9 showing the dispersion and frequency of the APP plume transport (high, medium, or low frequency) and the simultaneous ground-level SO2 concentrations as they relate to the individual effects

117

Relating Source-Specific Atmospheric Sulfur Dioxide Inputs

Table 4.4. Summary rankings of study sites by selected elements using aluminum as the enrichment factor. Plant species: Austrian pine, two-year-old needle samples Element

Rank by site no. S1

S2

S3

S4

S5

S6

S7

S8

S9

S10

S11

S12

S13

Be Cd Mo Na S Si Sr Ti

10 4 13 12 10 11 10 13

4 1 1 6 3 2 2 10

9 7 5 10 5 5 3 2

3 11 11 5 13 12 11 12

1 2 7 1 4 6 7 1

5 3 3 4 1 1 4 3

12 10 9 7 11 9 12 9

13 13 10 11 9 13 5 7

6 6 12 3 7 3 6 4

8 5 2 2 2 4 1 6

2 9 8 8 12 10 13 5

7 8 4 13 8 7 8 11

11 12 6 9 6 8 9 8

Total

83

29

46

78

29

24

79

81

47

30

67

66

69

Note: Site order by rank total: 6 W 2, 5 W 10 W 3 W 9 W 12 W 11 W 13 W 4 W 7 W 8 W 1. The smaller the total, the higher the rank order. (Compare these results with those presented in Fig. 9.)

study sites. Among the 13 sites, the only exception, where the results of EEA (Table 4.4) varied from the plume dispersion results presented in Fig. 9, was in regard to site no. 7, Fredes Boixar (1194 m ASL and 54 km from the power plant; Table 4.3). At the moment, no ready explanation can be found for this observed discrepancy. Nevertheless, the EEA approach can be used as a powerful preliminary step in source apportionment or receptor modeling studies to assess the effects of single- or multiple-source plumes on vegetation.

4.4. Conclusions

The present study represents a unique approach, from an ecological effects assessment viewpoint, to characterizing the real-time dynamics of a point-source plume transport pathway (as opposed to dispersion modeling) over an 8-year period and its relationships to ground-level SO2 concentrations in a highly complex mountainous terrain. The results allowed the selection of appropriate study sites for future assessment of the impacts of the plume on a major conifer species, Austrian pine. EEA, applied here for the first time, allowed soil contributions to the chemical composition of pine needles to be partitioned from atmospheric contributions, indicating the influence of the point-source emissions. Future studies should include: (1) more thorough chemical analysis of Austrian pine needles and the soils in the study plots; (2) use of that

118

N 30 NW

41.158

NE 15

1,900 1,700

0

W

E

Alcañiz

1,500

POWER_PLANT

40.958

1,300

Calanda

SW

1,100

SE 1

2

40.758

900

S

4 5

3 6

700

7Fredes

Corachar

500

Forcall Morella

8

40.558

10 Vallibona 300

9 San_Mateo

Villafranca Ares

Vinaroz

meter a.s.l.

40.358 -0.763

-0.563

-0.363

-0.163

0.037

0.237

> 20 % > 10 % < 10 %

0.437

Figure 4.9. Percent frequency of SO2 exposure on the Austrian pine study plots and Andorra Power Plant (APP) directional plume transport aloft (red line) and at ground level (blue line). Compare results in the figure with those presented in Table 4.4.

Jose´ Luis Palau et al.

12 13

100

11

Relating Source-Specific Atmospheric Sulfur Dioxide Inputs

119

database in receptor modeling, and (3) determination of the relationships between frequency of study-site-specific exposures to the plume; elemental accumulation (particularly S) in the pine needles, and changes in tree growth patterns, taking into account time periods before and after the installation of the SO2 control equipment in 1999.

REFERENCES Bell, J.N.B., and Treshow, M., eds. 2002. Air pollution and plant life, second ed. Wiley, Chichester, UK. ENDESA. 1994. Gestio´n Ambiental de Endesa. In: Martı´ nez, C., Razo´n y controversia sobre la Central Te´rmica Teruel, en Andorra. pp. 167–172. Also published in Tecno Ambiente 22, November 1992. Guderian. 1977. Air pollution. Phytotoxicity of acidic gases and its significance in air pollution control. Ecological Studies (Vol. 22). Springer Verlag, New York, NY. Huttunen, S., Laine, K., and Toruela, H. 1985. Seasonal sulphur contents of pine needles as indices of air pollution. Ann. Bot. Fenn. 22, 343–359. Krouse, H.R., Legge, A.H., and Brown, H.M. 1984. Sulfur gas emissions in the boreal forest: The West Whitecourt case study. V. Stable sulfur isotopes. Water Air Soil Pollut. 23, 61–67. Krupa, S.V., and Legge, A.H. 1998. Sulphur dioxide, particulate sulphur and their impacts on a boreal forest ecosystem. In: Ambasht, R.S., ed. Modern trends in ecology and environment. Backhuys Publishers, Leiden, The Netherlands, pp. 285–306. Krupa, S.V., and Legge, A.H. 1999. Foliar injury symptoms of Saskatoon serviceberry (Amelanchier alnifolia Nutt.) as a biological indicator of ambient sulfur dioxide exposures. Environ. Pollut. 106, 449–454. Krupa, S.V., and Legge, A.H. 2001. Saskatoon serviceberry and ambient sulfur dioxide exposures: Study sites re-visited, 1999. Environ. Pollut. 111, 363–365. Krupa, S., McGrath, M.T., Andersen, C.P., Booker, F.L., Burkey, K.O., Chappelka, A.H., Chevone, B.I., Pell, E.J., and Zilinskas, B.A. 2001. Ambient ozone and plant health. Plant Dis. 85, 4–12. Legge, A.H., and Krupa, S.V. 2002. Effects of sulphur dioxide. In: Bell, N.B., and Treshow, M., eds. Air pollution and plant life, second ed., Wiley, Chichester, UK, pp. 135–162. Legge, A.H., Jaques, D.R., Krouse, H.R., Rhodes, E.C., Schellhase, H.U., Mayo, J., Hartgerink, A.P., Lester, P.F., Amundson, R.G., and Walker, R.B. 1981. Sulphur gas emissions in the boreal forest: The West Whitecourt case study, I. Executive summary. Water, Air, and Soil Pollut. 15, 77–85. Legge, A.H., Bogner, J.C., and Krupa, S.V. 1988. Foliar sulphur species in pine: A new indicator of a forest ecosystem under air pollution stress. Environ. Pollut. 55, 15–17. Manninen, S., and Huttunen, S. 1995. Scots pine needles as bioindicators of sulphur deposition. Can. J. For. Res. 25, 1559–1569. Milla´n, M. 1978. Recent advances in correlation spectroscopy for the remote sensing of SO2. Proceedings of the 4th Joint Conference on Sensing of Environmental Pollutants, American Chemical Society, Washington, DC. Milla´n, M., Gallant, A.J., and Turner, H.E. 1976. The application of correlation spectroscopy to the study of dispersion from tall stacks. Atmos. Environ. 10, 499–511.

120

Jose´ Luis Palau et al.

Milla´n, M., Alonso, L., and Legarreta, J.A. 1986. Dispersio´n de contaminantes en la atmo´sfera: Parte I. Energı´ a, July–August, Madrid, 89–101. Milla´n, M., Navazo, M., and Ezcurra, A. 1987. Meso-meteorological analysis of air pollution cycles in Spain. In: Angeletti, G., and Restelli, G., eds. Physico-chemical behaviour of atmospheric pollutants. D. Reidel Publishing Co, Dordrecht, The Netherlands, pp. 614–626. (For the Commission of the European Communities.) Pe´rez-Landa, G., Palau, J.L., Mantilla, E., and Milla´n, M.M. 2002. A study of the dispersion of an elevated plume on complex terrain under summer conditions. In: 15th Symposium on Boundary Layers and Turbulence, 15–19 July, Wageningen, The Netherlands, 346–349. Prietzel, J., Mayer, B., and Legge, A.H. 2004. Cumulative impact of 40 years of industrial sulfur emissions on a forest soil in west-central Alberta (Canada). Environ. Pollut. 132, 129–144. Reff, A., Eberly, S.I., and Bhave, P.V. 2007. Receptor modeling of ambient particulate matter data using positive matrix factorization: Review of existing models. J. Air Waste Manage. Assoc. 57, 146–154. Sloof, J.E. 1995a. Lichens as quantitative biomonitors for atmospheric trace-element deposition, using transplants. Atmos. Environ. 29, 11–20. Sloof, J.E. 1995b. Pattern recognition in lichens for source apportionment. Atmos. Environ. 29, 333–343. van der Hoek, K.W., Erisman, J.W., Smeulders, S., Wisniewski, J.R., and Wisniewski, J., eds. 1998. Nitrogen, the confer-N-s, First International Nitrogen Conference 1998. Special issue. Environmental Pollution, 102 (Suppl. 1), 1–796.

Developments in Environmental Science, Volume 9 Allan H. Legge (Editor) Copyright r 2009 Elsevier Ltd. All rights reserved. ISSN: 1474-8177/DOI:10.1016/S1474-8177(08)00205-2

121

Chapter 5 Negative vs. Positive Functional Plant Responses to Air Pollution: A Study Establishing Cause–Effect Relationships of SO2 and H2S Luit J. De Kok, Liping Yang, C. Elisabeth E. Stuiver and Ineke Stulen Abstract Throughout the world, natural and agro-ecosystems are at risk from sulfurous air pollutants. However, establishing cause–effect relationships for these air pollutants, and determining acceptable atmospheric concentrations of them, are complicated by their paradoxical effects on plant functioning. Sulfurous air pollutants can act as both toxin and nutrient for plants. However, it is unclear to what extent metabolism contributes to the detoxification of absorbed sulfur gases, as there is no clear-cut transition in the level or rate of metabolism of the absorbed sulfur gases and their phytotoxicity. Moreover, the effects of sulfurous air pollutants on plant functioning are most probably strongly dependent on the sulfur status of the soil.

5.1. Introduction

The atmosphere in rural areas generally contains only trace concentrations of sulfur dioxide (SO2) and hydrogen sulfide (H2S). However, high concentrations may occur in areas with volcanic activity and in the vicinity of industry and/or bio-industry, and may be a matter of environmental concern. In developing countries, rapid economic growth, industrialization, and urbanization are generally associated with a strong increase in energy demand and emissions of various gaseous pollutants; in such locations vegetation is at risk from SO2 and other air pollutants and their negative side effects (Yang et al., 2002, 2006a). For instance, in China in 2002, 22.4% of the cities had an annual average SO2 concentration W0.024 mL L1, Corresponding author: E-mail: [email protected]

122

Luit J. De Kok et al.

and in one-third of the monitored cities, the annual average precipitation pH value was o5.6 (Yang et al., 2006a). In contrast, the atmospheric concentrations of sulfur gases in the USA and Europe have strongly declined over the last two decades as a result of strict regulations on sulfur gas emissions from fossil fuel combustion in order to diminish the negative effects of acid rain deposition. This has resulted in a diminished atmospheric sulfur deposition; in Western Europe, e.g., it has decreased from 70 kg ha1 year1 in the 1970s to o10 kg ha1 year1 currently (McGrath et al., 2002). However, in the direct vicinity of emission sources, sulfur gas concentrations may still exceed threshold concentrations for potential phytotoxic effects via both dry and wet deposition. More than 60 years ago, Thomas and co-workers established that sulfurous air pollutants might be metabolized upon foliar absorption (see Thomas, 1951) and, despite their potential toxicity, could be used as a source of sulfur to promote growth (De Kok, 1990; De Kok & Tausz, 2001; De Kok et al., 1998, 2007). It has become evident that dry and wet deposition of atmospheric sulfur gases substantially contributes to sulfur nutrition in agroecosystems, as modern fertilizers are generally low in sulfur. Throughout the world, the ongoing decline in atmospheric sulfur deposition appears to be a primary cause of sulfur deficiency in crop plants, and supplemental S fertilization is now needed to prevent economic losses (Ceccotti & Messick, 1997; Schnug & Evans, 1992; Thomas et al., 2003; Zhao et al., 1999). The paradoxical effects of atmospheric sulfur gases complicate the establishment of cause–effect relationships for these air pollutants and of what constitutes acceptable atmospheric concentrations of them in natural and agro-ecosystems. For instance, it is still unclear as to what extent directs metabolism (into the assimilatory sulfur pathway) contributes to the detoxification of absorbed sulfur gases. Furthermore, the physiological basis for the wide variation in susceptibility to atmospheric sulfur gases between plant species and cultivars is still largely obscure (De Kok, 1990; De Kok & Tausz, 2001; De Kok et al., 1998, 2007). Based on results of recent studies on the impact of foliar uptake (dry deposition) of SO2 and H2S on plant metabolism, we evaluate the negative vs. positive functional plant responses to air pollution for establishing cause–effect relationships. 5.2. Foliar uptake of sulfurous pollutants

The foliar uptake of atmospheric gases is determined by chemical and physical factors such as adsorption, diffusive path length, solubility, dissociation, and reactivity, and depends on physiological factors such as regulation of stomatal opening and metabolism. The foliar uptake of SO2 and H2S proceeds via the stomates, because the cuticle is impermeable to

A Study Establishing Cause–Effect Relationships of SO2 and H2S

123

Figure 5.1. Foliar gas exchange, where Jgas represents the gas exchange rate, ggas the diffusive conductance of the foliage, representing the result of stomatal and mesophyll conductance to the gas, and Dgas the gas concentration gradient between the atmosphere and leaf interior (Source: Derived from Baldocchi, 1993; De Kok & Tausz, 2001; De Kok et al., 1998).

these gases (Lendzian, 1984). The rate of uptake depends on stomatal and mesophyll conductance toward these gases and their atmospheric concentration, and can be described by Flick’s law for diffusion (Fig. 5.1; Baldocchi, 1993; De Kok & Tausz, 2001; De Kok et al., 1998, 2007). The factors determining the leaf interior (mesophyll) conductance toward SO2 and H2S are quite different. Mesophyll conductance toward SO2 is very high, whereas stomatal conductance is generally the limiting factor for foliar uptake of SO2; this is reflected by a nearly linear relation between uptake and atmospheric SO2 concentration (De Kok & Tausz, 2001; Tausz et al., 1998; Van der Kooij & De Kok, 1998). The high mesophyll conductance toward SO2 is predominantly determined by chemical/physical factors, as this gas is highly soluble in the aqueous phase of the mesophyll cells (in either the apoplast or cytoplasm). Furthermore, it is rapidly hydrated and dissociated, yielding bisulfite (HSO3 ) and sulfite (SO32 ), which either enter the assimilatory sulfur reduction pathway or are enzymatically or non-enzymatically oxidized to sulfate (Fig. 5.2; De Kok, 1990; De

124

Luit J. De Kok et al.

H2S

ATMOSPHERE

SO2

SHOOT GLUTATHIONE ADP+Pi glycine+ATP

ADP+Pi glutamate+ATP CYSTEINE

Acetyl CoA CoA

acetate

Serine serine acetyl transferase

METHIONINE

PROTEINS

O-acetylserine(thiol)lyase

O-acetylserine

SULFIDE 6Fdox sulfite reductase

6Fdred SQDG

SULFITE OXIDATION

AMP+GSSG APS reductase 2GSH APS PPi ATP sulfurylase

ATP vacuole

SULFATE SHOOT

SULFATE

ROOT

PEDOSPHERE SULFATE

Figure 5.2. Metabolism of sulfate, SO2, and H2S in plants. APS, adenosine 5u-phosphosulfate; Fdred, Fdox, reduced and oxidized ferredoxin; RSH, RSSR, thiol compound (reduced and oxidized, presumably glutathione). (Source: Adapted from De Kok et al., 2002b)

Kok & Tausz, 2001; De Kok et al., 2007). The liberation of free Hþ ions upon hydration of SO2 and/or the formed sulfate after its oxidation may be the basis of a possible acidification of the aqueous phase of the mesophyll cells, in case the buffering capacity is not sufficient. Generally, SO2

A Study Establishing Cause–Effect Relationships of SO2 and H2S

125

exposure results in an enhanced sulfur content of the foliage, mainly due to an accumulation of sulfate in the vacuole, even at relatively low atmospheric concentrations (De Kok, 1990; De Kok & Tausz, 2001; De Kok et al., 2007; Yang et al., 2006a). In contrast to SO2, mesophyll conductance toward H2S is predominantly determined by physiological factors, viz., the rate of its metabolism, and foliar uptake of H2S shows saturation kinetics with respect to atmospheric concentration, which can be described by the Michaelis– Menten equation (De Kok & Tausz, 2001; De Kok et al., 1998, 2002a, 2007; Stuiver & De Kok, 2001; Tausz et al., 1998; Van der Kooij & De Kok, 1998). At the apoplastic pH of the mesophyll cells, which varies from 5 to 6.4, the absorbed H2S is largely undissociated, and can easily pass through the membrane, where it is subsequently metabolized with high affinity into cysteine and subsequently into other sulfur metabolites. The H2S uptake is the result of O-acetylserine (thiol) lyase activity, the affinity of the enzyme for sulfide, and the in situ availability of O-acetylserine (Fig. 5.2; De Kok & Tausz, 2001; De Kok et al., 1998, 2002a, 2007). 5.3. Phytotoxicity of sulfurous air pollutants

In the USA, there are both short- and long-term primary National Ambient Air Quality Standards (NAAQS) for SO2 (http://www.epa.gov/air/ criteria.html). The short-term (24 h) standard of 0.14 mL L1 (365 mg m3) is not to be exceeded more than once a year. The long-term standard specifies an annual arithmetic mean not to exceed 0.03 mL L1 (80 mg m3). The secondary NAAQS (3 h) of 0.50 mL L1 (1300 mg m3) is not to be exceeded more than once a year. The standards for SO2 have undergone periodic review, but the science has not warranted a change since they were established in 1972. In Europe (EC), an annual mean of 0.008 mL L1 (20 mg m3) has been set for SO2 as an air quality standard for ecosystems (http://europa.eu.int/comm/environment/air/, last accessed on August 21, 2008). These air quality standards are close to the minimal effective concentration range of sulfurous air pollutants (Table 5.1). Table 5.1. Minimum effective concentration range of sulfurous air pollutants for plant injury Chronic injury 3 acute injury concentration [mL L1] SO2 H2S

0.01 3 0.03 0.03 3 0.3

Source: Derived from De Kok and Tausz (2001) and Posthumus (1998).

126

Luit J. De Kok et al.

To actually establish the phytotoxicity of atmospheric sulfur gases, one needs to make a distinction between the effects of wet and dry deposition. Wet deposition of SO2, together with that of other air pollutants (viz. NOx, NH3), contributes to soil acidification, which is harmful to natural ecosystems in general. In addition, when absorbed by foliage (dry deposition), sulfur gases may directly affect plant functioning and result in a disturbed metabolism, and reduced growth and fitness. Since the 1950s, the impact of air pollutants, including SO2, on plant metabolism has been studied in detail; however, the basis for differences in susceptibility of plant species and cultivars is still largely obscure. From several studies, it is unclear to what extent the observed changes in metabolism upon exposure are the primary cause or consequence of an already disturbed physiology or acute (pre-injury) effects. The use of molecular techniques has not as yet contributed to a better understanding of the underlying physiological mechanisms determining differences in susceptibility. The comment of Mudd (1998) that ‘‘as far as practical considerations and public policy decisions about pollutant control are concerned, research on ozone is essentially complete. The mechanism of toxicity of ozone to plants is an unsolved problem, and now remains as a fundamental problem of stress physiology’’ is not only valid for ozone, but also for other air pollutants, including SO2 and H2S. Undoubtedly, the physical/biochemical background of SO2 phytotoxicity can be ascribed to the negative consequences of acidification of tissue and cells upon dissociation of foliar-absorbed SO2 and/or the direct reaction of the formed sulfite with cellular constituents and metabolites. Likewise, the H2S phytotoxicity also may be ascribed to a reaction of sulfide with cellular components, e.g., metallo-enzymes appear to be particularly susceptible to sulfide, in a reaction similar to that of cyanide (De Kok, 1990; De Kok et al., 1998, 2002a, 2007).

5.4. Metabolism of sulfurous air pollutants

Sulfur is an essential element for plant growth (De Kok et al., 2002b). It has been recognized for several decades that foliar-absorbed sulfurous air pollutants may be metabolized and used as a plant nutrient (see De Kok, 1990; De Kok et al., 2007). The foliar-absorbed SO2 may enter the sulfur reduction pathway as either sulfite or, after oxidation, sulfate, as the shoot is the predominant site of sulfate reduction (Fig. 5.2). Excessive absorbed SO2 is presumably transferred into the vacuole as sulfate, where it is only slowly accessible for metabolism (Clarkson et al., 1993; Cram, 1990). Foliar uptake of H2S appears to be directly dependent on the rate of its

127

2

c b b

1

a

0 30 c

b

20

b a

10

0

+S +S+SO2

-S

-S+SO2

Root fresh weight (g)

3

S/R ratio (FW basis)

RGR of plant (% day-1)

Shoot fresh weight (g)

A Study Establishing Cause–Effect Relationships of SO2 and H2S 0.3

c c b

0.2

a 0.1

0.0 15 10

c

c b a

5

0

+S +S+SO2

-S

-S+SO2

Figure 5.3. The effect of SO2 exposure on growth of Chinese cabbage (Brassica pekinensis, cv. Beijing 3). Ten-day-old seedlings were grown on a 25% Hoagland nutrient solution with or without sulfate and simultaneously exposed to 0.06 mL L1 SO2 for 2 weeks. For experimental conditions see Buchner et al. (2004). Data on fresh weight represent the mean of two experiments with 10 measurements in each (7SD). Relative growth rate (RGR) of plants was calculated on a fresh weight basis and was determined over the exposure period. S/R ratio, shoot to root ratio. Different letters indicate significant differences between different treatments at pp0.01.

metabolism into cysteine, and subsequently, into other sulfur compounds, a reaction catalyzed by O-acetylserine (thiol) lyase (De Kok et al., 1998, 2002a, 2007; Fig. 5.2). Plants are able to switch from sulfate to foliar-absorbed SO2 or H2S as a sulfur source (De Kok, 1990; De Kok et al., 1998, 2002a, 2002b, 2007; Yang et al., 2003, 2006a, 2006b), and concentrations of X0.06 mL L1 appear to be sufficient (or nearly sufficient) to meet the sulfur requirements of, e.g., Brassica species for growth (Buchner et al., 2004; Yang et al., 2003, 2006a; Fig. 5.3). Sulfate deprivation in the root environment generally results in a shift in shoot-to-root biomass partitioning in favor of the root, resulting in a decreased shoot:root ratio (Buchner et al., 2004; De Kok et al., 1997, 2000; Stuiver et al., 1997; Yang et al., 2006a, 2006b; Fig. 5.3). However, in Brassica, the decreased shoot:root ratio caused by sulfate deprivation is generally not rapidly alleviated when the pedospheric sulfate is replaced by atmospheric SO2 or H2S as the sulfur source for growth. Root growth is even stimulated upon exposure to atmospheric sulfur gases, when compared with that of plants grown under sulfate-sufficient conditions (Buchner et al., 2004;

128

Luit J. De Kok et al.

Yang et al., 2006a, 2006b; Fig. 5.3). If these laboratory experiments could be extrapolated to outside field conditions, foliar sulfur absorption by plants growing on sulfur-deficient soils might cause a relatively greater investment in root development, especially when the plants are grown under stress conditions such as limited available nutrient resources and/or drought. The latter may have practical significance, especially in agro-ecosystems, where in recent years sulfur deficiency in soils has become a major problem in various parts of the world because of the imbalance of S in relation to nitrogen (N), phosphorous (P), and potassium (K) in fertilizers (Haneklaus et al., 2003; Yang et al., 2006a). Exposure of plants to atmospheric sulfur gases may increase total sulfur content of the shoot, which when exposed to SO2 is predominantly due to an accumulation of sulfate (De Kok, 1990; Van der Kooij & De Kok, 2000; Van der Kooij et al., 1997; Fig. 5.4). Exposure to H2S may also increase total sulfur content due to an accumulation of organic sulfur compounds, e.g., in spruce (Picea spp.; Tausz et al., 2003) or Allium spp., in the latter presumably due to an accumulation of secondary sulfur compounds (g-glutamyl peptides and alliins; Durenkamp & De Kok, 2002, 2004). Exposure to SO2 and H2S results in enhanced thiol levels in the shoot, which can be observed within a few hours after the start of exposure (De Kok, 1990; De Kok & Tausz, 2001; De Kok et al., 1998, 2007; Tausz et al., 1998; Van der Kooij et al., 1997; Fig. 5.4). Thiol accumulation depends on the atmospheric concentration, although it is generally greater after exposure to H2S than after exposure to SO2 at equal concentrations. The accumulation is generally not solely due to an enhanced level of glutathione, the predominant water-soluble non-protein thiol compound present in plant tissue, but also to strongly enhanced levels of its precursor cysteine (and even g-glutamyl cysteine under dark conditions; De Kok, 1990; De Kok & Stulen, 1993; De Kok & Tausz, 2001; De Kok et al., 1998). Apparently, if the regulation of sulfate uptake by the root and its assimilation in the shoot is bypassed and the shoot directly absorbs sulfur, then there is no strict regulation in the size and composition of the pool of thiol compounds (De Kok & Tausz, 2001). There is evidence for interaction between atmospheric and pedospheric sulfur utilization, because exposure to sulfurous air pollutants may repress the uptake of sulfate by the roots and its transport to the shoot (De Kok, 1990; De Kok et al., 1998, 2007; Herschbach et al., 1995; Tausz et al., 2003; Westerman et al., 2000). Exposure to SO2 or H2S may affect the sulfate reduction pathway by decreasing the activity of ATP-sulfurylase and adenosine 5u-phosphosulfate (APS) reductase (De Kok, 1990; Durenkamp et al., 2007; Westerman et al., 2001), with sulfide, O-acetylserine, cysteine, and/or glutathione being possible regulators (Brunold, 1990; Durenkamp

129

Thiols (µmol g-1FW)

Org-S (µmol g-1DW)

Sulfate (µmol g-1DW)

Total S (µmol g-1DW)

A Study Establishing Cause–Effect Relationships of SO2 and H2S 500 400

c

Shoot

Root b

b

b

300 200 100

a

a

a

a

a

a

a

a

0 500 c

400 300

b b

200

b

100 a

0 150

a b

100

b

a a a

50

a

0 0.6 b

0.4

c

b

d

a

b

0.2

a

a

0.0

+S

+S+SO2

-S -S+SO2

+S +S+SO2

-S -S+SO2

Figure 5.4. The effect of SO2 exposure on sulfur metabolites. Ten-day-old seedlings of Chinese cabbage (Brassica pekinensis, cv. Beijing 3) were grown on a 25% Hoagland nutrient solution with or without sulfate and simultaneously exposed to 0.06 mL L1 SO2 for 2 weeks. Data on total S and sulfate content (measured in freeze-dried material) represent the mean of five measurements with 9–12 plants in each (7SD). For analytical methods see Buchner et al. (2004). Data on total water-soluble non-protein thiol content (mmol g1 FW) represent the mean of three measurements with three plants in each (7SD). Different letters indicate significant differences between different treatments at pp0.01.

130

Luit J. De Kok et al.

Figure 5.5. The effect of H2S exposure on the growth of curly kale (Brassica oleracea, cv. Bornick F1). Eight-day-old seedlings were grown on a 25% Hoagland nutrient solution for 14 days and subsequently transferred to a fresh nutrient solution and exposed to various levels of H2S for 11 days (derived from De Kok et al., 2000). For experimental conditions and analytical methods see Buchner et al. (2004). Relative growth rate (RGR) was calculated on a fresh weight basis and was determined over the exposure period. Total sulfur content represents the mean of three measurements with three plants in each (7SD).

et al., 2007). Especially in curly kale (Brassica oleracea L.), there was a strong interaction between atmospheric H2S and pedospheric sulfate utilization. In this species, atmospheric H2S barely affected total sulfur content, not even at concentrations exceeding the plant’s sulfur requirement (X0.1 mL L1 H2S; Fig. 5.5). Exposure to H2S resulted in a decreased activity and expression of APS reductase, and a depressed sulfate uptake in Brassica oleracea (De Kok et al., 2002b; Durenkamp et al., 2007; Westerman et al., 2000, 2001) and in Brassica pekinensis (Koralewska et al., 2008). The shoot-to-root signal transduction pathway and the signal compounds involved therein are as yet unresolved (Buchner et al., 2004). Furthermore and perhaps even more confusing, the de-repressed sulfate uptake capacity and expression of genes encoding the different sulfate

A Study Establishing Cause–Effect Relationships of SO2 and H2S

131

transporters, which are characteristic of plants transferred to sulfatedeprived conditions, are not rapidly alleviated in Brassica when exposed to H2S (Buchner et al., 2004; Koralewska et al., 2008) or SO2 (Yang et al., 2006b). There is apparently (at least in Brassica) no strict, direct shoot-toroot signaling (e.g., via reduced S compounds) in the tuning of the factors involved in the regulation of the sulfate uptake capacity, expression of the sulfate transporters, or root development in sulfate-deprived roots (Buchner et al., 2004; Koralewska et al., 2008). 5.5. Sulfurous air pollutants: Toxicity vs. metabolism

The impact of sulfur gases on growth in relation to the sulfur status of plants is ambiguous. For instance, in some species, exposure to low concentrations of H2S (0.03–0.1 mL L1) results in a slightly enhanced biomass production, even under sulfur-sufficient soil conditions (De Kok, 1990). Other studies show that SO2 may be less toxic when the sulfur status of the plant is low, which supports the idea that metabolism of the absorbed sulfur might have significance in the detoxification of SO2 (De Kok, 1990). Indeed, laboratory studies support the idea that the effect of sulfurous air pollutants on plant growth strongly depends on the pedospheric sulfur supply. For example, under sulfate-sufficient conditions, growth of Chinese cabbage (Brassica pekinensis) was reduced after a 2 weeks’ exposure to concentrations as low as 0.06 mL L1 SO2 (Fig. 5.3). Sulfate deprivation not only resulted in a strong decrease in sulfate and thiol content of both shoots and roots, but also in strongly reduced plant growth, whereas organic sulfur content was hardly affected at this stage (Figs. 5.3 and 5.4). However, when sulfate-deprived plants were simultaneously exposed to 0.06 mL L1 SO2, growth was largely maintained (Fig. 5.3). On a plant basis, growth was still less than that of sulfate-sufficient plants and comparable to that of sulfate-sufficient SO2exposed plants. Still, it remains unclear to what extent these effects can be explained by an interfering toxicity of SO2 or a limitation in sulfur supply. The foliar uptake of H2S is largely determined by its rate of metabolism (De Kok et al., 1998, 2002a, 2007) and one might assume that variation in susceptibility between species might be related to differences in rates of metabolism of the absorbed H2S. However, in the species studied so far, there is no clear relationship between the rate of H2S uptake, its metabolism, and the growth response of plants (De Kok et al., 1998, 2002a). Hydrogen sulfide is a very reactive compound, and the in situ sulfide concentration and the degree of penetration of the absorbed sulfide in the meristems (e.g., vegetation point) are likely important factors in determining the phytotoxicity of H2S (De Kok et al. 2002a; Stulen et al., 2000).

132

Luit J. De Kok et al.

It has been suggested that enhanced intracellular cysteine levels are toxic, but it is unlikely that a disturbed deregulation of the size and composition of the thiol pool is directly involved in the phytotoxicity of H2S (De Kok, 1990; De Kok et al., 1998, 2002a). Several plant species tolerate high levels of thiols, including cysteine, when exposed to SO2 or H2S without any negative effects on biomass production (De Kok, 1990; De Kok et al., 1998, 2002a; Stulen et al., 2000). Likewise, there is no evidence for a direct relationship between the level of sulfate accumulation and the toxicity of sulfur gases (De Kok, 1990; Van der Kooij & De Kok, 2000; Van der Kooij et al., 1997). 5.6. Conclusions

It is evident that, based on current knowledge about the impact of atmospheric sulfur gases on plant functioning, it is hard to establish cause–effect relationships. First, the physiological basis for the variation in plant species’ susceptibility to SO2 and H2S is still largely obscure. Second, there is no clear-cut transition in the level or rate of metabolism of the absorbed sulfur gases and their phytotoxicity. Third, the impact of sulfurous air pollutants on plants may be strongly dependent on the sulfur status of the soil.

REFERENCES Baldocchi, D.D. 1993. Deposition of gaseous sulfur compounds to vegetation. In: De Kok, L.J., Stulen, I., Rennenberg, H., Brunold, C., and Rauser, W.E., eds. Sulfur nutrition and assimilation in higher plants; regulatory, agricultural and environmental aspects. SPB Academic Publishing, The Hague, The Netherlands, pp. 271–293. Brunold, C. 1990. Reduction of sulfate to sulfide. In: Rennenberg, H., Brunold, C., De Kok, L.J., and Stulen, I., eds. Sulfur nutrition and sulfur assimilation in higher plants. SPB Academic Publishing, The Hague, The Netherlands, pp. 13–31. Buchner, P., Stuiver, C.E.E., Westerman, S., Wirtz, M., Hell, R., Hawkesford, M.J., and De Kok, L.J. 2004. Regulation of sulfate uptake and expression of sulfate transporter genes in Brassica oleracea as affected by atmospheric H2S and pedospheric sulfate nutrition. Plant Physiol. 136, 3396–3408. Ceccotti, S.P., and Messick, D.L. 1997. A global review of crop requirements, supply, and environmental impact on nutrient sulphur balance. In: Cram, W.J., De Kok, L.J., Stulen, I., Brunold, C., and Rennenberg, H., eds. Sulphur metabolism in higher plants; molecular, ecophysiological and nutritional aspects. Backhuys Publishers, Leiden, The Netherlands, pp. 155–163. Clarkson, D.T., Hawkesford, M.J., and Davidian, J.C. 1993. Membrane and long-distance transport of sulfate. In: De Kok, L.J., Stulen, I., Rennenberg, H., Brunold, C., and Rauser, W.E., eds. Sulfur nutrition and assimilation in higher plants; regulatory

A Study Establishing Cause–Effect Relationships of SO2 and H2S

133

agricultural and environmental aspects. SPB Academic Publishing, The Hague, The Netherlands, pp. 3–19. Cram, W.J. 1990. Uptake and transport of sulfate. In: Rennenberg, H., Brunold, C., De Kok, L.J., and Stulen, I., eds. Sulfur nutrition and sulfur assimilation in higher plants. SPB Academic Publishing, The Hague, The Netherlands, pp. 3–11. De Kok, L.J. 1990. Sulfur metabolism in plants exposed to atmospheric sulfur. In: Rennenberg, H., Brunold, C., De Kok, L.J., and Stulen, I., eds. Sulfur nutrition and sulfur assimilation in higher plants; fundamental, environmental and agricultural aspects. SPB Academic Publishing, The Hague, The Netherlands, pp. 111–130. De Kok, L.J., and Stulen, I. 1993. Functions of glutathione in plants under oxidative stress. In: De Kok, L.J., Stulen, I., Rennenberg, H., Brunold, C., and Rauser, W.E., eds. Sulfur nutrition and assimilation in higher plants; regulatory, agricultural and environmental aspects. SPB Academic Publishing, The Hague, pp. 125–138. De Kok, L.J., and Tausz, M. 2001. The role of glutathione in plant reaction and adaptation to air pollutants. In: Grill, D., Tausz, M., and De Kok, L.J., eds. Significance of glutathione to plant adaptation to the environment. Kluwer Academic Publishers, Dordrecht, The Netherlands, pp. 185–201. De Kok, L.J., Stuiver, C.E.E., and Stulen, I. 1998. Impact of atmospheric H2S on plants. In: De Kok, L.J., and Stulen, I., eds. Responses of plants metabolism to air pollution and global change. Backhuys Publishers, Leiden, The Netherlands, pp. 51–63. De Kok, L.J., Westerman, S., Stuiver, C.E.E., and Stulen, I. 2000. Atmospheric H2S as plant sulfur source: Interaction with pedospheric sulfur nutrition—a case study with Brassica oleracea L. In: Brunold, C., Rennenberg, H., De Kok, L.J., Stulen, I., and Davidian, J.-C., eds. Sulfur nutrition and sulfur assimilation in higher plants: Molecular, biochemical and physiological aspects. Paul Haupt, Bern, Switzerland, pp. 41–44. De Kok, L.J., Stuiver, C.E.E., Westerman, S., and Stulen, I. 2002a. Elevated levels of hydrogen sulfide in the plant environment: Nutrient or toxin. In: Omasa, K., Saji, H., Youssefian, S., and Kondo, N., eds. Air pollution and biotechnology in plants. Springer, Tokyo, Japan, pp. 201–213. De Kok, L.J., Castro, A., Durenkamp, M., Stuiver, C.E.E., Westerman, S., Yang, L., Stulen, I. 2002b. Sulphur in plant physiology. Proceedings No. 500, The International Fertiliser Society, York, UK, pp. 1–26. De Kok, L.J., Durenkamp, M., Yang, L., and Stulen, I. 2007. Atmospheric sulfur. In: Hawkesford, M.J., and De Kok, L.J., eds. Sulfur in plants—an ecological perspective. Springer, Dordrecht, The Netherlsands, pp. 91–106. Durenkamp, M., and De Kok, L.J. 2002. The impact of atmospheric H2S on growth and sulfur metabolism of Allium cepa L. Phyton 42, 55–63. Durenkamp, M., and De Kok, L.J. 2004. Impact of pedospheric and atmospheric sulphur nutrition on sulphur metabolism of Allium cepa L., a species with a potential sink capacity for secondary sulphur compounds. J. Exp. Bot. 55, 1821–1830. Durenkamp, M., De Kok, L.J., and Kopriva, S. 2007. Adenosine 5u phosphosulphate reductase is regulated differently in Allium cepa L. and Brassica oleracea L. upon exposure to H2S. J. Exp. Bot. 58, 1571–1579. Haneklaus, S., Bloem, S., and Schnug, E. 2003. The global sulphur cycle and its link to plant environment. In: Abrol, Y.P., and Ahmand, A., eds. Sulphur in plants. Kluwer Academic Publishers, Dordrecht, The Netherlands, pp. 1–28. Herschbach, C., De Kok, L.J., and Rennenberg, H. 1995. Net uptake of sulfate and its transport to the shoot in spinach plants fumigated with H2S or SO2: Does atmospheric sulfur affect the ‘inter-organ’ regulation of sulfur nutrition? Botanica Acta 108, 41–46.

134

Luit J. De Kok et al.

Koralewska, A., Stuiver, C.E.E., Posthumus, F.S., Kopriva, S., Hawkesford, M.J., and De Kok, L.J. 2008. Regulation of sulfate uptake, expression of the sulfate transporters Sultrl;1 and Sultrl;2, and APS reductase in Chinese cabbage (Brassica pekinensis) as affected by atmospheric H2S nutrition and sulfate deprivation. Funct. Plant Biol. 35, 318–327. Lendzian, K.L. 1984. Permeability of plant cuticles to gaseous air pollutants. In: Koziol, M.J., and Whatley, F.R., eds. Gaseous air pollutants and plant metabolism. Butterworths, London, UK, pp. 77–81. McGrath, S.P., Zhao, F.J., Blake-Kalff, M.M.A., 2002. History and outlook for sulphur fertilisers in Europe. Proceedings No. 497, International Fertiliser Society, York, UK, pp. 1–26. Mudd, J.B. 1998. On ozone. In: De Kok, L.J., and Stulen, I., eds. Responses of plants metabolism to air pollution and global change. Backhuys Publishers, Leiden, The Netherlands, pp. xiii–xix. Posthumus, A.C. 1998. Air pollution and global change: Significance and prospectives. In: De Kok, L.J., and Stulen, I., eds. Responses of plants metabolism to air pollution and global change. Backhuys Publishers, Leiden, The Netherlands, pp. 15–22. Schnug, E., and Evans, E.J. 1992. Monitoring of the sulfur supply of agricultural crops in northern Europe. Phyton 32, 119–122. Stuiver, C.E.E., and De Kok, L.J. 2001. Atmospheric H2S as sulfur source for plant growth: Kinetics of H2S uptake and activity of O-acetylserine(thiol)lyase as affected by sulfur nutrition. Environ. Exp. Bot. 46, 29–36. Stuiver, C.E.E., De Kok, L.J., and Westerman, S. 1997. Sulfur deficiency in Brassica oleracea L.: Development, biochemical characterization, and sulfur/nitrogen interactions. Russ. J. Plant Physiol. 44, 505–513. Stulen, I., Posthumus, F., Amaˆncio, S., Masselink-Beltman, I., Mu¨ller, M., and De Kok, L.J. 2000. Mechanism of H2S phytotoxicity. In: Brunold, C., Rennenberg, H., De Kok, L.J., Stulen, I., and Davidian, J.C., eds. Sulfur nutrition and sulfur assimilation in higher plants: Molecular, biochemical and physiological aspects. Paul Haupt, Bern, Switzerland, pp. 381–383. Tausz, M., Van der Kooij, T.A.W., Mu¨ller, M., De Kok, L.J., and Grill, D. 1998. Uptake and metabolism of oxidized and reduced sulfur pollutants by spruce trees. In: De Kok, L.J., and Stulen, I., eds. Responses of plant metabolism to air pollution and global change. Backhuys Publishers, Leiden, The Netherlands, pp. 457–460. Tausz, M., Weidner, W., Wonisch, A., De Kok, L.J., and Grill, D. 2003. Uptake and distribution of 35S-sulfate in needles and roots of spruce seedlings as affected by exposure to SO2 and H2S. Environ. Exp. Bot. 50, 211–220. Thomas, M.D. 1951. Gas damage to plants. Annu. Rev. Plant Physiol. 2, 293–322. Thomas, S.G., Hocking, T.J., and Bilsborrow, P.E. 2003. Effect of sulphur fertilisation on the growth and metabolism of sugar beet grown on soils of differing sulphur status. Field Crops Res. 83, 223–235. Van der Kooij, T.A.W., De Kok, L.J., Haneklaus, S., and Schnug, E. 1997. Uptake and metabolism of sulfur dioxide by Arabidopsis thaliana. New Phytol. 135, 101–107. Van der Kooij, T.A.W., and De Kok, L.J. 1998. Kinetics of deposition of SO2 and H2S to shoots of Arabidopsis thaliana L. In: De Kok, L.J., and Stulen, I., eds. Responses of plant metabolism to air pollution and global change. Backhuys Publishers, Leiden, The Netherlands, pp. 479–481. Van der Kooij, T.A.W., and De Kok, L.J. 2000. Variation in response of Arabidopsis thaliana lines to atmospheric SO2 exposure. Phyton 40(3), 119–124.

A Study Establishing Cause–Effect Relationships of SO2 and H2S

135

Westerman, S., De Kok, L.J., and Stulen, I. 2000. Interaction between metabolism of atmospheric H2S in the shoot and sulfate uptake by the roots of curly kale (Brassica oleracea L.). Physiol. Plant. 109, 443–449. Westerman, S., Stulen, I., Suter, M., Brunold, C., and De Kok, L.J. 2001. Atmospheric H2S as sulfur source for Brassica oleracea: Consequences for the activity of the enzymes of the assimilatory sulfate reduction pathway. Plant Physiol. Biochem. 39, 425–432. Yang, L., Stulen, I., De Kok, L.J., and Zheng, Y. 2002. SO2, NOX and acid deposition problems in China—impact on agriculture. Phyton 42(3), 255–264. Yang, L., Stulen, I., and De Kok, L.J. 2003. Interaction between atmospheric sulfur dioxide deposition and pedospheric sulfate nutrition in Chinese cabbage. In: Davidian, J.-C., Grill, D., De Kok, L.J., Stulen, H., Hawkesford, M.J., Schnug, E., and Rennenberg, H., eds. Sulfur transport and assimilation in plants: Regulation, interaction and signaling. Backhuys Publishers, Leiden, The Netherlands, pp. 363–365. Yang, L., Stulen, I., and De Kok, L.J. 2006a. Sulfur dioxide: Relevance of toxic and nutritional effects for Chinese cabbage. Environ. Exp. Bot. 57, 236–245. Yang, L., Stulen, I., and De Kok, L.J. 2006b. Impact of sulfate nutrition on the utilization of atmospheric SO2 as sulfur source for Chinese cabbage. J. Plant Nutr. Soil Sci. 169, 529–534. Zhao, F.J., Hawkesford, M.T., and McGrath, S.P. 1999. Sulphur assimilation and effects on yield and quality of wheat. J. Cereal Sci. 30, 1–17.

Developments in Environmental Science, Volume 9 Allan H. Legge (Editor) Copyright r 2009 Elsevier Ltd. All rights reserved. ISSN: 1474-8177/DOI:10.1016/S1474-8177(08)00206-4

137

Chapter 6 Hormesis—Its Relevance in Phytotoxicology Hans-Ju¨rgen Ja¨ger and Sagar V. Krupa Abstract In toxicology, the concept of ‘‘hormesis’’ states that many nonessential chemicals, including tropospheric ozone, stimulate plant growth and other biological processes at low doses, but inhibit such processes at higher levels. In addition, differential species-dependent hormetic responses can be very important in inter-plant competition and in shaping the plant community structure. Because of the critical concern for the environment, up to now virtually all of the attention regarding air pollutant effects on plants has been directed to the quantitative aspects of their phytotoxicity. However, a complete numerical definition of the broad range of dose and the corresponding receptor response surface, including the capture of hormetic effects in particular, is required for establishing sound quantitative and predictive cause–effect relationships. Achieving that goal would necessitate a change in the use of traditional experimental designs.

6.1. Introduction

The term ‘‘dose’’ is defined as the ‘‘exact amount of medicine or extent of some other treatment to be given or taken at one time or at stated intervals’’ (Webster’s New World Dictionary, Third College Edition, 1994). Complete numerical definition of the range of dose and the receptor response surface is a prerequisite for establishing quantitative cause–effect relationships. According to one of the most fundamental principles of toxicology, ‘‘the dose determines the poison.’’ The German alchemist Paracelsus (1493–1541) recognized that the efficacy of medicinal use of small amounts of toxic chemicals depended on their dose. More than three

Corresponding author: E-mail: [email protected]

138

Hans-Ju¨rgen Ja¨ger and Sagar V. Krupa

centuries later, Schulz (1888) showed that many chemical agents had the effect of stimulating the growth and respiration of yeast. The phenomenon became known as the Arndt-Schulz Law. At about the same time, in Chicago, Hueppe (1896), apparently not being familiar with Schulz’s publications, made similar observations on bacteria, and his generalization became known as Hueppe’s Rule. Much later, Southam and Ehrlich (1943) found that sub-inhibitory concentrations of a natural antibiotic stimulated growth of wood decay fungi, and they named that stimulation by low levels of inhibitors as ‘‘hormesis’’ or ‘‘hormetic response.’’ The word hormesis is derived from the Greek language, meaning, ‘‘to excite’’ and should be viewed as being different from: ‘‘the dose determines the poison.’’ Subsequently, the concept of hormesis was used by Luckey (1975) and Stebbing (1981, 1982, 1987) and is described as many chemicals stimulate growth and other biological processes at low doses, but inhibit such processes at higher levels following the widely recognized b-curve (Fig. 6.1). Examples of hormesis have been reported for many types of chemicals and biological responses, including plant responses (Calabrese, 2005). Several databases have been compiled from peer-reviewed publications to quantify aspects of hormetic responses in toxicological studies. Davis and Svensgaard (1990, 1994) estimated the incidence of hormetic responses based on the frequency of deviation from control responses, independent of study design, and the No Observed Adverse Effect Level (NOAEL). The most comprehensive databases are those of Calabrese and Baldwin (1997, 2000, 2001a, 2001b) and Calabrese et al. (1999), created to assess those studies in the literature satisfying rigorous criteria for evidence of hormesis. According to Calabrese and Baldwin (2003), the toxicological research community made an error of historic dimensions believing in the threshold model, and should, therefore, rethink its central belief. The traditional

Figure 6.1. The most common dose–response curve, the b-curve showing hormesis (Source: Calabrese & Baldwin, 1997).

Hormesis—Its Relevance in Phytotoxicology

139

threshold model is used in risk assessment of non-carcinogens (Fig. 6.2a) and the linear non-threshold model is used to extrapolate risks to very low doses of carcinogens, including radiation (Fig. 6.2b). The evaluation by Calabrese and Baldwin of the evidence of hormesis for non-carcinogens shows that the dose–response curve is characterized by either a U-shaped or an inverted U-shaped (b-curve) dose response, depending on how the effect is defined (Fig. 6.3). That subject is discussed in greater detail in a subsequent section of this chapter.

Figure 6.2. Traditional dose–response curve for: (a) non-carcinogens showing a threshold and (b) linear non-threshold for carcinogens.

Figure 6.3. A dose–response curve depicting characteristics of the chemical hormetic zone. LOEL: Lowest Observable Effect Level, NOEL: No Observable Effect Level, ZEP: Zero End Point Value (Source: Calabrese & Baldwin, 1997).

