Scale, Governance and Change in Zambezi Teak Forests : Sustainable Development for Commodity and Community [1 ed.] 9781443889261, 9781443887151

Citation preview

Scale, Governance and Change in Zambezi Teak Forests

Scale, Governance and Change in Zambezi Teak Forests: Sustainable Development for Commodity and Community By

Michael Musgrave

Scale, Governance and Change in Zambezi Teak Forests: Sustainable Development for Commodity and Community By Michael Musgrave This book first published 2016 Cambridge Scholars Publishing Lady Stephenson Library, Newcastle upon Tyne, NE6 2PA, UK British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Copyright © 2016 by Michael Musgrave All rights for this book reserved. No part of this book 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 permission of the copyright owner. ISBN (10): 1-4438-8715-3 ISBN (13): 978-1-4438-8715-1

Dedicated to all the people of Mulobezi and south western Zambia

v

I keep six honest serving-men (They taught me all I knew) Their names are What and Why and When And How and Where and Who. I send them over land and sea, I send them east and west; But after they have worked for me, I give them all a rest. Rudyard Kipling

vi

Contents Page List of Figures

ix

List of Tables

xi

Preface

xiii

Acknowledgements

xvii

List of Acronyms Chapter 1:

xix

Forests, Climate Change and the Socio-ecological System

1

Chapter 2:

Methodology, Ontology and Epistemology for SD

15

Chapter 3:

Sustainable Forest Management (SFM) for Local and Global Outcomes

35

Towards a Holistic Synthesis of a Social-Ecological System

97

Chapter 5:

Scale Mismatches and Implications for SD and SFM

149

Chapter 6:

Land Cover Change and Above Ground Biomass

173

Chapter 7:

Forest Governance and Institutions

211

Chapter 8:

The Future of Forests and Sustainable Development 253

Chapter 4:

References:

275 vii

viii

Index

333

List of Figures 1.1 1.2

Concept Map showing integration of theory . . . . . . . . . 9 Components of a holistic synthesis . . . . . . . . . . . . . . 13

2.1 2.2

Differing cosmology . . . . . . . . . . . . . . . . . . . . . 28 Unifying Humans with Nature through Darwin . . . . . . . 32

3.1 3.2 3.3 3.4 3.5 3.6 3.7

Participatory Rural Appraisal: Research Methodologies . . . Participatory Rural Appraisal: Three Pillars . . . . . . . . . The Forest Carbon cycle . . . . . . . . . . . . . . . . . . . Biomass accumulation in dry miombo . . . . . . . . . . . . Deep rooted trees and atmospheric circulation . . . . . . . . The Forest Transition Curve: Baikiaea plurijuga vs other areas Forest transition theory . . . . . . . . . . . . . . . . . . . .

4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 4.15 4.16

Zambia and neighbouring countries . . . . . . . . Map of Zambia showing study area . . . . . . . . . Mature closed canopy Zambezi Teak forest . . . . Kalahari Woodland . . . . . . . . . . . . . . . . . Recording data in Mopane Woodland . . . . . . . Felling a large Zambezi Teak tree . . . . . . . . . . Abandoned steam engine . . . . . . . . . . . . . . Tribal and Ethnic Map of Zambia . . . . . . . . . . Aerial view of differences in leaf phenology . . . . Climate change effects on phenology . . . . . . . . Mean Annual Temperature change for Zambia . . . The five capital assets in a livelihood framework . . Drums are used in many aspects of daily life . . . . Conceptual framework for livelihoods approach . . Participatory mapping in progress . . . . . . . . . Ranking the relative importance of income sources ix

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

53 54 57 59 61 70 71 101 102 103 107 108 115 116 117 122 123 125 130 131 132 133 134

x

L IST OF F IGURES

4.17 4.18 4.19 4.20

A traditional Bellows (or Mvubu) . . . Collecting Bwili tubers from forest edge Collecting and processing Munkoyo . . Maize and sorghum on a drying rack . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

137 138 140 142

5.1 5.2 5.3 5.4

Concepts of Scale . . . . . . . . . . . . . . . . . . . . . . A hierarchy of rules and institutions . . . . . . . . . . . . Reconciling different epistemological concepts using scale Concepts of Scale . . . . . . . . . . . . . . . . . . . . . .

. . . .

151 163 167 168

6.1 6.2 6.3 6.4 6.5 6.6

Landsat image selection process and analysis procedure . . . 182 Maps used for accuracy assessment for Bombwe Forest . . . 184 Bombwe Forest Map 1970 . . . . . . . . . . . . . . . . . . 185 Nyangombe Forest . . . . . . . . . . . . . . . . . . . . . . 187 Frost Hollows near Sesheke . . . . . . . . . . . . . . . . . . 188 Maps used to measure map accuracy for frost hollows near Sesheke . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 6.7 Flowchart of image analysis procedures . . . . . . . . . . . 190 6.8 Location of sample plots for measuring Above Ground Biomass (AGB) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 6.9 The climbing vine Fockea multiflora . . . . . . . . . . . . . 194 6.10 Errors of commission and errors of omission for 1975 map . 197 6.11 Box plot showing range of AGB measures . . . . . . . . . . 202 7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 7.9 7.10 7.11

Group discussion underway . . . . . . . . . . . . . . Proximate and Ultimate Causes of Deforestation . . . Causative patterns of deforestation . . . . . . . . . . Scales of Governance . . . . . . . . . . . . . . . . . Triangulation of results . . . . . . . . . . . . . . . . Fires in Zambia study area between 2000 and 2005 . Fires in Tsholotsho District between 2000 and 2005. Making string from bark . . . . . . . . . . . . . . . Maize storage bins . . . . . . . . . . . . . . . . . . Ladder of participation in forest management . . . . Reporting our findings to HRH Chief Moomba . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

212 214 215 218 221 240 241 244 246 248 250

8.1 8.2

Dense thickets have replaced Zambezi Teak forests . . . . . 259 Evidence of past logging of Zambezi Teak forests . . . . . . 261

List of Tables 2.1 2.2

Research Approach and Methods . . . . . . . . . . . . . . . 16 Mode 1 and Mode 2 knowledge production . . . . . . . . . 23

3.1

Measures of Above Ground Biomass (AGB) for different dry forests in Africa . . . . . . . . . . . . . . . . . . . . . . . . 56 Competitiveness, corruption and development for Zambia . . 65

3.2 4.1 4.2 4.3 4.4

Livelihood variable scoring criteria . . . . . . . . . . . . . Sources of income for Mabwe community . . . . . . . . . Expenses requiring cash payment for Mabwe community members . . . . . . . . . . . . . . . . . . . . . . . . . . . Livelihood variable scores . . . . . . . . . . . . . . . . .

. 141 . 143

6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9

Allometric equations used to calculate AGB . . . . . . . Changes in forest/woodland area between 1975 and 2005 Percentage change in area between 1975 and 2005. . . . Agreement and Error percentages for 1975 map . . . . . Measures of map accuracy for 1975 map . . . . . . . . . Confusion matrix for 1975 map . . . . . . . . . . . . . . Confusion matrix for 2005 map . . . . . . . . . . . . . . Statistical measures of map accuracy for 2005 map. . . . Differences in mean AGB between 1975 and 2005. . . . .