140

Hans-Ju¨rgen Ja¨ger and Sagar V. Krupa

The b-dose–response curve results from some disruption of homeostasis leading to overcompensation, and discerns the toxicological nature from the concept of essential nutrients and from direct compensation. Certainly, the hormetic dose–response curve has many similarities to that of trace elements (e.g., copper (Cu)) where low concentrations are beneficial, but higher levels result in toxicity. Hormesis represents an advantage gained by the individual species from the overall resources and energy initially allocated for detoxification and repair, but in excess of that needed to repair the immediate damage. Over-shootings of responses such as these have been interpreted as the cost of adaptation (Burton, 1939), and characterized by transitions from an old to a new steady state under changed signal conditions (von Bertalanffy et al., 1977). Such transitions most likely occur in stress-triggered metabolic (e.g., increased photosynthesis, synthesis of stress proteins, and synthesis of metabolites, such as compatible solutes in the case of salinity or phyto-chelatins in the case of heavy metal burden) and developmental responses, and may cause acclimation or increased fitness. From an eco-physiological viewpoint, stress can be described as a state in which increasing demands on a plant lead to an initial perturbation of functions, followed by normalization and improved fitness. If the limits of tolerance are exceeded and the adaptive capacity is overtaxed, permanent damage, or even death may result (Larcher, 1987; Levitt, 1972). According to Selye (1936) ‘‘y all agents can act as stressors, producing both stress and specific reactions.’’ Effects of stressors are, therefore, accompanied by specific and non-specific reactions. Another distinctive feature of the stress syndrome concept is that stress induces physiological and biochemical reactions, which regulate homeostasis through different degrees of compensation, leading to increased defense. In accordance with the general adaptation syndrome (GAS) (Leshem & Kuiper, 1996), a short exposure to sub-lethal or lethal concentrations of a toxicant can increase the plant’s tolerance to another stressor. Larcher’s concept of stress (Larcher, 1987), considers a hierarchy of sequential phases of stress responses that include states of stability, instability, and transitions to alternative states of stability and functionality. This definition or concept assembles basic characters of stress responses (hormesis, overcompensation, over-shootings) that can easily be translated into the conceptual language of quantitative control of complex dynamic systems, for example, metabolic networks and growth (Heinrich & Rapoport, 1974; Kascer & Burns, 1973). From a dynamic viewpoint, adaptation is a self-organized process that adjusts metabolic networks (and growth) to varying environments using feedback mechanisms. In the past, the actions of stressors have too often been recognized in a negative

Hormesis—Its Relevance in Phytotoxicology

141

way, overlooking the values of adaptation. Stress responses are a fundamental aspect of evolutionary biological processes and, as has been frequently stated, ‘‘y without evolution, biology makes no sense.’’ 6.2. Numerical description of hormesis

Hormetic responses have been shown to occur over a limited range of dose and magnitude of response (Fig. 6.3). In general, the average maximum stimulatory response was roughly 30–60% greater than the control. When the response is seen as a percentage rather than in the order of a fold, it may often be interpreted as normal variability as opposed to real stimulation or as false positive response, and reproducibility becomes a real issue (Calabrese et al., 1999). The ability to assess a strong conformity to the b-curve ideally requires the establishment of an endpoint-specific Lowest Observable Effect Level (LOEL; Fig. 6.3) and a No Observable Effect Level (NOEL), with multiple treatments or doses within two orders of magnitude immediately below the NOEL; in other words, there should be at least four treatments distributed specifically relative to the NOEL. According to van Ewijk and Hoekstra (1993), a standard logistic model cannot be used when hormesis occurs. They described a model that included the Effective Concentration (EC50) as one of the model parameters, giving the advantage of defining its confidence intervals through the application of non-linear regression techniques. Certainly there are other approaches, particularly in evaluating higher, whole-plantpopulation responses to air quality, but such strategies must recognize and define the full spectrum of the b-curve. 6.3. Hormetic responses of plants to some environmental factors 6.3.1. Alcohols and organic solvents, fluoride, heavy metals, herbicides, and salinity

A number of studies indicate hormetic plant responses with several nonessential elements and molecules (Table 6.1). They include alcohols and organic solvents, fluoride (F), heavy metals (such as cadmium (Cd), nickel (Ni), lead (Pb), and arsenic (As)), herbicides, and salinity. Among these, the effects of herbicides have been the most studied. Among all the independent variables, shoot growth is the most studied response. Clearly, as stated earlier, the hormetic plant response depends on the nature, type, and dose of the independent variable, the plant species, and the experimental conditions. It is most interesting to note that F-stimulated plant response under hormetic doses both in ambient and experimental conditions

142

Hans-Ju¨rgen Ja¨ger and Sagar V. Krupa

Table 6.1. Examples of studies indicating a hormetic effect on plants, induced by different stress factorsa Stressor

Plant species

Response parameter

Alcohols Organic Solvents

Oats (Avena sativa) Corn (Zea mays)

Shoot growth Root growth

Ber and Moskwa (1951) Morre´ et al. (1965)

Fluoride

Pinto bean (Phaseolus vulgaris) Corn (Zea mays)

Plant dry weight

Treshow and Harner (1968)

Shoot growth

Bertrand (1969) Jensen (1907), Antonovics et al. (1971), Ernst (1974), Brooks et al. (1979), Brown and Martin (1981), Hertstein and Ja¨ger (1986), Ernst et al. (1992), Prasad (1995), and Brooks (1998) Meharg et al. (1993)

Heavy metals Cd, Ni, Pb

Several plant species (crops, wild plants)

Shoot growth

As

Yorkshire fog (Holcus lanatus) Cordgrass (Spartina patens, Spartina alterniflora) Chinese braken fern (Pteris vittata, and other Pteris spp.)

Shoot growth

Herbicides

Salinity

a

Oat (Avena sativa) Cucumber (Cucumis sativus) Peppermint (Mentha piperita) Cotton (Gossypium hirsutum) Several plant species (crops, wild plants)

Reference

Shoot growth

Carbonell-Barrachina et al. (1998)

Shoot growth

Tu and Ma (2005)

Shoot growth Root growth

Wiedman and Appleby (1972)

Shoot growth

Calabrese and Howe (1976)

Shoot growth

Allender (1997)

Shoot growth, root growth, yield

Yeo (1983), Epstein (1985), Marschner (1995), Kreeb (1996), Shannon and Grieve (1999), La¨uchli and Lu¨ttge (2002)

Only factors non-essential to plants are included.

(Table 6.1). However, in higher doses, it is an accumulative poison to plants and is important in the food chain as a toxicant in forage consumed by animals (McCune & Weinstein, 2002; National Academy of Sciences, 1971). 6.3.2. Ozone

A number of studies have also indicated the hormetic effects of ozone on plants (Table 6.2). Based on the evidence of dry weights of shoots or the yield values alone, for both cultivated and native vegetation, exposures to

143

Hormesis—Its Relevance in Phytotoxicology Table 6.2. Examples of studies indicating an O3-induced hormetic effect on plantsa Plant Species

Exposure

Response Parameter

Reference

Tomato (Lycopersicon esculentum)

30 pphm per 2 h

Stem elongation

Neil et al. (1973)

Bean (Phaseolus vulgaris) Barley (Hordeum vulgare) Smartweed (Polygonum lapathifolium)

3 pphm per 8 h day per 12 days

Plant dry weight or specific organ dry weight

Bennett et al. (1974)

Geranium (Geranium fremontii), Porter’s wild lovage (Langusticum porteri), Prostrate knotweed (Polygonum aviculare), Variable-leaf scorpion weed (Phacelia heterophylla)

15 pphm per 3 h per day per 5 days per week per 3 months per 3 years

Top dry weight

Harward and Treshow (1975)

Annual ryegrass (Lolium multiflorum) Crimson clover (Trifolium incarnatum)

3 or 9 pphm per 8 h per day per 6 weeks

% Dry weight

Bennett and Runeckles (1977)

Soybean (Glycine max)

4.6 pphm per 6 h intermittent per 113 days ¼ total 341 h

Growth and yield

Endress and Grunwald (1985)

Pumpkin (Cucurbita moschata)

2.0 or 4.0 pphm per 6 h day per 3 consecutive days

Shoot fresh and dry weights

Rajput and Ormrod (1986)

Spring Wheat (Triticum aestivum)

1 pphm. 24 h mean per 8 h exposure, whole growing season

Growth and grain yield

Adaros et al. (1991)

Bean (Phaseolus vulgaris)

2.0, 2.6 or 3.2 pphm per seasonal 7 h mean per 8 h day O3

Yield

Sanders et al. (1992)

Spring wheat (Triticum aestivum)

Ambient þ 2.5 pphm per 6 h day per 5 days per week per approximately 50–70 days

Grain yield

Finnan et al. (1996)

a

Only studies on growth or yield parameters are included and only cases reporting a statistically significant effect or cases where the effect was Z10% with reference to the control treatment are included.

pphm (or other statistical definitions of dose levels) of ozone in various experiments resulted in positive effects (Figs. 6.4, 6.5, and 6.6). Most interestingly, several unrelated plant species used in those studies are known to be ozone (O3) sensitive regarding their expression of foliar injury (Krupa et al., 1998). However, such data on O3-induced hormetic effects

144

Hans-Ju¨rgen Ja¨ger and Sagar V. Krupa

Figure 6.4. (a) Growth stimulation in single plants of Molinia caerulea exposed to increasing O3 concentrations: 3, 34, 54, 77 ppb (from right to left) and (b) monocultures exposed to charcoal-filtered air (left) and air with 50 ppb O3 (right) (Source: European Biostress Program, Agricultural University of Wageningen, The Netherlands). (http://www.plant.wageningenur. nl/expertise/cropecology/biostress_experiments.htm, July 17, 2008).

are very limited and are dependent, as noted elsewhere, on the plant species, exposure kinetics, and plant growth conditions. Thus, a generalized concept or response surface cannot be derived at this time. Nevertheless, the sparse data argue in favor of future experimental designs that will consider defining the entire plant response surface and

145

Hormesis—Its Relevance in Phytotoxicology 120

Star

Turbo

Dry Weight (%)

90

Total Biomass 1988 1989

60

Grain Yield 1988 1989 30

0 0

0.015

0.03

0

0.015

0.03

0.045

Ozone Concentration (ppm) Figure 6.5. Yield responses of two different cultivars of wheat (Triticum aestivum) subjected to chronic O3 exposures (see Table 6.2 for details on exposures) (Source: Adaros et al., 1991).

not just a part of it, as has been the practice (Heck et al., 1988; Ja¨ger et al., 1992; Kickert & Krupa, 1991).

6.4. Implications of hormesis in understanding air pollutant exposure–plant response relationships

Hormetic responses can be very important in inter-plant competition and in shaping the community structure. For example, among 27 native herbs and grass species examined by Pleijel and Danielsson (1997), as opposed to the others, growth rate of Festuca ovina was significantly stimulated by an above-ambient O3 treatment. Similar results were obtained with Centaurea jacea and Lychnis flos-cuculi, but not with 22 other members of the seminatural, temperate grasslands in Switzerland (Bungener et al., 1999). In addition to the variability in the adverse responses, the extent of the hormetic effect itself is very variable, being dependent on the plant species, location, and time, the stressor under consideration, its dose, and the response parameter measured. Thus, it is a highly stochastic process and will, therefore, result in an under- or over-estimation or assignment of a generalized threshold value. Woodwell (1975) and Odum et al. (1979)

146

Hans-Ju¨rgen Ja¨ger and Sagar V. Krupa

Figure 6.6. Relationship between growing season fourth highest daily maximum 8 h O3 concentration and mean cross-sectional stem area growth response of an O3-sensitive aspen (Populus tremuloides) clone in a free-air (FACE) exposure experiment (Source: Percy et al., 2006).

concluded that the notion of applying a threshold concentration in ecological effects research is inappropriate. Hormetic effects have been reported in passing by plant effects scientists (Tables 6.1 and 6.2), but in recent years, that is changing for several variables, particularly with those other than air pollutants (e.g., herbicides and pesticides). Hormesis relates to the issue of the dose–response relationships of reciprocal bi-modality (positive or stimulation and negative or adverse effect components in the dose–response curve) of non-essential elements for plants. In our preoccupation with demonstrating the deleterious effects of air pollutants on plants, experimental designs have mostly been constructed to optimize exposure doses above an accepted or perceived level (Fig. 6.2). These include the National Crop Loss Assessment Network Program (NCLAN; Heck et al., 1988) and the European Open-Top Chambers Program (EOTC; Ja¨ger et al., 1992). Rightfully, Rawlings et al. (1988) argued that there are good statistical reasons for using a dose–response approach with dose levels well above the region of immediate interest. With respect to using higher than ambient dose levels, a concern has been raised that high levels may trigger a biological process different from that causing the yield reductions at ambient levels, and as a consequence, the response equations using higher dose levels would not be reliable for prediction at

Hormesis—Its Relevance in Phytotoxicology

147

the low dose levels. It is certainly reasonable to expect more than one biological process to be involved in the plant response to O3. Krupa and Nosal (1989) reached a similar conclusion. However, according to Rawlings et al. (1988), with respect to the NCLAN data, none of the response equations or model outputs showed discontinuities that suggest any abrupt change in the processes. Although this is true in a statistical sense, in our view, mechanistic changes in biological systems will be gradual (rate controlled), being dependent on the progression of the short-interval dose levels. Nevertheless, the question may be raised as to why hormetic effects are not being reported more frequently. According to Calabrese and Baldwin (1997), this is due to a combination of study design along with its influence on response evaluation, which emphasizes the upper end of the dose–response continuum (i.e., higher concentrations resulting in toxic responses that can be used in risk assessment). Calabrese and Baldwin concluded that a direct relationship has been shown between the strength of the evidence supporting hormesis and the number of treatments that include both the overall number of doses in the experiment and the number of doses in the hormetic zone. However, as the average range of the hormetic zone is about one order of magnitude above zero exposure, hormesis is difficult to discern when doses of wide intervals are used. Because, in general, hormetic responses are modest in nature, more than the conventional number of treatment groups is needed to define the complete structure of the dose–response relationship, especially at the low end of the treatments where the hormetic response would be expected to occur. Because of the limited nature of the hormetic response, greater attention should be directed to the sample size and data analysis and, therefore, it would be desirable to use more subjects in the low-dose treatment groups than at the higher levels, because the treatment effects limit the magnitude and the variability in the response (Calabrese et al., 1999). According to Krupa and Teng (1982), the relationship between pollutant dose and yield loss may be represented as a three-dimensional response surface in which yield loss is affected differently by the pollutant dose with changing phenological physiology of the crop. Similar temporal considerations are relevant to hormesis (Stebbing, 1987). When attempting to generate data to model the response surface, it would be more meaningful to create a broad range of exposure regimes than to intensively replicate a few treatments (Myers, 1971; Myers & Montgomery, 1995). A review of the published air pollution effects literature shows that most studies have provided data that only enable an understanding of part of the specific pollutant dose–plant response surface (U.S. EPA, 2006). A real complex question is: What should be the reference treatment to which others producing an adverse effect can be compared in risk

148

Hans-Ju¨rgen Ja¨ger and Sagar V. Krupa

assessment? Should it be (1) a threshold value (Fig. 6.2a) or (2) a Zero End Point value (ZEP in Fig. 6.3) or (3) the dose producing the maximum stimulation (Fig. 6.3). Option (1) involving a single threshold value, is most desirable from a regulatory perspective, but is scientifically inappropriate as noted earlier (Odum et al., 1979; Woodwell, 1975). Threshold values will vary with the plant species and within the same plant species under different growing conditions (interactions among physical, chemical, and biological factors). It is a product of a stochastic process. Identification of ZEP (option 2) is scientifically more precise, as it is a product of defining the complete response surface, but suffers from the same difficulties as option (1). If hormetic responses were to be shown as a generally occurring phenomenon in air pollutant exposure–plant effects research, the grower might consider option (3) as most desirable, because it would provide the maximum loss value, and thus a justification for initiating altered management practices, and a higher probability for economic subsidy (e.g., crop insurance). A major limiting factor here is the stochasticity of the occurrence of hormesis. Clearly there are differing values or importance associated with the three options. It is beyond the scope of this chapter to draw a conclusion regarding the desirability of any particular option given, because of the dearth of data on hormesis in air pollutant–plant response relationships and because of our very limited knowledge of that subject at this time. Nevertheless, the discussion presented here should serve as an impetus for future, more thorough, research of the hormetic effect and the science associated with it, through the examination of the complete response surfaces of key plant species.

6.5. Conclusions

Many non-essential chemicals (e.g., heavy metals, herbicides, salinity) stimulate plant growth and other biological processes at low doses (hormesis), but inhibit such processes at higher levels, following the widely recognized b-curve. Hormesis represents an advantage gained by the individual species from the overall resources and energy initially allocated for detoxification and repair, but in excess of that needed to repair the immediate damage. As hormetic effects vary with the plant species, it can result in selective advantage for certain members over others in mixed communities. Although some air pollutant-induced hormetic effects have been reported, but not in the context of the present discussion, the ability to assess a strong conformity to the b-curve and fully define the hormetic effect,

Hormesis—Its Relevance in Phytotoxicology

149

ideally requires the establishment of an endpoint-specific LOEL and a NOEL, with multiple treatments or doses (Z4) within two orders of magnitude immediately below the NOEL. However, in our preoccupation to demonstrate the deleterious effects of air pollutants on plants, experimental designs have mostly been constructed to optimize exposure doses above an accepted or perceived level and thus, show insufficient potential for detecting or describing hormesis and its impacts on the traditional dose– response functions. A real complex question is: What should be the reference treatment to which others producing an adverse effect can be compared in risk assessment? Should it be a threshold value, or a ZEP value, or the dose producing the maximum stimulation? Certainly there are differing values or importance associated with the three options. Because of the dearth of data on air pollutant exposure and the definition of complete plant response surfaces, it is not possible at this time to draw a conclusion regarding the desirability of any particular option. In the future, more thorough research of the hormetic effect and the associated science is required through the examination of broad response surfaces of key plant species. That would require a change in the use of traditional experimental designs. Such a shift is also critical in examining interactive effects of multiple plant-growth-regulating variables in the ambient environment.

ACKNOWLEDGMENT

We would like to thank Jose´ Machado-Caballero (University of Minnesota), for his invaluable assistance in the preparation of the illustrations.

REFERENCES Adaros, G., Weigel, H.-J., and Ja¨ger, H.-J. 1991. Impact of ozone on growth and yield parameters of two spring wheat cultivars (Triticum aestivum L.). Z. Pflanzenkr. Pflanzenschutz. 98, 113–124. Allender, W.J. 1997. Effect of trifluoperazine and verapamil on herbicide stimulated growth of cotton. J. Plant. Nutr. 20, 69–80. Antonovics, J., Bradshaw, A.D., and Turner, R.G. 1971. Heavy metal tolerance in plants. Adv. Ecol. Res. 7, 1–85. Bennett, J.P., and Runeckles, V.C. 1977. Effects of low levels of ozone on growth of crimson clover and annual rye grass. Crop Sci. 17, 443–445. Bennett, J.P., Resh, H.M., and Runeckles, V.C. 1974. Apparent stimulations of plant growth by air pollutants. Can. J. Bot. 52, 35–41. Ber, A., and Moskwa, Z. 1951. The influence of organic solvents on the growth of plants. Experientia 7, 136–137.

150

Hans-Ju¨rgen Ja¨ger and Sagar V. Krupa

Bertrand, D. 1969. Le fluor, oligo-e´le´ment dynamique pour le mais. C. R. Hebd. Seances. Acad. Sci. Ser. D, Sci. Nat. 269, 1767–1769. Brooks, R.R., ed. 1998. Plants that hyperaccumulate heavy metals. CAB International, Wallingford, UK. Brooks, R.R., Morrison, R.S., Reeves, R.D., Dudley, T.R., and Akman, Y. 1979. Hyperaccumulation of nickel by Alyssum Linnaeus (Cruciferae). Proc. R. Soc. Lond. B. 203, 387–403. Brown, H., and Martin, M.H. 1981. Pretreatment effects of cadmium on the root growth of Holcus lanatus. New Phytol. 89, 621–629. Bungener, P., Nussbaum, S., Grub, A., and Fuhrer, J. 1999. Growth response of grassland species to ozone in relation to soil moisture condition and plant strategy. New Phytol. 142, 283–293. Burton, A.C. 1939. The properties of the steady state compared to those of equilibrium as shown in characteristic biological behavior. J. Cell. Comp. Physiol. 14, 327–349. Calabrese, E.J. 2005. Paradigm lost, paradigm found: The re-emergence of hormesis as a fundamental dose response model in the toxicological sciences. Environ. Pollut. 138, 378–411. Calabrese, E.J., and Howe, K.J. 1976. Stimulation of growth of peppermint (Menta piperita) by phosfon, a growth retardant. Physiol. Plant. 37, 163–165. Calabrese, E.J., and Baldwin, L.A. 1997. The dose determines the stimulation (and poison): Development of a chemical hormesis database. Int. J. Toxicol. 16, 545–559. Calabrese, E.J., and Baldwin, L.A. 2000. Chemical hormesis: Its historical foundations as a biological hypothesis. Hum. Exp. Toxicol. 19, 2–31. Calabrese, E.J., and Baldwin, L.A. 2001a. The frequency of U-shaped dose responses in the toxicological literature. Toxicol. Sci. 62, 330–338. Calabrese, E.J., and Baldwin, L.A. 2001b. Scientific foundation of hormesis. Crit. Rev. Toxicol. 31(Special issue), 4–5. Calabrese, E.J., and Baldwin, L.A. 2003. Toxicology rethinks its central belief. Nature 421, 691–692. Calabrese, E.J., Baldwin, L.A., and Holland, C.D. 1999. Hormesis: A highly generalizable and reproducible phenomenon with important implications for risk assessment. Risk Anal. 19, 261–281. Carbonell-Barrachina, A.A., Aarabi, M.A., Delaune, R.D., Gambrell, R.P., and Patrick, W.H., Jr. 1998. The influence of arsenic chemical form and concentration on Spartina patens and Spartina alterniflora growth and tissue arsenic concentration. Plant Soil 198, 33–43. Davis, J.M., and Svensgaard, D.J. 1990. U-shaped dose–response curves: Their occurrence and implications for risk assessment. J. Toxicol. Environ. Health 30, 71–83. Davis, J.M., and Svensgaard, D.J. 1994. Nonmonotonic dose–response relationships in toxicological studies. In: Calabrese, E.J., ed. Biological effects of low level exposures: Dose–response relationships. CRC/Lewis, Boca Raton, FL, pp. 67–86. Endress, A.G., and Grunwald, C. 1985. Impact of chronic ozone on soybean growth and biomass. Agric. Ecosyst. Environ. 13, 9–23. Epstein, E. 1985. Salt-tolerant crops: Origins development, and prospects of the concept. Plant Soil 89, 187–198. Ernst, W.H.O. 1974. Schwermetallvegetation der Erde. Fischer, Stuttgart, Germany. Ernst, W.H.O., Verkleij, J.A.C., and Schat, H. 1992. Metal tolerance in plants. Acta Bot. Neerl. 41, 229–248. Finnan, J.M., Jones, M.B., and Burke, J.I. 1996. A time-concentration study on the effects of ozone on spring wheat (Triticum aestivum L.): 1. Effects on yield. Agric. Ecosyst. Environ. 57, 159–167.

Hormesis—Its Relevance in Phytotoxicology

151

Harward, M., and Treshow, M. 1975. Impact of ozone on the growth and reproduction of understory plants in the aspen zone of western U.S.A. Environ. Conserv. 2, 17–23. Heck, W.W., Taylor, O.C., and Tingey, D.T., eds. 1988. Assessment of crop loss from air pollutants. Elsevier Applied Science, London, UK. Heinrich, R., and Rapoport, T.A. 1974. A linear steady-state treatment of enzymatic chains. General properties, control and effector’s strength. Eur. J. Biochem. 42, 89–95. Hertstein, U., and Ja¨ger, H.-J. 1986. Tolerance of different populations of grass species to cadmium and other metals. Environ. Exp. Bot. 26, 309–319. Hueppe, F. 1896. Principles of bacteriology. Open Court, Chicago, IL. Ja¨ger, H.-J., Unsworth, M., De Temmerman, L., Mathy, P., eds. 1992. Effects of air pollution on agricultural crops in Europe. Air Pollution Research Report 46, Commission of the European Communities, Brussels, Belgium. Jensen, G.H. 1907. Toxic limits and stimulation effects of some salts and poisons on wheat. Bot. Gaz. 43, 11–44. Kascer, H., and Burns, J.A. 1973. The control of flux. In: Davies, D.D., ed. Rate control in biological processes. The University Press, Cambridge, UK, pp. 65–104. Kickert, R.N., and Krupa, S.V. 1991. Modeling plant response to tropospheric ozone: A critical review. Environ. Pollut. 70, 271–383. Kreeb, K.H. 1996. Salzstress. In: Brunold, C., Ru¨egsegger, A., and Bra¨ndle, R., eds. Stress bei Pflanzen. P. Haupt, Stuttgart, Germany, pp. 149–172. Krupa, S.V., and Teng, P.S. 1982. Uncertainties in estimating ecological effects of air pollutants. In: Proceedings 75th Annual Air Pollution Control Association Meetings, New Orleans, Air Pollution Control Association, Pittsburgh, PA, USA 82-6.1, 1–10. Krupa, S.V., and Nosal, M. 1989. Application of spectral coherence analysis to describe the relationships between ambient ozone exposure and crop growth. Environ. Pollut. 60, 319–330. Krupa, S.V., Tonneijck, A.E.G., and Manning, W.J. 1998. Ozone. In: Flagler, R.B., ed. Recognition of air pollution injury to vegetation: A pictorial atlas. Air & Waste Management Association, Pittsburgh, PA, pp. 2-1–2-28. Larcher, W. 1987. Stress bei Pflanzen. Naturwissenschaften 74, 158–167. La¨uchli, A., and Lu¨ttge, U. 2002. Salinity: Environment–plants–molecules. Kluwer Academic Publishers, Dordrecht, The Netherlands. Leshem, Y.Y., and Kuiper, P.J.C. 1996. Is there a GAS (general adaptation syndrome) response to various types of environmental stress? Biol. Plant. 36, 1–18. Levitt, J. 1972. Responses of plants to environmental stress. Academic Press, New York, NY. Luckey, T.D. 1975. Hormology with inorganic compounds. In: Cowston, F., and Korte, F., eds. Environmental quality and safety. Georg Thieme, Stuttgart, Germany, Suppl., Vol. 1, pp. 1–120. Marschner, H. 1995. Mineral nutrition of higher plants. Academic Press, London, UK. McCune, D.C., and Weinstein, L.H. 2002. Effects of fluorides. In: Bell, J.N.B., and Treshow, M., eds. Air pollution and plant life. Wiley, Chichester, UK. Meharg, A.A., Cumbes, Q.J., and MacNair, M.R. 1993. Pre-adaptation of Yorkshire Fog, Holcus lanatus L. (Poaceae) to arsenate tolerance. Evolution 47, 313–316. Morre´, J.D., Rogers, B.J., and Gamble, R. 1965. Promotion of plant growth by long-chain alcohols and organic solvents. Phyton 22, 7–12. Myers, R.H. 1971. Response surface methodology. Allyn and Bacon, Boston, MA. Myers, R.H., and Montgomery, D.C. 1995. Response surface methodology: Processes and product optimization using designed experiments. Wiley, New York, NY. National Academy of Sciences. 1971. Fluorides. National Academy Press, Washington, DC.

152

Hans-Ju¨rgen Ja¨ger and Sagar V. Krupa

Neil, L.J., Ormrod, D.P., and Hofstra, G. 1973. Ozone stimulation of tomato stem elongation. HortScience 8, 488–489. Odum, E.P., Finn, J.T., and Franz, E.H. 1979. Perturbation theory and the subsidy-stress gradient. Bioscience 29, 349–352. Percy, K., Nosal, M., Heilman, W., Dann, T., Sober, J., and Karnosky, D. 2006. The North American ozone air quality standard: Efficacy and performance with two northern hardwood species. In: Weiser, G., and Tausz, M., eds. Critical levels of ozone: Further applying and developing the flux-based concept. Obergurgl, Austria. UNECE Workshop Report, BFW-Bericht, Vienna, Austria. Pleijel, H., and Danielsson, H. 1997. Growth of 27 herbs and grasses in relation to ozone exposure and plant strategy. New Phytol. 135, 361–367. Prasad, M.N.V. 1995. Cadmium toxicity and tolerance in vascular plants. Environ. Exp. Bot. 35, 525–545. Rajput, C.B.S., and Ormrod, D.P. 1986. Stimulation of plant growth in pumpkin by ozone. HortScience 21, 498–499. Rawlings, J.O., Lesser, V.M., and Dassel, K.A. 1988. Statistical approaches to assessing crop loss. In: Heck, W.W., Taylor, O.C., and Tingey, D.T., eds. Assessment of crop loss from air pollutants. Elsevier Applied Science, London, UK, pp. 389–416. Sanders, G.E., Robinson, A.D., Geissler, P.A., and Colls, J.J. 1992. Yield stimulation of a commonly grown cultivar of Phaseolus vulgaris L. at near-ambient ozone concentrations. New Phytol. 122, 63–70. Schulz, H. 1888. U¨ber Hefegifte. Pflu¨gers Archiv Gesamte Physiol. Menschen Tiere 42, 517–541. Selye, H. 1936. A syndrome produced by various noxious agents. Nature 138, 32–34. Shannon, M.C., and Grieve, C.M. 1999. Tolerance of vegetable crops to salinity. Sci. Hortic. 78, 5–38. Southam, C.M., and Ehrlich, J. 1943. Effects of extract of western red-cedar heartwood on certain wood-decaying fungi in culture. Phytopathology 33, 517–524. Stebbing, A.R.D. 1981. Hormesis—Stimulation of colony growth in Campanularia flexuosa (Hydrozoa) by copper, cadmium and other toxicants. Aquat. Toxicol. 1, 227–238. Stebbing, A.R.D. 1982. Hormesis—The stimulation of growth by low levels of inhibitors. Sci. Total Environ. 22, 213–234. Stebbing, A.R.D. 1987. Growth hormesis: A by-product of control. Health Phys. 52, 543–547. Treshow, M., and Harner, F.M. 1968. Growth responses of Pinto bean and alfalfa to sublethal fluoride concentrations. Can. J. Bot. 46, 1207–1210. Tu, C., and Ma, L.Q. 2005. Effects of arsenic on concentration and distribution of nutrients in the frond of the arsenic hyperaccumulator Pteris vittata L. Environ. Pollut. 135, 333–340. US EPA. 2006. Available at http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=149923 van Ewijk, P.H., and Hoekstra, J.A. 1993. Calculation of the EC50 and its confidence interval when subtoxic stimulus is present. Ecotoxicol. Environ. Saf. 25, 25–32. von Bertalanffy, L., Beier, W., and Laue, R. 1977. Biophysik des FlieXgleichgewichts. Vieweg, Braunschweig, Germany. Wiedman, S.J., and Appleby, A.P. 1972. Plant growth stimulation by sublethal concentrations of herbicides. Weed Res. 12, 65–74. Woodwell, G.M. 1975. The threshold problem in ecosystems. In: Levin, S.A., ed. Ecosystem analysis and prediction. Society of Industrial and Applied Mathematics, Philadelphia, PA, pp. 9–21. Yeo, A.R. 1983. Salinity resistance: Physiology and prices. Physiol. Plant. 58, 214–222.

Developments in Environmental Science, Volume 9 Allan H. Legge (Editor) Copyright r 2009 Elsevier Ltd. All rights reserved. ISSN: 1474-8177/DOI:10.1016/S1474-8177(08)00207-6

153

Chapter 7 Evaluating Ozone Effects on Growth of Mature Forest Trees with High-Resolution Dendrometer Systems S.B. McLaughlin and Miloslav Nosal Abstract Both manual and electromechanical dendrometer techniques have been used to define growth patterns of mature forest trees at scales ranging from hourly to seasonal, and to evaluate the roles of ozone and physical climate as contributors to observed variability in growth. This chapter addresses the issues of quantifying and modeling the specific effects of ozone in the presence of co-varying influences of other important environmental variables. A variety of statistical models have been developed that provide strongly converging evidence that short-term variations in ozone exposure, although they contribute only about 2% to hourly scale variations, have strongly accumulative effects over the growing season. Model predictions of growth loss in the range of 50% in high ozone years agreed well with observed growth changes with similar levels of ozone increase for selected sample trees. Observed and predicted growth losses in a high ozone year greatly exceed levels typically assumed for mature forest trees based on controlled studies with seedling trees.

7.1. Introduction

The predicted effects of a warming climate include both increased growth and latitudinal range of some forests, but also an increased frequency of declining health of others in response to increased importance of biotic and abiotic stresses (International Panel on Global Climate Change (IPCC), 2001). Among the components of climate change that may play a significant role in such changes are regional air pollutants, such as ozone (O3). Corresponding author: E-mail: [email protected]

154

S.B. McLaughlin and Miloslav Nosal

At current ambient levels that occur across large regions of the agroindustrial world, ozone affects many physiological processes that are important to water, carbon (C), and nutrient cycles, which are fundamental to forest physiological and ecosystem function (McLaughlin & Percy, 1999; Ska¨rby et al., 1998). Recent efforts to consider the physiological effects of ozone and other biogeochemical changes in forest production models (Felzer et al., 2004; Ollinger et al., 2002) suggest that ozone limitations on forest production may largely offset projected positive responses to both increasing carbon dioxide (CO2) and nitrogen (N) deposition. Such models provide a useful vehicle for incorporating current levels of understanding of the integrated effects of multiple components of a changing environment on forest growth processes. However, there are major uncertainties regarding some of the biological assumptions on which they have necessarily had to rely to date. Controlled exposure studies with tree seedlings and saplings have contributed in many important ways to current understanding of mechanisms of effects of regional air pollutants on forest growth processes. However, a major need in assessing likely effects on forest productivity and health has been documentation of responses of mature forest trees under natural conditions (Fuhrer et al., 1997; McLaughlin et al., 2002; Samuelson & Kelly, 2001). Efforts to scale ozone impacts from seedlings to large trees have been largely unsatisfactory (Samuelson & Kelly, 2001). There are many reasons for this, including differences in energy production budgets, canopy: root ratios and architecture, and carbon allocation patterns between small and mature trees. Scaling efforts have highlighted the need for improved data on physiological responses from mature trees in natural settings (Fuhrer et al., 1997; Samuelson & Kelly, 2001). Because of high maintenance costs of large trees, net annual productivity of forests is typically only 10–15% of gross primary productivity (Kira, 1975). This leaves little margin for loss of gross production capacity, and makes larger trees very susceptible to external stresses. Current estimates of ozone-induced reductions in forest productivity of 1–10% in Europe (Broadmeadow, 1998) and the USA (Chappelka & Samuelson, 1998) have to date been largely based on seedling and sapling responses under controlled environments. In addition to scaling issues, there are some fundamental uncertainties in characterizing expected changes in forest water use under increases in ozone levels. Here, the problem of defining appropriate responses of plant stomata to ozone exposure is critical to understanding how ozone influences forest water use (Maier-Maerker, 1998; Mansfield, 1998) and forest growth responses to drought. Generalized water use patterns derived from small plants in small containers artificially exposed to high ozone concentrations

Evaluating Ozone Effects on Growth of Mature Forest Trees

155

can significantly misrepresent responses of mature trees in native soils. Stomatal closure and the associated protection of plants from drought and additional ozone exposure can occur in chamber environments at high ozone exposure levels (Tingey & Hogsett, 1985). By contrast, responses of foliage from larger trees (Grulke et al., 2004; Maier-Maerker, 1997; Maier-Maerker & Koch, 1992) as well as saplings and seedlings (Lee et al., 1990; Pearson & Mansfield, 1993; Reich & Lassoie, 1984; Ska¨rby et al., 1987; Wallin & Ska¨rby, 1992) provide increasing evidence that stomatal control of transpiration may be reduced following ozone exposure. Such changes can lead to either greater stomatal apertures and/or delayed stomatal closure at night. The results of reduced stomatal function would include increased forest water loss and the amplification rather than diminution of the effects of drought. These results have been supported by field-based empirical models of forest growth that suggest that drought effects are enhanced, not reduced by concurrent stresses posed by ozone and reduced water availability (McLaughlin & Downing, 1995; Peterson et al., 1995; Zahner et al., 1989). The uncertainties imposed by inadequate understanding of the nature and direction of stomatal responses to ambient ozone have promulgated significant uncertainties in efforts to model and predict forest responses to ozone. For example, Ollinger et al. (1997) noted that if ozone increased rather than decreased stomatal conductance, as assumed in their model, losses in forest productivity across the northeastern United States would have increased from 7% to 11%. The advent of advanced dendro-ecological techniques has greatly improved the measurement capabilities for examining responses of mature forest trees to interacting environmental stresses, including regional pollutants (McLaughlin et al., 2002). Among the techniques with greatest biological and statistical potential, high-resolution dendrometry has represented a significant recent advance. The term ‘‘dendrophysiology’’ has been applied to the analysis of high-frequency measurements (rhourly intervals) of tree stem expansion, contraction, and growth in response to contributing environmental variables (McLaughlin et al., 2002). Such techniques rely on electromechanical measurements of the patterns of shrinking and swelling associated with diurnal and seasonal patterns of water use and growth by trees (Herzog et al., 1995; McLaughlin et al., 2003; Zweifel & Ha¨sler, 2001a, 2001b; Zweifel et al., 2005). Because of their short time resolution, such techniques can examine responses to individual stress events within a growing season and thereby provide important insights into trees’ capacity to respond to repeated signals of both natural and anthropogenic origin within the same year. Here, we have included ozone and sap flow among the variables considered in field studies aimed at better

156

S.B. McLaughlin and Miloslav Nosal

understanding the importance of variations in ozone levels in influencing growth and water use by mature forest trees.

7.2. Materials and methods 7.2.1. Measurement systems

In these studies, we have used a combination of manual and automated dendrometers distributed across three diverse forested plots to characterize variations in tree growth and tree water use in response to seasonal and annual variations in ozone and meteorological variables. Our hypothesis was that ambient ozone levels would increase water stress by reducing trees’ capacity to maintain adequate moisture status during diurnal cycles of water loss and recovery. The forested sites (Oak Ridge (OR), Twin Creeks (TC), and Look Rock (LR)) were located near or in the Great Smoky Mountains National Park and were at elevations ranging from 250 m (OR) to 750 m (LR). They represented diverse stand histories (severely disturbed in 1999 (OR) to undisturbed for W65 years (TC)), and productivity levels (mesic cove hardwood (TC) to a more xeric higher elevation ridge site (LR)). Previous analyses suggested that episodic ozone exposures of 1–3 days could interact with soil moisture stress to cause short-term reductions in stem expansion of loblolly pine (Pinus taeda) at the OR site (McLaughlin & Downing, 1996). However, from these studies, the mechanisms of response could not be identified because of the lower frequency of the manual measurements employed and the absence of direct measures of water use by subject trees. The present study combined manual and high-resolution dendrometer systems. Manual measurements of circumference changes (see McLaughlin & Downing, 1996) at approximately 2-week intervals were recorded for a total of 79 trees across all sites. These were linked to electromechanical measurements of radius changes at 30 min intervals with automated bands (Agricultural Electronics, Tucson, AZ) (McLaughlin et al., 2003) for six trees representing four species at the high-elevation LR site. Sap flow was measured at the LR site at 30 min intervals using temperature conductance probes (Wullschleger & King, 2000). Soil water content was measured using temperature conduction rate through buried ceramic probes (Campbell Scientific, Inc.) buried at 10 and 22 cm depths, 1 m from the base of individual sampled trees. Meteorological data for each site (see Table 7.1) were obtained from ongoing monitoring systems maintained for longer-term ecological or air quality studies associated with each of the three sites.

157

Evaluating Ozone Effects on Growth of Mature Forest Trees

Table 7.1. Growth and environmental conditions during three years of studies of growth patterns of mature trees at the Look Rock site in East Tennessee, USA Parameter

Environmental conditions 2001

Rainfall (mm)

D121–180 D181–240

Temperature (1C)

D122–180 July (D181–213)

Vapor pressure deficit (Average 12 h maximum)

D134–275

Palmer Drought Severity Index

D92–305 Monthly minimum

203 277 18.6 23.9 0.79

July (D181–213) Ozone exposure

Sum 060a CumMaxh-60ppbb AOT40c AOT60c

ppmh ppbh ppmh ppmh

2002 228 258 18.8 25.2 1.12

0.22 May (0.03) 0.27

0.54 August (1.83) 1.11

147 764 62.5 11.5

171 1918 78 24.1

2003 315 340 18.0 24.4 0.43

4.56 May (4.13) 4.94 89 776 61.6 11.7

a

Sum of ozone levels at or above 60 ppb summed for DOY 91–304. Cumulative of the daily maximum hourly ozone concentration – 60 ppb. c These are true accumulations of hourly ozone concentrations over the respective threshold values of 40 ppb and 60 ppb, respectively, for DOY 91–304. b

Our analytical approach in these studies was to use the manual dendrometer data to describe the similarities and differences among species and years in the seasonal patterns of growth for a larger sample of trees within the area. High-resolution data for the smaller subset of six instrumented trees at LR were then used to test for significance of influences of environmental variables on observed seasonal growth patterns, and to develop growth-models as tools to better understand component influences on seasonal growth patterns. Because of interrelationships among environmental variables that follow solar-driven diurnal cycles, an important component of our analyses have been the development of statistical models to separate and quantify the effects of these variables on stem increment. Thus, this chapter emphasizes the nature of that variability and the experimental and statistical methods we have applied to evaluate its effects on stem growth. Other aspects of our study, including ozone-induced changes in sap flow, soil moisture status, and stream flow have been reported elsewhere (McLaughlin et al., 2007a, 2007b).

158

S.B. McLaughlin and Miloslav Nosal

7.2.2. Modeling changes in stem increment in response to environmental variables

We have approached the challenge of modeling stem increment with two primary objectives. First, we wanted to develop modeling techniques that quantified the specific and relative influences of contributing environmental variables, including ozone, on seasonal growth dynamics. Our second objective was to develop models that would be useful as tools in projecting the effects of changes in ozone on growth of mature forest trees across time and space. The following techniques are presented here: (1) Stepwise regression analyses at daily and hourly scales using standard stepwise multiple linear regression. (2) Time series analyses employing multivariate linear regression techniques. (3) General additive models (GAM) that involved curvilinear modeling techniques. For regression analyses, both individual tree data and data averaged across trees were considered in formulating response models. A statistical threshold of Pr 0.05 was used in selecting both predictor variables and acceptable regression equations in these analyses. The environmental variables examined included solar radiation, vapor pressure deficit (VPD), temperature, relative humidity (RH), and rainfall. Analyses were focused principally on six trees, including red oak (Quercus rubra; RO30 and RO11), hickory (Carya sp.; H21), pitch pine (Pinus rigida; PP4), and chestnut oak (Quercus montana; CO26 and CO28). 7.2.2.1. Stepwise multiple regression

Stepwise linear regressions were developed with both daily averaged data and hourly data to examine sequential contributions of each component variable to the regression equation after the effects of other significant variables most strongly correlated with stem increment were considered. A wide variety of environmental and response variables and time steps were examined in initial sensitivity analysis. Analysis of daily averaged data was followed by analyses of responses at the hourly scale. For daily averaged data, the strongest predictive models were developed with 3-day running averages of stem increment in response to environmental data during 1–3 days preceding the reference day. Empirical models of stem increment developed in this way were highly significant statistically for the six trees examined. The R2 values were typically in the range of 0.21–0.52, and rainfall and VPD were typically the strongest predictors (data not shown). Ozone typically had a negative effect on daily growth increment, which was highly significant for only two of six trees examined (PP4 and CO28). This approach produced relatively strong models based on

Evaluating Ozone Effects on Growth of Mature Forest Trees

159

the coefficient of determination (R2) values, but embodied the inherent risks associated with averaging strongly cross-correlated variables over time. Hourly data were examined next, using changes in stem increment in each hour relative to hourly values and/or changes in candidate independent environmental variables. Sensitivity analysis indicated that among the characterizations of ozone dose, the change in the 3 h average ozone exposure level beginning 2 h before the hour of stem increment measurement was typically the strongest predictor of changes in stem increment. Inclusion of only times when 3 h ozone averages were increasing was found to be essential to strong model performance. This suggested that periods of increasing ozone exposure were producing short-term adverse effects that were not reversed during short intervals of declining ozone levels immediately following those episodes. In addition, a derived variable, the product of ozone concentration (O3W 60 nL L1 or ppb and VPD), was identified as a useful indicator of hourly changes in stem increment. In these exploratory analyses, environmental data were treated both as state variables and as differenced hourly changes to evaluate the relative importance of both magnitude and rate of change as predictors of hourly changes in stem increment. Both individual tree and multi-species models were developed from these analyses. 7.2.2.2. Best multiple regressions of hourly time series

Time series of hourly measurements of ozone and meteorological variables monitored during the 2002 growing season at LR were initially examined together with tree increment responses. The purpose was to build a ‘‘best’’ dose–response model for assessment of ozone impact on tree growth. The optimal properties of such a model required that it should work uniformly for all investigated trees, capture the essence of ozone impact, and be statistically highly significant for all investigated trees. The Best Regression Algorithm was used to assess a great variety of models, but it became obvious that different trees have significantly different responses to ozone under varying environmental conditions. Spectral and Coherence Time Series analysis (Chambers & Hastie, 1992) demonstrated that all predictors were highly cross-correlated, with solar radiation being the driving force for other predictors displaying strong 12 and 24 h periodicities. Furthermore, it became apparent that tree response to ozone is significantly influenced by VPD level, particularly highest VPD levels. In order to develop a uniformly optimal predictive model, the data were initially stratified into hours with low VPD (VPDo1.2) and high VPD (VPDW 1.2). The low VPD stratum represented about 80% of the data or about 2032 hourly measurements. The general form of this

160

S.B. McLaughlin and Miloslav Nosal

model was: Y ¼aþ

p X

bj X j þ 

(1)

j¼1

where Y is the tree response in terms of hourly stem increment and Xjs are the predictors defined earlier. 7.2.2.3. General additive model (GAM) of hourly time series

The general tree response to ozone and meteorological variables is in principle non-linear. Despite the fact that the initial models of tree response to environmental predictors were found to perform very well, they must be considered only as a first linear approximation of a general non-linear model. For this purpose a GAM was investigated. Unlike the previous model, the GAM is intrinsically non-linear and determines the best fit by fitting and integrating a series of parabolic splines to the seasonal response surface (Hastie & Tibshirani, 1990). The predictors and response variables in this model were identical to those in the previous regression model. The GAM model has the following general form: ZðYÞ ¼ a þ

p X j¼1

bj X j þ

p X

f j ðX j Þ þ 

(2)

j¼1

where Y is the response variable, given by the differentiated time series of the monitored hourly tree radii; Xj are the predictors: cumulative AOT60, VPD, temperature, solar radiation, and rainfall; fj are cubic splines of the above named predictors, representing non-linear relationship between the response and predictors; a is the regression intercept; bj the regression coefficients corresponding to predictor Xj; and Z the identity link function. 7.3. Results 7.3.1. Inter-annual and intra-annual variations in environmental conditions

High variation in both ozone exposure and other associated environmental variables typically occurs both within and between growing seasons in the southeastern United States This variability was important to our efforts to characterize short-term responses of tree growth and physiology to environmental stress. A summary of variations in rainfall, temperature, VPD, Palmer Drought Severity Index (Palmer, 1965), and ozone exposures during the three years of measurement encompassed by this study is included in Table 7.1. Notable features in the 3-year study interval

161

Maximum Hourly Ozone Exposure (ppb)

Evaluating Ozone Effects on Growth of Mature Forest Trees 140 120 100 80 60 40 Look Rock Oak Ridge Cades Cove

20 0

160

170

180

190

200

Cumulative Daily O3 Exposure (MaxH/d-60ppb)

a

b

210

220

230

240

250

260

270

Day of Year 2500 2000 2001

1500

2002 2003

1000 500 0 -500 120

140

160

180

200

220

240

260

280

Day of Year

Figure 7.1. (a) Comparison on daily maximum hourly ozone exposure at three sites in 2002. (b) Seasonal variations for 2001–2003 at the Look Rock site in accumulative daily maximum hourly exposure levels W60 ppb. Respite periods when O3 exposure was o60 ppb have been subtracted to indicate potential recovery periods.

included comparable rainfall in 2001 and 2002 with much higher ozone levels in 2002, combined with a mild mid-season drought. Ozone levels in 2001 and 2003 were very similar, but rainfall was more abundant in 2003. Seasonal variations in the maximum daily ozone exposure concentration among the three sites over the growing season for 2002, the highest ozone exposure year, are shown in Fig. 7.1a. This figure demonstrates two

162

S.B. McLaughlin and Miloslav Nosal

important features of ozone exposure in the southeastern United States. First, highest multi-day exposure episodes were similarly timed at the three sites because they are synchronized by synoptic weather patterns; and second, episodic high exposures are followed by much lower exposures in intervening periods, during which vegetation may experience some recovery or at least respite from stress. Apparent differences in ozone exposure levels between years depended very much on the mode of quantifying ozone exposure. Using 2001 as a baseline, ozone was determined to have increased from as little as 25% in 2002 (based on accumulated dose over threshold 40 ppb (AOT40)) to as much as 151% for an accumulative maximum daily dose function that considered plant recovery potential during lower stress (o60 ppb O3) intervals (McLaughlin et al., 2007a). We have graphed the accumulative dose for 2001–2003 to contrast the acceleration and relaxation of stress associated with intraseasonal variations in ozone exposure during the three years of this study (Fig. 7.1(b)). These curves effectively identify the substantial difference between the three study years in the timing of both periods of maximum stress, defined by rapidly increasing exposure to high ozone levels, and potential recovery, where intervening maximum daily levels were o60 ppb.

7.3.2. Intra- and inter-annual variations in growth rates

Seasonal growth data collected with the manual dendrometers documented a significant drop in tree growth rates in 2002, compared with 2001 and/or 2003. For red oak, a species that we will follow through several analytical stages, data in Fig. 7.2 show that growth rates in 2002 began to drop below those in 2001 and 2002, beginning around DOY 145, and differences accelerated in an accumulative fashion thereafter, leading ultimately to a growth reduction of about 50%. Thus, for this species, these curves identify a growth difference and the point at which the tree began to experience stresses that cause a growth reduction in 2002. Data for all other species examined, with the exception of chestnut oak, showed a growth reduction of around 25–75%. This was noted at all three sites (Table 7.2) and began at about the same time for all sites and species. From a review of the environmental data summarized in Table 7.1 and Fig. 7.1, it was apparent that growth slowdown in 2002 began at a time of comparable physical climatic conditions between years and at a time of rapidly increasing exposure of trees to ozone (Fig. 7.1(b)). Thus, from these data, ozone can be seen as a possible contributor to the consistent slowdown across sites. However, at this scale of analysis, there was only one readily definable period of departure in the growth curves for 2002

163

Evaluating Ozone Effects on Growth of Mature Forest Trees 16

Circumference Increment (mm)

14 12 10 8 6 4 2

2001 2002 2003

0 -2 80

100

120

140

160

180 200 Day of Year

220

240

260

280

Figure 7.2. Comparison of circumference growth of RO11 over three years at Look Rock based on manual band data.

with the alternate years, thus only one obvious temporal marker can be found from this comparison and repeatability cannot be assessed By examining the automated dendroband data for 2002 and contrasting growth rates among trees, one can see that there are many periods of increasing and decreasing rates of stem expansion when viewed at hourly averaged (Fig. 7.3) timescales. The opportunities for associating changes in growth rate with corresponding sets of environmental data within a year go up dramatically as the dynamics of growth are more clearly resolved. This is particularly true of the hourly scale data because one sees that variations in the normal diurnal cycle of stem expansion and contraction during concurrent growth and stem hydration cycles (McLaughlin et al., 2003) are amplified for some trees in association with multi-day stress cycles. In Fig. 7.3, we have superimposed the timing of the highest ozone exposure episodes on the seasonal growth curves, and one can see amplification of both diurnal cycles and periods of slowdown in stem expansion and/or stem contraction during or immediately following these high ozone episodes. In examining hourly scale increment data for the two red oaks in Fig. 7.4, RO11 and RO30 respectively, one can detect significant differences in these patterns even for two trees of the same species. Clearly, RO11 and RO30 had differences in relative sensitivity to early and late-season stresses that cause a mid-season reversal in relative growth rates.

164

S.B. McLaughlin and Miloslav Nosal

Table 7.2. Differences in annual circumference growth among tree species at three mixed species sites in East Tennessee, USA Site

Look Rock Elevation 800 m

Twin Creeks Elevation 700 m

Oak Ridge Elevation 300 m

Species

Annual circumference growth % change from baseline (100)a

Yellow Poplar Bottom site Ridge site Red Oak Pine Hickory Chest. Oak N

18 3 11 6 2 2 42

Yellow Poplar Red Oak Pine Hemlock S. Maple R. Maple B. Cherry N

9 3 5 3 3 4 7 25

Yellow Poplar

12

2001

2002

2003

100 100 100 100 100 100 Negative responses All responses

24 42 58 48 14 72 37 19

38 42 13 2.9 30 55 8 2.5

62 44 17 22 64 59 76 49

100 100 100 100 100 100 100

50

8

100

a

2001 is used as the baseline year for both Look Rock and Oak Ridge sites, whereas 2003 represents the baseline (‘‘control’’) year for Twin Creeks, at which measurements were not initiated until 2002. Relative ozone exposures (AOT60) at Look Rock and Oak Ridge sites were 24.1 PPMH, and 18 PPMH, respectively. The remote Cades Cove site, a surrogate for Twin Creeks, experienced 6 PPMH of ozone exposure in 2002.