. . . . . . . . .

. . . . . . . . .

193 195 196 198 198 199 201 201 203

7.1 7.2 7.3 7.4

Grading criteria for governance themes . . . . . . . . Grading scale for governance scoring criteria . . . . Ranking institutional compliance with Ostrom (1990) Timeline of decline in natural resources . . . . . . .

. . . .

. . . .

224 226 239 243

xi

. . . .

. . . .

. 135 . 139

Preface

The history of timber logging in south western Zambia in the twentieth century has a lot to teach us about sustainable forest management in the twenty first century. In 1918 a tramline was laid to transport Zambezi teak logs from Mapanda forest, 27 miles south east back to the mill in the frontier town of Livingstone. The rails were made of Zambezi Teak in the absence of steel and replaced the slow and expensive alternative of ox–wagon and river barge for transporting logs to the sawmill. By 1927 the wooden rails had been replaced with standard 3 feet 6 inch Cape gauge steel rails and the first bridge had been built across the Sinde river. By 1934 the line of rail had reached Mulobezi, 101 miles north west of Livingstone. This marked the start of industrial exploitation of the highly productive Zambezi Teak forests. Estimates from Zambesi [sic] Saw Mills records indicate that extraction was maintained at roughly 85,000 m3 of round logs per year for 40 years. This exploitation set in place a chain of events that ultimately led to the destruction of these forests. A few small patches of virgin forest remain. This was not the intended outcome. The Livingstone Mail of 30th October 1924 carried an article on the proposed new steel railway line which indicated that anticipated traffic was around 200 tons of timber per day and that “due to cutting being confined to mature trees over 14ins diameter, which formed but a small percentage of the forest stand, this quantity should continue or increase for all time.” When a group of investors recently attempted to revive the logging of Zambezi Teak in Mulobezi they struggled to maintain a harvest of 1000 m3 per year and after four years the company collapsed. The lessons to be learned are not simple and in simplifying them we risk missing out on an opportunity to learn how ecology and society interact to produce failure from the potential for success. Trees are a renewable resource, and although Zambezi Teak only increases in diameter by between 1 and 3 millimetres per year, the vast forests could have produced a sustainable harvest in perpetuity. So why didn’t they? This book attempts to answer that xiii

xiv

P REFACE

question, but more importantly it attempts to address the complex methodological, philosophical and theoretical problems that surround the study of forests and natural resources and their interaction with society in a more general sense. To illustrate this complexity we could try to answer the question about the decline of the Zambezi Teak forests by asking experts in certain disciplines to explain why this renewable resource was harvested to destruction. The forest manager will tell us that the harvest exceeded the annual increment and he or she would draw an analogy between capital and interest. Only the interest on the capital should be harvested or we risk incrementally drawing on our capital which, if we keep harvesting a fixed amount, will result in the forest becoming depleted. The ecologist will explain that young Zambezi Teak trees germinate readily but are vulnerable to fire and survived because fire was not able to penetrate mature Zambezi Teak forest. When logging opened the mature forests a thick undergrowth was able to grow and this supported hot fires which killed off young trees. South western Zambia has one of the highest incidences of wildfire in the world and therefore no regeneration takes place after logging unless fire is excluded for a few years. The anthropologist will tell you that Lozi and Nkoya people or their forebears have lived in the area for thousands of years. They harvest honey from Zambezi teak forests, and the alcoholic beverage they brew from honey plays an important part in their cultural life. Fire used to smoke bees from their nests is left burning over a wide area so that it clears the snakes and the long grass, making it easier and safer for the next expedition to harvest honey. The political scientist will identify the tension between chiefs and traditional institutions on the one hand and the state authorities on the other. The claims to authority by both parties over the same resource constitute one of the fundamental problems of natural resource governance in Zambia. The social scientist will point out the result of the state claiming tenure of all natural resources and their inability to police the claim. This has created a de facto open access commons where people try to harvest as much as they can so they don’t lose out to others who are doing the same thing. All of these explanations are correct. This book is about the Zambezi teak forests of south western Zambia but as Kipling’s verse suggests, it is an attempt to answer a difficult question: how do we study such a complex system of interactions in order to produce a solution which ensures long term sustainable forest management? There is no simple answer. It is an exciting time to be working on problems of natural resource management because it appears that professionals from disparate fields are finally acknowledging the need to take a multidisciplinary approach to problems around natural resources. The new discipline of Sustainable

P REFACE

xv

Development has been founded to accommodate this vision. Nevertheless the difficulties of successfully achieving this integration are far from being overcome. This requires a profound change in the views not only of academics, scientists and managers, but also of politicians, traditional leaders and civil servants. There is a need to change fundamental methodological, philosophical and theoretical approaches to how we study natural resources. Just as important is the need to communicate this knowledge to those responsible for making and implementing laws and governance. I hope this book will contribute to renewed interest in the Zambezi Teak forests and contribute to the changes which are needed to manage what remains of this once important resource.

Acknowledgements

My work on Zambezi Teak has its genesis in my first visit to Mulobezi in 2004. The Zambezi Sawmills railway, the history and the ecology of the Zambezi Teak forests fascinated me. The people of Mulobezi were a revelation. The generosity of HRH Chief Moomba over the years, the way he has welcomed me to Mulobezi and allowed me complete freedom of movement within his Chiefdom have allowed me to form the ideas which make up the bulk of this book. The genuine friendships I have in Mulobezi are a treasured outcome of my visits. I owe him, and the people of Mulobezi Game Management Area a great debt of gratitude. The interaction I have had with HRH Senior Chief Inyambo Yeta and the Kuta at Mwandi has made a lasting impression on me. The serene and ordered conduct of affairs at the palace at Mwandi is a rare occurrence in a region so affected by turmoil over the last 100 years. As an example of leadership and local administration, I know of no better system than that which exists in Barotseland. I hope that it can be put to good use in managing the natural resources of the region in the future. During the course of this work a number of people have been of assistance, and without them, a lot of this work would not have been possible. Kevin Leo-Smith, Stuart Christie, Rob Morley and Carl Oellerman of Sustainable Forest Management Africa have provided valuable support in the form of data, general insight and criticisms. Carl’s patient responses to panicked emails about GIS problems, and Stuart’s references have been invaluable. In Zambia, the people of Mabwe, specifically Induna Simbeza, Induna Kapanza and Induna Moyandulwa generously welcomed us to Mabwe and assisted with our research. Area Councillor Davidson Sikasivi is always a wise counsel and a valuable barometer of opinion in Mulobezi. Jason Banfield and Zambezi Sawmills provided welcome relief from the bush in the form of hot showers and a comfortable camp. Pete and Annabel xvii