In examining hourly scale data, it is clear that there are strong diurnal trends in radiation, temperature, and VPD that co-vary with ozone and with the diurnal cycles in stem expansion. Diurnal patterns of radiation, VPD, ozone, and stem increment (Fig. 7.5) for both high and lower ozone days (DOY 154 and 182, respectively) reveal obvious differences in the midday contraction and recovery cycle that reflect the tree’s ability to match water demands from sap flow with water uptake by roots. Recovery of stem water potential at night is critical to stem growth, because cell division and cell expansion are very sensitive to cell hydration (Hsaio et al., 1976). In Fig. 7.5, one can see that on DOY 182, the tree experienced a net loss of increment over the 24 h cycle due to an incomplete recovery of stem increment from midday losses.

165

Radial Increment Ratio to Seasonal Midpoint

Evaluating Ozone Effects on Growth of Mature Forest Trees

2.0 1.5 1.0 0.5 Ppine 4 RO 30 Ch Oak 26

0 -0.5 121

130

140

150

160

170

180

190

200

Hick 21 RO11 Ch Oak 28

210

220 227

Day of Year Figure 7.3. Comparison of seasonal growth trends of six trees at Look Rock based on hourly scale measurements. Note superimposition of highest O3 exposure days in Fig. 7.1b.

1.8

Circumference Increment Ratio Normalized to Series Midpoint

1.6 1.4 1.2 1.0 0.8 0.6 0.4 RO 30 RO 11

0.2 0 -0.2 121

130

140

150

160

170

180

190

200

210

220 227

Day of Year Figure 7.4. A contrast in hourly scale time series of radial increment growth between RO11 and RO30 revealed responses to environmental stimuli that are synchronous but very different in magnitude.

Stem Growth (mmC *100); VPD (mPa); Radiation (w/m2); Ozone (µl/1000l)

166

S.B. McLaughlin and Miloslav Nosal 140 120 100

R11Grow VPD*50 RAD/20 O3

80 60 40 20 0 0 2 4 6 8 10 12 14 16 18 20 22 24 0 2 4 6 8 10 12 14 16 18 20 22 24 Day and Hour - DOY154 (left) and DOY182 (right)

Figure 7.5. Hourly scale variations in O3, VPD, solar radiation, and stem radial increment on two days with high radiation, ozone, and VPD during June 2002 demonstrate the issues of environmental covariance inherent in these data. Note the failure of stem increment to reach its pre-dawn levels on DOY 182.

Careful characterization of temporal variations of contributing environmental variables is critical to developing strategies to quantify their effects on stem growth. We have included a wide range of variable formulations in these modeling exercises in an effort to capture the biologically relevant features of environmental effects on growth trends. This was especially true with ozone exposure dose, because of its known phytotoxicity and the importance of capturing potential lagged responses. Partial correlations determined in initial multiple regression analyses indicate that the specific influence of ozone on stem growth was uniformly negative among the species examined. 7.3.3. Empirical models of hourly and seasonal stem growth 7.3.3.1. Stepwise multiple regression models

The initial model of changes in hourly stem increment in response to environmental stimuli considered a total of 10 variables as potential contributors to hourly changes in stem increment. Only changes in hourly ozone levels (differenced values) were considered at this stage. Four formulations of ozone exposure (hourly concentration, change in hourly O3 W60 ppb, cumulative AOT60, and change in 3 h O3) were included. Radiation, rainfall, temperature, VPD, and RH were included as hourly state variables. For RO30, which we will follow through sequential modeling

167

Evaluating Ozone Effects on Growth of Mature Forest Trees 12

Circumference Growth (mm)

10

8

6

4 RO30MC9 RO30CGr RO30MC6

2

0 121

130

140

150

160

170 180 Day of Year

190

200

210

220 227

Figure 7.6. A stepwise regression model of seasonal accumulated circumference growth based on measured hourly radial increment identified significant influences of several environmental variables, including hourly changes in 3 h AOT60 exposure dose. Actual R030 circumference growth (RO30CGR) is compared with simulated growth based on six significant variables (RO30C9) and 10 significant variables (RO30C10) selected by stepwise regression from 10 and 17 candidate variables sets, respectively.

iterations, the hourly-based empirical model identified six significant variables from the 10 variable candidates and had a predictive R2 of 0.28 (Pr0.04, n ¼ 2534). Both ozone (AOT60), and VPD had significantly negative influences on hourly scale stem increment. Further addition of values of hourly change of climate variables and one interaction term (the product of hourly ozone and change in hourly VPD) increased the list of independent variables to 17, from which nine were selected as significant, and improved the model R2 for RO30 to 0.325 (Pr0.04, n ¼ 2516). A comparison of simulated vs. actual seasonal growth of RO30 with both sixvariable and nine-variable models is shown in Fig. 7.6. Simulated growth with each model is shown in relationship to actual growth of RO30 at the LR site in 2002. The VPD and rainfall were the strongest predictors, contributing 83% of the R2 term of the final model. Ozone (as O3*VPD) contributed only about 5%, but was highly significant (Po0.0001). Such sensitivity can be attributed to the high statistical power associated with over 2500 data points, and in part, to inclusion of independent variables expressed in biologically relevant dimensions of time and level.

168

S.B. McLaughlin and Miloslav Nosal

Several generalizations were derived from the iterative testing of variables and model performance at this early stage of model development. Increasing the list of environmental variables from 10 to 17 and including hourly differentials improved the model R2 by 5% for RO30. Clear differences in significant variable sets were found among species and even trees within species. However, models of averaged hourly increment across six trees representing four species were highly significant, had R2 values comparable to those for individual trees, and identified the same general variables, including ozone, as significant. Testing of model relationships over shorter time intervals (DOY 121–160, 121–196 vs. DOY 121–227) indicated that models of similar or greater strength could be developed around much less than a full season of data. This suggested that intra-seasonal contrasts might be useful in future analyses. At this stage, however, the need to apply modern time series techniques to more effectively address both serial correlations among variables and non-linear response surfaces led us to test more sophisticated time series techniques as described later. 7.3.3.2. Time series multiple regression models

As in previous analyses, many ozone parameters were explored as potential predictors of ozone impact on tree growth. Comparisons of time series model performance with various parameters indicated that systematically the best overall ozone parameter for assessment of ozone impact in time series analyses was seasonal cumulative sum of ozone concentrations exceeding 60 ppb, which we refer to as CumAOT60. Interestingly, an analogous CumAOT40 performed similarly well as a predictor. Table 7.3 presents the main results of the regression (REG) Table 7.3. Summary of performance characteristics among trees for multiple regressions of hourly time series of tree responses to O3, temperature, VPD, radiation, and rain Tree species

Pitch Pine 4 Red Oak30 Red Oak 11 Chestnut Oak Hickory 21

Ozone effect

Ozone significance

Model R2 (adjusted) and significance (%)

Relative O3 contribution to the total model effect (%)

Negative Negative Negative Negative Negative

p ¼ 0.000* p ¼ 0.000* p ¼ 0.000* p ¼ 0.012** p ¼ 0.001**

32.4* 40.0* 30.7* 32.1* 38.3*

1.89 1.52 1.82 0.70 0.93

*Significance p ¼ 0.000 means the model or ozone effect is ‘‘statistically highly significant’’. **p ¼ 0.012 or p ¼ 0.001 means that the ozone effect is ‘‘statistically significant’’.

169

Evaluating Ozone Effects on Growth of Mature Forest Trees

models developed around these analyses and may be summarized as follows: (i) All regressions were very highly significant (P ¼ 0.001) and thus the GAM model provided a uniformly optimal and statistically highly significant predictive tool for assessment of ozone and meteorological variables on growth of the four tree species examined. (ii) All regression coefficients for ozone effect prediction (quantified as CUMAOT60) on tree growth were negative and statistically significant. Thus, this model clearly indicated that the ozone effect on the tree growth was systematically negative and statistically significant. (iii) The adjusted coefficient of determination R2 for the models of individual tree performance ranged uniformly between 31% and 40%, and thus, demonstrates quite high predictive power of this model and very good fit of the model. (iv) The ozone contribution to the overall power of the model to predict hourly increment was in the order of about 2%, and all contributions to seasonal stem growth were negative.

7.3.3.3. Generalized additive model

The GAM, like the time series multiple regression model (REG) described earlier, was initially applied to the 80% of the hourly data for which VPD was r1.2. The application of the GAM model significantly improved model performance over the REG model in terms of goodness of fit, and provided significantly greater predictive power for empirically derived treelevel models. The improved R2 values of the non-linear models developed with GAM techniques, compared with REG models, are apparent in Table 7.4. However, this improvement occurred at the cost of a significantly greater degree of complexity and a massive computational effort. Table 7.4. Comparison of model performance between multiple regression and generalized additive models (GAMs) Tree species Pitch Pine 4 Red Oak 30 Red Oak 11 Chestnut Oak Hickory 21

Regression R2 (%)

GAM R2 (%)

32.5 40.1 30.7 32.3 38.4

46.4 49.0 45.5 43.9 50

170

S.B. McLaughlin and Miloslav Nosal 2500 RO30 Monitored

2000

RO30 REG Predicted LVDT RO30 GAM Predicted LVDT

LVDT

1500

1000

500

0 Day of Year

Figure 7.7. Time series regression analysis was focused on 80% of the hourly data to identify significant effects of ozone and other environmental variables during times when VPD was Z1.2. This model for seasonal radius growth of RO30 focused on describing stem expansion during the more favorable periods following the highest water stress levels. These were associated with the highest levels of VPD, O3, and temperature. LVDT units are in microvolts, where 1 mV is approximately 0.8 mm.

Fig. 7.7 presents graphical comparisons of goodness of fit between observed growth data, and the predictions of seasonal growth from both the REG and GAM. A visual inspection clearly shows that the monitored data and GAM predictions are in very close agreement. However, restricting analyses of both the REG and GAM analyses to the 80% of the dataset for which VPD was r1.2 created some issues that had to be addressed. Analysis of environmental and growth data with respect to this VPD threshold reveals that at VPDo1.2, average ozone, temperature, and radiation are substantially lower. However, for all species except chestnut oak, a drought-tolerant species, and average hourly stem increments were strongly positive at VPD r 1.2 and negative at VPD Z 1.2, where moisture demands are the highest. So, although a less variable dataset is created that is more easily evaluated statistically with this approach, the meaning of the results must be considered as well. Thus, simulated seasonal increment patterns based on this stratified dataset, although they did a good job of predicting actual growth during these lower stress periods (Fig. 7.7), greatly overestimated observed seasonal growth for the entire dataset (season). A solution to this dilemma resulted from exploratory analyses with the stepwise regression analyses, from which it was determined that

171

Evaluating Ozone Effects on Growth of Mature Forest Trees

4,000

Monitored Growth Predicted Growth in Ambient O3

Tree Radial Increment (µm)

Predicted Growth at half AOT60 Predicted Growth at half O3

3,000

2,000

1,000

0

120

130

140

150

160

170

180

190

200

210

220

230

Day of Year Figure 7.8. A generalized additive model (GAM) was used to address interactive curvilinear effects of environmental variables on seasonal circumference growth at Look Rock in 2002. This approach included all hourly data points and averaged data for six trees, including two chestnut oaks, one pine, one hickory, and two red oaks (including RO30). Actual and simulated growth under ambient conditions in 2002 and at two reduced O3 levels (see text) are presented.

differencing the hourly environmental values improved results and reduced cross-correlations among variables. In addition, an interaction term (product of hourly AOT60 and differenced VPD) was identified as an important predictor variable that was compatible with proposed mechanisms of ozone effects on tree water loss. Inclusion of this term, combined with averaging hourly responses across trees, allowed 100% of the hourly data to be analyzed and, in addition, improved the stability and predictability of those analyses. Using six-tree (four species) average responses and the complete dataset produced strong R2 values of 0.50 and 0.60 for REG and GAM analyses, respectively. In Fig. 7.8, the multispecies GAM predictions of seasonal growth are compared with observed growth for the same six trees. In addition, the expected responses to a 50% reduction in both hourly ozone levels and AOT60 are shown (after

172

S.B. McLaughlin and Miloslav Nosal

McLaughlin et al., 2007a). The GAM model results shown in Fig. 7.8 provided a very good reproduction of the actual seasonal growth pattern and level achieved by the six-tree average in 2002. The upper curve in Fig. 7.8 represents expected growth under ‘‘control’’ conditions (AOT near 0), which approximated the AOT60 level of 0.5 mL L1 h1 in 1989 (the lowest of the past 24 years). With an increase in AOT60 from the ‘‘control’’ level to approximate average exposures in 2001/2003 (AOT60 ¼ 12 mL L1 h1) seasonal growth decreased by about 33%. Approximately doubling ambient ozone levels (vs. 2001/2003) to ambient levels in 2002 (AOT60 ¼ 24 mL L1 h1) resulted in a further growth reduction of 48% compared with those more average years. The difference between modeled growth in 2002 and modeled growth at the level of a 50% reduction in AOT60 can be compared with actual growth differences between 2002 and 2001 as one measure of both model performance and the effect that simulated differences in ozone exposure between years (2 ozone in 2002) could have in explaining observed growth differences between the two years. Model projections were of a 48% growth loss for the multi-species model in response to a doubling of AOT60 from 2001 levels. This simulated response compares favorably to the measured growth of the 40 trees at LR, which slowed by an average of 43% in 2002 vs. 2001. For the 79 trees in the study at all sites, the average reduction in 2002 was 40%. The dynamics of growth pattern divergence between 2001 and 2002 (a reduction in growth rate that became most apparent after DOY 145 in 2002) were similar to observed patterns of differences between 2001 and 2002 (Fig. 7.3). A similar pattern of divergence was also simulated with the less complex stepwise regression model (data not shown).

7.4. Discussion

We have analyzed stem increment data over 2–3 years for several tree species in natural forests at three locations. Our analysis focused on characterizing the roles of short-term variations in climatic variables, including ozone, on observed growth responses. The data collection interval included large variations in environmental conditions both within and between years examined, and subject trees exhibited wide variations in growth rates in response to these conditions. The substantial reduction in growth of most species in 2002, a high ozone year, provided us with an opportunity to compare analytical techniques in addressing the causes of this reduction as measured at different time resolutions, ranging from weekly to hourly. Manual dendrometer data, collected at 2- to 3-week intervals over the three years, provided inferential evidence that ozone was a

Evaluating Ozone Effects on Growth of Mature Forest Trees

173

contributor to slower growth in 2002, based on the timing of the reduction relative to strong increases in ozone exposure and the absence of differences in patterns of other potential contributing environmental parameters. High-resolution dendrometer data for a subset of six trees were examined over a 106-day interval in 2002 at scales ranging from daily (n ¼ 106) to hourly (n ¼ 2500) to compare the strength of various analytical approaches and the conclusions that could be derived from each. As the analysis time interval was reduced to daily and then hourly intervals, the resolution of tree growth dynamics increased dramatically, and the powers of statistical detection were significantly increased. Our analyses at multiple scales provided converging lines of evidence that ozone was a significant contributor to the significant growth reductions observed in 2002. Daily averaged data identified ozone as a negative influence on five of six trees examined using stepwise multiple regression, but effects were highly significant for only two of the six trees examined. Analysis of both complete and stratified hourly data with multiple regression and GAM were able to detect a small (about 2%) but highly significant (Po0.001) negative influence of ozone exposures on the hourly variation in stem expansion rate for all trees. This had significant effects when aggregated over the growing season. The strongest predictive models based on R2 were provided by the GAM, which addressed curvilinear responses. Empirical models developed from the stepwise multiple regression model and from the GAM were used to explore the effects of a 50% reduction in AOT60 ozone levels, a difference that paralleled the actual difference in exposure levels between 2002 and 2001 or 2003. Increases in growth predicted by these models provided comparable levels of growth to levels observed during the alternate years when ozone levels were 50% lower. Interestingly, linear models (Figs. 7.6 and 7.7) overestimated actual growth during the early season (DOY 121–160). Separate analysis of this 40-day interval using stepwise multiple regression analysis and comparison of regression coefficients with those derived over the 161–227 interval suggest that the seasonal models may have underestimated the negative effects of VPD, and to a lesser degree VPD*ozone interactions, early in the season. A higher sensitivity of trees to drier atmospheric conditions at a stage when they have just completed canopy regeneration is plausible and bears further investigation with these techniques. This period of overestimation was reduced, but not entirely eliminated by using the curvilinear GAM modeling approach (Fig. 7.6). Our analyses have used ozone dose formulations that were empirically determined by iterative testing to provide the best fit to observed growth responses at the intensively monitored LR site. For hourly scale analyses, changes in ozone exposure level (AOT 60) at an interval of 1–3 h

174

S.B. McLaughlin and Miloslav Nosal

preceding the hourly response interval were found to provide the most biologically and statistically significant growth response parameters. Our previous analyses (McLaughlin et al., 2007a) indicate that the peak hourly ozone levels during each day and perhaps a relatively few high ozone episodes during the growing season may play a dominant role in altering tree water use patterns. They also may play an important role in the growth limiting potential of the seasonal ozone exposure. For the limited time interval of this study, the relative growth responses observed at three sites in 2002 as described in Table 7.2, were generally comparable despite large differences in cumulative seasonal exposure levels at the three sites. In contrast, a comparison of the daily maximum ozone exposure levels (DMaxH) at the three study sites (see Fig. 7.1), shows that the seasonal pattern of occurrences of higher multi-day peak ozone exposures was much more comparable between sites than were seasonally averaged exposure levels. The range in seasonal average DMaxH was only 19% of the minimum among the three sites in 2002, whereas variations in AOT40 (183%) and AOT60 (293%) were much larger. Our data suggest that biological responses of mature forest trees in this study were more dependent on rates of increase in ozone exposure levels within and between seasons than on absolute levels themselves. Differences in site quality and stand characteristics will doubtlessly influence the translation of ozone exposure dose to growth responses at individual sites, however, in this study, rather similar responses were observed in 2002 across widely divergent stand conditions. Finally, the identification of an accumulative effect of ozone through integration of the small but statistically significant effects on stem increment that were detected with hourly scale analyses is of interest from a mechanistic perspective. Two primary hypotheses appear plausible: (1) increased water stress and (2) reduced accumulation of carbohydrates due to direct or indirect effects of ozone on net photosynthetic production and/or utilization. Two lines of evidence from this study support the first and underlying hypothesis for these studies. First visual and statistical examination of diurnal increment curves, such as those seen in Figs. 7.3 and 7.5, indicate that subject trees had a reduced capacity to recover predawn stem radius on days of and immediately following higher ozone exposures. Second, direct measurements of stem water flow in this study have revealed that higher ozone levels lead to increased water usage by trees (McLaughlin et al., 2007b). One mechanism for such changes is loss of stomatal control over the rate and/or duration of transpiration (Mansfield, 1998). One aspect of this loss of control can be delayed stomatal closure at night (Grulke et al., 2004; Matyssek et al., 1995). The net effect of these responses is that cell growth, one of the most sensitive

Evaluating Ozone Effects on Growth of Mature Forest Trees

175

of cell processes to moisture stress, will be more frequently reduced by water stress when higher ozone levels occur. The second hypothesis can be supported by this study as an indirect effect of increased water stress, but direct effects on carbohydrate production cannot be evaluated from our data. However, cumulative exposures to ozone at the LR site in 2002 were 234 mL L1 h (24hSum0) over the 15 April–15 October interval, well within range of the 120–200 mL L1 h (24hSum0) at which Grulke et al. (2002) measured cumulative reduction of gross photosynthetic rates of ponderosa pine (Pinus ponderosa) in California using GAM procedures. 7.5. Conclusions

Manual and automated measurements of seasonal growth dynamics of mature forest trees at three East Tennessee sites indicate that episodic high ozone levels can significantly reduce stem growth in natural forests. The influence of ozone on observed growth rates has been supported by evaluation of the timing and consistency of observed responses relative to increased ozone exposure as well as by three types of statistical models that identify highly significant influences of ozone on hourly stem increment rates. Sensitivity analyses support the concept that rates of increase in ozone exposure are an important component of phytotoxicity at both hourly and longer scales. Although ozone contributed a relatively small component (2%) of the highly dynamic processes apparent in diurnal stem expansion cycles, the effects of ozone are cumulative and can lead to losses in the range of 25–75% in annual growth capacity of individual trees in higher ozone years. We obtained similar and very useful results with three types of models that varied greatly in mathematical sophistication. However, we found that the iterative process of evaluating multiple indicators of biological stress, coupled with the capacity to describe non-linear biological responses provided the best capability to detect the small, but cumulative, effects that ozone exposure exerts on growth processes. Analyses of sub-seasonal intervals is certainly possible with the high-resolution data and may be a productive way to examine seasonal changes in sensitivity of trees to climatic stress as well as differences among species in sensitivity to those stresses. ACKNOWLEDGMENTS

This research was supported by a grant from the U.S. Forest Service, Southern Global Change Research Program through the University of Tennessee, and by the National Park Service through a grant to Auburn

176

S.B. McLaughlin and Miloslav Nosal

University. The Oak Ridge National Laboratory provided technical support.

REFERENCES Broadmeadow, M. 1998. Ozone and forest trees. New Phytol. 139, 123–125. Chambers, J.M., and Hastie, T.J. 1992. Statistical models. In: Chambers, J.M., and Hastie, T.J., eds. Statistical models. S. Wadsworth and Brooks/Cole, Pacific Grove, CA. Chappelka, A.H., and Samuelson, L. 1998. Ambient ozone effects on forest trees of the eastern United States: A review. New Phytol. 139, 91–108. Felzer, B., Kicklighter, D., Melillo, J., Wang, C., Zhuang, Q., and Prinn, R. 2004. Effects of ozone on net primary production and carbon sequestration in the conterminous United States using a biogeochemistry model. Tellus 56B, 230–248. Fuhrer, J., Ska¨rby, L., and Ashmore, M. 1997. Critical levels for ozone effects on vegetation in Europe. Environ. Pollut. 97, 91–106. Grulke, N.E., Preisler, H.K., Rose, C., Kirsch, J., and Balduman, L. 2002. O3 uptake and drought stress effects on carbon acquisition of ponderosa pine in natural stands. New Phytol. 154, 621–631. Grulke, N.E., Alonso, R., Nguyen, T., Cascio, C., and Dobro, W. 2004. Stomata open at night in pole-sized mature ponderosa pine: Implications for O3 exposure metrics. Tree Physiol. 24, 1001–1010. Hastie, T.J., and Tibshirani, R.J. 1990. Generalized additive models. Chapman & Hall, London, UK. Herzog, K.M., Ha¨sler, R., and Thum, R. 1995. Diurnal changes in the radius of a subalpine Norway spruce stem: Their relationship to sap flow and their use to estimate transpiration. Trees 10, 94–101. Hsaio, T.C., Acevedo, E., Fereres, E., and Henderson, D.W. 1976. Stress metabolism. Water stress, growth, and osmotic adjustment. Philos. Trans. R. Soc. Lond., B, Biol. Sci. 273, 479–500. International Panel on Global Climate Change (IPCC). 2001. Technical summary. Report of the Panel on Climate Change, IPCC Secretariat, Geneva, Switzerland. Kira, T. 1975. Primary production of forests. In: Cooper, J.P., ed. Photosynthesis and productivity in different environments. Cambridge University Press, New York, NY, pp. 5–41. Lee, W.S., Chevone, B.I., and Seiler, J.R. 1990. Growth and gas exchange of loblolly pine seedlings as influenced by drought and air pollutants. Water Air Soil Pollut. 51, 105–116. Maier-Maerker, U. 1997. Experiments on the water balance of individual attached twigs of Picea abies (L.) Karst. in pure and ozone-enriched air. Trees 11, 229–239. Maier-Maerker, U. 1998. Predisposition of trees to drought stress. Tree Physiol. 19, 71–78. Maier-Maerker, U., and Koch, W. 1992. The effect of air pollution on the mechanism of stomatal control. Trees 7, 12–25. Mansfield, T. 1998. Stomata and plant water relations: Does air pollution create problems. Environ. Pollut. 101, 1–11. Matyssek, R., Gunthardt-Goerg, M., Maurer, S., and Keller, T. 1995. Nighttime exposure to ozone reduces whole-plant production in Betula pendula. Tree Physiol. 15, 159–165. McLaughlin, S.B., and Downing, D.J. 1995. Interactive effects of ambient ozone measured on mature forest trees. Nature 374, 252–257. McLaughlin, S.B., and Downing, D.J. 1996. Interactive effects of ambient ozone and climate measured on mature loblolly pine trees. Can. J. For. Res. 26, 670–681.

Evaluating Ozone Effects on Growth of Mature Forest Trees

177

McLaughlin, S.B., and Percy, K. 1999. Forest health in North America: Some perspectives on actual and potential roles of climate and air pollution. Water Air Soil Pollut. 116, 151–197. McLaughlin, S.B., Shortle, W.C., and Smith, K. 2002. Dendroecological applications in air pollution research and environmental chemistry: Research needs. Dendrochronologia 20(1–2), 133–137. McLaughlin, S.B., Nosal, M., and Wullschleger, S.D. 2003. Diurnal and seasonal changes in stem increment and water use by yellow poplar trees in response to environmental stress. Tree Physiol. 23, 1125–1136. McLaughlin, S.B., Nosal, M., Wullschleger, S.D., and Sun, G. 2007a. Interactive effects of ozone and climate on tree growth and water use in a southern Appalachian forest in the USA. New Phytol. 174, 109–124. McLaughlin, S.B., Wullschleger, S.D., Sun, G., and Nosal, M. 2007b. Interactive effects of ozone and climate on water use, soil moisture content and streamflow in a southern Appalachian forest in the USA. New Phytol. 124, 125–136. Ollinger, S.V., Aber, J.D., and Reich, P.B. 1997. Simulating ozone effects on forest productivity: Interactions among leaf and stand-level processes. Ecol. Appl. 7, 1237–1251. Ollinger, S.V., Aber, J.D., Reich, P.B., and Freuder, R. 2002. Interactive effects of nitrogen deposition, tropospheric ozone, elevated CO2, and land use history on the carbon dynamics of northern hardwood forests. Glob. Chang. Biol. 8, 545–562. Palmer, W.C. 1965. Meteorological drought. U.S. Weather Bureau, Washington, DC. Pearson, M., and Mansfield, T.A. 1993. Interactive effects of ozone and water stress on the stomatal resistance of beech (Fagus sylvatica L.). New Phytol. 123, 351–358. Peterson, D.L., Silsbee, D.G., Poth, M., Arbaugh, M.J., and Biles, F.E. 1995. Growth response of big-cone Douglas fir (Pseudotsuga macrocarpa) to long-term ozone exposure in southern California. J. Air Waste Manage. Assoc. 45, 36–45. Reich, P.B., and Lassoie, J.P. 1984. Effects of low level O3 exposure on leaf diffusive conductance and water-use efficiency in hybrid poplar. Plant Cell Environ. 7, 661–668. Samuelson, L., and Kelly, J.M. 2001. Scaling ozone effects from seedlings to forest trees. (Tansley Review no. 21.) New Phytol. 149, 21–41. Ska¨rby, L., Troeng, E., and Bostro¨m, C.-A˚. 1987. Ozone uptake and effects on transpiration, net photosynthesis, and dark respiration in Scots pine. For. Sci. 33, 801–808. Ska¨rby, L., Ro-Poulsen, H., Wellburn, F.A.M., and Sheppard, L.J. 1998. Impacts of ozone on forests: A European perspective. New Phytol. 139, 109–122. Tingey, D.T., and Hogsett, W.E. 1985. Water stress reduces ozone injury via a stomatal mechanism. Plant Physiol. 77, 944–947. Wallin, G., and Ska¨rby, L. 1992. The influence of ozone on the stomatal and non-stomatal limitation of photosynthesis in Norway spruce, Picea abies (L) Karst exposed to soil moisture deficit. Trees 6, 128–136. Wullschleger, S.D., and King, A.W. 2000. Radial variation in sap velocity as a function of stem diameter and sapwood thickness in yellow poplar. Tree Physiol. 20, 511–518. Zahner, R., Saucier, J.R., and Myers, R.K. 1989. Tree-ring model interprets growth decline in natural stands of loblolly pine in the southeastern United States. Can. J. For. Res. 19, 612–621. Zweifel, R., and Ha¨sler, R. 2001a. Dynamics of water storage in mature subalpine Picea abies: Temporal and spatial patterns of change in stem radius. Tree Physiol. 21, 561–569. Zweifel, R.H., and Ha¨sler, R. 2001b. Link between diurnal stem radius changes and their relation to tree water relations. Tree Physiol. 21, 869–877. Zweifel, R., Zimmermann, L., and Newbery, D.M. 2005. Modeling tree water deficit from microclimate: An approach to quantifying drought stress. Tree Physiol. 25, 147–156.

Developments in Environmental Science, Volume 9 Allan H. Legge (Editor) Copyright r 2009 Elsevier Ltd. All rights reserved. ISSN: 1474-8177/DOI:10.1016/S1474-8177(08)00208-8

179

Chapter 8 Methods for Measuring Atmospheric Nitrogen Deposition Inputs in Arid and Montane Ecosystems of Western North America M.E. Fenn*, J.O. Sickman, A. Bytnerowicz, D.W. Clow, N.P. Molotch, J.E. Pleim, G.S. Tonnesen, K.C. Weathers, P.E. Padgett and D.H. Campbell Abstract Measuring atmospheric deposition in arid and snow-dominated regions presents unique challenges. Throughfall, the flux of nutrients transported in solution to the forest floor, is generally the most practical method of estimating below-canopy deposition, particularly when monitoring multiple forest sites or over multiple years. However, more studies are needed to relate throughfall fluxes to total atmospheric deposition, particularly in seasonally dry regions. In seasonally snow-covered regions, the distribution of atmospheric deposition and subsequent nitrogen (N) fluxes are highly sensitive to the temporal and spatial dynamics of snow accumulation and melt. Recent developments in passive monitoring techniques for throughfall and measurement of gaseous pollutants greatly facilitate monitoring of atmospheric deposition and ambient pollutant concentrations over broader spatial scales than was previously possible. Here we focus primarily on N fluxes as N is both a limiting nutrient and a pollutant in many terrestrial ecosystems, and because sulfur (S) deposition is not a widespread problem in the West. Methods suggested for estimating spatially distributed atmospheric deposition in arid and snow-dominated systems include simulation modeling, inferential method, throughfall collection, branch rinsing, N accumulation in surface soils of arid zones, and snowpack sampling methods. Applying more than one approach is often necessary to capture the various atmospheric deposition pathways and the spatial and temporal variability of N deposition. Corresponding author: E-mail: [email protected]

180

M.E. Fenn et al.

8.1. Introduction

In recent years, our understanding of nitrogen (N) deposition effects in ecosystems of western North America has increased considerably, although in some areas of the West, near large emission sources, deposition rates and ecological effects are poorly understood. In order to understand the effects of atmospheric deposition on ecosystems or to determine the effectiveness of emissions control strategies, suitable methods of measuring deposition inputs are needed. In many instances, N loading to ecosystems of western North America is either unknown or highly uncertain. This is largely because of the technical and logistical difficulties of measuring atmospheric deposition in systems that are either dominated by dry deposition or in the case of high-elevation systems, where measuring deposition is difficult because of remoteness, lack of electric power, severe weather conditions, widely fluctuating snow levels, and dilute pollutant concentrations in large volumes of precipitation (Weathers et al., 2000, 2006). Various approaches are needed to measure N deposition in arid and montane ecosystems of western North America (Table 8.1). Selection of appropriate techniques is based on the predominant climate and ecosystem conditions, the number of sites to be monitored, the accuracy needed in the measurement, the geographic scale of the area to be monitored, cost considerations, and the level of technical expertise available. Existing national deposition networks (see Section 8.2.6) are useful in tracking long-term trends in atmospheric deposition; however, data from these networks are often of limited usefulness in studies of N deposition effects in western ecosystems. The purpose of this chapter is to provide an overview of methodologies for measuring atmospheric deposition under conditions prevalent in western North America, with particular emphasis on high-elevation ecosystems that are seasonally snow covered, and on arid and semiarid ecosystems with a major dry deposition component. Information on chemical, spatial, and temporal aspects of N deposition in snow or dry deposition onto snow is another aspect that is important in evaluating ecological effects and will also be treated in this chapter. The major focus of the chapter will be on N deposition, but deposition of sulfur (S) is also briefly considered. Dry deposition methods have been reviewed previously (Lovett, 1994; Lovett & Lindberg, 1993; Wesely & Hicks, 2000); thus, our focus will be on particular challenges and methods to apply in western systems, on the use of passive samplers for gaseous pollutants and for measuring deposition in throughfall and precipitation, and on

Table 8.1. Comparative advantages and disadvantages of the atmospheric deposition methods discussed in this chapter Method

Application

Disadvantages

References Da¨mmgen et al. (2005), NADP site: http:// nadp.sws.uiuc.edu/, Erisman et al. (1994), Glaubig and Gomez (1994) (flip-top collectors that do not require electric power), Krupa (2002), Williams et al. (1998) Da¨mmgen et al. (2005), Erisman et al. (1994), Erisman and Draaijers (1995), Mosello et al. (1988)

To measure deposition in precipitation, excluding dry deposition

Is the method most amenable to comparative measurements at multiple sites, including networks

When only wet deposition is measured such as in the NADP network, total deposition is grossly underestimated in arid systems. Samplers usually require electric power. Under collection of precipitation is common, especially under windy or snowy conditions

Bulk deposition

Also used to measure deposition in precipitation, but the collector opening is continuously exposed to the atmosphere

Electric power not required; inexpensive collectors that can easily be replicated

Throughfall

Usually collected as bulk throughfall, in which the collectors are continuously open

Provides data on nutrient solution fluxes to soil. Deposition fluxes are based on surface area properties of native vegetation. Deposition fluxes include inputs from precipitation, dry deposition, and cloudwater. Total atmospheric deposition can be calculated from throughfall flux data, but the canopy budget models in use are still highly uncertain

Mostly collects wet deposition, but varying amounts of dry deposition to the sampler are also collected. Under collection of precipitation is common, especially under windy or snowy conditions. Contamination from bird droppings can be a problem Unknown amounts of deposited compounds are retained by the canopy, thus fluxes are underestimates of total atmospheric deposition. Underestimates may be greater in arid zones where long precipitation-free periods occur. Atmospheric deposition to the forest floor and understory vegetation usually not measured. Difficult to collect throughfall in arid zones with low-lying vegetation due to dust contamination

Bleeker et al. (2003), Draaijers and Erisman (1995), Erisman and Draaijers (1995), Fenn et al. (2000), Lovett and Lindberg (1993), Thimonier (1998), Weathers et al. (1995, 2006)

181

Wet deposition

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

Advantages

Table 8.1. (Continued ) Advantages

Disadvantages

Branch washing

To measure dry deposition fluxes to foliar and branch surfaces

During precipitation-free periods, deposition fluxes to native vegetation can be estimated; this is particularly advantageous in arid zones

Inferential

Used to calculate dry deposition fluxes

With adequate data collection is considered the best method of estimating dry deposition to natural ecosystems

Soil accumulation

Used to measure accumulation of drydeposited pollutants in bare soils of arid systems

Provides data on accumulation of pollutants in surface soils during dry periods. An important deposition pathway for ecosystems with bare soils

Requires considerable amount of ancillary data to calculate annual deposition fluxes to a forest stand. Not applicable if fog or a rain event occurs during the monitoring period. As with throughfall, unknown amounts of atmospherically deposited compounds are retained by vegetation, leading to underestimates of total deposition. In arid zones, during extended dry periods, branch rinsing may overestimate dry deposition fluxes Requires intensive data collection (meteorology, LAI by species, and frequent monitoring of all major pollutants). As a result, it is costly to implement at multiple sites, although the use of passive samplers can extend the number of sites monitored (Schmitt et al., 2005). Deposition velocity values used in calculations are often uncertain Mainly useful only during the dry season as precipitation leaches ions to deeper soil layer

References Fenn and Bytnerowicz (1997), Marques et al. (2001), Shanley (1989)

Erisman et al. (1994), Lovett (1994), Lovett and Lindberg (1993), Wesely and Hicks (2000)

Padgett et al. (1999)

M.E. Fenn et al.

Application

182

Method

Passive samplers

A thin soil layer in a plate for measuring dry fluxes to native soil; also soil samplers with thin soil layer glued to a disk (see Section 8.2.7) Used to obtain a timeaveraged atmospheric concentration

Ion exchange resin (IER) column samplers

Used to collect ions from bulk deposition and throughfall; ions are collected by IER columns

Simulation modeling

Simulated deposition based on atmospheric emissions, atmospheric transport and chemical transformations, and deposition processes

Only useful for dry deposition flux measurements

Padgett and Bytnerowicz (2001), M.E. Fenn (unpublished data)

Simple to use, inexpensive, does not require electric power; allows for replication at multiple and remote sites, thus providing information on spatial trends in atmospheric exposures Obviates the need for event-based or frequent collection of liquid samples, making it possible to expand the number of collectors installed in the field with reduced logistical and analytical costs Provides complete spatial coverage over the modeled domain; can simulate deposition through multiple pathways; can forecast future conditions including the effects of emissions reductions

Gives only average concentrations, which obscures diurnal and other temporal trends

Bytnerowicz et al. (2001, 2005), Krupa and Legge (2000), Tang (2001)

Cannot measure pH of throughfall or bulk deposition samples. Not designed to measure ionic concentrations of samples

Fenn and Poth (2004), Simkin et al. (2004)

Validity of results must be verified with field sampling; Models are incomplete and can have compensating errors. Elaborate statistical methods needed to compare model grid output to point-based observations

Pleim et al. (2001), Swall and Davis (2006), Tonnesen et al. (2003)

183

Can estimate deposition fluxes to native soil, even over short time periods (days)

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

Passive soil sampler

184

Table 8.1. (Continued ) Method

Application

Advantages

Disadvantages

Empirical deposition modeling

To estimate deposition over heterogeneous landscapes ranging in size from 100s to 1000s of kilometers

Over larger scales (e.g., 500 km up to continental scale) uncertainties become large

Holland et al. (2005), Weathers et al. (2006)

Snowpack sampling

Snowpack is sampled prior to snowmelt to estimate wintertime wet deposition and dry deposition to the snowpack in sites where snow accretion occurs in winter Concentrations of N and S in lichen tissue provide a surrogate measure of deposition

Based on field data; uses existing monitoring data to anchor the landscape deposition maps; results are spatially explicit, identifying hot and cold spots Best method for estimating atmospheric deposition in winter in high-elevation remote sites

Deposition estimates may be in error due to sublimation losses or snowmelt leachate losses in some sites. Logistically difficult. Dilute concentration of samples requires special care in sample collection and analysis Does not provide direct wet or dry deposition estimates; Nutrient accumulation is species specific

Clow et al. (2002), Turk et al. (2001)

Lichen biomonitoring

Provides biologically relevant data, particularly when combined with data on changes in lichen communities

References

Fenn et al. (2003a, 2007, 2008), Geiser and Reynolds (2002)

M.E. Fenn et al.

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

185

simulation and empirical modeling approaches for estimating deposition fluxes over broad landscapes. 8.2. Measuring atmospheric N deposition in arid systems

In arid systems, prolonged periods without measurable precipitation constitute the major factor influencing deposition processes and approaches to quantifying deposition. Dry deposition processes predominate under these conditions. Dry deposition occurs to all surfaces in an ecosystem, and deposition varies based on vegetation type and the amount of surface area. The major difficulty in measuring dry deposition of atmospheric pollutants is that deposition fluxes are influenced by factors such as the dynamic mix of pollutant species (each with its own range of deposition velocities and concentrations), physical and chemical characteristics of surfaces, and meteorological conditions (Weathers et al., 2006). Because of the large amount of data needed to calculate these deposition fluxes, throughfall is commonly used as an integrative technique to evaluate pollutant deposition (Bleeker et al., 2003). Throughfall is the water that washes from canopies to the forest floor. Its chemistry is a combination of wet, dry, and fog inputs through atmospheric deposition and what has been leached from the canopy. When there is little or no precipitation for successive months, no throughfall is collected. Therefore, throughfall measurements likely underestimate total deposition in arid systems to a greater degree than in mesic systems (Fenn et al., 2000). Thus, multiple approaches used in conjunction are sometimes needed to quantify total annual N deposition in western ecosystems. 8.2.1. Inferential method

The inferential method will not be covered in detail because it is well described in the literature (Erisman et al., 1994; Lovett, 1994; Lovett & Lindberg, 1993; Wesely & Hicks, 2000). In brief, the inferential method consists of calculations of dry-deposition fluxes to the ecosystem based on measurements of atmospheric concentrations of the pollutants of interest, the deposition velocities of these pollutants, meteorological conditions, the surface area, and the physical characteristics of surfaces. The inferential method is highly appropriate for arid and semiarid systems because of the importance of dry deposition in these systems. The inferential method has also been used effectively to better understand canopy interactions of dry-deposited species, and in

186

M.E. Fenn et al.

particular, to determine the extent to which throughfall measurements of deposition differ from total deposition in polluted environments because of canopy–pollutant interactions (Erisman & Draaijers, 1995; Lovett & Lindberg, 1993). Total deposition to the ecosystem can be determined by including precipitation inputs as wet deposition measurements. In areas where deposition in fog or cloud water occurs, this must also be measured if total deposition is to be determined. Deposition in fog can be important even in semiarid regions influenced by marine fogs or at high elevations where clouds come into contact with vegetation (Weathers, 1999). Data requirements are intensive when using the inferential method to quantify N deposition. Many pollutants must be measured continuously or semicontinuously throughout the year if annual inorganic N deposition inputs are to be calculated. Continuous measurements of nitric acid (HNO3) vapor and ammonia (NH3) concentrations are very difficult to make, and deposition velocities for the various pollutant–plant species combinations are often unknown, requiring the use of estimated values from the literature. Also, considerable ancillary data are needed concerning meteorological conditions and surface area properties such as leaf area index (LAI) by species (Baumgardner et al., 2002). Lastly, organic N loading is not quantified, but can be an important component of annual N loading in many montane systems of the West (Sickman et al., 2001). 8.2.2. Throughfall and branch washing

Throughfall deposition is a commonly applied method of estimating pollutant deposition to forest ecosystems (Bleeker et al., 2003) and allows for the quantification of all forms of N deposition to the forest floor. Many studies have shown that throughfall deposition of S is a reasonable estimate of total S deposition (Erisman & Draaijers, 1995; Johnson & Lindberg, 1992; Weathers et al., 2006). However, in many areas, N deposition in throughfall is a lower-bound estimate of total N deposition because of canopy uptake of atmospheric N. In mesic forests, throughfall has often been shown to underestimate total N deposition by 25–40% (Fenn & Bytnerowicz, 1997; Lovett & Lindberg, 1993). In arid zones with prolonged dry periods, foliar surfaces saturate with respect to N deposition, at least in the case of HNO3 vapor. In chamber studies, deposition velocity of HNO3 decreased with increased exposure (Cadle et al., 1991), and washable deposition reached a maximum after 2–4 weeks, whereas continued exposure did not yield any further soluble nitrate (NO3 ) (Padgett et al., 2004). Little is known about how

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

187

dry-deposition fluxes are affected by foliar surface saturation, resuspension of particles, or volatilization of surface-deposited compounds such as ammonium nitrate (NH4NO3) during extended rain-free periods. Also unknown is to what degree throughfall may underestimate dry-deposition fluxes when no precipitation occurs for several weeks or months, meaning that no throughfall samples are collected. The fate of compounds deposited to the canopy during prolonged dry periods and methods of quantifying canopy uptake of N and other atmospheric pollutants under such conditions warrant further study. Throughfall can also be collected under low-lying vegetation such as shrubs or under forest understory vegetation, using trough-style (Bleeker et al., 2003) or miniaturized throughfall resin collectors (Savard, 2005). However, in arid zones, wind-blown dust becomes a problem. For example, throughfall collectors connected to ion exchange resin (IER) columns (see Section 8.3.2) were installed in a desert scrub ecosystem in Joshua Tree National Park, but dust contamination was too severe to allow for throughfall collection. However, miniaturized throughfall resin columns have been used successfully under krummholz in the eastern U.S. (Savard, 2005). Periodic branch rinsing seems to be a more appropriate technique in arid conditions, but this method may overestimate dry-deposition fluxes as described below. The number of throughfall collectors needed for a particular degree of precision is specific to a given forest site. The number of collectors needed also varies widely depending on the ions being measured. Fewer collectors are needed for deposition than for chemical concentration (mainly because throughfall volumes used to calculate deposition are much less variable than chemical concentrations), and the number needed decreases with longer collection periods (Houle et al., 1999; Lawrence & Fernandez, 1993; Thimonier, 1998). The following formula has frequently been used to calculate the number of collectors to use: n¼

t2 CV2 E2

where n is the number of collectors needed to determine the mean throughfall deposition within a predetermined error and confidence level, t the Student’s t value for a given confidence level, CV the coefficient of variation as a percentage, and E the acceptable error (expressed as a percentage of the mean). For example, in a mixed hardwood forest in Quebec, the number of collectors needed to estimate semiannual deposition with a confidence interval of 95% with a 10% error was 13, 14, and 25 for SO42 , NO3 , and NH4þ , respectively (Houle et al., 1999).

188

M.E. Fenn et al.

Only five collectors were needed to estimate deposition of these same ions with a 20% error. Lawrence and Fernandez (1993) concluded that, for a spruce–fir (Picea–Abies spp.) forest in Maine, deposition of most ions on a seasonal or annual basis can be estimated within 20%, a level of precision that is generally adequate, using 20–30 collectors. These rules of thumb are likely to be applicable for plot-level flux estimates or over areas of similar vegetation, topography, and pollution exposure. When vegetation type changes, or with varying topographical features, deposition over the landscape is likely to become more variable (Weathers et al., 2006). Periodic aqueous rinsing of branches is another simple approach for measuring potential deposition fluxes to vegetation during dry periods. With this method, branches are pre-rinsed at the beginning of the study to remove washable compounds from branch surfaces and to create a starting point for measuring deposition fluxes. Then, at regular intervals (generally every 2 weeks in our studies), the same branches are washed and the rinse solution is collected. This approach is limited to the measurement of ions that do not leach from branches during washing (e.g., NH4þ , NO3 , and SO42 ). Alternatively, at the end of the exposure period, the branches can be collected and washed in the laboratory and new branches prewashed in the field. The foliar surface area of the rinsed branches is measured to calculate deposition fluxes per leaf area. From this information, it is possible to scale up and calculate potential deposition to the larger canopy based on canopy leaf area (Shanley, 1989). However, unless a tower is erected for access throughout the vertical profile of the canopy (Bytnerowicz et al., 1999), it must be assumed that deposition fluxes are similar at all heights. This may not be a problem in ecosystems with open or low-lying canopies such as desert, coastal sage, or subalpine ecosystems. In mixed-species forests, branch-rinsing data and the leaf area for all the major species in the stand are needed. Seasonal deposition fluxes are needed to calculate annual deposition fluxes. If precipitation occurs during an exposure period, the dry-deposition flux cannot be calculated for that time period. The branches must be rinsed again and a new dry-deposition exposure initiated. It can be seen that several critical data parameters are needed to calculate stand-level deposition from branch-rinse flux data and that each one of these parameters entails uncertainty. It should also be remembered that, as with throughfall, retention or uptake of N by branches and foliage might also result in underestimation of total N deposition when using the branch-rinsing method. In arid or semiarid regions, branch-rinse data may often not be comparable to throughfall deposition data. During periods with infrequent or

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

189

no rain for several months, periodic branch rinsing removes a much larger proportion of dry-deposited pollutants than are removed in throughfall (Fenn et al., 2000). Thus, dry-deposition flux data from sequential rinses of the same branches may be more properly considered as a measure of the dry deposition potential. This is because each time branches are washed; a new clean surface is created that is a more effective receptor than a canopy that has not been recently washed, either artificially or by rain. Presumably, dry deposition fluxes are greater to washed than unwashed foliage, which may explain why at two sites in the San Bernardino Mountains over a 7-month period during the dry season (6 months with no precipitation), N deposition fluxes determined from branch rinses were 50% higher than throughfall fluxes (Fenn et al., 2000). These findings suggest that branch-rinse and throughfall data are likely to be more comparable in areas or in seasons when precipitation occurs on a regular basis. In the semiarid San Bernardino Mountains, deposition fluxes to the forest during the 1993 summer dry season, as determined from sequential branch rinses, were comparable to throughfall fluxes measured over the same time period. In this case, it appears that both methods were comparable because periodic lowvolume rain events (59 mm precipitation) occurred during the sampling period from late May to mid-November (Fenn & Bytnerowicz, 1997). During extended dry periods, branch rinsing is expected to overestimate throughfall deposition. However, this hypothesis needs further testing. 8.2.3. Wet deposition and bulk deposition

Wet deposition measurements refer to deposition in precipitation with dry deposition excluded. The so-called wet/dry bucket collectors [Aerochem Metrics model 301 (Fig. 8.1; Bushnell, Florida) or Loda Electronics model 2001 (Loda, Illinois)] are AC or battery-powered, and are used in the National Atmospheric Deposition Program (NADP) network to collect wet deposition only; a motorized lid covers the collection bucket except during precipitation events. Accurate estimates of wet deposition require efficient collection of precipitation samples and proper handling of samples to prevent degradation of samples before chemical analysis (Krupa, 2002). As far as inorganic N species are concerned, NH4þ is particularly susceptible to volatilization or chemical transformation losses during storage. Longer field incubation or sample storage times increase the risk of sample degradation. A flip-top collector that can collect wet deposition only (or throughfall when placed under a canopy) and does not require electric power to

190

M.E. Fenn et al.

Figure 8.1. Aerochem Metrics model 301 wet/dry bucket collector of the type used in the NADP monitoring network. These collectors are AC or battery powered and are used to collect wet deposition only.

operate is described by Glaubig and Gomez (1994) and Fenn and Bytnerowicz (1997). The lid is held closed over the funnel collector with ‘‘magic paper’’ that rapidly dissolves upon wetting, causing a counter weight to open the lid and allow collection of precipitation or throughfall samples. After the sample solutions are collected, the lid must once again be placed on top of the funnel opening and secured with the dissolvable paper. In the absence of electric power, deposition in precipitation is usually collected as ‘‘bulk deposition,’’ a term that refers to precipitation samples collected in samplers—often funnels—in which the opening is continuously exposed to the atmosphere, such that some dry-deposited compounds are inevitably collected during intervening dry periods. Throughfall is also usually collected as ‘‘bulk throughfall’’ (Weathers et al., 2001). In European studies, wet-only : bulk deposition ratios for NH4þ , NO3 , and SO42 were frequently 0.75–0.85, although ratios for NH4þ and SO42 were sometimes as high as 0.96–1.25 (Erisman & Draaijers, 1995). It is not clear why wet deposition of NH4þ and SO42 was sometimes greater than bulk deposition, but it could be due to error, greater volatilization losses, or microbial assimilation with bulk collectors, or differing collection efficiencies of the wet-only and bulk collectors.

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

191

Wet-only : bulk deposition ratios were lower for components such as base cations (Ca2þ, Mg2þ, Naþ, and Kþ), which deposit as large particles. Similarly, at a montane site in northern Italy, bulk deposition measurements of NH4þ , NO3 , and SO42 were 6–12% higher than wet deposition, but deposition of Ca2þ, Mg2þ, and Kþ was 30–50% higher in bulk deposition (Mosello et al., 1988). In arid or semiarid ecosystems, where extended dry periods occur, studies are needed to determine to what extent dry deposition, leaf litter, and insects falling into bulk collectors influence wet deposition measurements. The largest uncertainty in wet and bulk deposition measurements is the under-collection of precipitation because of air flow distortion caused by the sampler, which results in small droplets following the distorted air flow, thus preventing quantitative capture of rainfall. Under-collection of precipitation worsens with greater wind speed and smaller raindrop diameter (Da¨mmgen et al., 2005). Some workers have corrected for under-collection of precipitation by using precipitation gauges mounted with the aperture level with the ground and surrounded by an anti-splash grid. Samples for chemical analysis, particularly in arid regions, must still be collected from a height of 1.0–1.5 m to avoid collection of locally resuspended material. Bulk deposition collectors should be replicated at each site in case any collectors are disturbed or to identify samples that are contaminated, for example, with bird droppings. For more details on collection and analysis of wet and bulk deposition samples, see the reviews by Da¨mmgen et al. (2005) and Krupa (2002). Precipitation occurring as snowfall is particularly difficult to measure accurately because snow is transported horizontally and its sedimentation is slow and because snowfall is frequently associated with highly turbulent crosswinds at ground level (Krupa, 2002). As a result, snowfall is commonly underestimated, although at Niwot Ridge, in the Colorado Front Range, snowfall and associated N deposition were overestimated in wet deposition collectors because of oversampling, resulting from blowing snow events (Williams et al., 1998).

8.2.4. Deposition in cloud water and rime ice

Atmospheric deposition of N in fog or cloud water can constitute more than 35% of total deposition (Anderson et al., 1999), even in arid or semiarid zones where fogs occur as a result of orographic or coastal influences (Fenn & Poth, 2004; Fenn et al., 2000; Waldman et al., 1985; Weathers & Likens 1997; Weathers et al., 2000). Deposition of N and S in fog is an important input in southern California and in the Sierra Nevada

192

M.E. Fenn et al.

range in central California (Collett et al., 1990). When super-cooled cloud droplets impinge on vegetation or other surfaces, rime ice is formed. Ionic concentrations in rime and cloud water are typically many times higher than in rain or snow. Rime ice can be collected on plastic line or nettingtype passive collectors (Berg et al., 1991; Duncan, 1991). In polluted areas, fog or cloud water deposition needs to be accounted for by collecting fog water with active or passive fog samplers (Collett et al., 1990; Fenn et al., 2000; Weathers et al., 1988) or by collecting throughfall. In the latter case, throughfall deposition is an estimate of deposition through fog, precipitation, and dry deposition (Weathers et al., 1992, 1995, 2006). However, in areas of low to moderate N deposition, throughfall fluxes of N can be similar to or less than bulk deposition as a result of canopy uptake of atmospheric N. In such cases, throughfall data cannot be used to estimate inputs of dry and fog deposition. This is particularly true in coastal or other high rainfall areas where canopies are heavily colonized by epiphytic vegetation, which can increase canopy N uptake capacity.

8.2.5. Spatial extrapolation of deposition fluxes

Determining the spatial variation in deposition across a landscape or within a watershed is a key issue for evaluating ecosystem effects of atmospheric deposition (Weathers et al., 2006). For winter deposition in seasonally snow-covered environments, this topic is discussed below in Section 8.5. Recent advances in passive monitoring of atmospheric pollutant concentrations and of deposition in precipitation and throughfall have allowed sufficiently intensive data collection to estimate deposition across selected watersheds. For evaluating deposition over larger regions (e.g., hundreds to thousands of square kilometer), simulation or empirical models (see Section 8.4) are needed, but in the former case, ground verification is still required to evaluate model output. Total N deposition to forest stands was estimated by Schmitt et al. (2005) with the inferential method using passive monitors to measure gaseous pollutant concentrations, in combination with a regression model for deriving the other terms of the total deposition of N from measurements of bulk N deposition. By implementing a passive monitoring network over the area of interest, it may be possible to estimate dry deposition fluxes of gaseous N compounds across complex terrain such as an entire montane watershed. For total dry N deposition, flux of particulate nitrate and ammonium should also be added. Although theoretically possible, in practice very limited information on

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

193

concentrations of these pollutants is available. Such information can be obtained on a small scale from denuder/filter-pack systems, such as in the mixed-conifer forest stand at Barton Flats in the San Bernardino Mountains (Bytnerowicz et al., 1999). In that study, flux from nitrate and ammonium was calculated by the inferential method, multiplying ambient concentrations of nitrate and ammonium in fine particulate matter by literature values of their deposition velocity (Davidson & Wu, 1990). With present technologies, no large-scale evaluation of fine particulate contribution to total dry deposition of N based on field monitoring is possible. Total deposition can be obtained by adding wet deposition inputs to the estimated dry deposition, although where fog deposition occurs, this must also be measured. Total N deposition can also be estimated from throughfall measurements by calculating the canopy uptake of N using canopy budget models (Draaijers & Erisman, 1995; Schmitt et al., 2005). When using IER throughfall collectors (Section 8.3.2) rather than aqueous samplers, samples do not need to be collected after each event (see Section 8.3.2), and it is now practical to measure throughfall deposition with greater spatial intensity and over much larger land areas (Fenn & Poth, 2004; Simkin et al., 2004). For example, in the Kings River watershed study in the southern Sierra Nevada in California, IER throughfall collectors are being used to measure throughfall deposition in three watersheds with a grid spacing of 150 m throughout the watershed (Hunsaker & Eagan, 2003). A network of passive monitors of gaseous N pollutants has also been established within these watersheds, providing an opportunity to compare N deposition as determined by the inferential (plus wet deposition) and throughfall methods. Both the inferential and throughfall canopy budget model methods entail a high degree of uncertainty in estimating total N deposition, but the use of both methods provides more robust estimates of deposition. Schmitt et al. (2005) found reasonable agreement between the two methods. Bytnerowicz et al. (1999) also found good agreement between summertime estimates of dry deposition in the San Bernardino Mountains using branch rinsing, throughfall, and a quasi-empirical model based on branch-rinse data collected on a tower along the vertical profile of a relatively open, mixed-conifer forest canopy. For arid systems in particular, the canopy budget models (designed to estimate total N deposition from throughfall data) will likely require adaptation because of the unknown fate of dry-deposited pollutants during long dry periods without canopy wash-off. Furthermore, the canopy models assume little canopy exchange of nitrogen oxides (NOx), although this is known to be an incorrect assumption (Draaijers & Erisman, 1995).

194

M.E. Fenn et al.

Canopy exchange may be particularly problematic in western ecosystems exposed to urban photochemical smog, because of the predominance of oxidized forms of N pollutants. Nitric acid vapor is a highly reactive constituent of urban emissions, and controlled exposure studies suggest that 25–50% of the HNO3 deposited to trees and shrubs native to southern California is not easily washed from foliage. Tracer studies with 15HNO3 indicated that 1–5% of the dry-deposited HNO3 is absorbed and assimilated into amino acids and proteins by leaves, but a substantially larger portion is retained in a labile form, neither solubilized by water nor isolated from the biologically active pool (Padgett, 2004). 8.2.6. CASTNET, NADP, and IMPROVE monitoring networks

The NADP/National Trends Network (NTN) was established in the late 1970s and has since grown to over 200 monitoring sites in the U.S., and 78 sites in the western U.S. It is the most spatially and temporally extensive atmospheric deposition network in North America. The NADP data have proved invaluable in evaluating spatial and temporal trends in deposition, particularly of pollutants such as inorganic N and S. However, only wet deposition in precipitation is monitored, and thus, the data are of limited usefulness for estimating total deposition inputs, particularly in the montane West where dry deposition, and in some instances fog deposition, constitutes large proportions of total deposition. Other limitations of the NADP network for the western U.S. include low station density, few high-elevation stations, and no measurement of atmospheric organic N loading. The Clean Air Status and Trends Network (CASTNET) was established in 1987 to collect air chemistry data and model dry deposition inputs to the vegetation near each monitoring station. Currently, there are over 70 monitoring locations, 26 of them in the western states. Weekly average atmospheric concentrations of NH4þ , NO3 and SO42 , SO2, and HNO3 are measured using filter packs, along with hourly concentrations of ambient ozone (O3) levels and meteorological conditions (http:// www.epa.gov/castnet/). Filter-pack techniques, although widely used, have been criticized (Pathak & Chan, 2005) because of large errors in determinations of gaseous vs. particulate components of dry deposition (e.g., HNO3 vs. NO3 and NH3 vs. NH4þ ). Rates of dry deposition are determined at the CASTNET sites with an inferential method using atmospheric concentrations and modeled deposition velocities (Kolian & Haeuber, 2004). Model inputs also include meteorological data and information on land use, vegetation, and surface conditions. Inferential model flux calculations for CASTNET are generally biased low due to the

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

195

weekly integrated sampling protocol (Kolian & Haeuber, 2004). Total N deposition is also underestimated because NO, NO2, NH3, and organic N are not measured (Baumgardner et al., 2002). A promising new gas–particle ion-chromatographic system for continuous monitoring of soluble gases and ionic constituents of particulate matter is currently being tested by CASTNET. With this technology, semicontinuous measurements of NH4þ and anions in ambient aerosols and soluble acidic gases and NH3 can be made with a high degree of robustness and flexibility. This methodology is more expensive than current methods, but if implemented, should improve deposition estimates by providing time-resolved atmospheric concentrations to be used in calculating dry deposition. CASTNET sites are usually co-located with NADP sites, so that total atmospheric inputs can be estimated as the sum of dry and wet deposition. Data from the larger NADP network are frequently used to extrapolate wet deposition over the U.S. However, dry deposition data from CASTNET are mainly considered as an estimate for that monitoring site only, although measurements will often be indicative of relative dry deposition inputs to similar land-cover types in the area, assuming similar pollution exposures. A comparison of CASTNET dry deposition and total deposition (NADP þ CASTNET) with resin throughfall estimates of deposition has been carried out in eastern North America (Weathers, personal communication). These comparisons show that, for this region, throughfall and total deposition compare well over the time period of the summer growing season. Similar comparisons have not been performed on a regional basis in western North America. However, although annual N deposition at Giant Forest in Sequoia National Park was comparatively low, ranging from 2.8 to 3.9 kg ha1 in 1999, 2001, 2003, and 2004 based on NADP þCASTNET data (http://www.epa.gov/castnet/sites/ sek402.html), throughfall N deposition at the same site was 11.6 kg ha1 year1 (Fenn et al., 2003b). In an earlier study, throughfall deposition was 6.2 and 10.8 kg ha1 year1 under nearby mixed fir and Sequoia (Sequoiadendron giganteum) stands (Chorover et al., 1994), also much higher than the NADP þ CASTNET data. These comparisons indicate that CASTNET þNADP estimates of N deposition in semiarid forests of the West can be several-fold lower than throughfall fluxes. It appears that drydeposition fluxes of N are grossly underestimated with the CASTNET protocol at Giant Forest for the reasons described earlier and possibly because of high uncertainty in the inferential model (Kolian & Haeuber, 2004). Likewise, deposition in fog, which is important in Sequoia National Park (Collett et al., 1990), is not accounted for in the CASTNET and NADP data.

196

M.E. Fenn et al.

The Interagency Monitoring of Protected Visual Environments (IMPROVE) network was established in 1985 to measure the key aerosol species affecting visibility in federal Class I areas. Fine particulates NO2 , NO3 , and SO42 are measured, although total atmospheric NO3 is probably underestimated because HNO3 vapor and NO 3 in the coarse particulate fraction are not measured (Fenn et al., 2003b; http:// vista.cira.colostate.edu/improve/, last accessed on July 18, 2008). The Speciation Trends Network (STN) is a national network of approximately 1200 PM2.5 monitoring sites including samplers that provide mass data, visibility-related measurements, and chemically speciated particulate matter data. The STN database includes data collected from IMPROVE, State, and local air-monitoring stations, and other special study sites. Although the STN and IMPROVE data do not provide direct estimates of deposition, they are useful for validating numerical air quality models that do simulate the deposition of N (http://www.epa.gov/oar/oaqps/pm25/ general.html, last accessed on December 30, 2008). In summary, although existing monitoring networks provide useful data for characterizing some aspects of N deposition, data from ongoing monitoring networks will generally not provide the data needed to estimate total N deposition or to evaluate spatial and temporal variability in N deposition needed in support of ecosystem effects studies in western North America. 8.2.7. Nitrogen accumulation in surface soils of arid systems

In arid ecosystems, sparse vegetation exposes soil surfaces to direct dry deposition. Padgett et al. (1999) demonstrated that inorganic N accumulates on the surfaces of soils of coastal sage scrub ecosystems in southern California during dry smoggy periods. Nitrogen accumulation was far greater in areas with high levels of atmospheric pollutants. Deposition was also greater in the interspaces between plants than on soil under shrub canopies, which are typically higher in organic N, supporting the argument for an atmospheric source for surface accumulation of inorganic N. Following the onset of the wet season, nitrate concentrations in the top 2 cm of soil rapidly decreased, but could be detected further down the soil profile as N was leached through the soil water profile (Padgett & Bytnerowicz, 2001; Padgett et al., 1999). In controlled HNO3 fumigation exposures, thin soil layers were found to function as effective N deposition receptors for monitoring HNO3 deposition fluxes to soil. Deposition velocities of HNO3 vapor to isolated soil fractions of sand, silt, and clay followed established ion exchange principles of soil chemistry: levels of extractable NO3 increased with increasing surface area of the mineral soil particles (Padgett & Bytnerowicz, 2001).

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

197

Monitoring of N accumulation in surface soils has also been used in a desert ecosystem in Joshua Tree National Park in southern California for monitoring N deposition to bare soil during the summer drought season (M.E. Fenn, unpublished data). We have developed a soil plate sampler for estimating deposition fluxes of N to exposed soil. This technique could presumably be used for other pollutants as well. A thin layer of soil is placed in a small dish such as a Petri plate and exposed to the atmosphere for a given time, typically 1 or 2 weeks. In many areas, wind tends to disturb the soil in the plate, making it necessary to fix the soil to a solid surface before field deployment. We found that this can be done effectively by applying a multipurpose aerosol spray adhesive to a plastic disk, over which a thin layer of native soil is spread. The disk is then mounted on a small jar as a lid insert of the style used in glass canning or fruit jars in such a way that the soil disk is exposed to the atmosphere. After the field exposure, nitrate and ammonium (or any other ions of interest) are extracted with 2 N KCl and analyzed. Following blank correction, the deposition flux to soil can be calculated. 8.3. Passive monitoring techniques

In general, passive samplers measure long-term (weeks or months) average concentrations of air pollutants. Passive samplers for gaseous pollutants are devices for measuring pollutant concentrations based on passive diffusion of the pollutant through barriers (filters, screens) or diffusion tubes, and onto the collection medium, the latter chosen on the basis of its affinity to the pollutant gas to be sampled. In this section, we will discuss passive samplers that measure time-averaged atmospheric concentrations of key gaseous pollutants and a collector that accumulates ions from precipitation or throughfall solutions using IER methods. Lichens as bio-monitors will also be briefly discussed. 8.3.1. Passive samplers of gaseous pollutants

Krupa and Legge (2000) published a comprehensive review of the use of passive samplers for monitoring ambient air pollution from an ecological effects perspective. In this chapter, we present only selected information on the samplers that have been used in mountain settings. The main advantages of using passive samplers are easy deployment in field conditions, simplicity of design and operation, low requirements for field labor and maintenance, relatively low cost, no need for electric power or

198

M.E. Fenn et al.

Figure 8.2. Schematic diagram of the Ogawa passive sampler. Depending on the absorbent used on the coated collection filter, these samplers have been used to sample a number of gaseous pollutants including O3, NH3, NO2, NO, NOx, and SO2.

air-conditioned shelters, and feasible use across dense networks allowing for landscape or regional characterization of air pollution distribution patterns. Major disadvantages are low temporal resolution, possible interference from other pollutants, significant effects of meteorological conditions such as wind, temperature, or relative humidity, low detection limits, and poor relationship with vegetation responses (Bytnerowicz et al., 2000; Koutrakis et al., 1993; Krupa & Legge, 2000; Tang, 2001). Koutrakis et al. (1993) designed a passive sampler, now commonly known as the Ogawa passive sampler, which is widely used in North America, Europe, and Asia (Fig. 8.2; http://www.ogawausa.com/). The sampler is made of Teflons, with a Teflon cap with precision-drilled holes followed by a stainless steel mesh serving as the diffusion barriers. The pollutant to be sampled is collected on cellulose filters that are coated with reagents designed to react with a particular pollutant. Initially, this sampler was used almost exclusively for sampling ambient O3 on pads coated with sodium nitrite (Alonso et al., 2001; Brace & Peterson, 1998; Bytnerowicz et al., 2001; Fraczek et al., 2003; Gertler et al., 2006; Ray, 2001). The Ogawa samplers are now also used to determine NH3, NO2, NO, and SO2 concentrations (Fig. 8.3). Like earlier sampler designs, NO2 and SO2 are collected on filters coated with triethanolamine (TEA), whereas NOx are collected on filters coated with TEA plus an oxidizing agent (Ogawa, 1998). Ammonia is collected on filters coated with citric acid, resulting in the formation of ammonium citrate [(NH4)2C6H6O7], which is extracted from the filters and concentrations of NH4þ determined by colorimetric methods (Roadman et al., 2003). Concentrations of NH3

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

199

Figure 8.3. Photograph of an array of passive samplers installed for field monitoring of gaseous pollutants. Although not shown, SO2 is also measured with the Ogawa sampler.

are calculated based on calibrations against reference methods such as annular denuders. Passive samplers for measuring HNO3 and HNO2 concentrations (Fig. 8.3) based on quantitative absorption of these gases by nylon filters have been developed by Bytnerowicz et al. (2001, 2005). In the initial sampler design, three nylon filters in PVC rings hang underneath a PVC cap protecting them from rain and wind (Bytnerowicz et al., 2001). A new, improved badge-type HNO3 sampler uses a Teflon membrane for controlling air-flow to a nylon filter that absorbs HNO3 (Bytnerowicz et al., 2005). Passive monitors have been used to characterize air pollution distribution at scales ranging from individual trees to forest stands to mountain ranges or other large-scale applications (Arbaugh & Bytnerowicz, 2003; Bytnerowicz et al., 2004; Gertler et al., 2006; Schmitt et al., 2005). By combining passive sampler data and geostatistical models, large-scale air pollutant exposure patterns can be determined. Such information can point to ‘‘hot spots’’ of air pollutant exposure. For instance, in 1999, approximately 90 O3 monitoring plots were established in the Sierra Nevada mountain range in California. Measuring O3 over such a large network revealed some surprising results, such as much higher than expected concentrations during certain meteorological conditions in the ‘‘pristine’’ eastern Sierra Nevada (Fraczek et al., 2003). Similarly, the 2002 study in the Lake Tahoe Basin with about 30 monitoring sites clearly indicated that elevated levels of O3 and HNO3 in the Basin resulted mostly

200

M.E. Fenn et al.

from local emissions, not from long-range transport of polluted air masses from the California Central Valley (Gertler et al., 2006). Passive samplers are useful for understanding both horizontal and vertical distribution of air pollutants. Brace and Peterson (1998) reported a linear increase in O3 concentrations with elevation on Mt. Rainier in the Cascades, and Alonso et al. (2003) found that O3 concentrations in the San Bernardino Mountains in southern California did not change significantly between 1223 and 2667 m elevation. Information gained from passive monitoring networks can guide decisions on where and when active air pollution monitoring equipment should be deployed for understanding distribution of real-time concentrations of air pollutants. Information on concentrations of NO, NO2, NH3, HNO2, and HNO3 can also be used to estimate dry deposition of N in forest stands (Bytnerowicz et al., 1999). Atmospheric concentration data from passive samplers can be used to calculate landscape-level N deposition with the inferential method, assuming that additional information such as deposition velocities to various components (trees, understory vegetation, soils, rocks, etc.), LAI, and land-use data are also available (see Section 8.2.1 and Schmitt et al., 2005). Information from passive samplers presented as monthly averages of pollutants can also be used as input for models of air pollution dispersion such as CALPUFF. This is critical for remote areas where real-time air pollution monitoring does not exist.

8.3.2. Throughfall and precipitation monitoring with ion exchange resins

The use of IERs in throughfall and bulk deposition collectors can dramatically reduce the number of trips to field sites, sample numbers, and analysis costs. Fenn and Poth (2004) reported deposition estimates using IER throughfall collectors that compared well with co-located conventional solution collectors for measuring deposition of ammonium and nitrate in throughfall and bulk deposition. The IER collectors functioned well with field deployment times as long as 12 months (the longest time period tested), although IER columns are usually changed out every 6 months. Simkin et al. (2004) showed similar comparisons and utility for sulfate, nitrate, and chloride sampling using IER columns in eastern North America (Fig. 8.4). The IER collector is built of inexpensive materials and is designed and used similarly to conventional throughfall collectors, except that instead of collecting and analyzing the solution on an event or periodic basis, the solution is funneled through an IER column where the anions and cations are reversibly adsorbed on a mixed-bed resin and are collected every 6 weeks to 6 months.

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

201

Figure 8.4. A bulk deposition collector using an anion exchange resin column to collect NO3 , SO42 , and Cl in precipitation or throughfall (Simkin et al., 2004). The resin columns are usually exchanged every 6 weeks.

Snow tubes can be attached to the collector funnels (Fig. 8.5) during winter for snow collection, generally at the same time that IER columns are exchanged. Ions in snow are then adsorbed by the IER as the snow melts and percolates through the resin column, although in regions with high snowfall rates other methods must be employed, for example, collecting snow samples, melting them in the laboratory, and passing the solution through resin columns (Weathers et al., unpublished data).

202

M.E. Fenn et al.

Figure 8.5. A bulk deposition collector using mixed bed (anion and cation) ion exchange resin to collect NH4þ , NO3 , and SO42 in precipitation or throughfall—based on the design described by Fenn and Poth (2004). The resin columns are usually exchanged after 6 months of field exposure, but in some instances they are changed after 12 months. The collector on the right has a snow tube attached for winter collections.

In sites with extremely low temperatures, for example, boreal forests or subalpine locations, the IERs may lose their effectiveness and should not be used without testing the functionality of the resin under conditions occurring at the field sites. In high-elevation sites with deep or persistent snowpack, the IER collectors are best deployed in late spring or early summer after the bulk of the snowpack has melted; winter deposition can be estimated by sampling the snowpack as described in Section 8.5. At mid- or low-elevation sites, where the use of snow tubes is more appropriate, the collection efficiency of the snow tubes in open areas is expected to be low in many instances (Ranalli et al., 1997; Williams et al., 1998) and to vary temporally and spatially, particularly under windy conditions. Elaborate windshields have been designed to improve snow collection (Hansen & Davies, 2002), but these would not be practical for replicate IER collectors. Snow collection with throughfall collectors is likely to be more efficient under canopies than in open areas because of the sheltering effect of the forest canopy. The IER collectors can also be used to monitor the deposition of Ca2þ, Mg2þ, Naþ, and Kþ and, with the selection of appropriate resins,

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

203

possibly to measure organic N deposition and N isotopes. Data on the deposition of base cations are needed to calculate critical loads of deposition for acidity, and N and S deposition inputs are needed for critical loads for acidity and eutrophication effects (De Vries and Posch, 2003). It is anticipated that IER throughfall collectors will play a key role in collecting deposition data needed to calculate critical loads and in determining areas in which critical loads for deposition are exceeded (Weathers et al., 2006). Detailed instructions on the construction of IER collectors and field and laboratory procedures can be found at http://www.fs.fed.us/ psw/topics/air_quality/resin_collectors/index.shtml (last accessed on December 30, 2008) and in Simkin et al. (2004). 8.3.3. Lichens as bio-monitors/bio-accumulators of atmospheric deposition

Lichens have been used extensively in western North America as biomonitors of atmospheric deposition and as early warning signals of adverse ecological effects (Fenn et al., 2003a, 2008; Geiser & Reynolds, 2002). Although lichen monitoring does not provide quantitative total deposition fluxes, lichen data are useful in following spatial and temporal deposition patterns because lichens are good accumulators of pollutants such as N, S, and metals. Lichens have consistent elemental ranges in clean sites, which contrast with pollutant levels in polluted sites. Element analysis thresholds (97.5% quantiles for element concentrations in lichens from background sites) have been developed for lichen species in the Pacific Northwest. These thresholds are used to identify sites with elevated deposition. The relative severity of atmospheric deposition can also be evaluated based on lichen community analyses (http://www. nacse.org/lichenair/, last accessed on August 17, 2008). When lichen communities are altered by atmospheric deposition, this is a clear demonstration that ecological effects are already occurring (Fenn et al., 2003a, 2008). Lichen monitoring data are most useful when used in combination with conventional measurements of pollutant concentrations or deposition (Fenn et al., 2007, 2008). 8.4. Simulation modeling of deposition

Models for simulating atmospheric deposition serve useful functions such as providing complete spatial coverage over the modeled domain, simulating long periods with hourly resolution, forecasting future conditions, and simulating the benefits of emissions reductions. However, observations are needed to validate model simulations. Models might be

204

M.E. Fenn et al.

incomplete and can have compensating errors. Errors sometimes result from poor spatial resolution. Mesoscale models are expensive to operate for long periods and can be ‘‘gamed’’ to get a particular answer. Despite these concerns, atmospheric deposition models combined with observations for validation offer a good approach for quantifying deposition inputs and mechanisms. Models can represent both wet and dry deposition and can simulate deposition through multiple pathways including stomata, cuticle, and ground surfaces. 8.4.1. Empirical deposition modeling over heterogeneous landscapes

Recently, an empirical modeling approach has been developed to estimate deposition across heterogeneous (e.g., montane, coastal, variable land-use/land-cover) landscapes that range in spatial extent from hundreds to thousands of kilometers (Weathers et al., 2000, 2006). The approach quantifies deposition as a function of landscape variables such as elevation, vegetation type, and topographic exposure. It is based on statistical models developed from empirical data and their application within a GIS to create S and N deposition maps. Total deposition (wet þ dry þ fog) estimates are scaled up from monitoring station data (CASTNET and NADP). This model, the LANDMod, was developed from extensive field data collected from two national parks in the eastern U.S., where elevation and vegetation type were identified as the key variables that controlled deposition over the scale of the parks (250– 2500 km2). Resultant maps showed ‘‘hot’’ and ‘‘cold’’ spots of S and N deposition that varied up to 10-fold from regions of high to low deposition. High-elevation conifer forests were predicted to receive the highest deposition and low-elevation open areas (i.e., without forest) the lowest deposition. Validation for the LANDMod showed that it underestimated deposition to high-elevation/high-deposition regions and overestimated for low-elevation/low-deposition regions. The advantages of this type of empirical modeling and GIS approach are (1) it is based on field data, (2) it uses existing monitoring data to anchor the landscape deposition maps, (3) it is spatially explicit (hot and cold spots can be identified), (4) the approach can be tested in other heterogeneous landscapes, and (5) the modeling results can be updated. It will be especially useful to determine which landscape factors are likely to control deposition in arid and high-elevation western landscapes. We expect that topographic features (e.g., elevation and aspect, LAI, frequency and intensity of fog deposition, as well as snow deposition, accumulation, and redistribution) are likely to be some of the key controlling variables.

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

205

8.4.2. Simulation of dry deposition by comprehensive air quality modeling systems

Comprehensive air quality models generally use a three-dimensional (3D) Eulerian grid structure to represent the physical and chemical processes relevant to air quality for scales ranging from local to global. Grid cells can range from about 1 km for local-scale simulations up to several hundred kilometers for hemispheric or global simulations. Input data to these models include emissions inventories for the entire suite of air pollutants, meteorology data generated from numerical weather models, and atmospheric chemistry mechanisms designed to represent the chemical and phase transformations of air pollutants. Dry-deposition fluxes are included in the numerical processing as an important sink for airborne species as well as a model product for use in ecological impact assessments. The surface chemical flux is computed as the product of the chemical concentrations in the lowest model layer and the deposition velocity (F ¼ VdC; Seinfeld & Pandis, 1998). Note that this is a convenient approximation that allows Vd to be parameterized from meteorology and surface characteristics separately from the chemical model. A limitation of this approach is that the surface exchange is assumed to be to the surface only. Bi-directional fluxes, which are important for some chemical species (e.g., NH3), require chemical difference (DC) across the air–surface interface. Model development efforts are currently underway to add this capability. The dry-deposition velocity is usually parameterized using an electrical resistance analog where serial and parallel combinations of resistances represent various deposition pathways. There are three primary pathways: (1) to the ground, (2) to the leaf cuticles, and (3) through the stomata into the leaf tissue. Stomatal resistance depends on the degree of stomatal opening, which responds to sunlight as well as water stress (root zone soil moisture) and air temperature and humidity. Stomatal function will also be affected by other biological and environmental stresses, such as disease, parasites, nutrient availability, etc., but these are difficult to incorporate into atmospheric models. Because of similarity between the stomatal pathway for dry deposition and evapo-transpiration, drydeposition models can be integrated with land surface modeling components of meteorology models. In this way, long-term changes (over weeks and months) in soil moisture and vegetation can be simulated and their effects on dry deposition included (Pleim et al., 2001). In addition to dry deposition, wet deposition from clouds also plays a role in the removal of pollutants. The amount of wet deposition is typically calculated as the integral of the precipitation rate and the cloud water concentration over a certain time interval.

206

M.E. Fenn et al.

Deposition velocities vary substantially depending on the phase (i.e., gas or aerosol) and chemical species. For example, nitric oxide deposits very slowly but, when oxidized to form HNO3, it deposits rapidly, and when HNO3 condenses with ammonia to form ammonium nitrate aerosol, the deposition rate is reduced for both species. Thus, to simulate the deposition of these species accurately, it is important to represent accurately both the chemical and the phase transformations and to validate the model simulation by comparisons with speciated ambient monitoring data. One of the more widely used simulation models is the U.S. Environmental Protection Agency’s Community Multiscale Air Quality (CMAQ) model. We completed a CMAQ simulation for the calendar year 2002 using a 4-km grid resolution for central and southern California (Tonnesen & Wang, 2005). Meteorology for the simulation was generated using the National Center for Atmospheric Research (NCAR)/Penn State Mesoscale Model (MM5) (Grell et al., 1994), and emissions input data for this simulation were extracted from the Western Regional Air Partnership (WRAP) 2002 emissions datasets. Error and bias in the model performance are relatively large, and this is due in part to large errors in the meteorology and emissions data input to the model. However, errors tend to be smaller for longer averaging periods, and it is likely that errors in the annual simulated N deposition are considerably less than that calculated for hourly or daily averaged PM measurements. An alternative to 3D Eulerian grid models such as the CMAQ model is the use of Lagrangian models that also treat dry and wet deposition for both gas and aerosol species. Typical users of such models are air quality specialists from government agencies and consultant companies involved in the permit issuing process. In contrast to Eulerian models that simulate the chemistry, transport, and deposition within the 3D grid, Lagrangian models represent puffs of air flowing through the model domain. The spatial distribution of pollutants is approximated by simulating the varying trajectories of multiple puffs. The CALPUFF model (Scire et al., 2000) is a non-steady-state Lagrangian Gaussian puff model containing modules for complex terrain effects, over-water transport, coastal interaction effects, building downwash, wet and dry removal, and simple chemical transformation. It is capable of computing dry-deposition rates of gases and particulates as a function of geophysical parameters, meteorological conditions, and pollutant species through a full resistance model. Similar to the 3D Eulerian models, the surface pollutant deposition flux is calculated as the product of the deposition velocity and the pollutant concentration (F ¼ VdC). The deposition velocity is expressed as the inverse of a sum of ‘‘resistances’’ plus, for particles, gravitational settling terms (Scire et al., 2000). For gases, the ‘‘resistances’’

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

207

include (1) atmospheric resistance, which is determined by the predominant land-use type of each grid cell; (2) deposition layer resistance, which accounts for the molecular diffusion to the transport through the laminar deposition layer; and (3) canopy resistance, which is the resistance for gases in the vegetation layer. The three main pathways for canopy resistance (to the ground, to the leaf cuticles, and through the stomata) are identical to the approach described for Eulerian models earlier. For particulate matter, the resistance in the vegetation layer is not considered as particles are usually assumed to stick to the surface after they penetrate the deposition layer. The deposition velocities for particles are expressed in terms of atmospheric resistance, deposition layer resistance, and a gravitational settling term. In CALPUFF, an empirical scavenging coefficient approach is also included to compute the depletion and wet deposition fluxes due to precipitation scavenging. The primary advantage of Lagrangian models is their simplicity, reduced computational cost, and ability to more directly track the effects of pollutants originating from a particular emissions source. The advantages of Eulerian models include greater physical realism and a more direct representation of the spatial varying patterns of pollutant concentrations and deposition. Because Eulerian models typically include a more complete set of chemical reactions, emissions inventories, and spatial coverage, they are better suited for simulating deposition of air pollutants from many sources and over large regions. However, Eulerian models can exhibit large numerical dispersion and dilution of emissions from a single point source. Lagrangian models may be better suited for simulating deposition from a single, large point source, and for this reason, these models are frequently used to predict pollution exposures downwind from a proposed new emissions source during the permit application process. 8.4.3. Special issues in modeling arid, snow-covered, and mountainous areas

In arid environments, vegetation is sparse and soil moisture is usually very low. This means that the stomatal pathway is usually less important and ground deposition is generally more important. Uptake by dry surfaces is very uncertain and highly variable from one chemical species to the next. Measurements of surface resistances for each chemical species of interest are necessary for realistic parameterizations. Precipitation events are particularly important in arid ecosystems as many plant species grow and become active only for short periods when water is available. During desert bloom periods, the vegetation cover and activity can increase dramatically. Therefore, the stomatal pathway may become dominant at these times. Thus, the model system must be able

208

M.E. Fenn et al.

to accurately simulate precipitation events and the resulting growth of vegetation. Most current modeling systems, however, describe vegetation by land-use category and season only. Development of higher resolution land-use data and modeling techniques for desert vegetation are needed for better simulation of dry deposition in arid environments. Explicit modeling of the snowpack has been added to many mesoscale meteorology models in recent years. Dry-deposition models need to consider deposition to the snowpack. Uptake to snow depends greatly on temperature as melting snow provides an opportunity for dissolution of soluble chemical species. The fraction of melt water can be parameterized by temperature. Deposition models have parallel pathways to ice and liquid water. The liquid water pathway should have a serial resistance representing diffusion through the snowpack. Mesoscale grid models often have insufficient horizontal grid resolution to accurately represent complex terrain in montane landscapes. A consequence is that ambient boundary layer turbulence is usually underestimated, resulting in underestimated deposition velocities. To improve estimates of dry deposition, sub-grid terrain information can be used to enhance the aerodynamic roughness length. However, unresolved slope effects on the flow fields, particularly canopy infiltration on windward slopes, will not be taken into account. Hence, high-resolution modeling is especially important for modeling in complex terrain. 8.4.4. Comparison of simulated deposition to ground measurements and model evaluation methods

The CMAQ model outputs at the 36-km grid scale identified hotspots of elevated N deposition in the western U.S. (Tonnesen et al., 2003) and performed reasonably well in estimating deposition in the Sierra Nevada and Colorado Front Range where deposition inputs are moderately elevated. Measurements of annual throughfall deposition ranged from 1.4 to 13.4 kg ha1 at 9 mixed-conifer forest sites along a north–south transect in the Sierra Nevada, compared with simulated deposition of 5–10 kg ha1 year1 in the Sierra Nevada (Fenn et al., 2003b). In the Colorado Front Range, deposition ranges from 5 to 8 kg ha1 year1 and simulated deposition was 9–10 kg ha1 year1. Simulated deposition (a peak of 11 kg ha1 year1 in the Los Angeles Air Basin) was underestimated in the highly polluted western San Bernardino Mountains, where throughfall deposition, including deposition to canopy-free areas of the stand, can be as high as 71 kg ha1 year1 (Fenn et al., 2008). The underestimation of N deposition in the San Bernardino Mountains is partially due to the steep N deposition gradients

209

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

that are characteristic when dry deposition or fog deposition (i.e., in coastal or montane regions) is a major form of N deposition; a factor generally lost in the 36-km grid scale simulations. In recent 4-km grid simulations for California, simulated N deposition fluxes were approximately 40 kg ha1 year1 in parts of the San Bernardino Mountains (Fig. 8.6), suggesting that modeled deposition values are more realistic in these polluted areas with models run at finer grid resolution. The underestimated deposition in the western San Bernardino Mountains is probably also due to underestimation of deposition in fog, and because NH3 emissions from dairy farms in the region were not adequately accounted for in the original 36-km CMAQ emissions data. Another N deposition hotspot not picked up in the 36-km simulations is the Phoenix area, where average deposition was estimated to be 13.5 kg ha1 year1, and in the most exposed desert and forested regions located downwind, N deposition was as high as 29 kg ha1 year1. Deposition estimates for the Phoenix area from the 36-km CMAQ simulations were only 6–7 kg ha1 year1 (Fenn et al., 2003b). Another disadvantage of the 36-km grid

Dry N Deposition

WET N Deposition

CMAQ 2002 Simulation California

CMAQ 2002 Simulation California

30.00 225

5.00

26.25

4.38

22.50

3.75

18.75

3.12

15.00

2.50

11.25

1.88

7.50

1.25

3.75

0.62

0.00

(a)

0.00

1

kg-N/ha/year 1

144

January 1,2002 1:00:00 Min = 0.06 at (95,225), Max = 98.83 at (92,51)

225

1

kg-N/ha/year 1

(b)

144

January 1,2002 1:00:00 Min = 0.00 at (131,23), Max = 9.54 at (83,51)

Figure 8.6. Annual dry and wet N deposition in California for the CMAQ air quality model 2002 base case, simulated at a 4-km scale. To best show the variation in deposition over the modeled area, the color scale in the figure was limited to 0–30 kg N ha1 year1 for dry deposition and 0–5 kg N ha1 year1 for wet deposition. Actual peak N deposition values were 98.83 kg ha1 year1 for dry deposition in a limited area near Chino Hills in Orange County and ca. 9.54 kg ha1 year1 for wet deposition in montane regions of southern California and in the southern Sierra Nevada in central California.

210

M.E. Fenn et al.

simulation is that it does not adequately resolve orographic features in montane regions. Two major problems are encountered when evaluating numerical model output by comparisons with observed data. One is the sparseness of observational data used to compare with model output. The second problem is that observational data and model output are at different spatial resolutions, resulting in the comparison of model-generated grid cell averages with point-referenced monitoring data. The latter problem is known as a ‘‘change of support problem’’ (Fuentes & Raftery, 2005; Swall & Davis, 2006). This problem is acknowledged in comparisons of modeled and monitored deposition fluxes across Europe at the many ICP-forest sites, but comparisons are being made notwithstanding the change of support problem (Simpson et al., 2006). In this case, monitoring data from 30  30 m2 plots are compared with modeled output for 50  50 km2 grids. Interpolation or extrapolation methods applied to the point measurements can be used to compare observations to model output, but such comparisons are questionable because of the sparseness of observations and because uncertainty in the interpolated values is not accounted for. Fuentes and Raftery (2005) developed formal Bayesian statistical methods for combining sources of information with different spatial resolutions and for the evaluation of numerical models. The Bayesian methods reported by Fuentes and Raftery (2005) remove the bias in the model output and combine model output with observations in a coherent way, leading to improved air quality maps. On the basis of the work of Fuentes and Raftery, Davis and Swall (2006) developed a Bayesian hierarchical model for evaluations of CMAQ that can be viewed as a type of Bayesian kriging technique. From the point-wise observed monitoring data, statistical methods were used to estimate the values expected from the CMAQ simulations. These statistical estimates were then compared for each grid cell with values simulated by CMAQ (Davis & Swall, 2006). Point measurements of atmospheric deposition or pollutant concentrations can also be compared with CMAQ by kriging the CMAQ output fields for locations where measurements were taken (Davis & Swall, 2006). However, traditional stationary kriging predictions ignore the uncertainty in the covariance parameters and do not produce appropriate prediction errors (Fuentes & Raftery, 2005). Other statistical approaches have also been used (Fuentes & Raftery, 2005; Mebust et al., 2003), and this is an area of ongoing research. Whereas the aforementioned methods apply to spatial processes at a fixed time point, techniques have also been developed to assess the ability of CMAQ to capture spatiotemporal patterns in pollutant concentrations (Jun & Stein, 2004). In summary, ongoing development of

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

211

statistical methods for evaluating numerical models will allow researchers to derive the greatest informational value from observational data and simulation models, while also providing a tool for further improvement of atmospheric simulation models.

8.5. Measuring deposition in high-elevation basins in the Western United States

The hydrology of high-elevation basins in the western U.S. is driven by the accumulation of a seasonal snowpack during winter (usually October–March) and subsequent melting during spring and early summer (usually April–June). Snow generally accounts for most of the precipitation, and the relative importance of snow increases with elevation. In Colorado, for example, snow typically accounts for 50–80% of total precipitation at high-elevation snow monitoring sites in the snowpack telemetry (SNOTEL) program. At Emerald Lake in the Sierra Nevada (elevation 2800 m), snow accounted for 88% of annual precipitation from 1983 to 2000 (Sickman et al., 2001; Sickman and Melack, unpublished data). Summer monsoonal storms typically supply most of the rain in the Rocky Mountains, whereas in the Sierra Nevada, monsoonal rains are less common and most rain derives from low-pressure systems off of the Pacific Ocean during the late autumn (Melack & Sickman, 1997). Solute deposition in precipitation (wet deposition) is a function of precipitation amount and solute concentration. Although snow provides most of the water inputs to high-elevation basins, its relative importance in terms of solute deposition is less pronounced because solute concentrations in snow are lower than in rain. For example, the average nitrate concentration in snow at Emerald Lake is an order of magnitude lower than in rain. In Loch Vale in the Rocky Mountains, nitrate concentrations in snow are roughly half that in rain. At Emerald Lake, the winter:summer nitrate deposition ratio ranges from 2 to 4, whereas in Loch Vale, the winter:summer nitrate deposition ratio is close to 1. Total wet deposition is measured at NADP/NTN sites scattered throughout the western U.S., but few of the sites are at high elevation due to difficulties with access and reliable power supplies, as well as restrictions on installation of equipment in wilderness areas (Turk et al., 2001). Data from high-elevation sites often do not meet NADP/NTN quality assurance criteria during winter because the wetfall collectors have a relatively poor capture efficiency for snowfall, especially dry snow that falls during windy conditions. Bulk precipitation collectors, consisting of a continuously open bucket or plastic bag, provide a more reliable

212

M.E. Fenn et al.

source of data with small investment in equipment and installation (Ranalli et al., 1997). Snowpack chemistry surveys have been used to fill in data gaps at high elevation for the winter period (Fig. 8.7; Ingersoll et al., 2002; Melack & Sickman, 1997). Total deposition of water and solutes can be accurately determined from snowpack surveys conducted at maximum

Figure 8.7. Wet, dry, and winter (wet þ dry) deposition at Emerald Lake (upper figure) and wet deposition at seven high-elevation sites in the Sierra Nevada, California. Dry deposition was determined by the inferential method and wet deposition with precipitation collectors. Winter deposition (wet þ dry) was determined by sampling the accumulated snow pack in spring.

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

213

accumulation, typically on or near 1 April. Owing to relatively little or no winter snowmelt above ca. 2500 m (Leydecker & Melack, 1999), total winter deposition of solutes can be estimated with high accuracy (Berg et al., 1989). In addition, because the snow surface is an efficient scavenger of atmospheric aerosols and particles, total wet þ dry deposition of most elements can be estimated with a single, integrated sample of the spring snowpack (Sickman et al., 2001). New types of passive collectors (IER bulk deposition collectors; Section 8.3.2) hold promise for filling data gaps for dry deposition during the summer period (Fenn & Poth, 2004). Snow surveys and passive collectors sample a combination of wet and dry deposition, often referred to as ‘‘bulk deposition.’’ When installed in forested areas, passive collectors may also be installed below the canopy to sample ‘‘throughfall,’’ which reflects total deposition to vegetation. Dry deposition can contribute a significant fraction of total deposition, especially in mountainous areas that receive relatively little summer rain (Fig. 8.7). Approximately one-third of annual N and S deposition can occur as summer dry deposition in drought years in the Sierra Nevada (Sickman et al., 2001) and Rocky Mountains (Campbell et al., 2000). On an annual basis, temporal and spatial variability of wet deposition of N, S, and other solutes is primarily a function of variations in precipitation quantity; variations in chemical concentration explain a lesser amount of depositional variation (Lovett, 1994; Naftz et al., 1994). For example, over the past two decades, non-winter rainfall and winter snowfall at the Emerald Lake watershed have varied by factors of 16.4 and 4.6, respectively. In contrast, annual volume-weighted mean NO3 and SO42 concentrations varied by only 3- to 4-fold during the same period. Similarly, during any year, variations in snow water equivalence among catchments are typically greater than catchment-to-catchment variability of snow chemistry. For example, snow water equivalence ranged from 1200 to over 3000 mm among 15 Sierra Nevada watersheds during 1992–1993, whereas mean snowpack N concentrations varied only from 1.6 to 2.3 mEq L1 at these same sites (Sickman et al., 2001). 8.5.1. Snow sampling methods

Estimating solute deposition in snow involves collecting snow samples for measurement of solute concentration, and multiplying those concentrations by the snowpack water content, which can be manually measured or estimated from remote sensing data. Snow samples may be collected using either the coring method (Campbell et al., 2000) or the snowpit method (Figs. 8.8 and 8.9; Ingersoll et al., 2002; Naftz et al., 2002).

214

M.E. Fenn et al.

Figure 8.8. A snowpit dug for collecting snow samples.

Figure 8.9. Sampling snow from the side of a snowpit. Photos courtesy of K. Elder.

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

215

In either case, it is critical to sample the snowpack before the initiation of snowmelt because of preferential loss of solutes during the first phase of melting (Berg, 1992; Harrington & Bales, 1998). Careful attention also must be paid to proper cleaning of sampling equipment and use of ultraclean sampling techniques due to the very dilute nature of snow (Ingersoll et al., 2002). Strong acids and detergents should be avoided in the cleaning procedure, which primarily consists of soaking apparatus in 18 MO de-ionized water for several days. The QA/QC should include an adequate number of blanks (field and lab) and replicates to ensure that contamination was avoided. The methods described later are pertinent to collection for analysis of major solute chemistry (see references for detailed descriptions of methods). The snow-coring method of snow sampling involves pushing a clean snow-coring device vertically into the snowpack until just above the ground, then pulling the device and (hopefully) the snow core out of the snowpack. Samples are then poured out of the top of the coring device, and into clean plastic or Teflon bags. Bagged samples are then frozen until processed for analysis. Snow samples are generally filtered through 0.45-mm membrane filters (polycarbonate) into bottles that have been soaked in de-ionized water; glass fiber filters should be avoided as they leach cations into solution (see Melack and Sickman, 1997; Ingersoll et al., 2002, for processing and analysis methods). A Federal Sampler can be used as the coring device as long the inside of the sampler has not been treated with chemicals or wax, which is sometimes used to prevent snow from sticking to the inside of the device. One advantage of using a Federal Sampler as the coring device is that it is designed to measure snowpack water content, permitting calculation of total deposition. Although the snow-coring method is simple and fast, it has the disadvantage that contamination by soil, vegetation, and even the coring device is more difficult to avoid than when using the snowpit method. The Federal Sampler is not appropriate for trace metal sampling because it is made of aluminum. Other disadvantages of the snow-coring method are that it provides no information on whether the snowpack has started melting, and deep snowpacks can be extremely difficult to sample because of ice lenses and dense snow that cause under-sampling. The snowpit method of snow sampling involves digging a snowpit from the top of the snow surface to the ground (Fig. 8.8). The face of the pit that faces away from the sun is then shaved back using a clean plastic shovel. A depth-integrated snow sample is extracted from one wall of the pit, avoiding the top and bottom 10 cm of the snowpack to prevent contamination. Tools to perform the actual sampling may include clean plastic shovels, such as those designed for avalanche safety, or clear 2-in

216

M.E. Fenn et al.

PVC pipe. The shovel technique has been extensively tested and used for many years in the annual Rocky Mountain Snowpack Survey and by researchers in the Sierra Nevada (Ingersoll et al., 2002; Sickman et al., 2001). Samples extracted from the pit wall are placed in clean high-density polyethylene or Teflon bags and then frozen until processed for analysis. Snowpack temperatures typically are measured every 10 or 20 cm to verify that the snowpack has not begun to melt. Snow density is also measured at regular intervals (e.g., 10 or 20 cm) to allow calculation of total snowpack water content and subsequently total solute deposition. The snowpit method can be modified slightly to allow collection of largevolume samples for analysis of S or N isotopic composition (Mast et al., 2001) or pesticides and organic contaminants (Landers et al., 2003). Isotopic tracers provide a very useful technique for determining the sources of NO3 and SO42 in environmental solutes (Kendall & McDonnell, 1998; Kester et al., 2003). These methods are often used to ascertain the proportion of NO3 or SO42 in solution or in runoff that comes from atmospheric deposition and from nutrient cycling sources (e.g., nitrification). Nitrate has commonly been traced using the dual isotope technique in which the 15N and 18O signal of NO3 are measured in the major NO3 sources and in watershed runoff samples. More recently, the D17O signal of NO3 has been used to trace atmospheric NO3 and was found to be a more robust tracer of atmospheric NO3 (Michalski et al., 2004) than 15N and 18O methods. As with nitrate, multiple isotopic techniques exist to examine S sources in atmospheric deposition, and biogeochemical processes involving S in watersheds. Measurements of stable isotopes of S and O of SOx have been employed to apportion atmospheric deposition sources (Mast et al., 2001; Mcardle & Liss, 1995; Wadleigh et al., 1996). Radioactive 35S is the most abundant and longest lived of the radioisotopes of S with a half-life of 87 days. The isotope’s primary advantage is that it allows a direct determination of the amount of the present year’s sulfate deposition in surface and groundwater (Michel et al., 2002); older sources of sulfate, whether derived from biogeochemical processes in catchment soils or prior year’s atmospheric deposition, have no 35S. Combining stable and radioisotope measurements of sulfate in catchment studies can provide considerable information regarding the sources and fate of atmospherically deposited S (Kester et al., 2003). 8.5.2. Spatial distribution of snow water equivalent

Because of the strong relationship between snowpack accumulation and atmospheric deposition rates in high-elevation systems, determination of deposition amounts over the landscape requires quantification of

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

217

Figure 8.10. Distribution of point measurements of snow depth and density within the Tokopah Basin located in the alpine region of Sequoia National Park, California. Reproduced from Molotch et al. (2005). Copyright 2004, with permission from John Wiley & Sons Limited.

snowpack or snow water equivalent (SWE). A simple average of spatially distributed observations of snow depth and snow density (Fig. 8.10) can provide a good approximation of basin mean SWE (Cline et al., 1998); however, in many applications, the spatial distribution of SWE is needed as feedbacks with the nitrogen cycle occur spatially. Spatial estimates of SWE have been obtained using statistical interpolation of point measurements (Elder et al., 1998) and using remote sensing data (Carroll & Vadnais, 1980; Chang et al., 1991). The methods used in any particular application largely depend on the scale of the application and the available data within the study region. Given the physiographic heterogeneity of mountainous landscapes, statistical models are limited by a lack of ground-based data for the development of interpolation algorithms that are transferable over regional scales. Similarly, remotely sensed SWE data are limited in mountainous terrain due to the coarse resolution of the available data (e.g., 25 km2) relative to the sub-pixel physiographic heterogeneity of mountainous landscapes. A discussion of remote sensing techniques for measuring SWE is beyond the scope of this chapter, as these techniques have not proved particularly useful in complex rugged terrain. For a review of SWE measurement techniques from satellite and aircraft, the reader is referred to Konig et al. (2001) and Carroll and Vose (1984). Significant progress toward realizing relationships between snow accumulation and physiographic variables has been made where detailed snowpack measurements are available (Fig. 8.11; Balk & Elder, 2000; Elder, 1995; Elder et al., 1995, 1998; Erxleben et al., 2002; Molotch et al.,

218

M.E. Fenn et al. snow water equivalent, cm 200 a. April 180 150 120 90 60 30 0 240 170

b. May

140 120 100 70

N

30 0

0

1

2 km

150 140

c. June

120 100 80 40 20 0 Figure 8.11. Simulated snow water equivalent in the Tokopah Basin (Sequoia national park) in April, May, and June 1997. Reproduced from Molotch et al. (2005). Copyright 2004, with permission from John Wiley & Sons Limited.