xviii

ACKNOWLEDGEMENTS

Hemingway provided a bed, a workshop and great company in Livingstone. In the field Mr Francis Mwanakobwa Liwakala was a master of tree identification and general knowledge of the Mulobezi area. His help in identification was important for expediting the field work and making accurate identifications of trees. George Coubrough has been a huge source of support. I thank him for repairs to my Landcruiser, unequalled insight into what really happens in the timber industry in Sesheke area, as well as constant infectious enthusiasm for the trees and the people of south western Zambia, and his friendship. His introductions in Sesheke formed an important part of my data collection. The nights around his campfire will always be fond memories of my field trips. Alan Sparrow, forester, conservationist and friend, has always provided valuable insight into CAMPFIRE, local forestry and more generally included me in news of developments in conservation in the KAZA region which are a valuable source of information. In Zimbabwe Neil Rix, Mark Butcher and Geoff Calvert gave me plenty of their time to discuss forestry and community conservation issues. Neil’s contacts in the forestry industry were a source of valuable data. Without Mark Butcher’s offer of a place to stay at Bomani Camp in Tsholotsho, and the help of Big Boy, I simply couldn’t have worked there. I’m not sure I will ever be able to thank Geoff enough for the old maps of Mulobezi he gave me, but they will be safe and form the basis for future work which is perhaps the best gift I could give him. The Soddy Foundation, The Royal Geographic Society, The Russel Trust and The Rufford Small Grants Trust all provided financial assistance for field work which, quite simply, made the project possible. I am grateful to Rehema White for the freedom to pursue this research, for her support, and for the support of everyone at the Department of Geography and Sustainable Development at the University of St Andrews. They have at all times provided solid encouragement, sympathetic criticism and principled integrity which previous experience had led me to believe were entirely absent from academia. “Now faith is the substance of things hoped for, the evidence of things not seen”(Hebrews 11:1). Lastly I would like to thank my family for the sacrifices they have made during the research and writing of this book. My wife Bryony has made a supreme effort, is an excellent proof reader and has given a lot of herself to stay the course under difficult circumstances. Max and Georgia haven’t had as much of my time as they deserve.

List of Acronyms

Agricultural Fields

AF

Above Ground Biomass

AGB

Analysis of Variance

ANOVA BAU

Business-As-Usual

BRE

Barotse Royal Establishment

BPF

Baikiaea plurijuga forest Communal Areas Management Programme for Indigenous Resources

CAMPFIRE

Community Resource Board

CRB

CBNRM CITES

Community-Based Natural Resource Management

Convention on Trade in Endangered Species of Wild Flora and Fauna

COP

Conference of Parties

CPR

Common-Pool Resource

D

Dambo

DBH

Diameter at Breast Height

DBPF

degraded Baikiaea plurijuga forest

ETM+

Enhanced Thematic Mapper

EKC

Environmental Kuznets Curve

ENSO

El Niño Southern Oscillation xix

xx

ACRONYMS

Fire Information for Resource Management System

FIRMS F

Fire Scars

FTT

Forest Transition Theory

FAO

Food and Agricultural Organisation

GDP

Gross Domestic Product

GIS

Geographic Information System

GLAS

Geoscience Laser Altimeter System

GMA

Game Management Area

GCM

General Circulation Model

ICESat

Ice, Cloud and land Elevation Satellite

Intergovernmental Panel on Climate Change

IPCC IAD

Institutional Analysis and Development

IFRI

International Forestry Resources and Institutions

JFM

Joint Forest Management Joint Forest Management Area

JFMA KIA

Kappa Index of Agreement Keyhole Markup Language

KML KW

Kalahari Woodland

LiDAR

Light Detection and Ranging

LIRDP

Luangwa Integrated Rural Development Programme

MSS

Multispectral Scanner

MODIS MAUP MRV MW

Moderate Resolution Imaging Spectroradiometer Modifiable Areal Unit Problem

Monitoring, Reporting and Verification Colophospermum mopane Woodland

ACRONYMS

NDVI

Normalized Difference Vegetation Index

NGO

Non-Governmental Organisation Non-Timber Forest Products

NTFP PRA

Participatory Rural Appraisal

PFT

Plant Functional Type

PTA

Parent Teachers Association

RDC

Rural District Council Reduced Emissions from Deforestation and forest Degradation

REDD

Reduced Emissions from Deforestation and forest Degradation plus

REDD+ SD

xxi

Sustainable Development

SOM

Self Organising Map

SFM

Sustainable Forest Management

SRTM TM

Shuttle Radar Topography Mission

Thematic Mapper United Nations Framework Convention on Climate Change

UNFCCC

United Nations-Reduced Emissions from Deforestation and forest Degradation

UN-REDD

United States Geological Survey

USGS VAG

Village Area Group

WBCSD

World Business Council for Sustainable Development

WCRP

World Climate Research Programme

WCED

World Commission on Environment and Development

ZAWA

Zambia Wildlife Authority

Chapter One Forests, Climate Change and the Socio-ecological System

As humanity reaches a critical stage in the transition from an industrial society to a post-industrial information age, our relationship with nature presents human society with stark choices. Sound environmental stewardship will allow biological and cultural evolution of life on earth to continue, whilst further environmental degradation will have severe consequences for human survival and may ultimately result in extinction (Laszlo, 1994). The possibilities for research to make a contribution to the transition to a sustainable society are complex and the questions that must be answered are challenging to conventional research practice. Theoretical approaches to simultaneously maintaining sustainable ecosystems and achieving sustainable and equitable global development have provided the impetus for a postmodern “scientific revolution” over the last twenty years (Naveh, 2000). Attempts to understand the interaction of economic, ecological and social systems are central to a new holistic science, which recognises that the Total Human Ecosystem (sensu Naveh, 2000), cannot be understood by reducing it to its parts (Holling, 2001; Li, 2000). Complexity science acknowledges the influence of feedbacks in structuring non-linear relationships between factors which implies a strong role for the local context in mediating these interactions (Cilliers, 1998; Green & Sadedin, 2005; Larsen-Freeman, 1997; Liu et al., 2007; Walsh et al., 2008). The application of this new paradigm in a holistic synthesis of the problems of climate change and its multiple potential effects on environment and society is at the heart of this study of the Zambezi Teak 1

2

C HAPTER ONE

forests of south western Zambia. The demand for a holistic synthesis, in the spirit of this postmodern scientific movement, exposes traditional disciplinary approaches as insufficient for the task of anticipating the future outcomes of human economic and social activity on the environment (Stock & Burton, 2011). My approach to studying the decline of the Zambezi Teak forests has been to attempt a holistic synthesis of the problems surrounding Sustainable Forest Management (SFM) and Sustainable Development (SD) in the dry, deciduous forests of south western Zambia. Forests provide a focus for addressing one of the most important consequences of more than a century of industrial activity: the warming of the global climate system. The human causes for this are unequivocal (IPCC, 2013) and will result in a loss of 5% of GDP per year to the global economy unless action is taken to reduce the emission of CO2 into the atmosphere (Stern, 2006). There are other potential costs that measures of GDP do not take into account. The emphasis on forests in the context of climate change is twofold: 1. To prevent the conversion of forested land to other land cover types to reduce the emissions from land cover change (Bonan, 2008) 2. Forests represent one of the most cost effective methods of sequestering carbon from the atmosphere (Stern, 2006).