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

219

2005; Winstral et al., 2002). When these relationships are statistically significant, binary regression trees (Breiman et al., 1984) have proven to provide the most accurate spatial estimates of SWE (Erxleben et al., 2002). The most commonly used independent variables in these models are net solar radiation, slope, elevation, and in some cases, vegetation type. Improvements have been obtained by including additional independent variables representing wind redistribution of snow (Molotch et al., 2004b; Winstral et al., 2002) and aspect (Erxleben et al., 2002). At the regional scale, or in situations where detailed snowpack observations are not available, statistical interpolation of SNOTEL data is the most widely used technique to estimate the spatial distribution of SWE (Carroll & Carroll, 1993; Carroll & Cressie, 1996; Daly et al., 2000; Fassnacht et al., 2003; Ling et al., 1996; Molotch et al., 2004a). Energy and mass balance models are used by the National Operational Hydrologic Remote Sensing Center (NOHRSC) to estimate SWE across the U.S. (Carroll et al., 2001). In addition to a suite of model forcings, the NOHRSC model requires SWE observations made at SNOTEL sites to update model state variables. The purpose of the SNOTEL network is to provide information for runoff forecasting using empirical relationships between point values of SWE and observed runoff and, therefore, may not provide representative measurements for these spatial applications (Molotch & Bales, 2005). The recent modeling efforts of Nanus et al. (2003) are an excellent example of integration of multiple datasets to assess spatial patterns in atmospheric deposition. Similar work is underway in the eastern U.S. (Weathers, personal communication) and has been published for parts of the U.S. and Europe (Holland et al., 2005). In these studies, precipitation quantity maps were combined with spatially modeled precipitation chemistry to produce deposition maps of N and S for the entire regions. In the work by Nanus et al. (2003), deposition maps were produced for the Rocky Mountain region at a 1-km resolution. Maps of precipitation amounts were created by the Parameter-Elevation Regressions on Independent Slopes (PRISM) model. The model used various precipitation data sets spatially interpolated with a digital elevation model (Daly et al., 2000). Precipitation chemistry was not well correlated to elevation or other terrain variable, so a simple spatial kriging model was used that incorporated both NADP/NTN and snowpack chemical concentrations measured during surveys. Snowpack concentrations were adjusted to annual concentrations by modeling annual vs. snow accumulation season concentrations (Nanus et al., 2003). The deposition maps produced for the Rocky Mountains are an example of creative use of available data on precipitation and wet deposition chemistry. The major

220

M.E. Fenn et al.

limitations of the deposition maps produced by Nanus et al. (2003) are that NH4þ deposition and dry deposition were not included in the analysis. A slight bias for higher concentrations of NH4þ in bulk vs. wet deposition suggests a significant contribution from winter dry deposition of NH4þ (Clow et al., 2002). However, for sulfate and nitrate, co-located snowpack and wetfall measurements have shown little contribution from dry deposition during the winter months in the Rocky Mountains (Clow et al., 2002). The overall approach used by Nanus et al. (2003) can be improved by addition of NH4þ and organic nitrogen and by using additional spatial interpolation techniques and data at finer spatial scales. All the techniques described earlier—and future techniques using remote sensing—rely, to some degree, on ground-based rainfall and SWE observations to update model state or to evaluate precipitation estimates. Hence, the inability to scale point observations to the resolution of remotely sensed data and/or model grid elements is one of the most pressing issues for estimating rainfall and SWE distribution at the regional scale or in inaccessible mountainous regions of the western U.S. 8.6. Conclusions

Innovations in passive monitoring methods for measuring atmospheric pollutant concentrations and deposition inputs in throughfall and bulk deposition have increased our capacity to quantify deposition inputs and variability across the landscape. More studies are needed on methods of estimating total N deposition from throughfall deposition fluxes, particularly in arid zones. Improved atmospheric deposition simulation models and techniques for determining real-time concentrations of gaseous and particulate pollutants are also expected to lead to improved estimates of dry-deposition inputs. Because of the difficulty in accurately measuring dry-deposition inputs to complex landscapes in western North America, the implementation of more than one method is often recommended. Furthermore, the deposition monitoring methods to be used in each situation vary greatly based on spatial scale, climatological conditions, topography, the physical and chemical forms of atmospheric deposition, and vegetative cover properties. ACKNOWLEDGMENTS

We thank Susan Schilling for invaluable assistance in final preparation of several figures. The preparation of this work was supported in part by a National Science Foundation grant (NSF DEB 04-21530).

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

221

REFERENCES Alonso, R., Bytnerowicz, A., and Arbaugh, M. 2001. Vertical distribution of ozone and nitrogenous pollutants in an air quality Class I area, the San Gorgonio Wilderness, southern California. ScientificWorld 1, 10–26. Alonso, R., Arbaugh, M.J., and Bytnerowicz, A. 2003. Forest health research on a natural air pollution gradient in the San Bernardino Mountains, southern California. Ekologia (Bratislava) 22, 18–23. Anderson, J.B., Baumgardner, R.E., Mohnen, V.A., and Bowser, J.J. 1999. Cloud chemistry in the eastern United States, as sampled from three high-elevation sites along the Appalachian Mountains. Atmos. Environ. 33, 5105–5114. Arbaugh, M.J., and Bytnerowicz, A. 2003. Introduction to a regional passive ozone sampler network. In: Bytnerowicz, A., Arbaugh, M.J., and Alonso, R., eds. Developments in environmental science, Volume 2: Ozone air pollution in the Sierra Nevada: Distribution and effects on forests. Elsevier, Amsterdam, The Netherlands, pp. 157–164. Balk, B., and Elder, K. 2000. Combining binary decision tree and geostatistical methods to estimate snow distribution in a mountain watershed. Water Resour. Res. 36, 13–26. Baumgardner, R.E., Lavery, T.F., Rogers, C.M., and Isil, S.S. 2002. Estimates of the atmospheric deposition of sulfur and nitrogen species: Clean Air Status and Trends Network, 1990–2000. Environ. Sci. Technol. 36, 2614–2629. Berg, N., Marks, D., Dawson, D., McGurk, B., Melack, J., Setaro, F., and Bergman, J. 1989. Evaluation of methods for measurement of snowfall and collection of snow for chemical analysis. Final Report, Contract A6-078-32. California Air Resources Board, Sacramento, CA, USA. Berg, N., Dunn, P., and Fenn, M. 1991. Spatial and temporal variability of rime ice and snow chemistry at five sites in California. Atmos. Environ. 25A, 915–926. Berg, N.H. 1992. Ion elution and release sequence from deep snowpacks in the central Sierra Nevada, California. Water Air Soil Pollut. 61, 139–168. Bleeker, A., Draaijers, G., van der Veen, D., Erisman, J.W., Mols, H., Fonteijn, P., and Geusebroek, M. 2003. Field intercomparison of throughfall measurements performed within the framework of the Pan European intensive monitoring program of EU/ICP Forest. Environ. Pollut. 125, 123–138. Brace, S., and Peterson, D.L. 1998. Spatial patterns of tropospheric ozone in the Mount Rainie region of the Cascade Mountains. Atmos. Environ. 32, 3629–3637. Breiman, L., Friedman, J., Olshen, R., and Stone, C. 1984. Classification and Regression Trees. Wadsworth and Brooks, Pacific Grove, CA. Bytnerowicz, A., Fenn, M.E., Miller, P.R., and Arbaugh, M.J. 1999. Wet and dry pollutant deposition to the mixed conifer forest. In: Miller, P.R., and McBride, J.R., eds. Oxidant air pollution impacts in the montane forests of southern California: A case study of the San Bernardino Mountains, Ecological Studies 134. Springer, New York, pp. 235–269. Bytnerowicz, A., Carroll, J.J., Takemoto, B.K., Miller, P.R., Fenn, M.E., and Musselman, R.C. 2000. Distribution and transport of air pollutants to vulnerable California ecosystems. In: Scow, K.M., Fogg, G.E., Hinton, D.E., and Johnson, M.L., eds. Integrated assessment of ecosystem health. Lewis Publishers, Baca Raton, FL, pp. 93–118. Bytnerowicz, A., Padgett, P.E., Arbaugh, M.J., Parker, D.R., and Jones, D.P. 2001. Passive sampler for measurements of atmospheric nitric acid vapor (HNO3) concentrations. ScientificWorld 1, 815–822. Bytnerowicz, A., Arbaugh, M.J., Padgett, P.E., Alonso, R., Fraczek, W., Takemoto, B., Procter, T., Pronos, J., Reiner, J., and Ahuja, S. 2004. Evaluation of ozone and

222

M.E. Fenn et al.

HNO3 vapor distribution and ozone effects on conifer forests in the lake Tahoe Basin and Eastern Sierra Nevada. Final Report to the California Air Resources Board, Sacramento, CA, USA. Bytnerowicz, A., Sanz, M.J., Arbaugh, M.J., Padgett, P.E., Jones, D.P., and Davila, A. 2005. Passive sampler for monitoring ambient nitric acid (HNO3) and nitrous acid (HNO2) concentrations. Atmos. Environ. 39, 2655–2660. Cadle, S.H., Marshall, J.D., and Mulawa, P.A. 1991. A laboratory investigation of the routes of HNO3 dry deposition to coniferous seedlings. Environ. Pollut. 72, 287–305. Campbell, D.H., Baron, J.S., Tonnessen, K.A., Brooks, P.D., and Schuster, P.F. 2000. Controls on nitrogen flux in alpine/subalpine watersheds of Colorado. Water Resour. Res. 36, 37–47. Carroll, S., and Cressie, N. 1996. A comparison of geostatistical methodologies used to estimate snow water equivalent. Water Resour. Bull. 32, 267–278. Carroll, T., and Vadnais, K. 1980. Operational airborne measurement of snow water equivalent using natural terrestrial gamma radiation. Proceedings of the Western Snow Conference, vol. 48, pp. 97–106. Carroll, T., and Vose, G.D. 1984. Airborne snow water equivalent measurements over a forested environment using terrestrial gamma radiation. Proceedings of the Eastern Snow Conference, vol. 29, pp. 101–115. Carroll, T., and Carroll, T.R. 1993. Increasing the precision of snow water equivalent estimates obtained from spatial modeling of airborne and ground-based snow data. Proceedings of the Eastern Snow Conference, vol. 50, pp. 83–87. Carroll, T.R., Cline, D., Fall, G., Nilsson, A., Li, L., and Rost, A. 2001. NOHRSC operations and the simulation of snow cover properties for the coterminous U.S. Proceedings of the Western Snow Conference, vol. 69, pp. 1–14. Chang, A.T.C., Foster, J.L., and Rango, A. 1991. Utilization of surface cover composition to improve the microwave determination of snow water equivalent in a mountain basin. Int. J. Remote Sens. 12, 2311–2319. Chorover, J., Vitousek, P.M., Everson, D.A., Esperanza, A.M., and Turner, D. 1994. Solution chemistry profiles of mixed-conifer forests before and after fire. Biogeochemistry 26, 115–144. Cline, D.W., Bales, R.C., and Dozier, J. 1998. Estimating the spatial distribution of snow in mountain basins using remote sensing and energy balance modeling. Water Resour. Res. 34, 1275–1285. Clow, D.W., Ingersoll, G.P., Mast, M.A., Turk, J.T., and Campbell, D.H. 2002. Comparison of snowpack and winter wet-deposition chemistry in the Rocky Mountains, USA: Implications for winter dry deposition. Atmos. Environ. 36, 2337–2348. Collett, J.L., Jr., Daube, B.C., Jr., and Hoffmann, M.R. 1990. The chemical composition of intercepted cloudwater in the Sierra Nevada. Atmos. Environ. 24A, 959–972. Daly, S.F., Davis, R., Ochs, E., and Pangburn, T. 2000. An approach to spatially distributed snow modeling of the Sacramento and San Joaquin basins, California. Hydrologic. Proc. 14, 3257–3271. Da¨mmgen, U., Erisman, J.W., Cape, J.N., Gru¨nhage, L., and Fowler, D. 2005. Practical considerations for addressing uncertainties in monitoring bulk deposition. Environ. Pollut. 134, 535–548. Davidson, C.I., and Wu, Y.-I. 1990. Dry deposition of particles and vapors. In: Lindberg, A.S., Page, E.L., and Norton, S.A., eds. Acidic precipitation, vol. 3. Sources deposition and canopy interactions. Springer, New York. Davis, J.M., and Swall, J.L. 2006. An examination of the CMAQ simulations of the wet deposition of ammonium from a Bayesian perspective. Atmos. Environ. 40, 4562–4573.

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

223

De Vries, W., and Posch, M. 2003. Critical levels and critical loads as a tool for air quality management. In: Hewitt, C.N., and Jackson, A.V., eds. Handbook of atmospheric science: Principles and applications. Blackwell Science Publishing, Oxford, UK, pp. 562–602. Draaijers, G.P.J., and Erisman, J.W. 1995. A canopy budget model to assess atmospheric deposition from throughfall measurements. Water Air Soil Pollut. 85, 2253–2258. Duncan, L.C. 1991. Chemistry of rime and snow collected at a site in the Central Washington Cascades. Environ. Sci. Technol. 26, 61–66. Elder, K. 1995. Snow distribution in alpine watersheds. PhD dissertation. University of California, Santa Barbara, CA, 309 pp. Elder, K., Michaelsen, J., and Dozier, J. 1995. Small basin modeling of snow water equivalence using binary regression tree methods. In: Tonnessen, K. A., Williams, M.W., and Trantor, M., eds. Biogeochemistry of seasonally snow-covered catchments. IAHS Publication, Wallingford, CO, 228, 129–139. Elder, K., Rosenthal, W., and Davis, R.E. 1998. Estimating the spatial distribution of snow water equivalence in a montane watershed. Hydrologic. Proc. 12, 1793–1808. Erisman, J.W., and Draaijers, G.P.J. 1995. Atmospheric deposition in relation to acidification and eutrophication. Studies in environmental science 63. Elsevier, Amsterdam, The Netherlands. Erisman, J.W., Beier, C., Draaijers, G., and Lindberg, S. 1994. Review of deposition monitoring methods. Tellus 46B, 79–93. Erxleben, J., Elder, K., and Davis, R. 2002. Comparison of spatial interpolation methods for estimating snow distribution in the Colorado Rocky Mountains. Hydrologic. Proc. 16, 3627–3649. Fassnacht, S.R., Dressler, K.A., and Bales, R.C. 2003. Snow water equivalent interpolation for the Colorado River Basin from snow telemetry (SNOTEL) data. Water Resour. Res. 39, doi:10.1029/2002WR001512. Fenn, M.E., and Bytnerowicz, A. 1997. Summer throughfall and winter deposition in the San Bernardino Mountains in southern California. Atmos. Environ. 31, 673–683. Fenn, M.E., and Poth, M.A. 2004. Monitoring nitrogen deposition in throughfall using ion exchange resin columns: A field test in the San Bernardino Mountains. J. Environ. Qual. 33, 2007–2014. Fenn, M.E., Poth, M.A., Schilling, S.L., and Grainger, D.B. 2000. Throughfall and fog deposition of nitrogen and sulfur at an N-limited and N-saturated site in the San Bernardino Mountains, southern California. Can. J. Res. 30, 1476–1488. Fenn, M.E., Baron, J.S., Allen, E.B., Rueth, H.M., Nydick, K.R., Geiser, L., Bowman, W.D., Sickman, J.O., Meixner, T., Johnson, D.W., and Neitlich, P. 2003a. Ecological effects of nitrogen deposition in the western United States. BioScience 53, 404–420. Fenn, M.E., Haeuber, R., Tonnesen, G.S., Baron, J.S., Grossman-Clarke, S., Hope, D., Jaffe, D.A., Copeland, S., Geiser, L., Rueth, H.M., and Sickman, J.O. 2003b. Nitrogen emissions, deposition, and monitoring in the western United States. BioScience 53, 391–403. Fenn, M.E., Geiser, L., Bachman, R., Blubaugh, T.J., and Bytnerowicz, A. 2007. Atmospheric deposition inputs and effects on lichen chemistry and indicator species in the Columbia River Gorge, USA. Environ. Pollut. 146, 77–91. Fenn, M.E., Jovan, S., Yuan, F., Geiser, L., Meixner, T., and Gimeno, B.S. 2008. Empirical and simulated critical loads for nitrogen deposition in California mixed conifer forests. Environ. Pollut. 155, 492–511. Fraczek, W., Bytnerowicz, A., and Arbaugh, M.J. 2003. Use of geostatistics to estimate surface ozone patterns. In: Bytnerowicz, A., Arbaugh, M.J., and Alonso, R., eds. Ozone

224

M.E. Fenn et al.

air pollution in the Sierra Nevada: Distribution and effects on forests. Developments in environmental science 2. Elsevier, Amsterdam, The Netherlands, pp. 215–247. Fuentes, M., and Raftery, A.E. 2005. Model evaluation and spatial interpolation by Bayesian combination of observations with outputs from numerical models. Biometrics 61, 36–45. Geiser, L., and Reynolds, R. 2002. Using lichens as indicators of air quality on federal lands. Workshop Report. USDA Forest Service, PNW Region, R6-NR-AG-TP-01-02. http://ocid.nacse.org/research/airlichen/workgroup/ Gertler, A.W., Bytnerowicz, A., Cahill, T.A., Arbaugh, M., Cliff, S., Kahyaog˘lu-Koracˇin, J., Tarnay, L., Alonso, R., and Fraczek, W. 2006. Local air pollutants threaten Lake Tahoe’s clarity. Calif. Agric. 60, 53–58. Glaubig, R., and Gomez, A. 1994. A simple, inexpensive rain and canopy throughfall collector. J. Environ. Qual. 23, 1103–1107. Grell, G.A., Dudhia, J., and Stauffer, D.R. 1994. A description of the fifth-generation Penn State/NCAR Mesoscale Model (MM5). National Center for Atmospheric Research Technical Note NCAR/TN-398_STR, 138 pp. Available from NCAR, P.O. Box 3000, Boulder, CO, 80307, USA. Hansen, S., and Davies, M.A. 2002. Windshields for precipitation gauges and improved measurement techniques for snowfall. Tech Tip 0225-2325-MTDC. Missoula, MT: U.S. Department of Agriculture, Forest Service, Missoula Technology and Development Center, Missoula, MT, USA. Harrington, R., and Bales, R.C. 1998. Modeling ionic solute transport in melting snow. Water Resour. Res. 34, 1727–1736. Holland, E.A., Braswell, B.H., Sulzman, J., and Lamarque, J.F. 2005. Nitrogen deposition onto the United States and western Europe: Synthesis of observations and models. Ecol. Appl. 15, 38–57. Houle, D., Ouimet, R., Paquin, R., and Laflamme, J.G. 1999. Determination of sample size for estimating ion throughfall deposition under a mixed hardwood forest at the Lake Clair Watershed (Duchesnay, Quebec). Can. J. Res. 29, 1935–1943. Hunsaker, C.T., and Eagan, S.M. 2003. Small stream ecosystem variability in the Sierra Nevada of California. In: Renard, K.G., McElroy, S.A., Gburek, W.J., Canfield, H.E., Scott, R.L., eds. First interagency conference on research in the watersheds, 27–30 October 2003. U.S. Department of Agriculture, Agricultural Research Service, pp. 716–721. http:// www.tucson.ars.ag.gov/icrw/Proceedings/Hunsaker.pdf Ingersoll, G.P., Turk, J.T., Mast, M.A., Clow, D.W., Campbell, D.H., and Bailey, Z.C. 2002. Rocky Mountain snowpack chemistry network: history, methods, and the importance of monitoring mountain ecosystems. U.S. Geological Survey Open-File 01466. Geological Survey Acid Rain Quality-Assurance Project. U.S. Geological Survey Water-Resources Investigations Report 90-4029. Johnson, D.W., and Lindberg, S.E., eds. 1992. Atmospheric deposition and forest nutrient cycling. Ecological studies 91. Springer, New York. Jun, M., and Stein, M.L. 2004. Statistical comparison of observed and CMAQ modeled daily sulfate levels. Atmos. Environ. 38, 4427–4436. Kendall, C., and McDonnell, J.J., eds. 1998. Isotope tracers in catchment hydrology. Elsevier Science, Amsterdam, The Netherlands. Kester, C.L., Baron, J.S., and Turk, J.T. 2003. Isotopic study of sulfate sources and residence times in a subalpine watershed. Environ. Geol. 43, 606–613. Kolian, M., and Haeuber, R. 2004. Clean air status and trends network (CASTNet)—Airquality assessment and accountability. In: Wiersma, G.B., ed. Environmental monitoring. CRC Press, Boca Raton, FL, pp. 685–718.

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

225

Konig, M., Winther, J.G., and Isaksson, E. 2001. Measuring snow and glacier ice properties from satellite. Rev. Geophys. 3, 1–27. Koutrakis, P., Wolfson, J.M., Bunyaviroch, A., Froelich, S.E., Hirano, K., and Mulik, J.D. 1993. Measurement of ambient ozone using a nitrite-coated filter. Anal. Chem. 65, 209–214. Krupa, S.V. 2002. Sampling and physico-chemical analysis of precipitation: A review. Environ. Pollut. 120, 565–594. Krupa, S.V., and Legge, A.H. 2000. Passive sampling of ambient, gaseous air pollutants: An assessment from an ecological perspective. Environ. Pollut. 107, 31–45. Landers, D.H., Simonich, S.L., Campbell, D.H., Erway, M.M., Geiser, L.H., Jaffe, D.A., Kent, M.L., Schreck, C.B., Blett, T.F., and Taylor, H.E. 2003. Western airborne contaminants assessment project research plan: U.S. environmental protection agency, Office of Research and Development, NHEERL, Western Ecology Division, EPA/600/R-03/035. Lawrence, G.B., and Fernandez, I.J. 1993. A reassessment of areal variability of throughfall deposition measurements. Ecol. Appl. 3, 473–480. Leydecker, A., and Melack, J.M. 1999. Evaporation from snow in the central Sierra Nevada of California. Nordic Hydrol. 30, 81–108. Ling, C.H., Josberger, E.G., and Thorndike, A.S. 1996. Mesoscale variability of the upper Colorado River snowpack. Nordic Hydrol. 27, 313–322. Lovett, G.M. 1994. Atmospheric deposition of nutrients and pollutants in North America: An ecological perspective. Ecol. Appl. 4, 629–650. Lovett, G.M., and Lindberg, S.E. 1993. Atmospheric deposition and canopy interactions of nitrogen in forests. Can. J. Res. 23, 1603–1616. Marques, M.C., Gravenhorst, G., and Ibrom, A. 2001. Input of atmospheric particles into forest stands by dry deposition. Water Air Soil Pollut. 130, 571–576. Mast, M.A., Turk, J.T., Ingersoll, G.P., Clow, D.W., and Kester, C.L. 2001. Use of stable sulfur isotopes to identify sources of sulfate in Rocky Mountain snowpacks. Atmos. Environ. 35, 3303–3313. Mcardle, N.C., and Liss, P.S. 1995. Isotopes and atmospheric sulfur. Atmos. Environ. 29, 2553–2556. Mebust, M.R., Eder, B.K., Binkowski, R.S., and Roselle, S.J. 2003. Models-3 community multiscale air quality (CMAQ) model aerosol component, 2. Model evaluation. J. Geophys. Res. 108, 4184, doi:10.1029/2001KD001410. Melack, J.M., and Sickman, J.O. 1997. Monitoring of wet deposition in alpine areas in the Sierra Nevada, Final report, contract, A932-081, California Environmental Protection Agency, Air Resources Board, Research Division, Sacramento, CA, USA. Michalski, G., Meixner, T., Fenn, M., Hernandez, L., Sirulnik, A., Allen, E., and Thiemens, M. 2004. Tracing atmospheric nitrate deposition in a complex semiarid ecosystem using D17O. Environ. Sci. Technol. 38, 2175–2181. Michel, R.L., Turk, J.T., Campbell, D.H., and Mast, M.A. 2002. Use of natural 35S to trace sulphate cycling in a small lake in the Flattops Wilderness Area, Colorado, USA. Water Air Soil Pollut.: Focus 2, 5–18. Molotch, N.P., and Bales, R. 2005. Scaling snow observations from the point to the grid element: Implications for observation network design. Water Resour. Res. 41, W11421, doi:10.1029/2005WR004229. Molotch, N.P., Fassnacht, S.R., Bales, R.C., and Helfrich, S.R. 2004a. Estimating the distribution of snow water equivalent and snow extent beneath cloud-cover in the SaltVerde River basin, Arizona. Hydrologic. Proc. 18, 1595–1611. Molotch, N.P., Painter, T.H., Bales, R.C., and Dozier, J. 2004b. Incorporating remotely sensed snow albedo into a spatially distributed snowmelt model. Geophys. Res. Lett. 31, doi:10.1029/2003GL019063.

226

M.E. Fenn et al.

Molotch, N.P., Colee, M.T., Bales, R.C., and Dozier, J. 2005. Estimating the spatial distribution of snow water equivalent in an alpine basin using binary regression tree models: The impact of digital elevation data and independent variable selection. Hydrologic. Proc. 19, 1459–1479. Mosello, R., Marchetto, A., and Tartari, G.A. 1988. Bulk and wet atmospheric deposition chemistry at Pallanza N. Italy. Water Air Soil Pollut. 42, 137–151. Naftz, D.L., Schuster, P.F., and Reddy, M.M. 1994. Assessment of spatial variability of major-ion concentrations and DEL oxygen-18 values in surface snow, upper Fremont Glacier, WY, USA. Nordic Hydrol. 25, 371–388. Naftz, D.L., Susong, D.D., Schuster, P.F., Cecil, D., Dettinger, M.D., Michel, R.L., and Kendall, C. 2002. Ice core evidence of rapid air temperature increases since 1960 in alpine areas of the Wind River Range, Wyoming, United States. J. Geophys. Res. 107(D13), 4171, doi:10.1029/2001JD000621. Nanus, L., Campbell, D.H., Ingersoll, G.P., Clow, D.W., and Mast, M.A. 2003. Atmospheric deposition maps for the Rocky Mountains. Atmos. Environ. 37, 4881–4892. Ogawa & Co., USA, Inc. 1998. NO, NO2, NOx and SO2 sampling protocol using the ogawa sampler. Users’ Guide, version 4.0, February, 1998. Ogawa & Company, Pompano Beach, FL. Padgett, P.E. 2004. Vegetation as passive collectors y maybe not. In: Douglas, K.E., and Bedient, P.S., eds. Proceedings of National Atmospheric Deposition Program Technical Committee Meeting and Scientific Symposium. 21–23 September 2004, Halifax, Nova Scotia, Canada, 46 pp. http://nadp.sws.uiuc.edu/lib/proceedings/NADPpro2004.pdf Padgett, P.E., and Bytnerowicz, A. 2001. Deposition and adsorption of the air pollutant HNO3 vapor to soil surfaces. Atmos. Environ. 35, 2405–2415. Padgett, P.E., Allen, E.B., Bytnerowicz, A., and Minich, R.A. 1999. Changes in soil inorganic nitrogen as related to atmospheric nitrogenous pollutants in southern California. Atmos. Environ. 33, 769–781. Padgett, P.E., Bytnerowicz, A., Dawson, P.J., Riechers, G.H., and Fitz, D.R. 2004. Design, evaluation and application of a continuously stirred tank reactor system for use in nitric acid air pollutant studies. Water Air Soil Pollut. 151, 35–51. Pathak, R.K., and Chan, C.K. 2005. Inter-particle and gas-particle interactions in sampling artifacts of PM2.5 in filter-based samplers. Atmos. Environ. 39, 1597–1607. Pleim, J.E., Xiu, A., Finkelstein, P.L., and Otte, T.L. 2001. A coupled land-surface and dry deposition model and comparison to field measurements of surface heat, moisture, and ozone fluxes. Water Air Soil Pollut.: Focus 1, 243–252. Ranalli, A.J., Turk, J.T., and Campbell, D.H. 1997. The use of bulk collectors in monitoring wet deposition at high-altitude sites in winter. Water Air Soil Pollut. 95, 237–255. Ray, J.D. 2001. Spatial distribution of tropospheric ozone in national parks of California: Interpretation of passive-sampler data. ScientificWorld 1, 483–497. Roadman, M.J., Scudlark, J.R., Meisinger, J.J., and Ullman, W.J. 2003. Validation of Ogawa passive samplers for the determinations of gaseous ammonia concentrations in agricultural settings. Atmos. Environ. 37, 2317–2325. Savard, M. 2005. Sulfate and nitrate deposition to balsam fir measured across an elevational gradient on Mt. Washington, White Mountains, New Hampshire. Undergraduate Thesis. Princeton University, Department of Ecology and Evolutionary Biology, Princeton, NJ, USA. Schmitt, M., Thoni, L., Waldner, P., and Thimonier, A. 2005. Total deposition of nitrogen on Swiss long-term forest ecosystem research (LWF) plots: Comparison of the throughfall and the inferential method. Atmos. Environ. 39, 1079–1091.

Methods for Measuring Atmospheric Nitrogen Deposition Inputs

227

Scire, J.S., Strimaitis, D.G., and Yamartino, R.J. 2000. A user’s guide for the CALPUFF dispersion model (Version 5.4). Earth Tech, Inc., Concord, MA. Seinfeld, J.H., and Pandis, S.N. 1998. Atmospheric chemistry and physics. Wiley, New York. Shanley, J.B. 1989. Field measurements of dry deposition to spruce foliage and Petri dishes in the Black Forest, F.R.G. Atmos. Environ. 23, 403–414. Sickman, J.O., Leydecker, A., and Melack, J.M. 2001. Nitrogen mass balances and abiotic controls on N retention and yield in high-elevation ecosystems of the Sierra Nevada, California, USA. Water Resour. Res. 37, 1445–1461. Simkin, S.M., Lewis, D.N., Weathers, K.C., Lovett, G.M., and Schwarz, K. 2004. Determination of sulfate, nitrate, and chloride in throughfall using ion-exchange resins. Water Air Soil Pollut. 153, 343–354. Simpson, D., Fagerli, H., Hellsten, S., Knulst, J.C., and Westling, O. 2006. Comparison of modelled and monitored deposition fluxes of sulphur and nitrogen to ICP-forest sites in Europe. Biogeosciences 3, 337–355. Swall, J.L., and Davis, J.M. 2006. A Bayesian statistical approach for the evaluation of CMAQ. Atmos. Environ. 40, 4883–4893. Tang, H. 2001. Introduction of Maxxam all-season passive sampling system and principles of proper use of passive samplers in the field study In: Proceedings of the International Symposium on Passive Sampling of Gaseous Air Pollutants in Ecological Effects Research, 9 April 2001, Riverside, California, USA. ScientificWorld 1, 463–474. Thimonier, A. 1998. Measurement of atmospheric deposition under forest canopies: Some recommendations for equipment and sampling design. Environ. Monit. Assess. 52, 353–387. Tonnesen, G., and Wang, Z. 2005. Assessment of nitrogen deposition: Modeling and habitat assessment, Draft Report, California Energy Commission, Sacramento, CA, USA. Tonnesen, G., Wang, Z., Omary, M., and Chien, C.J. 2003. Formulation and application of regional air quality modeling for integrated assessments of urban and wildland pollution. In: Bytnerowicz, A., Arbaugh, M.J., and Alonso, R., eds. Developments in environmental science, volume 2: Ozone air pollution in the Sierra Nevada: Distribution and effects on forests. Elsevier, Amsterdam, The Netherlands, pp. 299–324. Turk, J.T., Taylor, H.E., Ingersoll, G.P., Tonnessen, K.A., Clow, D.W., Mast, M.A., Campbell, D.H., and Melack, J.M. 2001. Major-ion chemistry of the Rocky Mountain snowpack, USA. Atmos. Environ. 35, 3957–3966. Wadleigh, M.A., Schwarcz, H.P., and Kramer, J.R. 1996. Isotopic evidence for the origin of sulphate in coastal rain. Tellus Series B 48, 44–59. Waldman, J.M., Munger, J.W., Jacob, D.J., and Hoffmann, M.R. 1985. Chemical characterization of stratus cloudwater and its role as a vector for pollutant deposition in a Los Angeles pine forest. Tellus 37B, 91–108. Weathers, K.C. 1999. The importance of cloud and fog to the maintenance of ecosystems. Trends Ecol. Evol. 14, 214–215. Weathers, K.C., and Likens, G.E. 1997. Clouds in southern Chile: An important source of nitrogen to nitrogen-limited ecosystems? Environ. Sci. Technol. 31, 210–213. Weathers, K.C., Likens, G.E., Bormann, F.H., Bicknell, S.H., Bormann, B.T., Daube, B.C., Eaton, J.S., Galloway, J.N., Keene, W.C., Kimball, K.D., McDowell, W.H., Siccama, T.G., Smiley, D., and Tarrant, R.A. 1988. Cloudwater chemistry from 10 sites in North America. Environ. Sci. Technol. 22, 1018–1026. Weathers, K.C., Lovett, G.M., and Likens, G.E. 1992. The influence of a forest edge on cloud deposition. In: Schwartz, S.E., and Slinn, W.G.N., eds. Precipitation scavenging and atmosphere–surface exchange. Vol. 3—The summers volume: Applications and appraisals. Hemisphere Publishing Corporation, Bristol, PA, pp. 1415–1423.

228

M.E. Fenn et al.

Weathers, K.C., Lovett, G.M., and Likens, G.E. 1995. Cloud deposition to a spruce forest edge. Atmos. Environ. 29, 665–672. Weathers, K.C., Lovett, G.M., Likens, G.E., and Lathrop, R. 2000. The effect of landscape features on deposition to Hunter Mountain, Catskill Mountains, New York. Ecol. Appl. 10, 528–540. Weathers, K.C., Cadenasso, M.L., and Pickett, S.T.A. 2001. Forest edges as nutrient and pollutant concentrators: Potential synergisms between fragmentation, forest canopies, and the atmosphere. Conserv. Biol. 15, 1506–1514. Weathers, K.C., Simkin, S.M., Lovett, G.M., and Lindberg, S.E. 2006. Empirical modeling of atmospheric deposition in mountainous landscapes. Ecol. Appl. 16, 1590–1607. Wesely, M.L., and Hicks, B.B. 2000. A review of the current status of knowledge on dry deposition. Atmos. Environ. 34, 2261–2282. Williams, M.W., Bardsley, T., and Rikkers, M. 1998. Overestimation of snow depth and inorganic nitrogen wetfall using NADP data, Niwot Ridge, Colorado. Atmos. Environ. 32, 3827–3833. Winstral, A., Elder, K., and Davis, R.E. 2002. Spatial snow modeling of wind-redistributed snow using terrain-based parameters. J. Hydrometeorol. 3, 524–538.

Developments in Environmental Science, Volume 9 Allan H. Legge (Editor) Copyright r 2009 Elsevier Ltd. All rights reserved. ISSN: 1474-8177/DOI:10.1016/S1474-8177(08)00209-X

229

Chapter 9 Air Quality Changes in an Urban Region as Inferred from Tree-Ring Stable Isotopes Martine M. Savard, Christian Be´gin, Joe¨lle Marion, Jean-Christophe Aznar and Anna Smirnoff Abstract This chapter constitutes the first assessment of d18O values of stem cellulose as an indicator of stress for trees exposed to pollution, and of d13C combined with d18O values to evaluate the relative impact of changes in climatic conditions and in air quality in a peri-urban region. The results obtained near Montreal (Canada) indicate that d13C and d18O characteristics in pine (Pinus spp.) trees record changes in air quality, whereas d13C patterns clearly change in beech (Fagus spp.) trees under stressed conditions. Therefore, dendroisotopic analyses offer a potential method for detecting past changes in air quality and evaluating forest responses to pollution.

9.1. Introduction

There is no comprehensive understanding of the past and current impacts of pollution on carbon (C) uptake by forests, but pollution studies foretell unprecedented ozone (O3) levels in the coming decades (e.g., Krupa et al., 2004). Under natural conditions, stem growth is mainly controlled by nutrient availability, climatic conditions, and ecological conditions (disease, forest dynamics). It is also known that the d18O (and d2H) values in trees are influenced by temperature (e.g., Feng & Epstein, 1995), precipitation (e.g., Saurer et al., 1997), and relative humidity (e.g., Shu et al., 2005), and that the d13C ratios can reflect changes in local amounts of precipitation (e.g., Lipp et al., 1991), or soil water (Dupouey et al., 1993). Numerous experiments have demonstrated that high concentrations of tropospheric

Corresponding author: E-mail: [email protected]

230

Martine M. Savard et al.

O3 and sulfur dioxide (SO2) can impair physiological functions of terrestrial vegetation and modify the fractionation of C isotopes (e.g., Martin et al., 1988; Saurer et al., 1995). The general isotopic response is a reduced discrimination against 13C due to stomatal closure, which creates higher d13C values in tree rings or leaf tissues (e.g., Matyssek et al., 1998). This is in accordance with a prediction of higher d13C related to lower conductance in tissues of plants exposed to stress (Farquhar et al., 1989). A few studies have also shown that trees exposed to point-source pollution under real field conditions increase their d13C values (Freyer, 1979; Martin & Sutherland, 1990; Savard et al., 2004). The water isotopes present in trees (2H/1H and 18O/16O) have never been the targets of controlled experiments simulating pollution stress. However, the study by Savard et al. (2005) under field conditions has shown that even hydrogen isotopes can be affected during periods of stress induced by high SO2 concentrations from a point source. These authors have proposed combining tree-ring chronologies with C and water isotopes to infer pollution stress through time. The developed model for a time series of trees exposed to smelter emissions comprises coinciding inverse shifts of d2H and d13C long-term trends (Fig. 9.1). Urban pollutants that are phytotoxic include: carbon monoxide (CO), SO2, nitrogen oxides (NOx), O3, peroxyacetyl nitrate (PAN), hydrogen sulfide (H2S), hydrogen fluoride (HF), chlorine (Cl), etc. Their combined impact on C stable isotope fractionation has only been evaluated in a few studies associated with diffuse gas aureoles of peri-urban regions (Niemela et al., 1997; Sakata & Suzuki, 2000). Hitherto, there has been no study

Figure 9.1. Inverse variation model of H (or O) and C isotope ratios through time developed for trees exposed to a phytotoxic gas from a point source (Savard et al., 2005). Pre-pollution period showing natural isotopic behavior contrasting with the syn-pollution period characterized by stressed responses. Vertical line indicates the beginning of critical exposure to pollutants.

Urban Air Quality and Tree-Ring Stable Isotopes

231

combining C and water isotopes to understand the impact of mixtures of urban pollution on trees and to discriminate isotopic changes generated by natural conditions from those generated by atmospheric pollution. Here, we determine whether the d13C and d18O values of field-growing trees exposed to urban pollution will shift according to the fashion documented by Savard et al. (2005; Fig. 9.1). The specific objectives of the present chapter are to (1) verify the sensitivity of oxygen (O) isotopes to urban pollution; (2) infer the periods during which the trees have grown under air pollution stress; and (3) assess the combined use of CþO isotope ratios to detect past changes in air quality in a peri-urban forest exposed to diffuse pollution, knowing that historical data on pollution are scarce.

9.2. A regional setting

The metropolitan Montreal area in southwestern Quebec (Fig. 9.2a) belongs to the St. Lawrence Lowlands physiographic region, in which the Windsor-Quebec City corridor is highly populated and industrialized, and is the locus of the highest pollution levels in Canada (Pollution Probe, 2002). Climatic conditions have been compiled from five meteorological stations located in a 25-km radius around Montreal to complete a series covering the period 1880–2002 (Environment Canada stations: 701490, 7024400, 7025250, 7025280, 7026839). Similarities of the overlapping periods between the individual climatic series were assessed before averaging the climatic data to obtain a representative regional series. Field measurements of atmospheric concentrations for some phytotoxic pollutants are only available for a maximum of 30 years in the area (Environment Canada data). Centenary and older trees of coniferous and deciduous species growing in large populations under field conditions were found at the Arboretum Morgan, a protected research facility of McGill University located in a low-density residential area on the west end of Montreal Island (Fig. 9.2a). The selected site covers a 0.3-km2 area, and is dominated by mixed old stands growing on brunisolic soils developed in sandy alluvial deposits. American beech (Fagus grandifolia Ehrh. and white pine (Pinus strobus L.), the co-dominant specimens selected for the study, had a healthy appearance and did not show visible symptoms of decline. At the Quebec City end of the Windsor-Quebec City corridor, 250 km downwind from Montreal, studies have shown that up to 50% of some pollutants (e.g., lead (Pb)) were transported over long distances from southwestern sources located in the U.S. (e.g., Gallon et al., 2005). In Montreal, pollutants from those distant sources mix with locally produced

232

Martine M. Savard et al.

Figure 9.2. (A) Location map of the study region (Montreal) within the highly polluted Windsor-Quebec City corridor (gray band). The inset map shows the Montreal area, the location of the studied site (Arboretum Morgan; black dot), and three air-quality monitoring sites of the Montreal network where ozone (a and b) and sulfur dioxide (c) have been monitored. (B) Graph showing historical data from the sites shown in (A) (O3 in a and b; SO2 in c). The dotted line shows the curve of modeled SO2 levels for the Montreal region (see text for details).

Urban Air Quality and Tree-Ring Stable Isotopes

233

atmospheric contaminants. Consequently, the Arboretum Morgan trees were likely exposed to toxic levels of various atmospheric pollutants during their growth. From 1970 to 2005, the number of cars in the province of Quebec rose from 1.58 million to 3.78 million (138% increase), significantly increasing emissions due to hydrocarbon combustion (Ministe´re des transports, 1994; Socie´te´ de l’assurance automobile du Que´bec (SAAQ), 2006). An increase of the same proportion may be assumed for the Montreal region. Recent air quality monitoring indicates that downtown Montreal SO2 concentrations were between 107.0 and 32.4 mg m3 during 1971–1982 (Environment Canada, curve not shown here), exceeding the Canadian Annual Maximum Desirable National Ambient Air Quality Objective of 30 mg m3 (Barker & Barker, 1988). The regional monitoring network for the large Montreal area recorded SO2 concentrations diminishing by 90% between 1971 and 1990 (Environment Canada data; Fig. 9.2b). However, during that same period, O3 levels were above the lowest observable adverse effect levels (LOAEL; SUM60) for acute and chronic impacts on tree species for several discontinuous time intervals (Environment Canada data). At monitoring station ‘‘a’’ (Fig. 9.2a), 1.7 km southeast of the Arboretum Morgan, summer concentrations of O3 exceeded the Canadian norm of 30 mg m3 (Fig. 9.2b). Briefly, air quality conditions during the 1980s and 1990s can be depicted as if the O3 level significantly increased while the SO2 concentration decreased below critical levels. There is no record of the atmospheric concentrations of SO2, O3, or other potentially toxic pollutants for the pre-1970 years in the Montreal area (Fig. 9.2b). Thus, the historical estimation of air quality cannot be based on measurements but only on pollution proxies. Sulfur dioxide emissions were modeled and presented as representative for Canada by Be´langer (2000). The modeled SO2 level is expressed as a percentage (0–100) of the maximum emissions from Lefohn et al. (1999) and Environmental Protection Agency (EPA) (1998) records. This modeled SO2 air content (Be´langer, 2000) is assumed here to reflect pollution levels in the Montreal area, and serves as a proxy for the complex mixture of air contaminants produced by industrial activities in the region and those transported over long distance from southwestern industrial centers. Given the historical changes during 1980s and 1990s described earlier, it is clear that modeled SO2 constitutes a good proxy for air quality for the pre-1980 period. The effect of individual contaminants on trees may have increased in combination with other chemicals as exposure of plants to mixtures of potentially phytotoxic pollutants can reduce the threshold at which impacts are individually detected (e.g., Darrall, 1989).

234

Martine M. Savard et al.

9.3. Analytical procedures 9.3.1. Sampling of trees and dendro-chronology

More than 12 specimens of each species were carefully chosen for dendrochronological analysis. Among those trees, three specimens were selected for isotope dendro-geochemistry for each species (Table 9.1). Six cores of 5 mm diameter equally distributed around the stem were collected at 1.4 m from the soil surface on the three specimens selected for dendro-geochemistry, and one or two cores on additional specimens for dendro-chronology. The beech and pine cores selected for tree-ring measurements were handled using standard dendro-chronological methods. Dating was subsequently verified using the COFECHA program (Holmes, 1992). Basal area increment (BAI) of stems was calculated using ring width, for which measurement precision was 0.001 mm. This procedure involved 6 and 10 oldest trees (103 years or older) of the beech and pine populations, respectively (Table 9.1), in order to compare trees representative of the oldest successional status. This way we avoid a bias related to growth rates typical of younger trees during BAI series construction. The suppression period related to youth was removed from individual BAI curves following the method described by Duchesne et al. (2003). The cut-off values for beech and pine specimens were 1 and 2 cm2, respectively. 9.3.2. Isotope geochemistry

Wood sampling involved mechanical separation of ring pairs with stainless steel blades, with the initial year of each pair being even-numbered. The same pair for all cores of a given specimen was pooled for geochemistry. Homogenization of the complete ring sub-samples and the analyses for C and O stable isotopes were carried out at the Delta-Lab of Natural Resources Canada (Geological Survey of Canada, Quebec). Ring pairs were assessed to yield a resolution sufficient for this pilot project on urban pollutant effects (Savard et al., 2005). The isotope values were obtained Table 9.1. Summary of statistics obtained for trees selected for the BAI and isotope studies Tree species

Beech Pine

Number of trees dendrochronology/BAI/ isotope analyses

Period covered

12/6/3 20/10/3

1838–2003 1867–2003

d13C

d18O

Number of subsamples

EPS

s

EPS

s

EPS

s

378 366

0.95 0.80

0.2 0.3

0.76 0.80

0.2 0.3

0.90 0.50

0.2 0.3

BAI

Urban Air Quality and Tree-Ring Stable Isotopes

235

from extracted cellulose in order to avoid inconsistent isotopic variations caused by changing proportions of wood constituents (Savard et al., 2004, 2005). All extracted cellulose sub-samples were analyzed for d13C using an elemental analyzer in continuous flow with an Isotope Ratio Mass Spectrometer (IRMS). Calibration was performed using two of three international standards (NBS-19, or IAEA-CH6 and IAEA-CH7). The international standards and our internal cellulose standard were used with each run to ensure that there was no analytical drift. The precision of the analyses (n ¼ 57) and the accuracy were both better than 0.2m over the study period. Analysis of O isotopes was performed using a pyrolysis-elementalanalyzer online with an IRMS. Calibration was performed using the IAEA-SO6 international standard and SELENA internal standard, and later verified with each run using numerous international standards (IAEA-SO5, IAEA-C3, and NBS-127) as well as another internal standard (Vanillin). The precision of the analyses was better than 0.1m (n ¼ 22), and the averaged accuracy of the standards was always better than 0.3m over the study period. The d18O and d13C results are all reported in values relative to VSMOW and VPDB, respectively. All results are normalized relative to the average values for each specimen so as to minimize differences induced by individual metabolic effects on isotope fractionation and to easily compare the isotopic behavior of the trees between species (Table 9.1). The d13C results were also corrected for the diminishing d13C values of atmospheric carbon dioxide (CO2) by simply subtracting the atmospheric value from the one obtained for each ring pair. 9.3.3. Statistics on time series

For each selected species, the expressed population signal (EPS) is calculated (Table 9.1). The EPS is an evaluation of the common signal between time series from different specimens (Briffa & Jones, 1989). It is used to assess the minimum number of trees necessary to calculate an average representative for a given stand. The standard deviation (SD) is also calculated for the entire time series as the difference between the maximum and minimum values for a given ring pair, divided by two (Table 9.1). Pearson correlation coefficients of pairs of selected variables were evaluated to illustrate statistics on specific graphs. For both temperature and precipitation, an analysis was carried out using individual months from the July–September period before the year of growth. Months of the previous year are used because there can be a strong relationship between ring growth and the previous summer’s climatic conditions. Afterward, a combination of

236

Martine M. Savard et al.

months was tested and the best fits were obtained for the average of the last 14 months for temperature (from July to August) and cumulative precipitation from January to September. Subsequently, statistical links were established with a regression analysis of the d13C, d18O, and BAI sets of data. To assess the influences of climatic fluctuations and SO2 levels on their variations, models with autoregressive errors were fitted Y t ¼ b0 þ b1 X 1 þ b2 X 2 þ Zt where Zt is an ‘‘autoregressive’’ term Z t ¼ pZ tn þ t The term et represents the portion of Zt that could not have been predicted from previous values (Monserud & Marshall, 2001). To avoid co-linearity problems, correlations between regressors were checked. The degree of non-independence in time-series data was estimated by autocorrelation between observations. First- and higher-order autocorrelations in the individual time series were determined with PROC ARIMA (SAS Institute, 1999). Both first- and second-order autocorrelations were significant so the lag for autoregressions was set equal to two. Datasets were examined for seasonal and other forms of trends that violate assumptions of standard autoregression analysis by performing a Phillips– Perron unit root test on each variable. Regressions were significant for climatic variables (temperature and precipitation), and the term ‘‘year’’ was included as an independent variable in the autoregressive models. Finally, equality of variances between observations was tested using the AUTOREG ARCHTEST option. Autoregressive estimates were calculated by the maximum likelihood method in PROC AUTOREG. 9.4. Results

An EPS of 0.8 or more is generally considered acceptable for expressing the common signal between time series from different specimens (Briffa & Jones, 1989). All EPS values obtained for the trees of the Arboretum Morgan site are above or near this threshold, except for d18O results for pine trees (Table 9.1; see later). The BAI, and C and O isotope variability presented by the selected specimens is 0.2 and 0.3 for beech and pine trees, respectively (s, Table 9.1). 9.4.1. Carbon isotope analysis

The carbon isotope ratios in ring pairs of pine trees generally match those obtained for beech trees for the 1880–2003 period (Fig. 9.3a).