In addition to the services that forests provide at a global scale, the dry deciduous forests of Africa supply local communities with a variety of essential Non Timber Forest Products (NTFPs) (Chikamai et al., 2009; Shackleton & Shackleton, 2004; Shackleton et al., 2011) and provide a range of ecosystem services (Marunda & Bouda, 2010). This global, regional and local demand for forests to sequester carbon, provide ecosystem services and supply the demands from local communities for forest products is a significant challenge for SFM. A collapse in rural livelihoods which are dependent on the harvesting of forest products during times of hardship, may result in social and political unrest as a result of a combination of forest destruction and climate change. The exact nature of what needs to be done to manage forests sustainably is complex, with multiple potential outcomes possible, and there is a need for solutions which are scale dependent, involve local communities in forest management and are effective at reducing emissions. This presents a significant challenge. If forests represent a cost effective way of ameliorating climate change as Stern (2006) suggests, they almost certainly do not represent a simple way of achieving this objective. The challenge lies in the fragmented nature of the theory which informs the process of deforestation and sustainable development, the diversity of

F ORESTS , C LIMATE C HANGE AND THE S OCIO - ECOLOGICAL S YSTEM

3

national and traditional governance arrangements for managing forests, the diversity of forest types, and the potential effects of climate change on forests around the world. A holistic synthesis of the many disciplines and theories which inform SFM is needed. The task is to connect functional social-ecological interrelationships which are spatially relevant and matched to an appropriate scale. I attempt such a synthesis in this study. To place the problem of SFM in such a context, this introduction outlines the background to the problem, the purpose of the study and its significance, and ends with the presentation of four research questions which address some of the problems associated with the SFM of dry deciduous forests in south western Zambia.

Forests and Sustainable Development In a landmark publication the World Commission on Forests and SD suggested that forest decline not only undermines social stability and local cultural diversity, but threatens global economic security in ways which transcend national boundaries (Salim & Ullstein, 1999). Forest decline has effects on the environment, the economy and society. These effects interact, frequently compounding cause and effect and leading to feedback loops that are difficult to predict and which imperil the well-being of future generations. The interactions, their consequences and their amelioration are recurring themes throughout this study and are at the heart of sustainability science (Clark & Dickson, 2003). Sustainable Forest Management (SFM) represents the integration that is needed between forestry practice and SD. The term SFM does not refer to a rigid set of aims and objectives and like SD, tends to have different interpretations which are dependent on one’s opinion on conservation, economic development and politics. Traditionally, forests have been managed to sustain a flow of timber. Gifford Pinchot, the first head of the Forest Service in the United States and an influential figure in forestry and conservation, was clear in advocating ‘wise use’ as the primary purpose of a national forest (Pinchot, 1907). In Pinchot’s view forests had a single purpose: they should be managed to supply timber in perpetuity in order to facilitate economic progress. Carbon, rather than timber, is potentially a new commodity for which forests are in demand, but this should not result in forest management policy falling into the trap of replacing timber with carbon in a Pinchot style, single objective management plan. The goals of SFM must be to integrate the principles of SD into the policies and practices of forest management so that in addition to supplying a commodity, forest resilience is sustained,

4

C HAPTER ONE

livelihood requirements of local people are satisfied, tenure and access rights of local people are clear and secure and there is equitable sharing of forest revenue (Campbell et al., 2007; Wunder, 2006). There is increasing evidence that if these requirements are not satisfied, and if plans to manage forests for carbon sequestration encourage centralisation of forest governance in pursuit of carbon revenue, then the outcomes of the global programme of Reduced Emissions from Deforestation and forest Degradation plus (REDD+) are not guaranteed (Phelps et al., 2010b; Poudyal et al., 2013). Forest management policy in Zambia is not capable of delivering community benefits through REDD+ (Leventon et al., 2014). A significant problem which is not addressed is the widespread unsustainable charcoal production which supplies the urban poor with a source of heat and cooking fuel and the rural poor with a cash income (Kutsch et al., 2011). The current policy and practice of the Zambia Forestry Department do not constitute an attempt at SFM (Jones, 2007; Leventon et al., 2014; Whiteman, 2013).

More fundamental problems with implementing REDD+ are of equal concern for successful SFM. There are limitations to the current techniques for measuring land cover change and biomass; the scale of research and theory application are not matched to the scale of the problem; the lack of appropriate governance structures which include local Zambian communities in forest governance is a threat to forest survival; and the medium to long term threat which climate change itself poses for forest survival is seldom assessed. Forests and SFM need to adopt the principles of SD to be successfully implemented, but also have the potential to inform theories of SD through the testing and implementation of these theories.

A continuing theme in this study of tropical forest change is that it demands an interdisciplinary approach to produce a holistic synthesis of the ecological, social and institutional factors which determine land cover change. There is clear recognition that interdisciplinary research is necessary for identifying solutions to resource use problems (Janssen & Goldsworthy, 1996). The concept of sustainability is integrative and the emerging field of sustainability research is partly defined by its interdisciplinarity (Clark, 2007). Interdisciplinary research involves a triangulation of depth, breadth and synthesis (Klein, 1996). In this book I offer breadth in a review of the social-ecological system, depth in the chapters which present findings from empirical research, and some elements of synthesis.

F ORESTS , C LIMATE C HANGE AND THE S OCIO - ECOLOGICAL S YSTEM

5

Climate Change Concerns about global climate change have reached unprecedented levels, and there is increased confidence that these changes are due to human activity (IPCC, 2013). There is evidence that forests can attenuate global warming through carbon sequestration and that in doing so tropical forests perform an environmental service worth hundreds of billions of dollars (Canadell & Raupach, 2008). Approximately 70%-80% of the forested area of Africa is covered by dry, deciduous forests (Murphy & Lugo, 1986). The Above Ground Biomass (AGB) of these forests is low relative to moist, tropical evergreen forests (Timberlake et al., 2010), but because the area they occupy is large, they represent huge potential as a carbon sink provided the correct management strategies are adopted (Marunda & Bouda, 2010). Indications are that the below ground biomass of these forests also comprises a significant carbon pool, the extent of which has not been adequately investigated. More than half of Africa’s population live in the area where these forests occur, and their livelihoods depend largely on the natural resources which the forests produce (Chidumayo & Gumbo, 2010). Poor forest management practices are widespread with the consequence that most deforestation in Africa is taking place in dry forests (Brink & Eva, 2009). Desanker & Justice (2001) highlighted the lack of case studies of land use change in Africa at the local level. Their focus on the potential which free Landsat data represents for developing detailed land use change models in Africa is a challenge partly taken up in this study. Documenting land use change is an important input to regional and global climate modelling and will contribute towards better predictive ability of these models (Willis et al., 2013).