Urban Air Quality and Tree-Ring Stable Isotopes

237

Figure 9.3. Analytical results for beech and pine trees growing in the Montreal metropolitan region as a function of time (years). (a) Carbon isotope values (average of three trees per species). (b) Oxygen isotope values (average of three trees per species). (c) Basal area increment (average of 6 beech and 10 pine trees).

The long-term trend of the two curves exhibits a flat trend until 1922–1923, followed by a gentle positive slope between 1924–1925 and 1956–1957. After that period, the curves are generally flat, with a slightly negative slope after 1984–1985. For both species, the long-term peak of the curve appears between 1958–1959 and 1984–1985. Short-term

238

Martine M. Savard et al.

fluctuations of 70.3m are superimposed on the long-term trend of the two species. 9.4.2. Oxygen isotope analysis

The d18O curves obtained for the pine and beech trees show notable shortterm differences superimposed on a gentle long-term rise for the ring pairs formed between 1880–1881 and 1932–1933 (Fig. 9.3b). After that, the beech curve is flat until 1964–1965, then displays low values until 1980–1981, whereas the pine curve exhibits a negative slope until 1972–1973. The variation among the three individual pine series is minimal after 1926–1927, and the low EPS value mentioned earlier reflects significant departures between the three curves occurring during the pre-1926 period. This observation suggests that the post-1927 portion of the pine d18O series can be used and interpreted along with the average d13C results of the two tree species. After 1981, the beech and pine d18O curves show oscillations, but their low d18O values appearing between 1958–1959 and 1980–1981 are largely synchronous with the episode of peak d13C values (Fig. 9.3a). 9.4.3. Basal area increments (BAIs)

The average variations of the pine BAI show short-term variations superimposed on a steady increase between 1880–1881 and 1932–1933, followed by a decreased incremental growth between 1934–1935 and 1948–1949 and a prolonged flat trend between 1950–1951 and 2001–2002 (Fig. 9.3c). In contrast, the average BAI for beech trees presents a gentle long-term rise between 1880–1881 and 1922–1923, followed by an increased growth rate until 1952–1953. The long-term changes after that include a negative slope until 1962–1963 and a slightly positive slope for the remaining series. The important aspect here is that the pine series exhibits a reduction in growth after 1932–1933, which is broadly coeval with relatively high d13C trends and low or flat d18O trends obtained for the pine and beech tree rings. 9.5. Interpretations and discussion

Identification of the periods during which the investigated trees have been impacted by air pollution requires partitioning isotope shifts caused by pollution (modeled SO2) from those caused by climate stresses. The regression analysis partly expresses these influences on the d13C and d18O results (Table 9.2). The beech trees present d18O variations mostly controlled by climatic conditions as these isotopic changes correlate directly with mean

Beech d18O r2 Intercept (b0) Year Precipitation (b1) Temperature (b2) SO2 (b3) AR1 AR2

0.38 3.67675.864 0.00170.003 0.00270.001 0.36070.132 0.27970.482 0.10970.148 0.24370.148

Pine d13C

0.534 0.776 0.005 0.009 0.566 0.580 0.064

0.41 11.75775.453 0.00770.003 0.00070.001 0.19170.110 1.67670.480 0.31870.152 0.09470.152

d18O

0.037 0.021 0.880 0.089 0.001 0.026 0.517

0.39 15.97575.327 0.00870.003 0.00270.001 0.16970.129 1.10470.453 0.10670.153 0.00770.153

d13C

0.005 0.010 0.017 0.197 0.019 0.476 0.942

0.67 12.17277.745 0.00670.004 0.00270.001 0.07570.089 2.13870.591 0.51770.152 0.06170.152

0.123 0.124 0.007 0.401 0.001 0.002 0.445

Urban Air Quality and Tree-Ring Stable Isotopes

Table 9.2. Regression analysis of isotopic time series

Note: r2 of model, coefficient, error, and probability are shown in sequence for the estimated parameters (January–September precipitation, 14-month average temperature, modeled SO2) in relation to each isotope ratio (corrected values). Bold probabilities indicate terms at po0.05.

239

240

Martine M. Savard et al.

Figure 9.4. Comparison of climatic conditions with isotopic results: (a) 14-month average temperature (previous July to August) and d18O values for beech trees as a function of time (years); (b) relationship between January–September total precipitation and d13C values for pine pairs of rings produced during the 1880–2003 period.

temperature and inversely with precipitation (Fig. 9.4a; Table 9.2). For the d13C results, the analysis reveals a direct correlation with level of pollution and temperature (Table 9.2; Fig. 9.5a). In pine trees, the d18O values vary inversely with modeled atmospheric SO2 levels and total precipitation, whereas the d13C values vary directly with the proxy for the level of pollution (Fig. 9.5b), but inversely with total precipitation (Table 9.2). It is clear from these statistics that climatic conditions controlled most d18O variations in beech trees, but that climatic conditions and pollution effects should be invoked to explain the changes in d13C values of beech trees, and in d13C and d18O values of pine trees. The regression analysis does not detect significant correlations of BAI with climate or pollution data for the two studied species (therefore not listed in Table 9.2), but a long-term decline in growth rate after 1932–1933 exists for pine trees (BAI change; Fig. 9.3c). Comparison with a control curve would have been useful to interpret this relative decline, but it is impossible to find trees growing under similar ecological conditions in regions not exposed to atmospheric phytotoxic pollutants. We can only note that the measurement of reduced BAI in Montreal pine trees broadly coincides with the period of high pollution levels (Figs. 9.3 and 9.5c). We believe that the isotopic fractionation of the Arboretum Morgan trees responded to pollution stress and climatic conditions concurrently. Beech trees show a relatively low long-term discrimination against 13C

Urban Air Quality and Tree-Ring Stable Isotopes

241

Figure 9.5. Stress-induced responses in trees. Values of d13C as a function of modeled SO2 level showing: (a) a direct relationship in beech trees; (b) a stronger direct correlation in pine trees; and (c) combined curves of d13C and d18O values in pine trees as a function of time (years) and compared with the modeled SO2 level. The arrows indicate isotopic responses to pollution stress.

after 1930–1931, and lower d18O values with flat and negative trends between 1940–1941 and 1980–1981 (Figs. 9.3a and 9.5a; Table 9.2). Note that the d13C and d18O values of pine trees, and perhaps the BAI as well, have been modified by pollution stress (Table 9.2; Figs. 9.3 and 9.5). This observation indicates that this species could represent a better archival system for changes in air quality than beech specimens. We propose that the persistence of foliage in coniferous trees is one of the factors explaining the sensitivity of this species to air pollution. After 1934–1935, the long-term d13C and d18O trends of pine trees vary inversely (Fig. 9.5c), as expected for specimens exposed to atmospheric pollution (Fig. 9.1). According to this isotopic model, pine trees recorded the most severe effects between 1942–1943 and 1984–1985. This hypothesis is supported by the fact that the mentioned period coincides with the highest pollution levels to this date (Fig. 9.5c). Responses of trees to high concentrations of pollutants tend to increase mesophyll resistance to diffusion of CO2, as well as change the photosynthetic rate, diminish the absorption of gaseous pollutants by premature senescence and abscission of foliage or by reduction of leaf conductance and stomatal density (Kropff, 1987; Matyssek et al., 1998;

242

Martine M. Savard et al.

Winner, 1989). Increased activities of PEP-carboxylase relative to rubisco and lowered root:shoot ratios can also occur as a result of gaseous pollution stress (Matyssek et al., 1998). We propose that the mechanisms responsible for the changes in the properties of the Arboretum trees not attributable to climate conditions include reduction of leaf conductance and perhaps modifications of enzymatic functions. The inverse isotopic long-term trends are certainly compatible with a lower leaf conductance, which could be caused by higher pH causing stomata closure (i.e., guardcell closure). Under these conditions, the amount of CO2 available for photosynthesis is reduced, reducing discrimination against 13C and increasing d13C values, and evapo-transpiration is reduced, retaining more 16O (or 1H) within the leaves and consequently reducing the d18O values (or d2H values; Savard et al., 2005). Such responses by trees translate into a lowered net assimilation of C by the exposed forests in the Montreal region, which is not necessarily expressed in their stems. The boreal and mixed forests transferring C from the atmosphere to the terrestrial biosphere constitute an important C sink. Recent studies suggest an increase in net photosynthetic production for trees in response to higher atmospheric concentrations of CO2 and N-chemicals (e.g., Ladeau & Clark, 2001). However, increased productivity might only occur in regions not exposed to phytotoxicity because the presence of such pollutants as SO2 and O3 can impair assimilation of C by trees (Sakata & Suzuki, 2000; Savard et al., 2004; Savard et al., 2005; present study).

9.6. Conclusions

This stable isotope study of trees exposed to urban diffuse pollution near Montreal illustrates how long-term stress responses are specific to species of trees after 1930. Beech trees show d18O and BAI ring series mostly controlled by climatic and ecological conditions, but d13C values clearly responding to both air quality changes and climatic variations. Pine trees, however, show d13C and d18O series all partly controlled by climate and pollution stress. In fact, pine trees show longterm isotopic shifts similar to those previously documented for pollution stress, and that identify changes generated by natural conditions and atmospheric pollution. This interpretation is supported by the fact that the inferred long-term period of stress broadly corresponds to a period of reduced growth rate for pines and to peak pollution caused by a complex mixture of pollutants using modeled SO2 levels as a proxy.

Urban Air Quality and Tree-Ring Stable Isotopes

243

Oxygen isotopes in pine tree rings are sensitive to urban pollution and, when combined with C isotopes, can be used to detect past changes in air quality. We propose that the pollution effects on trees include a reduction of foliar conductance and perhaps modifications of enzymatic functions. We hope that, despite the paucity of historical data concerning air pollution patterns generated by large cities, past urban effects on forest performance of other regions can be inferred using our approach combining dendro-chronology with stable isotope geochemistry. ACKNOWLEDGMENTS

We thank Marc Luzincourt for participating in the preparation of the cellulose extracts; Kristina Idziac and Benoıˆ t Coˆte´ from the Arboretum Morgan (McGill University) for their support; Tom Dann and Melanie Peris from Environment Canada for providing pollution data; and Louis Duchesne for his constructive pre-submission review of the manuscript. Comments from two reviewers helped improve the manuscript. NRCan/ESS contribution: 2005301.

REFERENCES Barker, I., and Barker, J., eds. 1988. Clean air around the world—The law and practice of air pollution control in 14 countries in 5 continents. IUPPA, Brighton, UK. Be´langer, N., 2000. Investigating the long-term influence of atmospheric acid deposition and forest disturbance on soil chemistry and cation nutrient supplies in forested ecosystem of southern Que´bec. Ph.D. dissertation. Department of Natural Resource Sciences, Faculty of Agriculture and Environmental Sciences, McGill University, Montreal, Quebec, Canada, 164 pp. Briffa, K., and Jones, P.D. 1989. Basic chronology statistics and assessment. In: Cook, E.R., and Kairiukstis, L.A., eds. Methods of dendrochronology: Applications in the environmental sciences. Kluwer Academic, Dordrecht, The Netherlands. Darrall, N.M. 1989. The effect of air pollutants on physiological processes in plants. Plant Cell Environ. 12, 1–30. Duchesne, L., Ouimet, R., and Morneau, C. 2003. Assessment of sugar maple health based on basal area growth pattern. Can. J. For. Res. 33, 2074–2080. Dupouey, J.-L., Leavitt, S., Choisnel, E., and Jourdain, S. 1993. Modelling carbon isotope fractionation in tree rings based on effective evapotranspiration and soil water status. Plant Cell Environ. 16, 939–947. Environmental Protection Agency (EPA). 1998. Emissions scoreboards. Acid rain program. http://epa.gov/acidrain/ (last accessed on July 20, 2008). Farquhar, G.D., Ehleringer, R.H., and Hubick, K.T. 1989. Carbon isotope discrimination and photosynthesis. Annu. Rev. Plant Physiol. Plant Mol. Biol. 40, 503–537. Feng, X., and Epstein, S. 1995. Climatic temperature records in dD data from tree rings. Geochim. Cosmochim. Acta 59, 3029–3037.

244

Martine M. Savard et al.

Freyer, H.D. 1979. On the 13C record in tree rings. Part II. Registration of microenvironmental CO2 and anomalous pollution effect. Tellus 31, 308–312. Gallon, C., Tessier, A., Gobeil, C., and Beaudin, L. 2005. Sources and chronology of atmospheric lead deposition to a Canadian Shield lake: Inferences from Pb isotopes and PAH profiles. Geochim. Cosmochim. Acta 69, 3199–3210. Holmes, R.L. 1992. Dendrochronology program library, version 1992-1. Laboratory of Tree-Ring Research, University of Arizona, Tucson, AZ, USA. Kropff, M.J. 1987. Physiological effects of sulphur dioxide. 1. The effect of SO2 on photosynthesis and stomatal regulation of Vicia faba L. Plant Cell Environ. 10, 753–760. Krupa, S., Muntifering, R., and Chappelka, A. 2004. Effects of ozone on plant nutritive quality characteristics for ruminant animals. Botanica 54, 129–140. Ladeau, S., and Clark, J.S. 2001. Rising CO2 levels and the fecundity of forest trees. Science 292, 95–98. Lefohn, A.S., Husar, J.D., and Husar, R.B. 1999. Estimating historical anthropogenic global sulphur emission patterns for the period 1850–1990. Atmos. Environ. 33, 3435–3444. Lipp, J., Trimborn, P., Fritz, P., Moser, H., Becker, B., and Frenzel, B. 1991. Stable isotopes in tree ring cellulose and climatic change. Tellus 43B, 322–330. Martin, B., Bytnerowicz, A., and Thorstenson, Y.R. 1988. Effects of air pollutants on the composition of stable carbon isotopes, d13C, of leaves and wood, and on leaf injury. Plant Physiol. 88, 218–223. Martin, B., and Sutherland, E.K. 1990. Air pollution in the past recorded in width and stable carbon isotope composition of annual growth rings of Douglas-fir. Plant Cell Environ. 13, 839–844. Matyssek, R., Gunthardt-Goerg, M.S., Schmutz, P., Saurer, M., Landolt, W., and Bucher, J.B. 1998. Response mechanisms of birch and poplar to air pollutants. J. Sustainable For. 6, 3–22. Ministe´re des transports. 1994. E´le´ments de proble´matique et fondements de la politique sur l’environnement du ministe´re des Transports du Que´bec. Direction des communications, Quebec City, QC, Canada, 39 pp. Monserud, R.A., and Marshall, J.D. 2001. Time-series analysis of d13C from tree rings. I. Time trends and autocorrelation. Tree Physiol. 21, 1087–1102. Niemela, P., Lumme, I., Mattson, W., and Arkhipov, V. 1997. 13C in tree rings along an air pollution gradient in the Karelian Isthmus, northwest Russia and southeast Finland. Can. J. For. Res. 27, 609–612. Pollution Probe. 2002. L’abe´ce´daire du smog. http://www.pollutionprobe.org (last accessed on August 3, 2008). Sakata, M., and Suzuki, M. 2000. Evaluating possible causes for the decline of Japanese fir (Abies firma) forests based on d13C records of annual growth rings. Environ. Sci. Technol. 34, 373–376. SAS Institute Inc. 1999. SAS/STAT User’s Guide, Version 8: SAS Institute Inc., Cary, NC, USA. Saurer, M., Maurer, S., Matyssek, R., Landolt, W., Gu¨nthardt-Goerg, M.S., and Siegenthaler, U. 1995. The influence of ozone and nutrition on d13C in Betula pendula. Oecologia 103, 397–406. Saurer, M., Borella, S., and Luenberger, M. 1997. d18O of tree rings of beech (Fagus silvitica) as a record of d18O of the growing season precipitation. Tellus 49B, 8092. Savard, M.M., Be´gin, C., Parent, M., Smirnoff, A., and Marion, J. 2004. The environmental impact of smelter SO2 emissions—A time and space perspective recorded by carbon isotope ratios in tree ring cellulose. J. Environ. Qual. 33, 13–26.

Urban Air Quality and Tree-Ring Stable Isotopes

245

Savard, M.M., Be´gin, C., Smirnoff, A., Marion, J., Sharp, Z., and Parent, M. 2005. Change in fractionation of hydrogen isotopes in trees due to the effects of atmospheric pollutants. Geochim. Cosmochim. Acta 69, 3723–3731. Shu, Y., Feng, X., Gazis, C., Anderson, D., Anthony, M.F., Tang, K., and Ettl, G. 2005. Relative humidity recorded in tree rings: A case study along precipitation gradient in the Olympic Mountains, Washington, USA. Geochim. Cosmochim. Acta 69, 791–799. Socie´te´ de l’assurance automobile du Que´bec. 2006. Le Bilan 2005: Accidents, parc automobile et permis de conduire. Direction des e´tudes et des strate´gies en se´curite´ routie`re, Quebec City, QC, Canada, 209 pp. Winner, W.E. 1989. Photosynthesis and transpiration measurements as biomarkers of air pollution effects on forests. Biologic markers of air-pollution stress and damage in forests. The National Academy of Sciences, Washington, DC. http://www.nap.edu/openbook. php?isbn¼ 0309040787&page¼ 305 (last accessed on August 3, 2008).

Developments in Environmental Science, Volume 9 Allan H. Legge (Editor) Copyright r 2009 Elsevier Ltd. All rights reserved. ISSN: 1474-8177/DOI:10.1016/S1474-8177(08)00210-6

247

Chapter 10 Lichen Monitoring of Urban Air Quality, Hamilton, Ontario D.P. McCarthy*, B. Craig and U. Brand Abstract In February 2004, a lichen species richness survey was conducted on 734 open-grown trees at 156 sites in Hamilton, Ontario. Index of atmospheric purity (IAP) values mapped with GIS software showed that residential areas had more lichen species, suggesting that they had cleaner air than in the downtown and heavy industry areas. To characterize ambient air quality differences across the city, the lichen Parmelia sulcata Hale was collected at a clean rural site, placed in 38 mesh bags, and hung on trees across the city from July to November 2004. Elemental analysis by ICP-MS found that bags located near heavy industry had accumulated higher levels of lead (Pb), zinc (Zn), vanadium (V), and chromium (Cr) than that were found in bags placed elsewhere in the city. Lichen deserts and elevated levels of some elements in residential areas downwind of industry are potential areas of environmental concern that merit more detailed investigation. The lichen monitoring protocols used here have considerable potential as a way to assess urban air quality in southern Ontario.

10.1. Introduction

Mapping of lichen species diversity is now widely used as a measure of the biological effect of air pollution (e.g., Kricke & Loppi, 2002; van Haluwyn & van Herk, 2002), and to investigate the link between environmental quality and human health (e.g., Cislaghi & Nimis, 1997). Lichen mapping is especially useful in urban areas, where it can be difficult and expensive to monitor a varied mix of pollutants and point

Corresponding author: E-mail: [email protected]

248

D.P. McCarthy et al.

sources using chemical–physical detectors (Showman, 1988). The main strength of lichen (floristic) mapping is its cost effectiveness and ability to provide a quick and useful measure of ecosystem disturbance and resiliency. This has been well demonstrated in Europe, where lichen monitoring programs have revealed lichen colonization of trees soon after the enforcement of clean air legislation, as in London (Hawksworth & McManus, 1989), Paris (Seaward & Letrouit-Galinou, 1991), Munich (Kandler & Poelt, 1984), and Turin (Piervittori et al., 1996). Similar work is also being done by the Forest Health Monitoring program in the United States (e.g., Blett et al., 2003; Jovan & McCune, 2005; McCune, 1988). These experiences have shown that correct evaluation of changes in lichen species richness is only possible when a standardized sampling protocol is applied and periodic observations are made. In such cases, it is possible to relate the quantity and species of lichens present with changes in bark pH and improvements in air quality over time (e.g., Loppi et al., 2002). Although studies have shown that many lichen species respond negatively to high sulfur dioxide (SO2) concentrations, the direct effects of ozone (O3) and nitrogen dioxide (NO2) on lichen communities are not well understood (e.g., Ruoss & Vonarburg, 1995; van Dobben & ter Braak, 1998). Studies have also linked the appearance of nitrophilous (nitrogen loving) species to increases in ammonia (NH3) loads (e.g., Loppi et al., 2002; van Dobben & ter Braak, 1998). The latter work suggests that the type as well as the number of lichen species in an area can sometimes be used to characterize changes in air quality. Many studies and national lichen monitoring programs include a measure of species diversity such as the index of atmospheric purity (IAP; e.g., DeSloover & Leblanc, 1968; Kricke & Loppi, 2002), index of poleo-tolerance (Trass, 1973), and the index of lichen abundance (Moore, 1974). These methods use a form of weighted averaging in which a different weight (importance value) is assigned to each species. Low IAP values will identify areas that are low in species diversity, whereas high IAP values will indicate areas with high species diversity. The IAP method has the potential to provide a quantitative measure of the differential between urban and rural lichen communities. Although IAP values are often interpreted as indicators of air pollution, differences in IAP or lichen species richness can also arise due to habitat disturbance. Consequently, it is sometimes combined with other indices such as an index of human impact (IHI; Gombert et al., 2004) to provide a more general assessment of biodiversity or environmental stress. Many studies have also shown how analysis of the trace-element content of lichens can be used to track spatial and temporal changes in atmospheric fallout

Lichen Monitoring of Urban Air Quality, Hamilton, Ontario

249

(e.g., Garty, 1993). Floristic mapping and lichen trace-element analysis, when combined, can be very powerful investigative tools for delineating, tracking, and characterizing spatial and temporal differences in urban airborne chemical emissions and loads.

10.2. Development of a lichen monitoring program for Canada

Although the IAP approach was pioneered in Canada in the late 1960s, lichen mapping has seen much less use in Canada than in Europe. During the 1970s, the IAP approach was used to map the impact of air pollution at several sites in eastern Canada (e.g., Arvida: LeBlanc et al., 1972a; Montreal: LeBlanc & DeSloover, 1970; Murdochville: LeBlanc et al., 1974; Sudbury: LeBlanc et al., 1972b). Since that time, few Canadian lichen-mapping studies have been published (e.g., Case, 1980), and until very recently, Canada has not had a national lichen-monitoring program. In the last few years, some governmental and not-for-profit agencies (e.g., the Environmental Monitoring and Assessment Network (EMAN) of Environment Canada and Citizen’s Environment Watch (CEW) in Toronto) have developed lichen monitoring methodologies and training programs that are suitable for use by school children and amateur scientists. These initiatives have been modeled in part on school-based lichen monitoring programs in Ireland (Richardson, 1987), Pennsylvania (Keim, 1975), and Oregon (Dennison, 1973), and programs such as AIRNET in Hancock (NH, USA). In 2003, a small group of lichenologists helped EMAN to identify suites of 15–20 arboreal lichen species that would be suitable for use in a Canadian lichen-monitoring program. The lichen species were selected primarily for their different sensitivities to air quality, but also because they are common in Canada’s mixed hardwood, boreal, and temperate forest regions. Most of these lichen species can readily be identified in the field by non-lichenologists on the basis of color and morphology (Brodo & Craig, 2004). Although EMAN has developed a variety of lichen measurement and monitoring protocols, they have not yet been fully articulated, tested, or formally adopted by Environment Canada or the scientific community. In this chapter, we demonstrate and test EMAN’s urban lichen monitoring methodology in Hamilton (Ontario, Canada). The objectives of this study were to (a) determine if there are clear differences in epiphytic lichen species richness across Hamilton, (b) relate differences in lichen species richness to air quality as assessed by chemical analysis of ‘‘lichen bags,’’ and (c) establish a baseline against which future changes

250

D.P. McCarthy et al.

can be measured. The study involved the collection of lichen diversity data, analysis of those data using the IAP approach, mapping lichen data with GIS software, and trace-element analysis of transplanted ‘‘lichen bags’’ to characterize and confirm gradients in air quality in the various IAP zones.

10.3. Methodology 10.3.1. Study area

The study area is broadly defined as road-accessible sites within the urban limits of Hamilton, Ontario (Latitude 43116u N, Longitude 79154u W). The city covers 1117 km2 along the western edge of Lake Ontario. Topographically, it is divided into upper and lower elevation zones by a 90 m high rock cliff called the ‘‘Niagara Escarpment.’’ Atop the escarpment, there are some low-density residential neighborhoods, but much of the escarpment is undeveloped, and is managed to maintain its natural cover of conifers and deciduous trees. Small farms, a horseracing track, and quarry operations occupy the western edge of the city. Housing and light industries give way to vineyards and tender fruit orchards to the east of the city. The central business district, two major highways, large steel mills, scrap yards, a port facility, and small- to medium-sized industries are situated between the base of the Niagara Escarpment and Lake Ontario within the city limits (Fig. 10.1). Hamilton has a continental climate that is moderated by the maritime effects of the Great Lakes. Prevailing winds from the west generally force industrial plumes toward the lake and away from residential areas. However, inversions and localized on-shore airflow from Lake Ontario can move industrial and highway traffic emissions into the downtown and some residential areas. Currently and in the past, Hamilton’s air quality has been monitored by chemical–mechanical air sensors. These air monitors are maintained by Ontario’s Ministry of the Environment and Hamilton’s heavy industries. Results from industry-owned sensors are not publicly available, but government reports indicate that the number of ‘‘smog days’’ has increased in recent years, whereas the annual loads of some of the main airborne pollutants (e.g., ozone, SO2, and NOx) have held steady or slightly decreased. Long-range transfer of smog pollutants from the American Midwest and the Ohio River Valley also contributes to smog events in the Great Lakes region and has been linked to the flow of slow-moving high-pressure cells (Ministry of the Environment, 2006). Consequently, there are concerns about air and water quality in the area,

Lichen Monitoring of Urban Air Quality, Hamilton, Ontario

251

Figure 10.1. Map of Hamilton and surrounding region showing major roads, and industrial, residential, and parkland uses.

and in recent years, citizen-based monitoring programs (e.g., Stack Watch, Lake Keepers) have been developed to keep a closer watch on local smokestack emissions and water quality in Lake Ontario and its contributing creeks and rivers. 10.3.2. Lichen biodiversity survey

Our lichen survey was carried out between February 14 and 18, 2004 by eight undergraduate Biology students (Brock University) and two lichen ecologists. The students received 8 h of lichen identification training and worked in two-person search teams. Each team had a car, a city map with numbered survey points, a handheld global positioning unit (GPS), hand lenses, a lichen identification key, a collection of lichen reference samples, a percentage cover estimation chart, a lichen survey data form, a tree diameter tape, and a cell phone. Lichen identification was limited to 18 arboreal lichen species listed in the EMAN suite for the mixed hardwood forest (Table 10.1). Surveyors were also asked to take voucher samples of any questionable or unidentified lichens for subsequent identification in the laboratory. At each tree, a complete search and inventory of arboreal lichens was done from approximately 30 cm to 1.5 m height, and a lichen survey form was used to help standardize data and information collection.

252

D.P. McCarthy et al.

Table 10.1. Summary of lichen biodiversity and IAP data Number of Total number % of sites with this of sites in sites with this species study species Candelaria concolor (Dickson) Stein Physcia millegrana Degel. Xanthomendoza fallax (Hepp) Søchting, Ka¨rnefelt & S. Kondr. Physciella chloantha (Ach.) Essl. Physcia aipolia (Ehrh. ex Humb.) Fu¨rnr. Physcia adscendens (Fr.) H. Olivier Parmelia sulcata Taylor Flavoparmelia caperata (L.) Hale Physconia detersa (Nyl.) Poelt Candellariella efflorescens R.C. Harris & W.R. Buck Punctelia rudecta (Ach.) Krog Melanelixia subaurifera (Nyl.) O. Blanco et al. Graphis scripta (L.) Ach. Phaeophyscia rubropulchra (Degel.) Essl. Usnea diplotypus Vainio Evernia mesomorpha Nyl. Lobaria pulmonaria (L.) Hoffm. Parmelia squarrossa Hale

Mean frequency

155

156

99

5

146 141

156 156

94 90

5 5

112 108

156 156

72 69

5 5

94 74 62 61 47

156 156 156 156 156

60 47 40 39 30

4 3 3 3 3

35 23

156 156

22 15

2 2

13 4

156 156

8 3

1 1

0 0 0 0

Visual comparison charts (Terry & Chilingar, 1955) were used to estimate percentage cover for each lichen species. The tree species was noted, tree diameter at 1.3 m height was measured, and tree location was determined using the GPS unit and a street address. We tried to restrict sampling to mature healthy maple trees that were free of bark damage and had a diameter of more than 30 cm at 1.3 m height. However, where large maples were unavailable (18% of the trees), we surveyed lichens on other tree species and on smaller trees (Fig. 10.2). We identified 156 sampling locations spaced roughly 1.5 km apart in residential areas and approximately 2–3 km apart in rural areas. This uneven placement of sampling sites reflects the wider spacing of roads in rural areas and our main interest in characterizing conditions in residential areas. Generally, surveying was done on secondary roads so as to collect ‘‘best case’’ estimates of lichen cover. Trees at bus stops, on busy street corners, and along busy roads were avoided.

Lichen Monitoring of Urban Air Quality, Hamilton, Ontario

253

Figure 10.2. Percentage of various tree species sampled in this study.

10.3.3. IAP calculation

The IAP formula uses the occurrence and frequency of lichens at the study sites and an ecological index for each species. The formula is IAP ¼

n X Qf 1

10

where n is the number of species found at a given station; Q the ecological index of a species or the average species richness; f the number of sampled trees on which this lichen species was found at the site. The value for (Q  f ) is divided by 10 to give a more manageable number (Johnsen & Søchting, 1973).

10.3.4. Map production

Inventory data were entered into a Microsoft Accesst database and Microsoft Excelt was used to calculate and rank lichen frequency values. The resulting data were exported to GIS software (ArcMapt and ArcInfot). The software was used to automatically identify five IAP classes defined by the natural breaks (jenks) command. Areas that had no sampling points were excluded (masked) from the analysis. The resulting map used a spatial interpolation and a tension spline function with a weight of 5 and a focus on the six nearest points. These settings produce smooth contours, and the approach is consistent with that recommended by ArcMap to map pollution gradients.

254

D.P. McCarthy et al.

10.3.5. Lichen bags

In late May 2004, approximately 1000 g of the lichen Parmelia sulcata Hale was collected from trees in a forested area with relatively good air quality approximately 75 km east of Hamilton. The attached bark was removed from the lichens, thalli were broken into 4 cm2 pieces, mixed together, and allowed to air dry for a week in a dry, dark, and solvent-free room. A total of 38 samples, each weighing approximately 300–500 mg, were removed from this mix. Ten subsamples (250 mg each) were placed in a desiccator to serve as the ‘‘clean standard,’’ and the geometric mean of these will characterize the baseline chemistry against which all other data will be compared. The remaining 28 samples were loosely placed in mesh bags (25  15 cm) made from fiberglass window screening that was cleaned with Aqua Regia and rinsed with de-ionized water. The bags were sewn shut using monofilament fishing line and were suspended with plastic ties from branches of maple trees (at 1.5–2 m height) in city parks and along roadways. Lichen bags were distributed along three transects that crossed Hamilton’s industrial and rural areas. The bags were placed in the field in early July and were collected and prepared for chemical analysis in late November 2004. 10.3.6. Lichen biogeochemistry

Our analysis of trace elements in lichens is modeled on procedures outlined by Nimis et al. (2000). After collection and transportation of the lichen bags to the laboratory, the samples were oven dried, and approximately 200 mg of each sample was ground to a fine powder using an agate mortar and pestle. To avoid cross-contamination, the mortar and pestle were cleaned with 10% HCl and allowed to air dry before each use. An acid solution [7.2 mL (HNO3 þ HF) þ 3.0 mL H2O2] was added to each of the powdered samples before they were digested at 601C in a heating block for 12–18 h (e.g., Rusu, 2002). Samples were then filtered through filter paper, volumetrically adjusted to 50 mL with de-ionized H2O, and refrigerated in Aqua Regia cleaned bottles. De-ionized water was then used to dilute the chilled samples by a factor of 10 before ICP-MS elemental analysis. Lichen-free blanks were prepared and BCRCRM-482 lichen (Certified Reference Material-482; Bureau of Community Reference, http://curem.iaea.org/catalogue/TE/TE_003360000.html, last accessed on July 20, 2008) was also digested and filtered, and included to check on laboratory QA/QC. The blanks, reference material, ‘‘clean samples,’’ and samples of the exposed lichen bags were sent for Inductively Coupled Plasma Mass Spectrometry (ICP-MS) elemental

Lichen Monitoring of Urban Air Quality, Hamilton, Ontario

255

analysis at Acme Laboratory in Vancouver (Canada). The precision and accuracy for certified and indicative elements (BCR-CRM-482) are Pb: 3.7, 13.4; Zn: 23.6, 9.6; Cr: 8.7, 3.6; and V: 16.1, 18.0 relative percent, respectively. 10.4. Results 10.4.1. Lichen mapping

In February 2004, lichen data were collected from 734 open-grown trees (Fig. 10.2) at 156 sampling locations (Fig. 10.3). Summary floristic data (Table 10.1) show that only 14 of the 18 lichen species listed in EMAN’s mixed hardwood forest suite were found in Hamilton. The most common of these species are known to be tolerant or insensitive to SO2 and NOx, whereas others listed in Table 10.1 but not found at the sampling sites are widely described as pollution sensitive. When we mapped the IAP data, we saw a clear overlap between low IAPs and surrounding industrial land use. However, after detailed evaluation of our first IAP map and return visits to verify scores at several sites, we noticed that two residential and parkland areas in eastern Hamilton had anomalously low IAP values. These sites are in new subdivisions where, until recently, trees were in shade and did not support the same lichen cover as trees growing in open spaces. We removed these ‘‘false positives’’ and raised their IAP values to that of their nearest neighbors before producing our final maps (Figs. 10.3 and 10.4). The final maps show ‘‘lichen deserts’’ that may have been created by dust (i.e., quarry operations) and pollution plumes from industrial sources and/or from sources of environmental concern downwind (cf. Figs. 10.1 and 10.4). Residential areas and areas with main thoroughfares and light industry either have high IAPs or are lichen ‘‘struggle zones’’ that signal air quality improvements in rural parklands and are upwind from heavy industry. 10.4.2. Lichen biogeochemistry transects

Overall, the lichen bags hung in trees formed three transects that crossed Hamilton’s industrial and rural areas. Transect A is a west–east crosssection on top of the escarpment, with transect B below it running to the east along the lakeshore, and transect C crossing the escarpment in a north-to-southeast direction, passing from rural and suburban areas into downtown Hamilton (Fig. 10.5). The bags were placed in trees at select

256 D.P. McCarthy et al.

Figure 10.3. Oblique view showing most of the Hamilton study area and the IAP zones. Darker areas represent higher IAP values and greater lichen diversity.

257

Lichen Monitoring of Urban Air Quality, Hamilton, Ontario

N

Lake Ontario Area not surveyed Escarpment IAP Index 62.1-69.6

48.6-54.5

58.5-62.0

37.3-48.5 False Positive

54.6-58.4

0

5 km

Figure 10.4. Lichen IAP map. Areas with greater lichen species diversity (higher IAP) are shown in a dark tone, whereas species-poor areas (lichen deserts) are white. Lichen deserts near the lower right edge of the map (Grimsby and Stoney Creek, Fig. 10.1) are artifacts of habitat changes associated with new subdivision construction and tree cutting.

intervals to monitor some specific sources or areas of potential environmental concern. All samples, including duplicates, blanks, and standards, were analyzed for a suite of elements, although only specific ones will be presented for each transect in the following discussions. In each case, an analysis of variance (ANOVA) and a Dunnett’s multiple comparison test was used to determine if there were significant differences between the geomeans of the clean samples and the lichen bags that were left in the field. 10.4.3. Transect A

Four metals absorbed by the transplanted lichens were chosen for detailed analysis by ICP-MS, and for this particular transect, we selected

258

Location of the three lichen bag transects.

D.P. McCarthy et al.

Figure 10.5.

Lichen Monitoring of Urban Air Quality, Hamilton, Ontario

259

lead (Pb), zinc (Zn), and vanadium (V). The values are reported as the percentage change (D%) from the baseline value based on the clean samples, which represent changes in concentrations accumulated by the lichens during their exposure at the specific localities for the selected time period. The highest Pb concentration on this transect was detected at station A8 (Fig. 10.6). Similarly, the D% Zn at locality A8 is different from the other lichen samples east and west of this particular location. In this case, we believe that Pb and Zn are airborne by-products of quarrying activities from the grinding of galena and sphalerite, a common accessory mineral in the carbonates of the Niagara Escarpment. Sampling station A8 is downwind (east) of a quarry and approximately 30 m from a paved four-lane road. Vehicle exhaust, dust raised by trucks on an unpaved road at the quarry, and the dust and exhaust from the rock crusher at the quarry are all potential sources of environmental concern at station A8 (Fig. 10.6). Statistical analysis found that the Zn values were significantly different from our clean reference samples (Po0.05). No statistical significance was found for the Pb values on transect A. Another metal of interest is V, which is used in the manufacture of gasoline and released to the atmosphere by gasoline engines. Its D% value for location A8 was not anomalous with respect to the other transect locations (Fig. 10.6) and it had one of the lower percentage changes over the baseline concentrations. However, the amount accumulated at this site was found to be significantly different (Po0.01) from the clean baseline values. This suggests that the exhaust of gasoline engines may be elevating V levels at A8 and elsewhere on transect A.

10.4.4. Transect B

Transect B traverses the industrial heartland of Hamilton (Fig. 10.5). The high metal uptake and concentrations of metals in lichen monitors is evidence of decreased air quality usually associated with heavy industry. Although Pb and Zn concentrations in the lichen monitors were higher than those observed on the other two transects (Fig. 10.7), only the Zn values were significantly different (Po0.001) from our clean baseline. The apparently ‘‘high’’ chromium (Cr) concentrations in the majority of lichen monitors on this transect was not found to be statistically significant, and Cr was not above baseline parameters on the other two transects (Fig. 10.7; Table 10.1). However, V levels were again found to be statistically different from baseline (Po0.01), suggesting that gasolinepowered vehicles and heavy industry in the western part of this transect are a source of environmental concern for metals in ambient air.

260 D.P. McCarthy et al.

Figure 10.6. Histograms of V, Pb, and Zn contents found in lichen bags along transect A. The values are percentage changes based on comparison with clean baseline values and that of lichen bag data for each site. The photograph shows the quarry that is located approximately 50 m from sampling site A-8.

200

400

150

300

∆% Zn

∆% Pb

500

100

50

200

100

0

0 Baseline: xg = 26.24 mg/kg (23.87-31.65)

Baseline: xg = 69.21 mg/kg (62.13-82.88)

500

B-1

B-2

B-3

B-4

B-5

B-6

B-7

B-8

∆% Cr

400

300

200

100

Lichen Monitoring of Urban Air Quality, Hamilton, Ontario

250

0 Baseline: xg = 10.35 mg/kg (9.41-12.21) B-1

B-2

B-3

B-4

B-5

B-6

B-7

B-8

261

Figure 10.7. Histograms of Cr, Pb, and Zn contents found in lichens along transect B. The values are percentage changes based on comparison with clean baseline values and lichen bag data for each site. The arrow in the photo shows the lichen bag. Steel manufacturing, metal recycling, and chemical storage facilities are found within a one-block radius.

262

D.P. McCarthy et al.

10.4.5. Transect C

Transect C lichen monitoring localities start above the escarpment and run into the downtown area of Hamilton below the escarpment (Fig. 10.5). Lichens from sampling station C4 exhibit higher than normal Pb and Zn values relative to those observed at the other locations along the transect (Fig. 10.8). This site was chosen because it is southeast of the quarry operations at sampling station A8 of transect A (Fig. 10.6) and the areal coverage from this potential source appears to be quite extensive. Of equal interest is the greater change in V uptake by the lichens at this sampling locality. The proximity of roads and the high traffic volumes directly upwind of locality C4 are the source of this environmental concern (Fig. 10.8). The elevated values reported for Pb, V, and Zn on transect C were significantly different from the clean baseline (Pb, Po0.05; V, Po0.01; Zn, Po0.05).

10.5. Discussion

There are clear differences in both lichen species abundance and traceelement loadings across the city of Hamilton. Although the spacing of our IAP survey points does not allow clear identification of point-source emitters or clearly show the influence of major roadways on ambient air quality, the IAP map (Fig. 10.4) shows that lichen deserts overlap parts of the downtown core and areas where heavy industry is the dominant land user. The shape of some lichen deserts (e.g., Fig. 10.4, center) suggests that dust from a quarry and/or other emissions of industrial or vehicular origin are delivered to residential areas by prevailing westerly winds. Lichen bags placed in transects across these zones allow a crude characterization of the chemistry of trace-element deposition in these locations. Although our data do not and were not intended to track the path or absolute amount of airborne fallout from specific industries, it would seem that this deposition is likely the cause of the lichen deserts. We believe that the lichen bag approach holds considerable promise as a low-cost approach to characterizing chemical loads of air in urban settings. Standardization of bag placement, replication and testing with different species of lichens, and calibration against chemical–mechanical sensors are logical next steps in our experimentation with the use of lichen bags and their chemistry as proxies of air quality. The absence of Usnea and other pollution-sensitive lichen genera from Hamilton, and the dominance of nitrophilous lichen species in the study area, indicate that industrialization has altered lichen communities in this

200

200

150

150

∆% Pb

∆% Zn

250

100

50

100

50

0

0

Baseline: xg = 69.21 mg/kg (62.13-82.88)

Baseline: xg = 26.24 mg/kg (23.87-31.65) C-1

250

C-2

C-3

A-5

A-6

C-4

C-5

C-6

C-7

C-8

200

∆% V

150

100

Lichen Monitoring of Urban Air Quality, Hamilton, Ontario

250

50

0 Baseline: xg = 3.72 mg/kg (3.17-4.45) C-1

C-2

C-3

A-5

A-6

C-4

C-5

C-6

C-7

C-8

263

Figure 10.8. Histograms of lichen V, Pb, and Zn contents found in lichens along transect C. The values are percentage changes based on comparison with clean baseline values. The lichen bag is in the lower branches of the tree at the right side of the image. This site is approximately 50 m west (upwind) of the quarry and 30 m from a road.

264

D.P. McCarthy et al.

urban forest. However, we note that the government-operated air monitoring machines are clustered near heavy industries along the shore of Lake Ontario. This makes it very difficult for us to determine how well our IAP data correlate with data collected by the chemical–mechanical sensors. Nonetheless, we suggest that the fact that our largest lichen desert overlaps a residential area is grounds for concern. Accordingly, we suggest that residents, industries, and local governments attempt to investigate the underlying cause(s) of the low IAP zones identified in our study to determine whether low IAP values are a function of average long-term conditions or are due to seasonal or short-lived emissions. Overall we are very pleased with the way our student surveyors were able to use the EMAN lichen suite for the mixed hardwood forest. The fact that the survey was completed in less than 1 week—and in the middle of a Canadian winter—speaks of the ease and cost effectiveness of the protocol and the tremendous energy of our students. Clearly, lichen mapping and the use of lichen bags hold considerable potential as a way to delimit and characterize IAP zones. Therefore, we recommend that the approaches espoused in this study be used by other cities in Canada and suggest that the survey of Hamilton’s lichen cover be repeated in a few years so as to measure historical changes of lichen biodiversity and ambient air quality.

10.6. Conclusions and recommendations

A lichen species richness survey was conducted in 2004 on 734 open-area grown trees at 156 sites in the city of Hamilton (Ontario, Canada). Index of air purity values based on lichen diversity, mapped with GIS software, showed residential areas had more lichen species, and thus cleaner air, than downtown and heavy industry areas. Ambient air quality differences across the city were characterized by the lichen P. sulcata, which was collected at a clean rural site; subsets were subsequently placed in 38 mesh bags and hung on trees across the city from July to November 2004. Elemental analysis recorded highest levels of Pb, Zn, V, and Cr in lichens from bags that were closest to heavy industry activities. Lichen deserts and comparatively high levels of some elements (metals) in lichens collected in residential areas downwind of such industry and activities are potential areas of environmental concern that merit more detailed investigation. The lichen bag ambient air quality monitoring method presented in this study can potentially be used to broadly assess urban air quality in Ontario. However, other suites of lichens would be needed in different parts of the country and more work should be done to

Lichen Monitoring of Urban Air Quality, Hamilton, Ontario

265

investigate how well the IAP method works in situations where nitrophytic lichen species are dominant and N deposition is a major component of the air pollution load.

ACKNOWLEDGMENTS

We thank Environment Canada, Hamilton Community Foundation, the city of Hamilton, and the Experience Works program at Brock University for their financial support and encouragement. Sara Beatty, Kiirsten VanWyck, Melissa Mason, Rachel Crowhurst, Jessica Greig, Sara Stewart, Lyndsay Elliott, and Andrij Fedj did a wonderful job of surveying. The GIS work was expertly done by Ryan Waterhouse. Lisa Neville, Vanessa Redwing, and Joan McCarthy helped with the lichen bags and Mike Lozon drafted the figures. We sincerely thank Dr David Richardon, Dr Linda Geiser, and Dr Allan Legge for providing very helpful and detailed feedback on an earlier version of this manuscript. Finally, we are deeply indebted to Dr Irwin Brodo and others who wisely selected the suite of lichen species and compiled the EMAN lichen identification guide. The financial assistance of NSERC (to U. Brand) for operation of the geochemical laboratory and sample preparatory work is gratefully acknowledged.

REFERENCES Blett, T., Geiser, L., and Porter, E. 2003. Air pollution-related lichen monitoring in national parks, forests, and refuges: Guidelines for studies intended for regulatory and management purposes. U.S. Department of the Interior Technical Report NPSD 2292. Brodo, I.M., and Craig, B. 2004. Identifying mixed hardwood forest lichens: A reference handbook. Environment Canada, Burlington, ON, Canada. Case, J.W. 1980. The influence of three sour gas processing plants on the ecological distribution of epiphytic lichens in the vicinity of Fox Creek and Whitecourt, Alberta Canada. Water Air Soil Pollut. 14, 45–68. Cislaghi, C., and Nimis, P.L. 1997. Lichens, air pollution and lung cancer. Nature 387, 463–464. Dennison, W.C. 1973. A guide to air quality monitoring with lichens. NW Mycological Consultants, 702 NW 4th St., Corvallis, OR, USA. DeSloover, J., and Leblanc, F. 1968. Mapping of atmospheric pollution on the basis of lichen sensitivity. In: Misra, R., and Gopal, B., eds. Proceedings of the symposium in recent advances in tropical ecology. International society for tropical ecology. Banaras Hindu University, Varanasi, India, pp. 42–56. Garty, J. 1993. Lichens as biomonitors for heavy metal pollution. In: Markert, B., ed. Plants as biomonitors. VCH, Weinheim, Germany, pp. 193–263.

266

D.P. McCarthy et al.

Gombert, S., Asta, J., and Seaward, M.R.D. 2004. Assessment of lichen diversity by index of atmospheric purity (IAP), index of human impact (IHI) and other environmental factors in an urban area (Grenoble, southeast France). Sci. Total Environ. 324, 183–199. Hawksworth, D.L., and McManus, P.M. 1989. Lichen recolonization in London under conditions of rapid falling sulphur dioxide levels, and the concept of zone skipping. Bot. J. Linn. Soc. 100, 99–109. Johnsen, I., and Søchting, U. 1973. Air pollution influence upon the epiphytic lichen vegetation and bark properties of deciduous trees in the Copenhagen area. Oikos 24, 344–351. Jovan, S., and McCune, B. 2005. Air-quality bioindication in the greater central valley of California, with epiphytic macrolichen communities. Ecol. Appl. 15, 1712–1726. Kandler, O., and Poelt, J. 1984. Wiederbeisiedlung der Innestadt von Mu¨nchen durch Flechen. Naturwiss. Rundsch. 37, 90–95. Keim, K. 1975. Lichensensor: A community school air quality survey. Bull. Penn. Lung Assoc. 58, 3–8. Kricke, R., and Loppi, S. 2002. Bioindication: The IAP approach. In: Nimis, P.L., Scheidegger, C., and Wolsley, P.A., eds. Monitoring with lichens—Monitoring lichens. Kluwer Academic Publishers, The Netherlands, pp. 21–37. LeBlanc, F., and DeSloover, J. 1970. Relation between industrialization and the distribution and growth of epiphytic lichens and mosses in Montreal. Can. J. Bot. 48, 1485–1496. LeBlanc, F., Rao, D., and Comeau, G. 1972a. Indices of atmospheric purity and fluoride pollution pattern in Arvida, Quebec. Can. J. Bot. 50, 991–998. LeBlanc, F., Rao, D., and Comeau, G. 1972b. The epiphytic vegetation of Populus balsamifera and its significance as an air pollution indicator in Sudbury, Ontario. Can. J. Bot. 50, 519–528. LeBlanc, F., Robitaille, G., and Rao, D. 1974. Biological response of lichens and bryophytes to environmental pollution in the Murdochville copper mines area, Quebec. J. Hattori Bot. Lab. 38, 405–433. Loppi, S., Giordani, P., Brunialti, G., Isocrono, D., and Piervittori, R. 2002. Identifying deviation from naturality of lichen diversity for bioindication purposes. In: Nimis, P.L., Scheidegger, C., and Worsley, P.A., eds. Monitoring with lichens—Monitoring lichens. Kluwer Academic Publishers, The Netherlands, pp. 281–284. McCune, B. 1988. Lichen communities along O3 and SO2 gradients in Indianapolis. Bryologist 91, 223–228. Ministry of the Environment. 2006. Air quality in Ontario 2004 report. Queen’s Printer for Ontario, Toronto, ON, Canada. Moore, C.C. 1974. A modification of the ‘‘Index of Atmospheric Purity’’ method for substrate differences. Lichenologist 6, 156–157. Nimis, P.L., Scheidegger, C., and Wolsley, P.A. 2000. Monitoring with lichens—Monitoring lichens. Kluwer Academic Publishers, The Netherlands. Piervittori, R., Meregalli, M., Maffei, S., and Montersino, M. 1996. Ricolonizzazione lichenica nella citta` di Torino. Allinoia 34, 63–65. Richardson, D.H.S. 1987. Lichens as pollution indicators in Ireland. In: Richardson, D.H.S., ed. Biological indicators of pollution. Royal Irish Academy, Dublin, Ireland, pp. 155–168. Ruoss, E., and Vonarburg, C. 1995. Lichen diversity and ozone impact in rural areas of Central Switzerland. Cryptogam. Bot. 5, 252–263. Rusu, A.M. 2002. Sample preparation of lichens for elemental analyses. In: Nimis, P.L., Scheidegger, C., and Wolsley, P.A., eds. Monitoring with lichens—Monitoring lichens. Kluwer Academic Publishers, The Netherlands, pp. 305–309.

Lichen Monitoring of Urban Air Quality, Hamilton, Ontario

267

Seaward, M.R.D., and Letrouit-Galinou, M.A. 1991. Lichen recolonization of trees in the Jardin Luxembourg, Paris. Lichenologist 23, 181–186. Showman, R.E. 1988. Mapping air quality with lichens, the North American experience. In: Nash III, T.H., and Wirth, V., eds. Lichens, bryophytes and air quality. Bibl. Lichenol Cramer, Berlin, Germany, pp. 67–89. Terry, R.D., and Chilingar, G.V. 1955. Summary of ‘‘Concerning additional aids in studying sedimentary formations,’’ by M.S. Shvetsov. J. Sediment. Petrol. 25, 229–234. Trass, H. 1973. Lichen sensitivity to air pollution and index of poleotolerance (I.P.). Folia Cryptogam. Est. 3, 19–22. van Dobben, H.F., and ter Braak, C.J.F. 1998. Effects of atmospheric NH3 on epiphytic lichens in the Netherlands: The pitfalls of biological monitoring. Atmos. Environ. 32, 551–557. van Haluwyn, C., and van Herk, C.M. 2002. Bioindication: The community approach. In: Nimis, P.L., Scheidegger, C., and Wolsley, P.A., eds. Monitoring with lichens— Monitoring lichens. Kluwer Academic Publishers, The Netherlands, pp. 39–64.

Developments in Environmental Science, Volume 9 Allan H. Legge (Editor) Copyright r 2009 Elsevier Ltd. All rights reserved. ISSN: 1474-8177/DOI:10.1016/S1474-8177(08)00211-8

269

Chapter 11 Ozone Exposure-Based Growth Response Models for Trembling Aspen and White Birch Kevin E. Percy*, Miloslav Nosal, Warren Heilman, Jaak Sober, Tom Dann and David F. Karnosky Abstract We developed free-air exposure regression-based models comprising annual growing season 4th highest daily maximum 8-h average ozone (O3) concentration, growing degree days (GDD), and average wind speed (WS). The models include 95% confidence bands for determining uncertainty of prediction. The models are statistically significant, provide a high goodness of fit, and can be used within the ambient air context. Trembling aspen (Populus tremuloides) clones 216, 42E, 271, and 259 responded negatively to O3. Aspen clone 8L responded positively to growing season 4th highest daily maximum 8-h average O3 concentration r90 ppb. White birch (Betula papyrifera) responded positively to growing season 4th highest daily maximum 8-h average O3 concentration o80 ppb and negatively at higher concentrations. These responses conform to the toxicological response concept of hormesis. Regression analysis demonstrated that annual growing season 4th highest daily maximum 8-h average O3 concentration performed much better as a single O3 exposure index for trembling aspen and white birch cross-sectional area growth than did W126, SUM06, AOT40, and maximum 1-h average O3 concentration. Growing season 4th highest daily maximum 8-h average O3 concentration is most closely associated with the actual measured response in the biological endpoint. The W126 index significantly overestimated the negative growth response of aspen and birch to O3. The growing season 4th highest daily maximum 8-h average O3 concentration, cumulative GDD, and average WS-based model may provide an underutilized opportunity for scientifically defensible risk analysis within the North American air quality context. Corresponding author: E-mail: [email protected]

270

Kevin E. Percy et al.

11.1. Introduction 11.1.1. Ozone, forests, and risk analysis

Over the past 50 years, a large volume of literature has documented ozone (O3) impacts on forest trees (see reviews by Ashmore, 2004; Bytnerowicz et al., 2003; Chappelka & Samuelson, 1998; Karnosky et al., 2007a; Kickert & Krupa, 1990; McLaughlin & Percy, 1999; Percy et al., 2003). Ozone effects are known to cascade through tree gene expression, biochemistry, and physiology, ultimately feeding back to productivity, predisposing trees to pest attack and causing changes in water-use efficiency (Karnosky et al., 2003c, 2005; Percy et al., 2002). Recent longterm, free-air investigations have confirmed earlier findings on productivity loss under O3, but do not provide evidence for altered patterns in allometry or carbon (C) allocation as previously reported in open-top chamber studies (King et al., 2005; Kubiske et al., 2006). In a retrospective review of the roles of air pollutants and climate in North American forest health, McLaughlin and Percy (1999) reported that O3 was deleteriously affecting forest ecosystem function across large and geographically widely separated areas of the continent. Ollinger et al. (1997) simulated the effects of O3 on hardwood forest types in the northeastern U.S. and estimated growth reductions between 3% and 22%. Later, Laurence et al. (2001) linked the mechanistic TREGRO model with the ZELIG stand model, parameterized them with biological and meteorological data from three sites, and simulated 100-year growth under five O3 exposure regimes. Change in Pinus taeda basal area ranged from þ44% to 87% depending on O3 exposure and precipitation, whereas basal area of Liriodendron tulipifera (generally considered O3 sensitive) was not affected. Weinstein et al. (2005) used the same models to simulate growth of Pinus ponderosa and Abies concolor under increased O3 exposures in the western San Bernardino and Sierra Nevada mountains. They predicted negative effects on P. ponderosa but little response in A. concolor due to differential sensitivities to O3, influences of competition, and soil moisture. Interestingly, simulations by Tingey et al. (2004) were among the first to demonstrate a link between improved emission control strategies and improved tree growth. However, there remain questions as to whether process models can be accurately parameterized to predict mature tree response (Samuelson & Kelly, 2001). Recently, physiological effects of O3 and biogeochemical changes have been scaled (Felzer et al., 2004; Ollinger et al., 2002) to the landscape. These models predicted that O3 levels in the United States could largely offset increased forest productivity caused by increasing atmospheric

Ozone Exposure-Based Growth Response Models

271

carbon dioxide (CO2) concentrations. Although certainly indicating the direction and magnitude of potential impact on forest productivity, these models are built partially on assumptions around linearity of response and O3 exposure indices that do not perform well within the North American ambient air context (Karnosky et al., 2005; Percy et al., 2006, 2007). Risk analysis to date has relied, for the most part, on dose–response and mechanistic research in chambered environments that have limited use in terms of extrapolation to risk analysis (Manning, 2005a). One key deficiency, identified earlier by Karnosky et al. (2003a), was the urgent need to couple air quality and meteorology measurements in time and space to effects analyses. There is a clear need for new approaches that can increase scientific certainty in dose–response knowledge so as to bring greater certainty to risk modeling. Importantly, there is a requirement that new approaches demonstrate how they contribute to increased ‘‘scientific literacy’’ (Orbach, 2005), thus enhancing usefulness within the context of ambient air quality management. 11.1.2. Objective

Our objective was to develop O3 exposure-based trembling aspen (Populus tremuloides) and white birch (Betula papyrifera) growth response models from 5 years’ co-measured indicator-response data. The data used were collected in a free-air exposure system designed to reflect the ambient air quality reality in North America. 11.1.3. Air quality standards to protect forest trees

As summarized in Percy and Karnosky (2007), the best current science, balanced by social, economic, and political considerations, is employed to establish North American ambient air quality standards. The United States and Canada have both established the O3 air quality standard as ‘‘the 3-year average of the annual fourth highest daily maximum 8-h average O3 concentration’’ (Canadian Council of Ministers of the Environment (CCME), 2000; US Federal Register, 2008). In the United States, there is a primary standard (human health-based) and a secondary standard (welfare-based) that can be different or the same. A legally binding, primary standard of 75 ppb O3 is now used for regulatory purposes, with the secondary standard set the same as the primary standard at this time. In Canada, the form and averaging time are the same as in the United States, but the level differs. A Canadian

272

Kevin E. Percy et al.

target value of 65 ppb O3 (human health-based, not legally binding) has been adopted. Establishing cause–effect relationships for ambient O3 exposure and tree growth has proved to be an elusive goal (Manning, 2005a), making scaling up to the landscape level difficult (Karnosky et al., 2005). Foley et al. (2003) stated that, for human effects, ‘‘Exposure-based metrics provide an information-rich tool in assessing relative effectiveness of alternative control strategies and introduce a higher degree of accountability in meeting NAAQS by augmenting air quality metrics with ones more closely associated with morbidity and mortality caused by air pollution exposure.’’ It is clear from the comprehensive review by Musselman et al. (2006) that, during the past 30 years, hourly averaged O3 data have been summarized in many different ways to assess risk to vegetation. Among indices receiving the most attention in analyses of exposure–response relationships in chambered studies are: the SUM06 threshold-based sum of daytime O3 concentrations Z60 ppb (Lefohn & Foley, 1992); the accumulated over a threshold (AOT)-based sum of hours of the day with a clear-sky global radiation above 50 W m2 (usually 07:00–21:00 h accumulated over 3 months for crops and 6 months for trees) O3 concentrations W40 ppb (Fuhrer et al., 1997); and the sigmoidally weighted W126 function (Lefohn & Runeckles, 1987; Lefohn et al., 1988) under previous discussion for potential use in a secondary standard (U.S. Environmental Protection Agency (US EPA), 1996, 2006). In the specific case of regulating surface-level O3 to protect vegetation, continued research to define our estimate of the level of exposure that will protect vegetation is still clearly needed (Laurence & Andersen, 2003). Recently, McLaughlin et al. (2007a, 2007b) and McLaughlin and Nosal (2008) have used a field-based open-air approach with electromechanical dendrometer techniques to model specific effects of O3 in the presence of co-varying influences of other environmental variables important to O3 flux. Regression coefficients for ambient O3 exposure (cumulative SUM06) prediction were negative and statistically significant for Pinus rigida, Q. rubra, Q. prinus, and Carya spp. Model predictions of growth loss in the range of 50% in high O3 years agreed well with observed growth. This approach also has great potential for determining the contribution of O3 to changes measured in tree growth, and for scaling hourly effects of O3 to cumulative impact over the growing season (McLaughlin et al., 2003). In reviewing the use of exposure- and flux-based ozone indices for predicting vegetation effects, Musselman et al. (2006) concluded that, at the moment, ‘‘y exposure-based metrics appear to be the only practical

Ozone Exposure-Based Growth Response Models

273

measure for use in relating ambient air quality standards [in North America] to vegetation response.’’ 11.2. Materials and methods 11.2.1. Analytical approach

Data from the Aspen Free Air Carbon Dioxide Enrichment (FACE) O3 exposure experiment, where response measurement was tightly coupled with meteorological measurements in both space and time, were used to build a matrix of 30 cases [5 years’ data  six FACE rings (three control, three O3)] for analysis. Each individual case comprised (1) a response variable (mean stem cross-sectional area); (2) an O3 indictor variable (annual growing season 4th highest daily maximum 8-h O3 concentration and four other O3 indices); and (3) meteorological indictor variables important in controlling O3 flux into plants and ambient O3 concentrations (Krupa et al., 2003; National Research Council (NRC), 1991). We tested five aspen clones (total ¼ 1723 trees in 1999) and white birch (total ¼ 222 trees in 1999) covering a range of documented (Karnosky et al., 1996, 2005) sensitivity to O3. Our O3 exposure–response models integrated end-of-season growth response over a 5-year growth period (1999–2003). During that time, aspen height (averaged across clones) within the aspen plantation half of the control rings increased from 2.8 m to 5.8 m and the stand reached (2002) canopy closure. 11.2.2. The Aspen FACE experiment

The Aspen FACE experiment (32 ha) is situated on sandy loam glacial outwash soil near Rhinelander, northern Wisconsin, US (45106uN; 89107uW; 490 m asl; www.aspenface.mtu.edu, last accessed on July 20, 2008). The experiment consists of a full factorial with 12 30-m diameter FACE rings: three controls, three elevated CO2, three elevated O3, and three elevated CO2þO3. The rings were planted in 1997 and treatments occurred from bud break to the end of growing season from 1998 to present. The eastern half of each ring was randomly planted in two-tree plots at 1 m  1 m with five trembling aspen (P. tremuloides Michx.) clones of known and widely varying tolerance to O3. The northwest quarter was planted with a mixture of aspen (clone 216) and sugar maple (Acer saccharum Marsh.). The southwest quarter was planted with a mixture of aspen (clone 216) and a range-wide, northern Lake States source of white

274

Kevin E. Percy et al.

birch. Since 2003, Aspen FACE has been a designated component of a distributed U.S. Department of the Environment (US DOE) User Facility. Complete details on baseline site physical and chemical characteristics, micrometeorology measurement, O3 measurement, selection of plant material, and experiment operation are published elsewhere (Dickson et al., 2000; Karnosky et al., 2003c). 11.2.3. Ozone fumigation

The Aspen FACE protocol for O3 fumigation prescribed a 07:00–19:00 h (12 h; based on zenith sun angle) daily exposure, 7 days a week from bud break to bud set. Elevated O3 was controlled so as to track ambient O3, yielding a repeatable diurnal increase to early afternoon followed by a decrease to late afternoon. Elevated hourly average O3 concentrations (maximum hourly average concentration achieved 13:00–14:00 h) followed ambient concentrations closely throughout the experiment (Karnosky et al., 2005). Ozone was not released if leaf surfaces were wet or if daily maximum temperature was predicted to be o151C. In this analysis, using 5 years (1999–2003) of co-measured response-predictor variables, growing seasons ranged from 136 to 144 days. In practice, during 1999–2003, O3 was fumigated on only 48.7–51.6% of potential growing season days as follows: 1999 (124 d, 820 h); 2000 (121 d, 800 h); 2001 (122 d, 777 h); 2002 (107 d, 787 h); 2003 (117 d, 893 h) (Percy et al., 2006, 2007). Target elevated O3 was 1.5  (1999) or 1.4  (2000–2003) ambient air. 11.2.4. Response and indictor variables

Building on earlier work (Percy et al., 2006, 2007), the list of O3 indices tested was expanded to include the W126 index (Lefohn & Runeckles, 1987). The response variable used in this study was aspen clone mean stem cross-sectional area (m2). End-of-growing season tree diameters were measured at 3 cm (1998–2001) or at 10 cm (2001–2003) above ground. Diameters used for 2001 were the averages at 3 cm and 10 cm as described by Kubiske et al. (2006). All measurements were collected on individual trees growing within the core area (about five rows inward from the free-air inlets toward the ring center). Diameters (dia 71 cm) were converted to cross-sectional area using the equation crosssectional area (m2) ¼ 0.00007854  (dia2) (Husch et al., 2003). Mean cross-sectional stem area for the five aspen clones and white birch was then calculated for each FACE ring used in this study (three control, three O3).

Ozone Exposure-Based Growth Response Models

275

Annual growing season 4th highest daily maximum 8-h average hourly O3 (modified U.S. and Canadian air quality standard metric form and averaging time) was calculated from continuous O3 monitoring at ring center above the canopy (10 m) for each elevated O3 ring. Ozone was not continuously monitored within Aspen FACE control rings. Spatial analysis (ESRI ARCt Map; data interpolated using a tension spline, weight 0.1) was completed for 1999–2003 from 24-h continuous hourly active fence line monitor data collected along the Aspen FACE perimeter fence lines. This analysis showed little within-season variation in growing season 4th highest daily maximum 8-h average hourly O3 across the site (Percy, unpublished). Therefore, control ring annual 4th highest daily maximum 8-h average hourly O3 was taken from the published on-site ambient monitor (EPA AIRS ID 5508500044420101; data available at http://oaspub.epa.gov/airsdata) and assigned to each control ring. Meteorological indictor variables were calculated from higher frequency sampling intervals described elsewhere (Dickson et al., 2000). Daytime temperature, solar radiation, wind speed (WS), relative humidity (RH), and precipitation data used in this study were measured at the onsite Aspen FACE 20-m meteorological tower. Growing degree days (GDD) or ‘‘heat units’’ were computed by subtracting a base temperature of 10 1C from the average of the maximum and minimum temperatures (5 min scan interval) for each day measured at 10 m. If the daily average temperature computed from the maximum and minimum temperatures was less than 10 1C, the average temperature was set to 10 1C so that the GDD contribution from that day was zero, and not negative. Accumulated growing season photosynthetically active solar radiation (PAR) (mmol m2 s1; 5 s scan interval) was calculated as the sum of halfhourly values. Average growing season WS (m s1; 5 s scan interval; 30 min average reporting) and average growing season 09:00 h RH (%; 5 min scan interval; 30 min reporting) were calculated from data collected at 10 m. Time-specific growing season precipitation (mm) was calculated from monthly sums at the base of the tower. Average growing season soil moisture content (SMC) (%; 2 h scan interval) was calculated from biweekly averages taken at 5–35 cm below the surface within the FACE ring aspen communities. 11.2.5. Statistical analysis

Exploratory statistical analysis included investigation of the relationship between O3 and mean cross-sectional area growth. Pearson correlation (Millard & Neerchal, 2001) was used to characterize the relationships

276

Kevin E. Percy et al.

between the dependent and the seven independent indictor variables. As correlation analysis showed that RH and precipitation were very highly correlated (r ¼ 0.799; p ¼ 0.0000) and co-linear with respect to other predictors, RH was omitted as a predictor variable in subsequent analyses. Complete multiple regression (Millard & Neerchal, 2001) models were developed using the remaining six indicators. Analysis of residuals for the 30 cases constructed for each of the five aspen clones and white birch indicated highly standardized residuals for only two observations (2002, O3 ring 3 and 2003, O3 ring 3; both clone 8L). These two observations (residuals equal to 0.0010) did not conform to the normal probability plot and were deleted from the analysis. To determine the most suitable regression models for impact assessment of O3 and the other indictor variables on cross-sectional area growth, we systematically applied the best regression algorithm (Millard & Neerchal, 2001) to each of the five aspen clones and white birch. Using this outcome, we developed multiple regression models optimizing (minimum number of indictors with highest r2 adjusted) on the best indictors of aspen cross-sectional area growth. Confidence intervals were computed using Monte Carlo techniques (Millard & Neerchal, 2001) to randomly generate various scenarios (n W 3000) for all relevant ranges of O3, GDD, and WS. Normal probability plots for the predictors indicated a perfect fit for their distribution. Resulting confidence (95%) bands in two-dimensional Euclidean spaces were represented by graphs in planes for ease of visualization.

11.3. Results 11.3.1. Five-year trend in indicator variables

Metadata for the indicator variables used in this study are graphed in Fig. 11.1. The growing season 4th highest daily maximum 8-h average O3 concentration ranged from 94 ppb (elevated O3 rings 1999) to 65 ppb (control rings 2002). There was no overall trend in cumulative growing season GDD. GDD decreased in the order 2002W1999W2001W2003W 2000 (Fig. 11.1). Growing seasons 1999 and 2002 were slightly (up to 16%) warmer than 2000, 2001, and 2003. There was a tendency for average growing season WS at Aspen FACE to decrease (except for 2002) over the 5 years. Average WS ranged from 1.18 m s1 (1999) to 0.98 m s1 (2003) (Fig. 11.1). There was no apparent trend in accumulated growing season PAR, which ranged from 3211.2 mmol m2 s1 (2001) to

60

Control

40

Ozone

20 0

950 900 850 800 750 1999

2000

2001

2002

2003

Year

1.2 Solar radiation (m mol/m2/sec)

3400

1.1 1 0.9 0.8

3200 3000 2800 2600 2400

1999

Precipitation (mm)

1000

2000

2001 Year

2002

2003

600 400 200 0 2000

2001 Year

2000

2001

2002

2003

2002

2003

Year

800

1999

1999

2002

2003

Soil moisture content (%)

Wind speed (m/sec)

1999 2000 2001 2002 2003 Year

Growing degree days (degrees)

80

Ozone Exposure-Based Growth Response Models

4th highest ozone (ppb)

100

26 24 22 20 18 1999

2000

2001 Year

277

Figure 11.1. Five-year (1999–2003) trends in Aspen FACE annual growing season 4th highest daily maximum 8-h average O3 concentration (means of three replicate rings), accumulated GDD, average PAR, average WS, total precipitation, and average biweekly SMC.

278

Kevin E. Percy et al.

2735.6 mmol m2 s1 (2002) (Fig. 11.1). Growing season precipitation alternated (52% maximum change) biannually between lower amounts (1999, 425.11 mm; 2001, 301.86 mm; 2003, 358.14 mm) and higher amounts (2000, 625.69 mm; 2002, 547.57 mm) (Fig. 11.1). Average growing season SMC beneath the aspen stands varied little during 1999–2003 and ranged from 19.42% (elevated O3 ring 1,3 in 1999) to 27.2% (elevated O3 ring 2,3 in 2000) (Fig. 11.1).

11.3.2. Exploratory statistical analysis

Pearson correlations between mean cross-sectional area and annual growing season 4th highest daily maximum 8-h average O3 concentration were negative, signifying an inhibitory effect of O3 on cross-sectional area growth (Table 11.1). The correlations for aspen clone 8L and white birch were not, however, statistically significant. WS was negatively and highly (p ¼ 0.000) significantly correlated with mean cross-sectional area growth over the 5-year period in all aspen clones and in white birch (Table 11.1). Mean cross-sectional area was negatively correlated with PAR in aspen and birch. Only in aspen clones 8L, 42E, 216, and 259, however, was the correlation statistically significant. There was a tendency for mean cross-sectional area to be negatively correlated with Table 11.1. Pearson correlations and their significance [r, (p ¼ )] for mean cross-sectional area response and six predictor variables in five aspen clones and white birch Species/clone Aspen clone 8L Aspen clone 42E Aspen clone 216 Aspen clone 259 Aspen clone 271 White birch

4th highest O3a

WSa

GDDa

PARa

SMCa

Precip.a

0.070 (0.713) 0.505 (0.004) 0.689 (0.000) 0.422 (0.020) 0.535 (0.002) 0.246 (0.190)

0.708 (0.000) 0.719 (0.000) 0.725 (0.000) 0.811 (0.000) 0.746 (0.000) 0.728 (0.000)

0.156 (0.411) 0.129 (0.496) 0.188 (0.319) 0.305 (0.101) 0.215 (0.253) 0.177 (0.349)

0.432 (0.017) 0.484 (0.007) 0.389 (0.034) 0.396 (0.030) 0.327 (0.078) 0.198 (0.295)

0.044 (0.816) 0.031 (0.872) 0.007 (0.973) 0.009 (0.962) 0.177 (0.348) 0.306 (0.100)

0.189 (0.318) 0.170 (0.370) 0.208 (0.270) 0.232 (0.218) 0.243 (0.195) 0.321 (0.083)

Note: WS ¼ average growing season wind speed; GDD ¼ seasonal growing degree days (heat units); PAR ¼ growing season cumulative photosynthetically active solar radiation; SMC ¼ biweekly averaged soil moisture content; Precip. ¼ cumulative growing season precipitation. a 4th highest O3 ¼ growing season 4th highest daily maximum 8-h average O3 concentration.

Ozone Exposure-Based Growth Response Models

279

GDD and with precipitation, but only weakly. SMC was always positively correlated with mean cross-sectional area, but not significantly so (Table 11.1). SMC was not correlated (p ¼ 0.747) with precipitation amount. 11.3.3. Evaluation of selected O3 exposure indices

The frequency distribution of all growing season (1999–2003) hourly average O3 concentrations in each of the three replicate elevated O3 FACE rings is represented here by ring 2 (Fig. 11.2). Three-quarters of all growing season 4th highest daily maximum 8-h average hourly O3 concentrations were r63 ppb (1999) or r60 ppb (2000–2003). In 1999, 95% of all O3 concentrations were r84 ppb. During 2000–2003, 95% of all concentrations were r80 ppb. Over the 5-year period, 99.9% of concentrations were r100 ppb (1999) or r90 ppb (2000–2003) (Fig. 11.2). The best performing (highest r2 adjusted) O3 exposure indices as single indicators of aspen cross-sectional area growth were growing season W126 (24 h) and growing season 4th highest daily maximum 8-h average O3 concentration (Table 11.2). The dependence of cross-sectional growth on O3 exposure calculated using W126 was statistically significant

Figure 11.2. Frequency distribution of all 5-year (1999–2003) annual growing season hourly average O3 concentrations (ppb) measured at the center of Aspen FACE elevated O3 replicate ring 3.

280

Kevin E. Percy et al.

Table 11.2. Evaluation of five O3 exposure indices as single indicators of cross-sectional area growth in trembling aspen clones and white birch. Data are r2 adjusted (p values) from cubic regression analysis of dependence of cross-sectional area growth (1999–2003) on growing season hourly average O3 concentrations in three replicate elevated O3 FACE rings Aspen clone

4th highesta SUM06b AOT40c Max 1 hd W126e

White birch

42E

216

271

259

8L

0.513 (0.012) 0.170 (0.180) 0.222 (0.130) 0.250 (0.109) 0.575 (0.006)

0.479 (0.017) 0.137 (0.217) 0.190 (0.159) 0.314 (0.069) 0.648 (0.002)

0.454 (0.021) 0.163 (0.187) 0.213 (0.138) 0.197 (0.152) 0.618 (0.003)

0.179 (0.170) 0.223 (0.130) 0.030 (0.374) 0.121 (0.236) 0.780 (0.000)

0.354 (0.078) 0.228 (0.160) 0.375 (0.067) 0.371 (0.069) 0.647 (0.006)

0.119 (0.112) 0.031 (0.251) 0.000 (0.877) 0.331 (0.015) 0.376 (0.009)

Source: Modified from Percy et al. (2007). Growing season 4th highest daily maximum 8-h average O3 concentration (ppb). b Threshold-based sum of all daytime (08:00–19:59 h) ozone concentration hours Z60 ppb (Lefohn and Foley, 1992). c AOT-based sum of all growing season daytime (07:00–20:59 h; W50 W m2) ozone concentrations W40 ppb (Fuhrer et al., 1997). d Growing season maximum 1-h average ozone concentration (ppb). e Growing season Weibull 126 concentration-weighted sum of 24 h average hourly ozone concentrations (Percy and Karnosky, 2007, Table 4). Multiple regression models of growing season 4th highest daily maximum 8-h average O3 concentration, average growing season WS and GDD. a

(po0.01) for all five clones as well as for white birch. However, and very importantly, there was no consistent association between the level of statistical significance achieved (p value) and the actual measured response of the biological endpoint. Dependence of growth on growing season 4th highest daily maximum 8-h average O3 concentration was statistically significant (po0.05) for three aspen clones (42E, 216, 271) that responded negatively to O3, but not for white birch (p ¼ 0.112). As the level of statistical significance was consistent with measured response, calculation of O3 exposure using the growing season 4th highest daily maximum 8-h average O3 concentration resulted in a plausible biological association with response in the biological endpoint. Growth in all aspen clones and white birch was not dependent on O3 exposure as calculated using the SUM06, AOT40 indices (Table 11.3). White birch growth was dependent (p ¼ 0.015) on maximum 1-h average O3 concentration, but growth in the aspen clones was not.

281

Ozone Exposure-Based Growth Response Models

Table 11.3. Multiple linear regression model statistics for dependence of aspen clone and white birch cross-sectional area growth on growing season 4th highest daily maximum 8-h average O3 concentration, average growing season WS, and growing degree days Species/ clone Aspen clone 8L Aspen clone 42E Aspen clone 216 Aspen clone 259 Aspen clone 271 White birch

Model 4th highest O3 effect significance

4th Highest O3 significance

R2 4th highest O3 R2 (%) adjusted

p ¼ 0.000

Negative

p ¼ 0.900

0.767

6.8

0.636

p ¼ 0.000

Negative

p ¼ 0.001

0.762

10.0

0.734

p ¼ 0.000

Negative

p ¼ 0.000

0.894

47.4

0.882

p ¼ 0.000

Negative

p ¼ 0.038

0.739

17.8

0.709

p ¼ 0.000

Negative

p ¼ 0.001

0.757

28.6

0.729

p ¼ 0.000

Negative

p ¼ 0.540

0.615

6.0

0.570

Notes: 4th highest O3 ¼ growing season 4th highest daily maximum 8-h average O3 concentration. % values indicate the percent contribution to the model.

11.3.4. Multiple regression models

Multiple linear regression models comprising the six indicator variables [growing season 4th highest daily maximum 8-h average O3 concentration (4th highest O3), GDD, WS, PAR, precipitation, and SMC] produced a best available fit (r2 adjusted ¼ 0.687–0.944) for the aspen clones and white birch. The highest value corresponded to aspen clone 216 (Eq. (11.1)). The lowest value corresponded to aspen clone 8L (Eq. (11.2)). Clone 216 mean cross  sectional area ðm2 Þ ¼ 0:0130  0:000038 4th highest O3 þ 0:00022 WS  0:000001 GDD

(11.1)

 0:000010 SMC  0:000002 PAR  0:000003 precipitation Clone 8L mean cross-sectional area ðm2 Þ ¼ 0:00796 þ 0:000026 4th highest O3  0:0117 WS þ 0:000004 GDD þ 0:000007 SMC  0:000000 PAR þ 0:000002 precipitation

(11.2)

282

Kevin E. Percy et al.

11.3.5. Best subsets regression

To balance this exceptionally high degree of goodness of fit against the practical utility requirements of our models, the best subset regression algorithm was systematically applied to the aspen clones and white birch. Best optimized models for the four aspen clones (42E, 271, 216, and 259) that responded negatively to O3 within the range of 4th highest daily maximum 8-h average O3 concentrations (62–96 ppb) measured during 1999–2003 in the six FACE rings were determined to be those comprising: (1) growing season 4th highest daily maximum 8-h average O3; (2) average growing season WS; and (3) growing season GDD. Although the choice of 4th highest O3, WS, and GDD cannot be considered the absolute best choice for the aspen clones and white birch, the models were highly statistically significant, had a very high goodness of fit (r2 adjusted ¼ 0.57–0.88), and were plausible from the biological point of view. The three-indicator (Eq. (11.3) for clone 216) model in the end was deemed considerably simpler, and easier to use in practical applications than the complete six-predictor model (Eq. (11.1)) listed earlier. Aspen clone 216 mean cross-sectional area ðm2 Þ ¼ 0:00684  0:000031 4th highest O3

(11.3)

 0:00551 WS þ 0:000003 GDD

11.3.6. Ozone exposure–response models

We next developed a three-indicator multiple regression model as a tool for assessment of the impact of O3 and two meteorological variables on trembling aspen and white birch cross-sectional area growth. For the aspen clones (271, 42E, 216, 259) that responded negatively to O3, the corresponding r2 adjusted ranged from 0.71 to 0.88 (Table 11.3). Regression coefficients at the growing season 4th highest daily maximum 8-h average O3 concentration were negative and statistically significant (po0.038) for aspen clones 42E, 216, 271, 259. Contribution of growing season 4th highest daily maximum 8-h average O3 concentration was 10– 47.4% of tree cross-sectional area growth, depending on relative sensitivity of the clone to O3. The coefficients for 8L and white birch were not statistically significant (Table 11.3), implying that there was no negative effect resulting from O3 exposure, or that exposure to O3 resulted in some degree of growth stimulation.

Ozone Exposure-Based Growth Response Models

283

11.3.7. Uncertainty in model prediction

The Monte Carlo method was used to randomly generate thousands of various scenarios of O3, GDD, and WS based on the actual frequency distributions of these indictors measured at Aspen FACE during 1999–2003. Here, we use the example of aspen clone 271 to show the 95% confidence bands for the prediction of the growing season 4th highest daily maximum 8-h average O3 concentration effect on mean cross-sectional area growth. At a given growing season 4th highest daily maximum 8-h average O3 concentration, a vertical line can be drawn from the x-axis to the intersections with the red, green, and black lines. The black line intersection corresponds to the single midpoint prediction of the average (mean) cross-sectional area response to the given value of the growing season 4th highest daily maximum 8-h average O3 concentration (Fig. 11.3). Using the exposure–response models produced for the five aspen clones and white birch, we calculated the mean forecast (black line in Fig. 11.3) over a range of 60–95 ppb growing season 4th highest daily maximum 8-h average O3 concentration. From a baseline of 60 ppb, the growth change predicted for aspen and birch as O3 increases to 95 ppb is shown in Fig. 11.4. Among the five aspen clones, there was a clear difference in predicted outcomes. Clone 8L demonstrated a (þ2.5% to þ4.3%) growth stimulation with increasing O3 to 90 ppb, followed by a –2.5%

Figure 11.3. Exposure–response model (mean prediction 795% confidence intervals) for effect of growing season 4th highest daily maximum 8-h average O3 concentration on aspen clone 271 mean cross-sectional area growth.

284

Kevin E. Percy et al.

Figure 11.4. Percent change in aspen clone and white birch cross-sectional area from O3 exposure at a growing season 4th highest daily maximum 8-h daily average O3 concentration of 60 ppb at 5 ppb O3 exposure increments. The response averaged across all five clones is also shown.

growth loss between 90 ppb and 95 ppb (Fig. 11.4). Between growing season 4th highest daily maximum 8-h average O3 concentrations of 60 ppb and 65 ppb, predicted growth loss in decreasing order by clone was: clone 216, –7%; clone 42E, –5.41%; clone 271, –4.17%; clone 259, –1.15%. Relative to a growing season 4th highest daily maximum 8-h average O3 concentration of 60 ppb, mean cross-sectional area growth at 80 ppb was predicted to have decreased by: clone 216, –28.5%; clone 42E, –24.3%; clone 271, –20.8%; clone 259, –6.9% (Fig. 11.4). Averaging negative (clones 42E, 271, 216, 259) and positive (clone 8L) responses across all five aspen clones, the change in growth between growing season 4th highest daily maximum 8-h average O3 at 5 ppb increments was predicted to be: –3.0% (65 ppb); –6.5% (70 ppb); –10.7% (75 ppb); – 15.2% (80 ppb); –19.3% (85 ppb); –24.6% (90 ppb); –31.1% (95 ppb). White birch, like aspen clone 8L, exhibited growth stimulation to O3 at lower concentrations followed by growth inhibition at higher O3 concentrations. Birch cross-sectional area growth was predicted to be stimulated at growing season 4th highest daily maximum 8-h average O3 concentrations of r75 ppb and reduced between –1.05% and –5.3% at growing season 4th highest daily maximum 8-h average O3 concentrations Z80 ppb (Fig. 11.4).

Ozone Exposure-Based Growth Response Models

285

11.4. Discussion

We based our work on a hypothesis advanced by Krupa et al. (2003) ‘‘y that it should be possible to build an appropriate and inclusive predictive model comprising all important meteorological predictors plus soil moisture data that, together, would yield a first-order approximation of atmospheric O3 flux and stomatal uptake.’’ We also built upon the earlier approach of Hogsett et al. (1997) and used an important endpoint in a key species as recommended by Laurence and Andersen (2003). In so doing, we used a multi-year dataset from our randomized block, ecosystem-scale, free-air experiment to develop realistic exposure– response models based on a modified (growing season only) version of the United States (Federal Register, 2008) and Canadian (CCME, 2000) ambient air quality standard for O3. The aspen clones used in this study represented a wide range of sensitivity to O3. They were originally selected based on foliar symptoms from some 220 clones representing 15 populations over the entire conterminous U.S. natural aspen range (Berrang et al., 1986) and later validated in field trials under conditions of varying ambient O3 (Berrang et al., 1989; Karnosky et al., 2003b) and open-top chamber experiments (Karnosky et al., 1996, 2006). Both white birch and aspen clone 8L have been previously demonstrated to be very tolerant of O3 at Aspen FACE over an 8-year growth cycle (Karnosky et al., 2003c, 2005). Our models confirmed their relative degrees of tolerance by predicting that aspen clone 8L growth was (Fig. 11.4) stimulated at growing season 4th highest daily maximum 8-h average O3 concentrations o95 ppb. White birch growth was reduced, but only at growing season 4th highest daily maximum 8-h average O3 concentrations W75 ppb. In demonstrating both positive and negative growth response, our exposure–response models conformed to the theory of hormesis. Calabrese (2005) has convincingly stated the case for the hormetic dose–response relationship as underlying the toxicological basis for risk assessment. It has been only rarely demonstrated to this point in time with O3 exposure–plant response (Ja¨ger & Krupa, 2008), possibly because of an overwhelming focus on identifying negative responses (Manning, 2005b). Co-measured response and indictor variables for aspen and white birch yielded regression models that were statistically significant. Our initial seven meteorological growing season accumulated (GDD, PAR, precipitation) and averaged (WS, SMC) indicator data were derived from scan intervals of varying lengths. Our decision to delete RH from subsequent regression analysis was based on (1) its co-linearity with precipitation and (2) the fact that precipitation was added earlier in best

286

Kevin E. Percy et al.

subsets regression analysis. The resulting six-indicator variable multiple regression models provided a statistically very highly significant goodness of fit in terms of r2 adjusted, regression ANOVA F-test significance, and performance relative to recorded productivity within the FACE rings. However, although they accounted for most of the variability in mean cross-sectional area growth, they were very complex and very data dependent. In addition, not all the variables (SMC, PAR) are routinely reported across the landscape and model utility would, thus, have been compromised. The three-indicator models identified in our best subsets regression had a high degree of goodness of fit, and should be very simple to use within a North American ambient air quality O3 risk analysis context. The confidence bands can, in practice, be used by regulators to define uncertainty in the prediction. We are aware that the intrinsic relationship between O3 and tree growth is, of course, non-linear. Although multiple linear regression models of aspen clone growth on O3 were very highly statistically significant, polynomial cubic regression (Millard & Neerchal, 2001) was used to evaluate whether the assumption of non-linearity in tree growth response to O3 exposure could be verified. The resulting bivariate cubic curves (Percy et al., 2006, 2007) of tree cross-sectional area growth response to growing season 4th highest daily maximum 8-h average O3 displayed a significant degree of curvature and an improvement in goodness of fit when compared with a simple bivariate linear model. However, although a non-linear model could slightly enhance goodness of fit and predictive power, it would certainly be less utilitarian. In other words, any increase in predictive power yielded by the more complex cubic regression model may not compensate for lowered ease of use by regulatory agencies (Percy et al., 2007). The importance of WS as a factor in ambient O3 formation (NRC, 1991) and O3 flux through stomata (cf. Ashmore, 2004) is well known. In FACE systems, ambient air is used to dilute higher concentrations of emitted O3 as the air stream is carried from outside the ring, into, and through the tree canopy. Ozone was fumigated in this experiment when WS measured at ring center was above 0.5 m s1 and below 4.0 m s1. It is interesting that the positive relationship of WS to O3 sensitive clone 216 growth in the six-predictor model (Eq. (11.1)) was opposite to that (negative) in the final three-variable (Eq. (11.3)) model. The relationship of peak hourly to seasonal average O3 concentrations (Karnosky et al., 2003c, 2005) within the elevated O3 rings was quite consistent during the 5-year study period. Maximum 1-h average concentration was higher (106 ppb O3 ring 3 in 1999) than in the succeeding years 2000–2003 (o93 ppb) (Fig. 11.3).

Ozone Exposure-Based Growth Response Models

287

We did not have multi-port continuous monitoring data available for the FACE rings. This is being addressed through a planned intensive colocated active and passive monitoring study to more completely assess vertical and horizontal O3 profiles within the tree canopy. However, we know from cumulative monthly exposure data collected by passive monitors that there was only a slight gradient in accumulated O3 exposure within the core area of the elevated O3 rings (Karnosky et al., 2007b) where growth measurements were taken. The pattern of these passive data did not seem to indicate a large influence of WS on O3 concentrations, but rather, possibly the combined influence of mixing with distance and the influence of canopy uptake. If this hypothesis is indeed valid, and it has not been tested here, there seem to be some complex interactions related to WS that should be considered in future analysis. In previously published analyses by Percy et al. (2007), four O3 exposure indices were evaluated for their efficacy as single indicators of aspen and white birch cross-sectional area growth. This work concluded that the annual growing season 4th highest daily maximum 8-h average O3 concentration was a much better single indicator of aspen growth than either SUM60, AOT40 or 1-h maximum O3 concentration. Using the same dataset, we have now extended previous analyses to include the W126 sigmoidally weighted cumulative index developed by Lefohn and Runeckles (1987). As is evident from Table 11.2, with one exception (white birch, maximum 1-h average O3 concentration), only growing season 4th highest daily maximum 8-h average O3 concentration (aspen clones 271, 216, 42E) and W126 (aspen clones 271, 216, 42E, 259, 8L plus white birch) were statistically significant single growth indicators. The theory and application of the W126 index as developed by Lefohn and Runeckles (1987) has recently been succinctly summarized (Lefohn, 2006). The W126 is based upon a sigmoidal weighting function that (1) focuses on hourly average concentrations as low as 40 ppb; (2) has an inflection point near 65 ppb; and (3) has an equal weighting of 1 for hourly average concentrations Z100 ppb. For any hourly average O3 concentration, that concentration is multiplied by the corresponding sigmoidal weighting value and then all concentrations are summed (Lefohn, 2006). The frequency distribution (Fig. 11.2) of hourly average O3 concentrations in this manipulative study resulted in a greater relative weight assigned to approximately 20% of the O3 concentrations that were at or above the designated W126 inflection point of 65 ppb. This weighting may have unduly enhanced the mathematical relationship between W126 O3 exposure and response of the biological endpoint. Pearson correlation analysis had previously indicated that there was no

288

Kevin E. Percy et al.

statistically significant relationship between O3 exposure (growing season 4th highest index) and aspen clone 8L (p ¼ 0.713) or white birch ( p ¼ 0.190) cross-sectional area growth (Table 11.1). There is a continued desire on the part of air quality regulators to move toward a ‘‘biologically based standard’’ to protect vegetation (US EPA, 2006). At this time, it is unclear why the W126 was a statistically significant single indicator of aspen and birch growth. To evaluate the biological relevance of the W126 index statistical significance, we compared, measured, and modeled growth responses. Our conclusion from this analysis is that the W126 index greatly overestimated the negative responses for aspen clones 8L ( p ¼ 0.006), 259 ( p ¼ 0.000), and white birch (p ¼ 0.009) (Table 11.2). This is further supported by the lack of a statistically significant contribution from O3 exposure to aspen 8L (p ¼ 0.900) and white birch (p ¼ 0.540) growth (Table 11.3). Aspen clone 8L and white birch have clearly been documented to be positively affected by O3 (Karnosky et al., 2005; King et al., 2005; Kubiske et al., 2006). In the case of white birch, this may be partly due to a competitive advantage conferred by its greater tolerance to O3 relative to its planted cohort, aspen clone 216. Aspen clone 259 in open-top chamber experiments has been considered to be highly O3 sensitive (Karnosky et al., 1996). However, when inter-planted with other aspen clones, clone 259 manifested very high rates of mortality within the O3 rings during the first two fumigation seasons (1998–1999). There is, of course, the possibility that the only most tolerant individuals within clone 259 were left to be measured during 1999–2003, and, therefore, any modeled estimates for this one clone may have been biased. In summary, at least for our data in this analysis, the ‘‘statistical fit’’ achieved by the W126 O3 exposure index certainly does not reflect the ‘‘biological fit’’ based on measured response. The growing season 4th highest daily maximum 8-h average O3 concentration index does not include a weighting function and, thus, may not be as influenced as W126 by exposure frequency distribution at our lower O3 site over the 5-year period. Rather, it may be more influenced more by the relative difference between peak and average concentrations. The fact that the growing season 4th highest daily maximum 8-h average O3 concentration was a statistically significant indicator of growth only in aspen clones that responded negatively to O3 during the life of the experiment is important. This fact appears to confer greater biological plausibility on it, than on the W126. On the basis of our data, in terms of potential index application within a secondary (welfare-based) standard should it ever be promulgated, the 4th highest daily maximum 8-h average O3 concentration (1) appears to

Ozone Exposure-Based Growth Response Models

289

have greater association than the W126 O3 exposure index with an economically and ecologically relevant biological endpoint (growth) for two widely distributed and important northern hardwood species and (2) has the advantage of requiring only a change in averaging time (to annual) and perhaps a slight change in form. The current primary NAAQS form (Federal Register, 2008) actually requires data from only the 2nd and 3rd quarters of the year (the ‘‘ozone season’’). It is important to note that the growing season 4th highest daily maximum 8-h average O3 concentration indicator used in our models in fact represents the biologically relevant portion of the NAAQS (Federal Register, 2008) and CWS (CCME, 2000). 11.5. Conclusions

We have developed regression-based O3 exposure tree response models comprising annual growing season 4th highest daily maximum 8-h average O3 concentration, accumulated GDDs, and average WS. The models predict extremely well within a wide range of 4th highest daily maximum 8-h average O3 concentration and have immediate relevancy to ambient exposure conditions experienced by two of North America’s most widely distributed tree species. The models are highly statistically significant, have a high degree of goodness of fit, are endpoint based, and should be simple to use within the North American context. The models include defined limits of uncertainty in prediction as required for risk analysis. Our data document that O3 exposure may result in both positive and negative growth responses in aspen and birch that conform to the theory of hormesis. Aspen clone 8L and white birch modeled and measured 5-year growth responses to O3 exposure in the ambient air context conformed to the theory of hormesis, or low dose stimulation followed by higher dose inhibition. Comparative evaluation of five O3 exposure indices demonstrated that the W126 index greatly overestimated the negative response to O3 and that the growing season 4th highest daily maximum 8-h average O3 concentration index has high statistical significance and a much greater association with the biological endpoint. ACKNOWLEDGMENTS

The authors thank Environment Canada—Environmental Economics Branch and Environment Canada—Meteorological Service of Canada for their financial support of this work. We are most appreciative of

290

Kevin E. Percy et al.

helpful discussions with Dr Allan Legge (Biosphere Solutions), Prof. Sagar Krupa (University of Minnesota), and Prof. Hans-Ju¨rgen Ja¨ger (University of Giessen). We are grateful to Dr Mark Kubiske (USDA Forest Service) for providing the diameter growth data and Dr Kurt Pregitzer (Michigan Technological University) for providing the SMC data. Aspen FACE is principally supported by the Office of Science (BER), U.S. Department of Energy (grant no. DE-FG02-95ER62125 to Michigan Technological University, contract no. DE-AC02-98CH10886 to Brookhaven National Laboratory), the U.S. Forest Service Northern Global Change Program and Northern Research Station, Michigan Technological University, Canadian Federal Panel on Energy and Research Development (PERD) and Natural Resources Canada, Canadian Forest Service—Atlantic Forestry Centre.

REFERENCES Ashmore, M. 2004. Assessing the future global impacts of ozone on vegetation. Plant Cell Environ. 28, 949–964. Berrang, P., Karnosky, D.F., Mickler, R.A., and Bennett, J.P. 1986. Natural selection for ozone tolerance in Populus tremuloides. Can. J. For. Res. 16, 1214–1216. Berrang, P., Karnosky, D.F., Mickler, R.A., and Bennett, J.P. 1989. Natural selection for ozone tolerance in Populus tremuloides: Field verification. Can. J. For. Res. 19, 519–522. Bytnerowicz, A., Arbaugh, M.J., and Alonso, R., eds. 2003. Ozone air pollution in the Sierra Nevada: Distribution and effects on forests. Developments in environmental science. Elsevier, Amsterdam, The Netherlands, Vol. 2. Calabrese, E.J. 2005. Paradigm lost, paradigm found: The re-emergence of hormesis as a fundamental exposure response model in toxicological sciences. Environ. Pollut. 138, 378–411. Canadian Council of Ministers of the Environment (CCME). 2000. Canada-wide standards for particulate matter (PM) and ozone. (http://www.ccme.ca, August 3, 2008). Chappelka, A.H., and Samuelson, L. 1998. Ambient ozone effects on forest trees of the eastern United States: A review. New Phytol. 139, 91–108. Dickson, R.E., Lewin, K.F., Isebrands, J.G., Coleman, M.D., Heilman, W.E., Riemenschneider, D.E., Soˆber, J., Host, G.E., Zak, D.R., Hendrey, G.R., Pregitzer, K.S., and Karnosky, D.F. 2000. Forest atmosphere carbon transfer storage-II (FACTS II)—The aspen free-air CO2 and O3 enrichment (FACE) project: An overview. USDA Forest Service, North Central Research Station General Technical Report NC-214, St. Paul, MN, USA. Federal Register, USA. 2008. National ambient air quality standards for ozone: Final rule. Rules and regulations. 40 CFR Parts 50 and 58, Vol. 73, No.60, Thursday, March 27, 2008. Felzer, B., Kicklighter, D., Melillo, J., Wang, C., Zhuang, Q., and Prinn, R. 2004. Effects of ozone on net primary production and carbon sequestration in the conterminous United States using a biogeochemistry model. Tellus, 56B, 230–248. Foley, G.J., Georgopoulos, P.G., and Lioy, P.J. 2003. Accountability within new ozone standards. Environ. Sci. Technol. 393A–399A.

Ozone Exposure-Based Growth Response Models

291

Fuhrer, J., Ska¨rby, L., and Ashmore, M.R. 1997. Critical levels for ozone effects on vegetation in Europe. Environ. Pollut. 97, 91–106. Hogsett, W.E., Weber, J.E., Tingey, D., Herstrom, A., Lee, E.H., and Laurence, J.A. 1997. An approach for characterizing tropospheric ozone risk to forests. Environ. Manage. 21, 105–120. Husch, B., Beers, T.W., and Kershaw, J.A. 2003. Forest mensuration (4th ed.). Wiley, Hoboken, NJ, USA. Ja¨ger, H.-J., and Krupa, S. 2008. Hormesis: Its relevance in phytotoxicology. In: Legge, A.H., ed. Air quality and ecological impacts: Relating sources to effects. Elsevier, Amsterdam, The Netherlands, (Chapter 6, this volume). Karnosky, D.F., Gagnon, Z.E., Dickson, R.E., Coleman, M.D., Lee, E.H., and Isebrands, J.G. 1996. Changes in growth, leaf abscission, and biomass associated with seasonal tropospheric ozone exposures of Populus tremuloides clones and seedlings. Can. J. For. Res. 26, 23–37. Karnosky, D.F., Percy, K.E., Chappelka, A.H., and Krupa, S.V. 2003a. Air pollution and global change impacts on forest ecosystems: Monitoring and research needs. In: Karnosky, D.F., Percy, K.E., Chappelka, A.H., Simpson, C., and Pikkarainen, J., eds. Air pollution, global change and forests in the new millennium. Developments in environmental science. Elsevier, Amsterdam, The Netherlands, Vol. 3, pp. 443–455. Karnosky, D.F., Percy, K., Mankovska, B., Prichard, T., Noormets, A., Dickson, R.E., Jepsen, E., and Isebrands, J.G. 2003b. Ozone affects the fitness of trembling aspen. In: Karnosky, D.F., Percy, K.E., Chappelka, A.H., Simpson, C., and Pikkarainen, J., eds. Air pollution, global change and forests in the new millennium. Elsevier, Amsterdam, The Netherlands, pp. 195–206. Karnosky, D.F., Pregitzer, K.S., Hendrey, G.R., Percy, K.E., Zak, D.R., Lindroth, R.L., Mattson, W.J., Kubiske, M., Podila, G.K., Noormets, A., McDonald, E., Kruger, E.L., King, J., Mankovska, B., Sober, A., Awmack, C., Callan, B., Hopkin, A., Xiang, B., Hom, J., Sober, J., Host, G., Riemenschneider, D.E., Zasada, J., Dickson, R.E., and Isebrands, J.G. 2003c. Impacts of interacting CO2 and O3 on trembling aspen: Results from the Aspen FACE experiment. Funct. Ecol. 17, 289–304. Karnosky, D.F., Pregitzer, K.S., Zak, D.R., Kubiske, M.E., Hendrey, G.R., Weinstein, D., Nosal, M., and Percy, K.E. 2005. Scaling ozone responses of forest trees to the ecosystem level in a changing climate. Plant Cell Environ. 28, 965–981. Karnosky, D.F., Darbah, J.N., Sober, A., Riikonen, J., Kets, K., Nelson, N., Kubiske, M., and Percy, K.E. 2006. Ozone effects on growth and productivity of Populus tremuloides Michx.: A comparison of results from OTC, FACE, and ozone gradient studies with a common set of genetic materials. In: Weiser, G., and Tausz, M., eds. Critical levels of ozone: Further applying and developing the flux-based concept. Proceedings on the Workshop November 15–19, 2005, Obergurgl, Tyrol, Austria. Federal Research and Training Centre for Forests, Vienna, Austria, pp. 325–329. Karnosky, D.F., Skelly, J.M., Percy, K.E., and Chappelka, A.H. 2007a. Perspectives regarding 50 years of research on effects of tropospheric ozone air pollution on U.S. forests. Environ. Pollut. 147, 489–506. Karnosky, D.F., Werner, H., Holopainen, T., Percy, K., Oksanen, T., Oksanen, E., Heerdt, C., Fabian, P., Nagy, J., Heilman, W., Cox, R., Nelson, N., and Matyssek, R. 2007b. Free-air exposure systems to scale up ozone research to mature trees. Plant Biol. 9, 1–10. Kickert, R.N., and Krupa, S.V. 1990. Forest responses to tropospheric ozone and global climate change: An analysis. Environ. Pollut. 68, 29–65.