Drivers of Land Cover Change The fastest land cover change to occur since the evolution of humans is happening in tropical forests (Achard & Blasco, 1990). Changes in forest cover in dry ecosystems of Africa between 1990 and 2000, showed that the Zambezian region accounted for 84% of total deforested area (Bodart et al., 2013). Figures for deforestation rates in Zambia vary widely. Between 298,000 ha per annum and 444,800 ha per annum were lost between 1996 and 2006 (Stringer et al., 2012) depending on the method used to measure deforestation. The Food and Agricultural Organisation (FAO) estimates the rate of forest loss in Zambia between 1990 and 2010 to be about 0.33% per annum which is a figure of roughly 167,000 ha per annum (FAO, 2010). A

6

C HAPTER ONE

large part of the discrepancy results from how different authors take into account forest degradation, which is a far bigger problem than deforestation (Kutsch et al., 2011). Regardless of the accuracy of the estimates, Zambia has one of the highest rates of deforestation in the world. Globally, clearance of land for agriculture is one of the major drivers of deforestation in tropical forests (Babigumira et al., 2014). Traditional agricultural practice in Zambia employs the chitemene system of shifting cultivation, which involves cutting, stacking and burning trees where a field is to be sited (Trapnell & Clothier, 1937). The ash contains nutrients and supports crops on the leached, nutrient poor soils for two to three years (Holden, 2001). This is the system currently in use in the study area. Prior to the introduction of colonial agricultural influences, the chitemene system constrained population growth because the overall productivity of the system was low (Holden, 2001). Hein et al. (2008) indicate that the chitemene system is sustainable in Miombo woodlands in Zambia with a population density of about four persons per square kilometre. Outside of the main villages of Sesheke, Sichili and Mulobezi, population density in the study area is estimated to be about four to five persons per square kilometre (Republic of Zambia Central Statistical Office, 2011). There is a history of commercial logging in the Zambezi Teak forests, which has had an impact on the forests by opening them and making them vulnerable to fire (Huckabay, 1986b; Martin, 1940). The long history of commercial timber exploitation in the Zambezi Teak forests provides an opportunity to document changes in land cover. The presence of rural subsistence farmers provides an opportunity to examine the human influences on land cover change. The close proximity of Kalahari Woodland, Zambezi Teak Forest and Mopane Woodland in a landscape which experiences broadly similar climatic conditions, provides an opportunity to compare the influence of different drivers of deforestation in different forest and woodland types. The adaptations of tree species to these climatic conditions, such as variation in leaf phenology, provide an opportunity to speculate on the potential influence of climate change on different forests and woodlands. The historical record of Zambezi Teak forest distribution, exploitation, management and tenure provides a unique opportunity to examine the relationship between forests and SD. In addition, the diversity of forest types in close proximity in the area, and the different drivers of deforestation to which they are vulnerable, represent a unique situation with respect to the narrative of deforestation in other areas of Zambia, and of dry, tropical African forests in general. The distinctive spatial and historical features of the area provide an opportunity to examine how different drivers of defor-

F ORESTS , C LIMATE C HANGE AND THE S OCIO - ECOLOGICAL S YSTEM

7

estation affect different forest types and this has important implications for the implementation of REDD+, SFM and SD. Many of the suggested causes of deforestation (and environmental degradation in general) confuse proximate with ultimate causes (Geist & Lambin, 2002; Paul, 1989) and solutions to the problem cannot be found if the causal basis of deforestation is not clearly established. South western Zambia provides a unique opportunity to examine different proximate causes of land use change such as commercial logging, expansion of rural agriculture, increased incidence of fire and commercial charcoal production. REDD+ and

the Social-Ecological System

The role which forests play in sequestering CO2 received international focus in 2009 when the United Nations-Reduced Emissions from Deforestation and forest Degradation (UN-REDD) programme was launched to ensure that REDD+ was incorporated in any future agreements under United Nations Framework Convention on Climate Change (UNFCCC) negotiations (Phelps et al., 2010a). The proposal to make payments to forested countries on the basis of the successful reduction of carbon loss from forests is currently under a trial phase in Zambia and eight other countries (www.un-redd.org) (Stringer et al., 2012). The scale of a global REDD+ programme is immense and the potential implications for forest communities are profound (Angelsen, 2008a). The implementation of REDD+ presupposes the existence of the technical and human capacity for Monitoring, Reporting and Verification (MRV) and the implementation of governance structures which encourage land use practices that foster the sequestration of carbon (Phelps et al., 2010a). The commodification of carbon and the demands this introduces with respect to forest management policy will need to be carefully managed to ensure that REDD+ is enacted using the principles of SFM and achieves the objectives of SD (Karsenty et al., 2014). A key component of climate negotiations, and the reason for the inclusion of REDD+ in a binding agreement, is to address issues of climate equity (Karsenty et al., 2014). Not only are the poorest at most risk with respect to climate change, but the mechanisms to ameliorate climate change via REDD+ will also have a direct effect on their lives (Morgan & Waskow, 2014). Forests are a source of food, medicine, building materials, and agricultural land and are of spiritual and cultural significance to millions of people in the tropics (Chikamai et al., 2009; Makonda & Gillah, 2007; Shackleton et al., 2011). This demands a case study of how local social institutions which regulate how people use forests will be affected by the implementation of the

8

C HAPTER ONE

largest global programme of payment for ecosystem services which has ever been proposed. Figure 1.1 indicates how the different theories included in the theoretical framework inform the research presented in the relevant chapters. Concept mapping is a general method for developing and structuring research which aims to display ideas and the interrelationships between them in a graphical format called a concept map (Novak, 1990; Trochim, 1989). The approach taken to the research in this study has been called a “landscape approach” and some of the recommendations of Sayer (2009) are applicable to this work (see Page 94). The term covers a wide variety of approaches in the literature and could equally be called an “ecosystem approach”, although issues of scale make it possible that a landscape could be envisaged as a subset of an ecosystem. Nevertheless one of the requirements for a landscape approach is the combination of positivist and constructivist approaches to research without excessive philosophical introspection regarding ontology (Sayer, 2009). The concept of a landscape as a multifunctional entity (Naveh, 2001) demands the selection of an interdisciplinary framework for conducting the research.

Scale, Governance and Change Scale

Scale mismatches are at the heart of many natural resource management problems (Cumming et al., 2006) and include mismatches of ecological, social and economic factors that determine successful natural resource utilisation. Inadequate considerations of scale are often the source of problems associated with analytical methodology, and both economic theory and political discourse commit the error of applying theory to inappropriate scales, leading to unnecessary disputes in the implementation of REDD+. The south central region of Africa is a diverse ecological and cultural landscape (Figure 4.8). History, socio-cultural practices and details of geomorphology and ecology are unique to a particular location and are all factors which influence the outcome of SD (Becker & Ostrom, 1995; Clark, 2007; Fiksel, 2006; Naveh, 2000; White, 2013). At the continental scale most of south western Zambia is mapped as undifferentiated woodlands and miombo woodland (Timberlake et al., 2010; White, 1983). This obscures the immense diversity in the ecological characteristics of these woodlands and forests, one of which is variation in leaf phenology which varies between sites and between species (pers. obs.). The history of the study area is unique in Zambia, and there are

Figure 1.1: Concept Map showing the integration of theory and how different theories inform the research in this study

F ORESTS , C LIMATE C HANGE AND THE S OCIO - ECOLOGICAL S YSTEM 9

10

C HAPTER ONE

socio-cultural practices which have unique relevance to sustainable natural resource management. In order to bring some clarity to issues of scale in south western Zambia I address the question of how scale and sampling methods affect the potential implementation of REDD+ in the deciduous forests of the area. In particular I examine how a case study approach can contribute to producing a holistic synthesis of the major factors which govern the functioning of the social-ecological system and highlight issues which are relevant for Sustainable Development (SD), SFM and REDD+. Governance

Ostrom’s (1990) eight design principles for Common-Pool Resource (CPR) institutions make a clear link between the management of these resources and their sustainable utilisation. The Zambezi Teak forests in Zambia and Zimbabwe are managed under different conditions of boundary definition, rules of appropriation, participation in decision making processes, monitoring, censure for rule breaking, mechanisms of conflict resolution, self-determination of the communities and organisation in multiple layers of nested enterprises. Zambezi Teak forests are one of the only CPRs I am aware of that occur in two different countries, and are utilised under different institutional arrangements. Zambian forests are owned by the state with little or no community participation in forest governance or revenue sharing, while Zimbabwean forests are owned by the community. Using comparisons of forest age structure between the two areas, I propose a causal link between adherence to Elinor Ostrom’s rules of CPR use and some criteria of good governance, and forest condition. Institutional factors determine the rules of use which mediate the relationship between people and a common-pool natural resource. The structure of institutions and governance arrangements is the target of the action that SD demands and an understanding of how they function is a key part of suggesting changes which are in line with the principles of SD. In Zambia the competing claims by traditional authorities and the state for control over who is able to access and collect revenue from natural resources constitute one of the most contentious issues affecting natural resource utilisation in the country. I briefly examine how traditional institutional structures may be adapted to comply with modern standards of transparency and accountability. Change

Much of the forested area of Zambia has only been studied from the early 1990s, and it is often concluded that most of the deforestation dates from

F ORESTS , C LIMATE C HANGE AND THE S OCIO - ECOLOGICAL S YSTEM

11

around this period. REDD+ projects require the setting of a baseline from which carbon sequestration can be calculated as additive with respect to historical levels of AGB. Given the economic and biological importance of the Zambezi Teak forests, the setting of a baseline for future carbon sequestration will need to take the historical impact of logging into account in south western Zambia and this differs from the impacts in other areas of the country. Global maps of AGB (e.g. Baccini et al., 2008; Saatchi et al., 2011) do not correspond to observed differences in AGB in different forest/woodland types in the study area (pers. obs.). An investigation of the accuracy of these maps (which claim authority and relevance for REDD+ at the project scale) is needed to assess the accuracy of the techniques for measuring AGB. Mapping of land cover change is the first step towards understanding the processes which lead to change. I examine the land cover changes which have occurred in the forests of south western Zambia over the last 30 years and show how AGB has been affected by these changes. Sustainable development differs from many theoretical frameworks by having action at the heart of its agenda (Van Kerkhoff & Lebel, 2006). This action involves the reform of practices, regulations, policies and institutions but does not necessarily emerge from the research which is directed at subject related questions. Action requires knowledge from diverse disciplines such as natural science, economics, politics and social science and the process of integrating these different strands and formulating a plan of action which fulfils the aims of sustainable development, is complex and frequently beyond the remit of individual researchers. However, given the theoretical framework for this case study of the problems of forest management in southwestern Zambia, linking knowledge from research to action emerges as an essential requirement. The research has the potential to inform theories of SFM and SD by using the findings to challenge the status quo regarding existing policy and institutional structures. The consideration of environmental, ecological, social and institutional aspects of the problem for assessing how this study of Zambian forests contribute to the theory and practice of SFM and SD and specifically to the action-based agenda that is central to SD in the context of local and global demands.

A Holistic Synthesis Developing a holistic synthesis approach to the problems surrounding the sustainable management of dry deciduous forests in southwestern Zambia. This is attempted by connecting functional social-ecological interrelationships which are spatially relevant and matched to an appropriate scale. I

12

C HAPTER ONE

used multiple methods and drew on a wide range of disciplines and different theories in order to attempt a grand scheme of triangulation from the findings which emerged. To begin that process the preceding sections have briefly reviewed the dominant issues which inform the process of synthesis. A more detailed review follows in Chapter 2. Although the idea of holism has ancient roots (von Bertalanffy, 1968), it was the South African statesman, soldier and philosopher Jan Smuts who first defined the term as “the tendency in nature to form wholes that are greater than the sum of the parts through creative evolution” (Smuts, 1926, pg. 88). von Bertalanffy (1968) tracks the development of systems thinking and the study of chaos and complexity from these early ideas of holism. In a holistic synthesis there is an emphasis on the importance of the whole and the interdependence of its parts, but more importantly this emphasis has ontological significance because the systems which are studied are held to be computationally irreducible (Wolfram, 2002). No matter how much data is available it is not possible to predict a system state as a result of simulating the interactions. In Figure 1.2, I show how key concepts are used in this study to inform a holistic synthesis. The keywords around the boxes represent themes which occur frequently throughout this study and which are used to inform conceptual frameworks. The holistic synthesis I attempt in this study is not definitive, nor does a definitive methodology exist for attempting a holistic synthesis. To some extent the approach used here is an attempt to explore the methodology. It would be impossible to examine all the factors which contribute to a holistic synthesis sensu stricto. Rather, in a pragmatic approach I select some components of the system which cover ecological, social, institutional and governance issues. Previous experiences of the study area, time and financial constraints and limitations on my own capacity to learn new data collection and analytical skills have combined to determine the factors which I use in this interdisciplinary holistic synthesis.

Conclusion In this chapter I have provided an introduction to some of the theory around forests and SD; forests, REDD+ and climate change; the drivers of deforestation; and the emerging complex systems paradigm which drives a holistic, interdisciplinary approach to the research. These are the main theoretical strands which inform the approach to the problem, the methodology and the data collection. The complexity of reconciling conservation and development needs constitutes what have become known as “wicked” problems, with no

F ORESTS , C LIMATE C HANGE AND THE S OCIO - ECOLOGICAL S YSTEM

13

Figure 1.2: Components of a holistic synthesis. The main conceptual frameworks which form the holistic synthesis are supported by quotes from the literature giving examples of how the framework component contributes to the whole. Keywords on the outside of the quote boxes represent themes which occur throughout this study and which inform the conceptual framework used in the holistic synthesis

14

C HAPTER ONE

clear or final solution (Conklin, 2006). At the beginning of this chapter I indicated that the significance of the research is linked to the role of REDD+ as a cost effective method for reducing CO2 emissions (Stern, 2006). This may be the case, but it does not appear to be a simple way of reducing emissions because of the way forests form a focal point for all of the complexities which have come to define “wicked” problems. It is only by attempting a holistic synthesis that the problems can be integrated so that a new vision for the role of tropical forests in ameliorating climate change emerges to complement the economic recommendations for REDD+ as a solution to reducing greenhouse gas emissions. Methodological issues are central to sustainability research (Franklin & Blyton, 2011; White, 2013). In the following chapter I discuss the methodological issues which arise from tensions between researchers from the global North and the researched in the global South (Bunzi, 2008), issues with conducting research in authoritarian states and the problems with reconciling different epistemological traditions.