292

Kevin E. Percy et al.

King, J.S., Kubiske, M.E., Pregitzer, K.S., Hendrey, G.R., McDonald, E.P., Giardina, C.P., Quinn, V.S., and Karnosky, D.F. 2005. Tropospheric O3 compromises net primary production in young stands of trembling aspen, paper birch and sugar maple in response to elevated atmospheric CO2. New Phytol. 168, 623–636. Krupa, S.V., Nosal, M., Ferdinand, J.A., Stevenson, R.E., and Skelly, J.M. 2003. A multivariate statistical model integrating passive sampler and meteorology data to predict the frequency distribution of hourly ambient ozone (O3) concentrations. Environ. Pollut. 124, 173–178. Kubiske, M.E., Quinn, V.S., Heilman, W.E., McDonald, E.P., Marquardt, P.E., Teclaw, R.M., Friend, A.L., and Karnosky, D.F. 2006. Inter-annual climatic variation mediates elevated CO2 and O3 effects on forest growth. Glob. Change Biol. 12, 1054–1068. Laurence, J.A., and Andersen, C.P. 2003. Ozone and natural systems: Understanding exposure, response and risk. Environ. Int. 29, 155–160. Laurence, J.A., Retzlaff, W.A., Kerm, J.S., Lee, E.H., Hogsett, W.E., and Weinstein, D.A. 2001. Predicting the regional impact of ozone and precipitation on the growth of loblolly pine and yellow poplar using linked TREGRO and ZELIG models. For. Ecol. Manage. 146, 251–267. Lefohn, A.S. 2006. The development of the W126 exposure index. (www.asl-associates.com, August 3, 2008.) Lefohn, A.S., and Runeckles, V.C. 1987. Establishing a standard to protect vegetation— Ozone exposure/dose considerations. Atmos. Environ. 21, 561–568. Lefohn, A.S., and Foley, J.K. 1992. NCLAN results and their application to the standardsetting process: Protecting vegetation from surface ozone exposures. J. Air Waste Manage. Assoc. 42, 1046–1052. Lefohn, A.S., Lawrence, J.A., and Kohut, R.J. 1988. A comparison of indices that describe the relationship between exposure to ozone and reduction in the yield of agricultural crops. Atmos. Environ. 22, 1229–1240. Manning, W.J. 2005a. Establishing a cause and effect relationship for ambient ozone exposure and tree growth in the forest: Progress and an experimental approach. Environ. Pollut. 137, 443–454. Manning, W.J. 2005b. Invited review on hormesis. Environ. Pollut. 138, 377. McLaughlin, S.B., and Percy, K. 1999. Forest health in North America: Some perspectives on actual and potential roles of climate and air pollution. Water Air Soil Pollut. 116, 151–197. McLaughlin, S.B., and Nosal, M. 2008. Evaluating ozone effects on growth of mature trees with high resolution dendrometer systems. In: Legge, A.H., ed. Air quality and ecological effects: Relating sources to effects. Elsevier, Amsterdam, The Netherlands, (Chapter 7, this volume). McLaughlin, S.B., Wullschleger, S.D., and Nosal, M. 2003. Diurnal and seasonal changes in stem increment and water use by yellow poplar trees in response to environmental stress. Tree Physiol. 23, 1125–1136. McLaughlin, S.B., Nosal, M., Wullschleger, S.D., and Sun, G. 2007a. Interactive effects of ozone and climate on tree growth and water use in a southern Appalachian forest in the USA. New Phytol. 174, 109–124. McLaughlin, S.B., Wullscheger, S.D., Sun, G., and Nosal, M. 2007b. Interactive effects of ozone and climate on water use, soil moisture content and stream flow in a southern Appalachian forest in the USA. New Phytol. 174, 125–136. Millard, S.P., and Neerchal, N.K. 2001. Environmental statistics with S-Plus. Applied environmental statistics series. CRC Press, Boca Raton, FL, USA.

Ozone Exposure-Based Growth Response Models

293

Musselman, R.C., Lefohn, A.S., Massman, W.J., and Heath, R.L. 2006. A critical review and analysis of the use of exposure- and flux-based ozone indices for predicting vegetation effects. Atmos. Environ. 40, 1869–1888. National Research Council (NRC), USA. 1991. Rethinking the ozone problem in urban and regional air pollution. National Academy Press, Washington, DC, USA. Ollinger, S.V., Aber, J.D., and Reich, P.B. 1997. Simulating ozone effects on forest productivity: Interactions among leaf-canopy and stand-level processes. Ecol. Appl. 7, 1237–1251. Ollinger, S.V., Aber, J.D., Reich, P.B., and Freuder, R.J. 2002. Interactive effects of nitrogen deposition, tropospheric ozone, elevated CO2, and land use history on the carbon dynamics of northern hardwood forests. Glob. Chang. Biol. 8, 545–562. Orbach, R.L. 2005. The case for science: Economic growth, scientific literacy and intellectual excitement. Presentation to AAAS Fellows Breakfast, February 19, 2005 (http://www.science.doe.gov/News_Information/speeches/speeches/05/2005_AAAS_ Fellows/AAAS_Fellows_Breakfast.htm, August 21, 2008). Percy, K.E., and Karnosky, D.F. 2007. Air quality in natural areas: Interface between the public, science and regulation. Environ. Pollut. 149, 256–267. Percy, K.E., Awmack, C.S., Lindroth, R.L., Kubiske, M.E., Kopper, B.J., Isebrands, J.G., Pregitzer, K.S., Hendrey, G.R., Dickson, R.E., Zak, D.R., Oksanen, E., Sober, J., Harrington, R., and Karnosky, D.F. 2002. Altered performance of forest pests under CO2- and O3-enriched atmospheres. Nature 420, 403–407. Percy, K.E., Legge, A.H., and Krupa, S.V. 2003. Tropospheric ozone: A continuing threat to global forests? In: Karnosky, D.F., Percy, K.E., Chappelka, A.H., Simpson, C., and Pikkarainen, J., eds. Air pollution, global change and forests in the new millenium. Elsevier, Amsterdam, The Netherlands, pp. 85–118. Percy, K.E., Nosal, M., Heilman, W., Dann, T., Sober, J., and Karnosky, D.F. 2006. The North American ozone air quality standard: Efficacy and performance with two northern hardwood species. In: Weiser, G., and Tausz, M., eds. Critical levels of ozone: Further applying and developing the flux-based concept. Proceedings of the Workshop November 15–19, 2005, Obergurgl, Tyrol, Austria. Federal Research and Training Centre for Forests, Vienna, Austria, pp. 85–90. Percy, K.E., Nosal, M., Heilman, W., Dann, T., Sober, J., Legge, A.H., and Karnosky, D.F. 2007. New exposure-based metric approach for evaluating O3 risk to North American aspen forests. Environ. Pollut. 147, 554–566. Samuelson, L., and Kelly, J.M. 2001. Scaling ozone effects from seedlings to forest trees. Tansley Review 21. New Phytol. 149, 21–41. Tingey, D.T., Hogsett, W.E., Lee, E.H., and Laurence, J.A. 2004. Stricter ozone ambient air quality standard has beneficial effect on Ponderosa pine in California. Environ. Manage. 34, 397–405. U.S. Environmental Protection Agency (US EPA). 1996. Air quality criteria for ozone and related photochemical oxidants. EPA report no. EPA/600/P-93/004bF. Environmental Protection Agency, Office of Research and Development, Research Triangle Park, NC. U.S. Environmental Protection Agency (US EPA). 2006. Air quality criteria for ozone and related photochemical oxidants. Report no. EPA/600/R-05/004 af. Office of Research and Development, Research Triangle Park, NC. Weinstein, D.A., Laurence, J.A., Retzlaff, W.A., Kern, J.S., Lee, E.H., Hogsett, W.E., and Weber, J. 2005. Predicting the effects of tropospheric ozone on regional productivity of ponderosa pine and white fir. For. Ecol. Manage. 205, 73–89.

Developments in Environmental Science, Volume 9 Allan H. Legge (Editor) Copyright r 2009 Elsevier Ltd. All rights reserved. ISSN: 1474-8177/DOI:10.1016/S1474-8177(08)00212-X

295

Chapter 12 Concluding Remarks Sagar V. Krupa* and Allan H. Legge

12.1. Introduction

In the early days of air pollution science, the primary emphasis of terrestrial vegetation effects research had been on emissions from single or specific, point sources, visible foliar injury on native and cultivated plants and forest damage caused by primary pollutants such as sulfur dioxide, SO2 (Legge & Krupa, 2002). In the last 50 years as photochemical smog and long-range transport of air pollutants became an increasing concern, emphasis shifted from local to regional scale studies. Currently, tropospheric ozone (O3) is considered to be the most important phytotoxic air pollutant worldwide (Krupa et al., 2001; Percy et al., 2004). In addition to field surveys for assessing the visible effects, hundreds of studies were conducted under laboratory conditions in controlled environment fumigation chambers to characterize dose– response relationships with both primary (e.g., SO2) and secondary (e.g., O3) air pollutants (Payer et al., 1990). Those were univariate (all growth-regulating variables except the pollutant (e.g., SO2) of interest being kept relatively constant) studies that mainly consisted of acute exposures (relatively high pollutant concentrations from a few to several hours) (Unsworth & Ormrod, 1982). As our knowledge of the range of air pollutant-induced foliar injury symptoms (e.g., occurrence of chlorosis, bronzing, premature senescence including fall coloration) coupled with changes in plant biology and functional physiology increased, the importance of chronic or whole growth season exposures (e.g., with O3) and responses (Table 12.1) became the focus of numerous studies requiring the use of greenhouse and field exposure chambers (from cuvettes to huge open-top chambers), although many of them continued Corresponding author: E-mail: [email protected]

296

Sagar V. Krupa and Allan H. Legge

Table 12.1. Air pollutant-induced ecological effects on terrestrial vegetation Visible foliar injury Changes in growth patterns Changes in biomass Changes in esthetic qualities Changes in nutritive qualities Changes in recreational qualities Changes in fitness and diversity Changes in ecosystem structure and function

to represent univariate investigations (Manning & Krupa, 1992). Nevertheless, yield data generated from field chamber studies in large national scale networks in the US and Europe were used to describe possible ambient air quality standards or objectives for O3 (Heck et al., 1988; Ja¨ger et al., 1993). Those efforts generated an ongoing debate about the validity of using chamber-based univariate data for regional scale crop loss assessment under ambient conditions (Krupa & Kickert, 1997). Recently, a few scientists have begun to use chamber-less, free air, trace gas exposure systems (e.g., see Karnosky et al., 2004). In Europe, emphasis changed from air concentrations to modeling pollutant fluxes from the atmosphere on to the plant canopy for uptake or the absorbed dose (Atmospheric Environment 2004, Volume 38 [15]). Ambient atmosphere is composed of highly variable combinations of multiple pollutants in time and in space and therefore, any observed negative effect(s) on vegetation is the result of exposure to those pollutant mixtures. However, a particular pollutant, because of its greater phytotoxicity may have a bigger impact at a given time and location (e.g., occurrence of typical visible foliar injury). Nevertheless, presence of two or more pollutants can result in additive, more than or less than additive effects (e.g., Fangmeier et al., 2002; Mansfield & McCune, 1988). That makes it very difficult to conduct artificial field exposure studies that are realistic and can explain the stochastic (random) relationships between cause and effect. Further, such studies can only define a portion of the total response surface. They are frequently limited by small number of treatments due to logistic and financial considerations (Kickert & Krupa, 1991). Alternative approaches to this critical issue are discussed in Section 12.5 of this chapter. In the context of the very important need to examine the entire response surface, Ja¨ger and Krupa (Chapter 6) discuss the phenomenon of ‘‘hormesis’’ where many non-essential chemicals (e.g., O3, heavy metals) stimulate plant growth and other biological processes at low doses, but inhibit such processes at higher levels. ‘‘Hormesis’’ represents

Concluding Remarks

297

an advantage gained by the individual species from the overall resources and energy initially allocated for detoxification and repair, but in excess of that needed to repair the immediate damage. As hormetic effects vary with the plant species, it can result in selective advantage for certain members over others in mixed communities. According to Ja¨ger and Krupa (Chapter 6), although some air pollutant-induced hormetic plant effects have been reported (e.g., Percy et al., 2006), the ability to fully define such an effect, ideally requires the establishment of an endpoint-specific lowest and no observable effects levels, with multiple treatments or doses (Z4) within two orders of magnitude immediately below the no effect level. So far experimental designs have mostly been constructed to optimize exposure doses above an accepted or perceived level to demonstrate adverse effects and thus show insufficient potential for detecting or describing ‘‘hormesis’’ and its impacts on the traditional dose–response functions. That would require a change in the use of traditional experimental designs. Such a shift is also critical in examining interactive effects of multiple plant growthregulating variables (both air pollutants and climate parameters such as air temperature, soil moisture, etc., that are required for normal growth and development) in the ambient environment. In that context, sulfur is an essential plant nutrient, with soil being its main source of supply. The stimulatory effects of atmospheric S on plants growing on soils that have marginal sulfur content is not considered to be an ‘‘hormetic’’ effect by classical definition (see Ja¨ger & Krupa, Chapter 6). Nevertheless, De Kok et al. (Chapter 5) note that sulfurous air pollutants can act as both toxins and as nutrients for plants. However, it is unclear as to what extent metabolism contributes to the detoxification of absorbed sulfur gases, as there is no clear-cut transition in the level or rate of metabolism of the absorbed sulfur gases and their phytotoxicity. Moreover, the effects of sulfurous air pollutants on plant functioning are strongly dependent on the sulfur status of the soil (e.g., see Legge & Krupa, 2002). On a global scale, fossil fuel combustion is the main source for both atmospheric SO2 and NOx (Finlayson-Pitts & Pitts, 1999). As with S, nitrogen is an essential element and a fertilizer. Although there are reasons to believe that at low concentrations gaseous N can be stimulatory, there are no specific studies conducted to address that issue. With the exception of ammonia (Krupa, 2003), in general ambient concentrations of gaseous N species do not exist at phytotoxic concentrations. Aside from its critical role in photochemistry and in the generation of O3 and other oxidants, excess bulk (wet and dry) deposition of total nitrogen is known to adversely alter plant population structure by allowing the invasion of

298

Sagar V. Krupa and Allan H. Legge

grasses into perennial, herbaceous plant communities (Bobbink & Lamers, 2002; Phoenix et al., 2006). Fenn et al. (Chapter 8) describe the development and application of a bulk cation-anion resin adsorption sampler that can be used in remote locations for estimating atmospheric total N deposition in ecological studies.

12.2. Relating source emissions to receptor sites

As indicated previously, early studies on air quality and terrestrial vegetation effects were directed to single or specific point sources (e.g., see Krupa & Legge, 1998). As secondary air pollutants and area or regional scale emissions from multiple sources became increasingly important, source apportionment methods were developed for air quality management (control strategies). According to Hopke (Chapter 1) source apportionment is the estimation of the contributions of elemental emissions from natural and anthropogenic sources to the airborne concentrations at a given location. Integrating source apportionment methods to ecological effects studies would represent a major step in establishing source–effect relationships under ambient condition, but would require very close collaboration between plant and atmospheric scientists. In the overall context, receptor models are applied to elicit information on the sources of air pollutants from the measured constituent air concentrations. Such models include Chemical Mass Balance (CMB), Target Transformation Factorial Analysis (TTFA) coupled to Monte Carlo computer simulation, Positive Matrix Factorization (PMF), etc. (for details see Hopke, Chapter 1). Typically, they use the chemical composition data from repeated sampling and analysis of airborne particulate matter at a given location. In such cases, the outcome is the identification of the pollution source types (e.g., power plant, petroleum extraction plant, mobile sources, vegetation) and estimates of the contribution of each source type to the observed air concentrations at the receptor location. It can also involve efforts to identify the locations of the sources through the use of ensembles of air parcel back trajectories (e.g., NOAA—HYbrid Single Particle Lagrangian Integrated Transport—HYSPLIT model; Draxler, 2003; Draxler & Hess, 1997, 1998). In recent years, there have been improvements in the factor analysis methods that are applied in receptor modeling, as well as easier application of trajectory methods (Hopke, Chapter 1). Overall, these are powerful tools and for a further discussion of the overall subject and for some new approaches to their applications, see Poirot (Chapter 2).

299

Concluding Remarks

12.3. Elemental tracers of source emissions and their accumulation in receptors

Under ambient conditions responses of sensitive plant species can be used to assess relative air quality (Tables 12.2 and 12.3). For example, development of pollutant-specific foliar injury symptoms on sensitive plant species has been used as a biological indication of the relative air quality at a given location and time (e.g., Heggestad, 1991). In some cases progressive disappearance of a particular species in a given geographic area has also been used as an indication of deteriorating air quality (Krupa et al., 1977; Environmental Protection Agency (EPA), following web page was last opened on September 17, 2008: http://cfpub.epa.gov/ ncea/cfm/recordisplay.cfm?deid=149923). Another indicator is shifts in the plant populations within a community (see Bell & Treshow, 2002). Traditionally in addition to foliar injury surveys, a number of investigators have used sulfur accumulation in plant tissues at various distances along directional transects from a point source (e.g., coal-fired power plant, metal smelter, petroleum refinery, natural gas extraction plant) to map zones of impact or no impact in predominant upwind and downwind areas (Linzon, 1978). Legge et al. (1988) and Krupa and Legge (1999, 2001) used differences in the concentrations of total, inorganic and organic S in the plant tissue to differentiate the relative point source plume impacts versus the contribution of the soil. In contrast, Palau et al.

Table 12.2. Types of plant indicators of air quality Indicator type Sentinels Detectors Exploiters Accumulators Ecosystem indicators

Description Introduced into the environment of concern Occurring naturally in an area of interest Presence indicates pollution or disturbance Accumulate pollutants in measurable quantities Changes in population structure and diversity

Table 12.3. Temporal responses of plant indicators to air quality Response type Acute, short-term Chronic, long-term

Indicator type Sentinels and detectors Biomarkers (e.g., change in chlorophyll fluorescence) Accumulators Exploiters Ecosystem indicators

300

Sagar V. Krupa and Allan H. Legge

(Chapter 4), used an Elemental Enrichment Analysis (EEA) to separate the contributions of the atmosphere from those of the soil to Austrian pine (Pinus nigra) foliar concentrations of total S and other elements. Based on the least amount of variance between several elemental concentrations in P. nigra needles and the corresponding soils in the plots, aluminium was chosen as the normalization element for computing elemental enrichment factors. Those results identifying sites with various levels of impacts were in close agreement with the results of measured plume transport and deposition in the complex terrain, before and after the installation of the SO2 control system. Palau et al. verified their results independently by plume tracking over a number of years using groundbased, but mobile Correlation Spectroscopy (CoSpec) and a fast response pulse fluorescence SO2 analyzer. Thus, EEA offers a significant advancement in the traditional application of S accumulation in plant tissues to map plume impacts. Earlier, using a different, but a more sophisticated approach, in the West White Court Case Study in Alberta, Canada, Legge et al. used stable S isotopes (32S: 34S) as plume tracers into the ecosystem (for a summary, see Krupa & Legge, 1998). Caution is warranted in using single elemental isotopes as tracers in impact assessment. For example, the 32S: 34S in the emission must be distinctly different from the background value and that was the case in the studies of Legge et al. In contrast, Krupa (1981) was unable to find a similar differentiation regarding a coal-fired power plant plume in Minnesota, US. The probability of success is increased with the application of multi-element stable isotopes. Savard et al. (Chapter 9) used dual elemental isotopes of 13 C and 18O in stem cellulose to examine the stress responses of tress to air pollution in an urban corridor in Quebec, Canada. Another step in the application of stable elemental isotopes in environmental research is the use of three elements, for example S, N, and O (i.e., SO42 and NO3). In addition to the West White Court Case Study of Legge et al., with S, recently the stable isotope 17O signal of NO3 has been used as a tracer of atmospheric NO3 and was found to be a more robust tracer of atmospheric NO3 than 15N and 18O methods (Michalski et al., 2004). Certainly the use of stable elemental isotopes allows the separation of anthropogenic from the influence of natural sources. It also allows the tracing of the fate of the element through the ecosystem components and consequently its impacts. However, where multiple source plumes are involved with not so distinctly different stable isotopic signals, other source signatures must be used. One aspect of receptor modeling involves the use of US EPA’s ‘‘Speciate’’ source finger print library on inorganic elemental composition

Concluding Remarks

301

of emissions from various types of sources (http://www.epa.gov/ttn/chief/ software/speciate/index.html, last opened on September 11, 2008). Here elements such as As, Be, Cd, Cr, Hg, Ni, Pb, Rb, Se, Sr, Ti, and V and others are included in the source apportionment methodology (see Hopke, Chapter 1, also EPA’s ‘‘Speciate’’ web site). Many of these and other elements (both essential and non-essential) accumulate in plant tissues, particularly in lower plants such as the lichens (e.g., see Aznar et al., 2008; also the review of Bates, 2002). The concept of source apportionment and receptor modeling are based on data gathered separately on the chemical composition of fine (o 2.5 mm) and coarse (W2.5 mm) particles. Where opportunities for those types of data collection do not exist, because of logistic and financial restrictions, accumulation of elements in plant tissues (receptor accumulation) can be used in source apportionment. An excellent example relates to the works of Sloof (1995a, 1995b) who used spatial variability in the elemental composition of lichens throughout the Netherlands to map air quality as influenced through source apportionment.

12.4. Relating elemental accumulation to vegetation effects

With the exception of O3, many other air pollutants such as SO2 (S), NOx–NOy (N), HF hydrogen fluoride (F), and trace metals accumulate in foliar tissues. It is very important to note that normally soil is the predominant source for many of the elements measured in plants. Palau et al. (Chapter 4) used the EEA to separate the role of the soil from the atmosphere. But in receptor modeling, as shown by Sloof (1995a, 1995b), patterns and variability of multi-elemental accumulation in plants over a region can be used in source apportionment. Here, the key is the use of multiple elements as source fingerprints (predominantly source-specific spectral patterns of elements) and not a single element such as S. The critical requirement here is, there is a need to demonstrate co-linearity between the occurrences of the phytotoxic element such as S (SO2) and the elements (trace metals) used in the receptor modeling. At this time, there are many studies for example, on the accumulation of S or its metabolic products (see Legge & Krupa, 2002; De Kok, Chapter 5) and plant physiological responses such as changes in photosynthesis. However, to our knowledge there are no studies relating the dynamics of atmospheric deposition, tissue elemental accumulation and irreversible effects such as yield reductions. Such studies will involve repeated measurements (time series of relating tissue elemental accumulation to growth or yield) and multi-point modeling of chronic relationships

302

Sagar V. Krupa and Allan H. Legge

of cause and effect. However, repeated measurements are one of the backbones of atmospheric receptor modeling (see Hopke, Chapter 1). In air pollution—plant biomass effects literature, virtually all of the studies relate to single point, season end correlations. However, as shown by Kuik et al. (1993a, 1993b), even in single point measurements, multifactorial analysis coupled with Monte Carlo simulation can be used in receptor modeling in partitioning the types of emission sources and their contributions to the elemental accumulation in plant (lichen) tissues distributed over a wide geographic area such as the Netherlands. Lower plants such as bryophytes and lichens are excellent accumulators of various elements (McCarthy et al., Chapter 10) and can be used in receptor modeling. Particularly species that are epiphytic or those that are representative of ombotrophic plant communities derive virtually their entire tissue elemental signature through atmospheric uptake. The major limitation here is a need for repeated measurements over multiple years to establish a recurring or changing patterns of multiple source contributions to the receptor. However, within the sampling limitations (single point), it is still possible to gain meaningful results as described by Kuik et al. (1993a, 1993b) and Sloof (1995a, 1995b).

12.5. Future perspectives

Although elemental accumulation in biological receptors such as in the lichens can be used in source apportionment, there are issues associated with biodiversity (loss of sensitive lichen species themselves, as with nitrophobic (Bobbink & Lamers, 2002) or SO2 sensitive (Bates, 2002) species). A major objective should include long-term impacts on the primary producers (the energy flow) in the ecosystem. According to Guderian (1977), pollutant accumulation in lichens is comparable to that in higher plants. Under various SO2 exposure regimes, tissue S accumulation rates in lichens and strawberry and white pine foliage was found to be similar. Comparison of the S uptake with concentration (c) and exposure time (t) under similar products of c and t showed that pollutant uptake of lichens was more dependent on the exposure time than on concentration. A main difference between the lichens and the higher plants is the lack of ability of lichens to dilute the absorbed pollutants through the formation of significant amounts of new plant material with low natural levels of the particular element in question. Thus, lichens are more sensitive and can be used as an early warning system in spatial mapping of impact and no impact zones to assist in the assessment of

Concluding Remarks

303

long-term air pollutant effects on the growth and productivity of the primary producers in the ecosystem. In the context of the primary producers, even repeated measurements of tissue elemental concentrations by themselves are not likely to account satisfactorily for the stochastic behavior of growth and biomass relationships under ambient conditions, because the combined effects of more than one or multiple growth-regulating factors (air pollutant mixtures, air temperature, precipitation, diseases, etc.) are in effect. Few scientists have addressed this complex problem of the ambient environment, although there are ways to do so. Many mechanistic or process models have been developed, but have not been validated with independent sets of data (see e.g., Kickert & Krupa, 1991). Recently Krupa et al. (web page last opened on September 20, 2008: http:// plpa.cfans.umn.edu/Bsagark/WCAS_Report_Krupa.pdf ) and Lin et al. (2007) offered potential approaches to addressing this issue (Table 12.4). Although these are first order efforts, they represent general time series methods that are multi-variant models that can accommodate repeated measurements of tissue elemental accumulation rates. That is in addition to other independent variables such as air temperature, precipitation Table 12.4. Empirical dynamic, time series relationships between multiple air pollutant exposures, climate variables, and alfalfa biomass yield 1998–2002 Global regression analysis: alfalfa yield versus harvest # (two harvests per growth season) and all other predictors. 78 cases included Nomenclature (examples): O3p95_1 ¼ hourly O3 concentration 95th percentile during alfalfa growth stage # 1 (early growth stage) O3 med_2 ¼ median hourly O3 concentration during alfalfa growth stage # 2 (exponential growth stage) SO2_3 ¼ integral (concentration  duration) of SO2 exposures during alfalfa growth stage # 3 (late growth stage) NOx exposures are also defined as an ‘‘integral’’. All others (T ¼ air temperature; RH ¼ relative humidity; GSR ¼ global solar radiation) are average values, except precipitation (precip.) in total depths. Harvest ¼ first and second harvests during each growth season Alfalfa shoot biomass per 100 stems ¼ 6506 – 17.2 O3med_1þ9.76 O3p95_1 – 88.0 O3med_2 þ 40.9 O3p95_2 þ 104 O3med_3 – 34.4 O3p95_3 þ 0.691 SO2_1 – 0.140 NOx_1 – 45.4 T_1 þ 18.1 RH_1 þ 2.72 GSR_1 þ 0.574 SO2_2 þ 0.148 NOx_2 þ 34.4 T_2 – 60.7 RH_2 – 4.71 GSR_2 þ 0.177 SO2_3 þ 0.260 NOx_3 – 176 T_3 – 10.3 RH_3 – 2.05 GSR_3 – 6.03 precip1 – 38.6 Harvest R2 ¼ 77.6%, R2 (adjusted) ¼ 68.1%, p ¼ 0.000 Source: Modified from Krupa, Nosal, Ryl and Legge (original web page last opened on September 20, 2008): http://plpa.cfans.umn.edu/Bsagark/WCAS_Report_Krupa.pdf

304

Sagar V. Krupa and Allan H. Legge

depth, etc. that effect changes in plant growth and biomass under ambient (not experimental) conditions. Thus, repeated measurements of biological responses can be coupled to repeated measurements of elemental deposition/accumulation in receptor modeling in deriving great benefits in source apportionment studies that has so far eluded plant scientists. Equally importantly, such efforts can allow an assessment of the efficacy of air quality regulatory policies (data from before and after their implementation of control strategy).

REFERENCES Aznar, J.C., Richer-Lafle`che, M., and Cluis, D. 2008. Metal contamination in the lichen Alectoria sarmentosa near the copper smelter of Murdochville, Que´bec. Environ. Pollut. 156, 76–81. Bates, J.W. 2002. Effects on bryophytes and lichens. In: Bell, J.N.B., and Treshow, M., eds. Air pollution and plant life, second ed., Wiley, Chichester, UK, pp. 309–342. Bell, J.N.B., and Treshow, M., eds. 2002. Air pollution and plant life (second ed.). Wiley, Chichester, UK. Bobbink, R., and Lamers, L.P.M. 2002. Effects of increased nitrogen deposition. In: Bell, J.N.B., and Treshow, M., eds. Air pollution and plant life (second ed.). Wiley, Chichester, UK, pp. 201–236. Draxler, R.R. 2003. Evaluation of an ensemble dispersion calculation. J. Appl. Meteorol. 42, 308–317. Draxler, R.R., and Hess, G.D. 1997. Description of the HYSPLIT 4 modeling system. NOAA Tech Memo ERL ARL-224. Draxler, R.R., and Hess, G.D. 1998. An overview of the HYSPLIT 4 modeling system for trajectories, dispersion, and deposition. Aust. Meteorol. Mag. 47, 295–308. Fangmeier, A., Bender, J., Weigel, H.-J., and Ja¨ger, H.-J. 2002. Effects of pollutant mixtures. In: Bell, J.N.B., and Treshow, M., eds. Air pollution and plant life (second ed.). Wiley, Chichester, UK, pp. 251–272. Finlayson-Pitts, B.J., and Pitts, J.N. 1999. Chemistry of the upper and lower atmosphere. Academic Press, New York. Guderian, R. 1977. Air pollution, ecological studies 22. Springer, Berlin. Heck, W.W., Taylor, O.C., and Tingey, D.T., eds. 1988. Assessment of crop loss from air pollutants. Elsevier Applied Science, London, UK. Heggestad, H.E. 1991. Origin of Bel-W3, Bel-C and Bel-B tobacco varieties and their use as indicators of ozone. Environ. Pollut. 74, 264–291. Ja¨ger, H.-J., Unsworth, M., De Temmerman, L., and Mathy, P., eds. 1993. Effects of air pollution on agricultural CROPS in Europe. Air Pollution Research Report 46. Commission of the European Communities, Brussels, Belgium. Karnosky, D.F., Percy, K.E., Chappelka, A.H., Simpson, C., and Pikkarainen, J., eds. 2004. Air pollution, global change and forests in the new millennium. Developments in environmental science (vol. 3). Elsevier, Amsterdam, The Netherlands. Kickert, R.N., and Krupa, S.V. 1991. Modeling plant response to tropospheric ozone: A critical review. Environ. Pollut. 70(4), 271–383.

Concluding Remarks

305

Krupa, S.V. 1981. Effects of dry deposition components of acidic precipitation on vegetation. In: D’Itri, F., ed. Effects of acidic precipitation on ecological systems. Ann Arbor Science, Ann Arbor, MI, pp. 471–484. Krupa, S.V. 2003. Effects of atmospheric ammonia (NH3) on terrestrial vegetation: A review. Environ. Pollut. 124, 179–221. Krupa, S.V., and Kickert, R.N. 1997. Considerations for establishing relationships between ambient ozone (O3) and adverse crop response. Environ. Rev. 5, 55–77. Krupa, S.V., and Legge, A.H. 1998. In: Ambasht, R.S., ed. Sulphur dioxide, particulate sulphur and their impacts on a boreal forest ecosystem, modern trends in ecology and environment. Backhuys Publishers, Leiden, The Netherlands, pp. 285–306. Krupa, S.V., and Legge, A.H. 1999. Foliar injury symptoms of Saskatoon serviceberry (Amelanchier alnifolia Nutt.) as a biological indicator of ambient sulfur dioxide exposures. Environ. Pollut. 106, 449–454. Krupa, S.V., and Legge, A.H. 2001. Saskatoon serviceberry and ambient sulfur dioxide exposures: Study sites re-visited, 1999. Environ. Pollut. 111, 363–365. Krupa, S.V., Kohut, R.J., and Laurence, J.A. 1977. Air pollution—vegetation effects studies in two national parks. In: Linn, R., ed. Proceedings of the first conference on scientific research on national parks. US Department of Interior, Washington, DC, pp. 189–193. Krupa, S.V., McGrath, M.T., Andersen, C.P., Booker, F.L., Burkey, K.O., Chappelka, A.H., Chevone, B.I., Pell, E.J., and Zilinskas, B.A. 2001. Ambient ozone and plant health. Plant Dis. 85, 4–12. Kuik, P., Blaauw, M., Sloof, J.E., and Wolterbeek, H.Th. 1993a. The use of Monte Carlo methods in factor analysis. Atmos. Environ. 27, 1967–1974. Kuik, P., Sloof, J.E., and Wolterbeek, H.Th. 1993b. Application of Monte Carlo-assisted factor analysis to large sets of environmental pollution data. Atmos. Environ. 27, 1975–1983. Legge, A.H., and Krupa, S.V. 2002. Effects of sulphur dioxide. In: Bell, J.N.B., and Treshow, M., eds. Air pollution and plant life (second ed.). Wiley, Chichester, UK, pp. 135–162. Legge, A.H., Bogner, J.C., and Krupa, S.V. 1988. Foliar sulphur species in pine: A new indicator of a forest ecosystem under air pollution stress. Environ. Pollut. 55, 15–27. Lin, J.C., Nosal, M., Muntifering, R.B., and Krupa, S.V. 2007. Alfalfa nutritive quality for ruminant livestock as influenced by ambient air quality in west-central Alberta. Environ. Pollut. 149, 99–103. Linzon, S.N. 1978. Effects of airborne sulfur pollutants on plants. In: Nriagu, J.O., ed. Sulfur in the environment, Part II: Ecological impacts. Wiley, New York, pp. 109–162. Manning, W.J., and Krupa, S.V. 1992. Experimental methodology for studying the effects of ozone on crops and trees. In: Lefohn, A.S., ed. Surface level ozone exposures and their effects on vegetation. Lewis Publishers, Inc., Chelsea, MI, pp. 93–156. Mansfield, T.A., and McCune, D.C. 1988. Problems of crop loss assessment when there is exposure to two or more gaseous pollutants. In: Heck, W.W., Taylor, O.C., and Tingey, D.T., eds. Assessment of crop loss from air pollutants. Elsevier Applied Science, London, pp. 317–344. Michalski, G., Meixner, T., Fenn, M., Hernandez, L., Sirulnik, A., Allen, E., and Thiemens, M. 2004. Tracing atmospheric nitrate deposition in a complex semiarid ecosystem using D17O. Environ. Sci. Technol. 38, 2175–2181. Payer, H.D., Pfirrmann, T., and Mathy, P. 1990. Environmental research with plants in closed chambers. Air Pollution Research Report 26. Commission of the European Communities, Brussels, Belgium. Percy, K.E., Legge, A.H., and Krupa, S.V. 2004. Tropospheric ozone: A continuing threat to global forests? In: Karnosky, D.F., Percy, K.E., Chappelka, A.H., Simpson, C., and

306

Sagar V. Krupa and Allan H. Legge

Pikkarainen, J., eds. Air pollution, global change and forests in the new millennium. Elsevier Science, Amsterdam, The Netherlands, pp. 85–118. Percy, K.E., Nosal, M., Heilman, W., Dann, T., Sober, J., and Karnosky, D.F. 2006. The North American ozone air quality standard: Efficacy and performance with two northern hardwood species. In: Weiser, G., and Tausz, M., eds. Critical levels of ozone: Further applying and developing the flux-based concept. Proceedings of the Workshop 15–19 November 2005, Obergurgl, Tyrol, Austria. Federal Research and Training Centre for Forests, Vienna, Austria, pp. 85–90. Phoenix, G.K., Hicks, W.K., Cinderby, S., Kuylenstierna, J.C., Stock, W.D., Dentener, F.J., Giller, K.E., Austin, A.T., Lefroy, R.D.B., Gimeno, B.S., Ashmore, M.R., and Ineson, P. 2006. Atmospheric nitrogen deposition in world biodiversity hotspots: The need for a greater global perspective in assessing N deposition impacts. Glob. Chang. Biol. 12, 470–476. Sloof, J.E. 1995a. Lichens as quantitative biomonitors for atmospheric trace-element deposition, using transplants. Atmos. Environ. 29, 11–20. Sloof, J.E. 1995b. Pattern recognition in lichens for source apportionment. Atmos. Environ. 29, 333–343. Unsworth, M.H., and Ormrod, D.P., eds. 1982. Effects of gaseous air pollution in agriculture and horticulture. Butterworth Scientific, London.

307

Author Index

Aznar, J-C. 229 Be´gin, C. 229 Brand, U. 247 Bytnerowicz, A. 179 Calatayud, V. 99 Campbell, D.H. 179 Cape, J.N. 61 Clow, D.W. 179 Craig, B. 247 Dann, T. 269 De Kok, L.J. 121 Fenn, M.E. 179 Heilman, W. 269 Hopke, P.K. 1 Ja¨ger, H-J. 137 Karnosky, D.F. 269 Krupa, S.V. 99, 137, 295 Legge, A.H. 295

Marion, J. 229 McCarthy, D.P. 247 McLaughlin, S.B. 153 Milla´n, M. 99 Molotch, N.P. 179 Nosal, M. 153, 269 Padgett, P.E. 179 Palau, J.L. 99 Percy, K.E. 269 Pleim, J.E. 179 Poirot, R.L. 35 Sanz, M. 99 Savard, M.M. 229 Sickman, J.O. 179 Smirnoff, A. 229 Sober, J. 269 Stuiver, C.E.E. 121 Stulen, I. 121 Tonnesen, G.S. 179 Weathers, K.C. 179 Yang, L. 121

309

Subject Index

Accumulation 61–64, 67, 70–77, 79–80, 82, 84–91, 100–101, 105, 112, 119, 125, 128, 132, 174, 179, 182, 184, 196–197, 204, 211, 213, 216–217, 219, 299–304 Accumulation, plant 299–301 Accumulation, pollutant 70–71, 73, 76–77, 80, 91 Accumulators 61, 63, 65, 67, 69, 71, 73, 75, 77, 79, 81, 83, 85, 87–89, 203, 299, 302 Aerosols 36, 195, 213 Air quality changes 229, 242 Air Quality Standards 125, 271, 273, 296 Ammonia 14, 57, 66, 77, 186, 198, 206, 248, 297 Ammonium 4, 44, 57, 66, 70, 87, 187, 192–193, 197–198, 200, 206 Apportionment 1–5, 7, 9, 11, 13, 15, 17, 19, 21, 23, 25, 27–28, 117, 298, 301–302, 304 Arid systems 181–182, 185, 193, 196 Asian dust 10, 15, 17–18 Assimilation 112, 128, 190, 242 Back trajectories 1, 15, 17, 19, 21, 44, 46, 48, 298 BAI 234, 236, 238, 240–242 Basal area increment 234, 237–238 Benzene 63–64, 66 Best multiple regression model 159–160 Best subsets regression 282, 286 Bioaccumulation 91 Biodiversity 248, 251–252, 264, 302 Bioindicators 70–71, 82, 88 Branch washing 182, 186 Bulk deposition 75, 181, 183, 189–192, 200–202, 213, 220 b-curve 138–139, 141, 148 d13C 229–231, 234–236, 238–242 CALPUFF 200, 206–207

CAPITA 36, 40, 42 CASTNET 194–195, 204 CATT 35–36, 40, 42, 44, 46–50, 57, 59 Cause-effect relationships 82–84, 100, 121–132, 137, 272 Chemical analysis 68, 71–72, 79, 86, 89, 117, 189, 191, 249, 254 Chemical Mass Balance 4, 298 Cloud water 70, 186, 191–192, 205 CMB 4, 298 Combined Aerosol Trajectory 35–36, 50 Conductance 63, 123, 125, 155–156, 230, 241–243 Correlation spectrometer 99, 105 COSPEC 99, 105–106, 300 Cultivated indicator 77, 80–81, 88 Dendro-chronology 234, 243 Dendrophysiology 155 Deposition fluxes 181–183, 185, 188–189, 192–194, 196–197, 203, 207, 209–210, 220 Deposition, plant 61 Deposition, rates 67–68, 71, 82, 85, 90–91 Deposition, soil 61, 70, 81 Dry deposition 85, 89–90, 122, 126, 180–185, 189, 191–196, 200, 204–205, 208–209, 212–213, 220 EC 9–13, 22–25, 125 Effects 27, 35–36, 61–63, 68, 72, 83–86, 88–91, 99–101, 103, 106–108, 111–113, 116–117, 121–122, 126, 131–132, 137, 140–143, 146–149, 153–155, 157–159, 161, 163, 165–167, 169–171, 173–175, 180, 183, 192, 196–198, 203, 205–208, 234–235, 240–241, 243, 248, 250, 270–272, 295–298, 301–303 Eigen value 4–5 Electrochemical measurements 155–156

310 Elemental accumulation 77, 105, 119, 301–303 Elemental carbon 5, 21, 27 Elemental enrichment analysis 99, 116, 300 Elemental tracer 100, 299 EMAN 249, 251, 255, 264 Empirical deposition modeling 184, 204 EPS 234–236, 238 Expressed population signal 235 FACE 35, 146, 215, 273–277, 279–280, 282–283, 285–287 FASTNET 35–36, 38–42, 44, 57, 59 Fingerprints 35, 37, 39, 41, 43, 45, 47, 49, 51, 53, 55, 57 Flick’s law 123 Fog 142, 182, 185–186, 191–195, 204, 209 Foliar injury 100, 113, 143, 295–296, 299 Foliar uptake 122–123, 125–126, 131 Forest trees 153–159, 161, 163, 165, 167, 169, 171, 173–175, 270–271 Formaldehyde 66 Free Air Carbon dioxide Enrichment 273 Functional plant responses 121–122 GAM 158, 160, 169–173, 175 General additive model 158, 160 Global climate change 153 Growth rate 72–73, 75, 80, 83, 88, 127, 130, 145, 162–163, 172, 175, 234, 238, 240, 242 Growth response model 269, 271, 273, 275, 277, 279, 281, 283, 285, 287 H2S 87, 121–132, 230 HCl 66, 254 Heavy metals 62, 81, 89–91, 141–142, 148, 296 High elevation ecosystems 180 High-elevation basins 211 High-resolution dendrometer 153, 156, 173 HNO3 14, 64, 66, 186, 194, 196, 199–200, 206 Hormesis 137–141, 143, 145–149, 269, 285, 289, 296–297 Hormetic effect 137, 142–143, 145–149, 297 Hormetic Response 137–138, 141, 145, 147–148 Hydrogen chloride 66 Hydrogen sulfide 87, 121, 131, 230 HYSPLIT 15, 17, 298

Subject Index IAP 247–250, 252–253, 255–257, 262, 264–265 IER 183, 187, 193, 197, 200–203, 213 IMPROVE 8–11, 17–18, 36, 38, 40, 44–54, 57, 59, 194–196, 202, 208 Index of air purity 264 Indicator variables 276, 281 Inferential method 179, 185–186, 192–194, 200, 212 Ion exchange resin 183, 187, 200–202 Isotope geochemistry 234, 243 LANDMod 204 Lichen bags 249–250, 254–255, 257, 260, 262, 264 Lichen biogeochemistry 254–255 Lichen mapping 247, 249, 255, 264 Lichens 71, 74, 197, 203, 248, 251–254, 257, 259, 261–264, 301–302 LOEL 139, 141, 149 Mature trees 154–155, 157 Mesophyll 123, 125, 241 Metabolism 63, 121–123, 125–127, 131–132, 297 Montane ecosystem 179–180 Monte Carlo Simulation 302 Multi-point models 301 N deposition 89, 122, 179–181, 183, 185, 187, 189, 191–197, 199, 201, 203, 205, 207, 209, 211, 213, 215, 217, 219 N fluxes 179 N loading 180, 186, 194 NADP 181, 189–190, 194–195, 204, 211, 219 NH3 64, 66–67, 88–89, 126, 186, 194–195, 198, 200, 205, 209, 248 NH+ 4 66, 88, 89 Nitrate 4, 10, 13–14, 27, 44, 56–57, 70, 186–187, 192–193, 196–197, 200, 206, 211, 216, 220, 230 Nitric acid 66, 186, 194 Nitrogen 62, 65, 70, 87–88, 100, 128, 154, 179–181, 183, 185, 187, 189, 191, 193, 195–197, 199, 201, 203, 205, 207, 209, 211, 213, 215, 217, 219–220, 230, 248, 297 Nitrogen accumulation 196

Subject Index Nitrogen deposition 179, 181, 183, 185, 187, 189, 191, 193, 195, 197, 199, 201, 203, 205, 207, 209, 211, 213, 215, 217, 219 Nitrogen dioxide (NO2) 64–65, 195, 198, 200, 248 NO3 300 NOAEL 138 d18O 229–231, 234–236, 238–242 O3 63–64, 100, 143–147, 153, 161–162, 165–168, 170–171, 194, 198–200, 229–230, 232–233, 242, 248, 269–289, 295–297, 301, 303 O3 exposure indices 271, 279–280, 287, 289 O3 exposure-response models 273 OC 9–15, 22–25, 44, 46–48 Ogawa sampler 198–199 Organic carbon 5, 21, 44–45 Ozone 63, 83, 100, 126, 137, 142–143, 145, 153–175, 194, 229, 232, 248, 250, 269–275, 277, 279–283, 285–287, 289, 295 Particles 4, 8, 17, 21, 27, 67–70, 72, 85–86, 89–90, 100, 187, 191, 196, 206–207, 213, 301 Particulate matter 1, 5, 7, 21, 101, 193, 195–196, 207, 298 Passive samplers 180, 182–183, 197, 199–200 PCA 4, 6–7 Pedospheric sulfate 127, 130 Phytotoxicity 121, 125–126, 131–132, 137, 166, 175, 242, 296–297 Plume dispersion 100–101, 103–104, 111–113, 117 PMF 4–9, 15, 19–21, 49, 298 Point source 27, 62, 77, 82–83, 87–88, 100, 207, 230, 295, 298–299 Pollutant absorbed dose 63 Pollutant exposure 71, 73–74, 81, 84–86, 91, 145, 148–149, 199, 303 Pollutant stress 91 Positive Matrix Factorization 4, 49, 298 Potential Source Contribution Function 17, 19, 23, 25–26, 49 Precipitation monitoring 200 Primary pollutants 100–101, 295 Principal Component Analysis 4 PSCF 17, 19–22, 24–26

311 Receptor modeling 1, 7, 27, 113, 117, 119, 298, 300–302, 304 Receptor models 1, 3, 27–28, 49, 100–101, 298 Regional patterns 83 Residence Time Analysis 17 Rime ice 191–192 Root uptake 61, 70, 87 RTA 17 Saharan dust 18 Sample 3, 6, 9, 15, 22, 40, 44, 47, 59, 71–73, 77, 80, 147, 153, 157, 184, 189–190, 198, 200, 213, 215, 254 Sampling 9, 17, 21–22, 27, 57, 68, 71–72, 75–77, 79–80, 83–85, 89–91, 179, 183–184, 189, 195, 198, 200, 202, 212–215, 234, 248, 252–253, 255, 259–260, 262, 275, 298, 302 Secondary pollutants 47 Seedlings 127, 129–130, 154–155 Shoot: root ratio 127 Simulation modeling 179, 183, 203 Single point models 302 Snowfall 191, 201, 211, 213 So 2 4 112 Source apportionment 1, 3, 5, 7, 9, 11, 13, 15, 17, 19, 21, 23, 25, 27, 117, 298, 301–302, 304 Spatial variability 72, 76, 79, 90, 213, 301 Speciate 4, 101, 300–301 Speciation Trends Network (STN), 196 Stable isotopes 216, 229, 231, 233–235, 237, 239, 241, 300 Stem cellulose 229, 300 Stem increment 157–160, 164, 166–167, 170, 172, 174–175 Step-wise multiple regression model 166–168 Stomata 61–64, 66, 68, 70–71, 85, 90, 154, 204–205, 207, 242, 286 Sulfate 4–5, 8–10, 13–14, 21–22, 26–27, 44–45, 52–57, 65, 67, 69–70, 86–87, 112, 124–132, 200, 216, 220 Sulfate: Selenium ratio 52, 56 Sulfur 8–9, 14, 49–50, 52, 62, 65, 71, 85–86, 99–101, 103, 105, 107, 109, 111, 113, 115, 117, 121–122, 124–132, 179–180, 230, 232–233, 248, 295, 297, 299

312 Sulfur dioxide, (SO2) 14, 22, 26, 47, 49, 51–52, 55, 64–65, 67, 72, 85–87, 99–106, 111–113, 116–119, 121–129, 131–132, 194, 198–199, 230, 232–233, 236, 238–242, 248, 250, 255, 295, 297, 300–303 Target transformation factorial analysis 298 TCA 64, 69 Temporal trends 75, 82, 84, 183, 194 Thiol 124–125, 127–129, 131–132 Threshold concentration 122, 146 Threshold value 145, 148–149, 157 Throughfall 70, 179–183, 185–190, 192–193, 195, 197, 200–203, 208, 213, 220 Time Series multiple regression model 168–169 Total S 55, 71, 128–130, 216 Toxicity 27, 89, 122, 126, 131–132, 140 Trace elements 27, 83, 85, 140, 254 Trace metals 35, 37, 39, 41, 43, 45, 47, 49, 51, 53, 55, 57, 100, 301 Transects 77, 83, 105, 254–255, 258–259, 262, 299 Transplants 75, 77, 80–81, 88 Tree ring analysis 234, 238

Subject Index Trembling aspen 269, 271, 273, 280, 282 Trichloroacetate 63, 67, 70 TTFA 298 UNMIX 4–5, 7, 49 Uptake 61–67, 69–71, 73–74, 76, 80–81, 83, 85–87, 89–91, 112, 122–123, 125–126, 128, 130–131, 164, 186–188, 192–193, 207–208, 229, 259, 262, 285, 287, 296, 302 Urban air quality 231, 233, 235, 237, 239, 241, 247, 249, 251, 253, 255, 257, 259, 261, 263–264 Urban region 229–230 Visible foliar injury 100, 295–296 Water use 154–156, 174 Wet deposition 69, 85, 87–91, 122, 126, 181, 184, 186, 189–191, 193–195, 205–207, 209, 211–213, 219–220 White birch 269, 271, 273–274, 276, 278, 280–285, 287–289 ZEP 139, 148–149