Chapter Two Methodology, Ontology and Epistemology for SD

One of the defining characteristics of sustainability science is the focus on dynamic interactions between nature and society (Clark & Dickson, 2003). To the extent that different methodologies are dominant in different disciplines, sustainability research can be said to be defined by an interdisciplinary methodology (White, 2013). Inductive and deductive approaches are used where appropriate to produce problem driven research directed at achieving sustainable goals. This study uses a theoretical lens which emerges from the aim of producing a holistic synthesis. This synthesis serves as a general, overarching framework for the investigation. The methodology which is required to address different aspects of the problem emerges from this framework. The methodologies used in this study include quantitative methods used to extract patterns from data such as reflectance values of radiation from vegetation or measurements of tree trunk diameter. Qualitative methods are used to collect data on institutions which determine how human influences on forests impact the structure and ecological characteristics of these forests. A summary of methodological approaches used in this study are shown in Table 2.1.

15

Methods Remote Sensing, Geographic Information Systems, Participatory Rural Appraisal (PRA), Participant Observation, Livelihood Analysis Remote Sensing, Geographic Information Systems, Spatial Analysis, Forest enumeration field measurements Key Informant Interviews, Semi-structured Interviews, Remote Sensing, Forest enumeration field measurements, Geographic Information Systems Critical Analysis, Data from previous chapters

Critical Analysis

Research Approach

1. Case study approach to produce a holistic synthesis of the major factors which govern the functioning of the socialecological system.

2. Measure land cover changes in the forests of south western Zambia over the last 30 years.

3. Examine how differences in institutional structure and governance between Zambia and Zimbabwe affect SFM.

4. Examine how scale and sampling methods affect the implementation of REDD+ in the forests of south western Zambia.

5. How can the research-based knowledge from this study contribute to the theory and practice of SFM and SD and specifically to the action-based agenda that is central to SD.?

Table 2.1: Research Approach and Methods used to collect data or conduct analysis

16 C HAPTER TWO

M ETHODOLOGY, O NTOLOGY AND E PISTEMOLOGY FOR SD

17

In using different methods, I make no claim for their ontological primacy, nor do I seek to establish this from the results which emerge. If anything emerges from the results with respect to methodology, it is that I use a grand scheme of triangulation, using a variety of data sources, methods and perspectives to produce a holistic overview of the problem of land cover change in dry deciduous forests of south central Africa. There are potential clashes of ontology and epistemology between proponents of different methods, but in my view these are disarmed by their application in the context of studying land cover change and the interactions between ecology and society. The arguments become more fractious at the edges when contrasting methodological approaches between disciplines. In land cover studies the advocates of the primacy of a positivist epistemology cannot avoid the fact that, “Landscapes” are the symbolic environments created by human acts of conferring meaning to nature and the environment, of giving the environment definition and form from a particular angle of vision and through a special filter of values and beliefs. Every landscape is a symbolic environment. These landscapes reflect our self-definitions that are grounded in culture. (Greider & Garkovich, 1994, 1). Conversely, the most determined advocates of constructivist ontology may need to consider that the social construction of nature can only take place when rivers are providing fresh water, trees are providing fuel, food and shelter and rain is falling on the portions of the landscape where food crops have been planted (Demeritt, 2002).

Fieldwork and Methodology Fieldwork in the global South by researchers from the global North is almost inevitably affected by ethical issues related to positionality, reflexivity and power relations (Sultana, 2007). In my own case, even though I am from the global South, and well known to people in the community where the research was conducted, the process of conducting research is still charged with political contextual issues, an imbalance in power relations and is complicated by layers of history and historiography which conflict and reverberate powerfully in south western Zambia. Despite these problems, the possibility of withdrawing from research in the global South because of fears of misrepresentation and inauthenticity, and concentrating only on textual analysis (Nagar, 2002) is not an option because of the demands that

18

C HAPTER TWO

places on the governments of forested countries and the communities that live in forested areas. The imperatives of international climate policy drive a need for research which will ensure that issues of climate equity and climate justice are met at diverse scales. Without research it is likely that a larger injustice will result than that which concerns and affects the individual researcher, not least because of the potential effects of climate change, which are global (Nagar, 2002). The link between local actions and the global concern around climate change drives the justification for research, albeit with emphasis on local context and reflexivity about local conditions (Sultana, 2007). Crucially, there is not only a static spatial element to these issues, but also a political-temporal positionality which changes how issues are researched and perceived by both parties. I suggest that as the UN-REDD process develops, and as climate change starts to become a factor in people’s lives, this timeline will introduce a change in the ethical considerations under which research in the global South is conducted by researchers or institutions in the global North. Unfortunately, injustice and misrepresentation by local institutions and governments in the global South with respect to their own people (e.g. Brockington, 2007; Brockington et al., 2004) are as much in need of ethical reflection as histories of colonialism, development, globalisation and other external forms of dominion and domination (Williams & Mawdsley, 2006). REDD+

Positionality

A person’s knowledge can only exist by virtue of a vast range of past experiences which have been lived through, often with the most intense feelings. These experiences, including textual experiences (books, lectures, lessons, conversation, etc), we have been taught to disguise so that our utterances are made to seem as though they emerge from no particular place or time or person but from the fount of knowledge itself. (Rosen, 1998, 30) In fulfilment of the challenge posed by the quotation from Rosen (1998), I acknowledge the influence that my lifetime of lived experience in southern Africa has on my choice of career, research interests, the methodology I use in this study, and ultimately what emerges from my attempts at a holistic synthesis in this study (England, 1994). My mother’s family left England for Africa in 1820, my father’s in 1949, and I was born in South Africa. Early childhood experiences in South Africa, Botswana, Zimbabwe, Zambia and Mozambique generated an abiding interest in African natural history,

M ETHODOLOGY, O NTOLOGY AND E PISTEMOLOGY FOR SD

19

culture, language and tradition. The early African explorers, Livingstone, Speke and Selous (Hugon, 1993) were early influences, and later the work of scientists such as Austin Roberts (Roberts, 1957), Reay Smithers (Skinner & Smithers, 1990), Vivian Fitzsimons (FitzSimons, 1962), Don Broadley (Branch, 1997) and Keith Coates Palgrave (Coates Palgrave, 1977) shaped my interest in natural history. In my pursuit of these interests, either birding, snake catching or attempting tree identification from botanical keys, I frequently encountered the rock paintings of the San people, some of which had never been documented and which are abundant throughout southern Africa (Lewis-Williams, 1982). This led me to an interest in African prehistory and anthropological enquiry. I remember seeing a makishi ceremony (Jordán, 2007) in Victoria Falls in 1979 and I still have a collection of makishi masks which I have assembled over the years. Discoveries in physical anthropology in Africa through the work of Raymond Dart (Dart, 1925) and Louis and Richard Leakey (Brown et al., 1985; Leakey et al., 1964) were always of interest. I took a first year undergraduate class in anthropology as part of my Honours degree in Zoology and Entomology (only one arts subject was permitted as part of a science degree) which introduced me to the arguably more complex field of social anthropology through the work of Max Gluckman (Brown, 1979), Elizabeth Colson (Colson, 1967), Thayer Scudder (Scudder, 1962) and Eileen Krige (Krige, 1943). In 1994 I read for a Masters in African Ecology. It was the increasing ability of computers to analyse large data sets, the emergence of Geographical Information Systems and the potential they represented for spatial analysis of patterns, that attracted me to the emerging fields of conservation biology and landscape ecology. However, the publication by Peters (1991) of his detailed Critique for Ecology had already planted the seeds of doubt about the ability of ecological knowledge to contribute to conservation. The elimination of the black rhinoceros in the Luangwa Valley of Zambia between 1979 and 1985 (Leader-Williams et al., 1990) and later in the Zambezi Valley in Zimbabwe from 1985 until 1993 (Tatham, 1988) were events I was aware of in the case of Zambia, and directly affected by in the case of Zimbabwe when a friend working for the Department of National Parks and Wildlife Management was shot and wounded by poachers. The exasperation I felt at the failure of a dedicated team of people to prevent the extinction of the black rhinoceros from these areas was compounded when, 10 years later, I witnessed entire forested landscapes in eastern Zambia transformed to scrub and cropland over a period of 5 years by the cutting of trees for the charcoal trade (Petit et al., 2001). The prevailing academic approach to conservation and the restrictive campaign mentality of conservationists had

20

C HAPTER TWO

both failed to prevent widespread environmental impact. I formed the opinion that conservation would be better served by increasing understanding about economic, institutional and political factors than by building more complex ecological models or starting more successful fundraising campaigns (e.g. Dyer et al., 2014). Recently ecology has evolved to incorporate the resilience perspective as an approach for understanding the dynamics of social-ecological systems (Folke, 2006). Although this is an encouraging development, the problem with scientists restricting advances in ecology to the domain of positivist enquiry emerges as one of the themes in this study. There is an African epistemology of nature which includes traditional knowledge, spirituality and cultural practice which is essential for inclusion in a holistic synthesis of forest conservation (Mavhunga, 2014). Ecologists and conservationists pay lip service to the relevance of these alternative epistemologies (Mavhunga, 2014; Shinn, 2014), but seldom allow for the dominant role that positivist, and indeed, postcolonial influences have in contemporary knowledge production (Jackson, 2014; Shizha, 2007). The choice of methodology for gaining an understanding of the social and institutional structure around management of the Zambezi Teak forests was heavily influenced by the layered political and social discord in the two countries (e.g. Derman & Kaarhus, 2013). In Zambia, the source of this discord is the contentious relationship between communities and the government over the use of natural resources, although a background of historical political issues is important, as discussed later in Chapter 3. Discord in Zimbabwe centres on the ongoing political disputes, and because land and resources are central to this discord, collecting information about institutions and the role they play in Community-Based Natural Resource Management (CBNRM) is viewed with suspicion. Gentile (2013) discusses some of the restrictions of field research in authoritarian states. His most pertinent observation is that ‘the greatest risk is being unaware of the risks . . . ’ (Gentile, 2013, pg. 432). Field data collection was conducted with a clear understanding (gained from extensive past lived experience in both countries) of the risks present in both Zambia and Zimbabwe. Under these circumstances it is impossible to conduct a data collection exercise through the use of questionnaires or other survey techniques. Besides the fact that many people are not literate, the act of writing and recording information is almost universally viewed with suspicion in both countries, although for different reasons. Given that any direct questioning on the part of a researcher who could not establish his legitimacy would also be received with apprehension, a careful approach was required

M ETHODOLOGY, O NTOLOGY AND E PISTEMOLOGY FOR SD

21

to collect qualitative data. In many respects the extent to which a particular methodology was able to be used was in itself informative about the relationship between people and institutions involved with CBNRM or forest management. In Zambia, where state ownership of resources dominates, people at every level of involvement had to be carefully approached and the legitimacy and neutrality of the researcher established before questioning could begin, despite the existence of a relatively open and democratic society. In Zimbabwe, despite the presence of an oppressive political regime, people were more receptive to having discussions about the problems around CBNRM, although anyone asking questions about what amounts to the functioning of local government invites suspicion. In Zimbabwe, community ownership of natural resources has been in place for nearly 20 years. Nevertheless it was not possible to approach ordinary villagers who live near the forests. Interviews had to be conducted in Bulawayo, mostly with key informants with whom legitimacy could be established and who understood that the research process did not necessarily represent a threat, or imply affiliation with opposition political parties or Western Non-Governmental Organisations (NGOs). Similar problems have been encountered by other researchers in Zimbabwe and Mandiyanike (2009) gives a detailed account of these issues. During his fieldwork in rural Zimbabwe, Mandiyanike (2009) did not use electronic recording equipment to avoid attracting attention and arousing suspicion. Mandiyanike (2009) had previously spent 12 years working at a Rural District Council (RDC) in Zimbabwe, but this relationship was still not sufficient to dispel suspicion about his research into the performance of rural local authorities in the polarised political environment with its general climate of fear. An important concern with almost every participant in this study was to avoid suspicion that the information disclosed in the interview would result in some form of retribution, either from an organisation, or from his or her community. I had to present myself to the interviewee in ways which provided assurance. It was helpful to have eight years of experience in the field, as well a comprehensive network of colleagues, friends, and acquaintances, when attempting to collect data from a person who was previously unknown to me. Any pretence at collecting data as a disinterested observer in order to discover what people thought about the relationship between natural resources and institutions was not going to be possible. The methodological approach was adopted on the basis of the strong ethnographic context which my background brought to the project. The ability to make people feel comfortable discussing contentious issues came from my understanding of their position in their community, and with the situation in the country at large. I have lived and worked in both countries and

22

C HAPTER TWO

am familiar with local and national issues. The small size of the community which made it possible to make the appropriate connections between myself and well known hierarchies both appropriate and benevolent to the individual concerned, and was a key factor in facilitating the research. Experience and Reflexivity

Reflexivity in research involves reflecting on the relationship between researcher and researched and the disparities of power with respect to researcher accountability in data collection, interpretation and dissemination (Greenbank, 2003). Reflexivity can be applied to multiple scales. These range from individual relationships between the researchers and the participants in a study and large scale issues such as the ethics of remotely sensed data collection and the restrictive access which the traditional model of scientific publishing imposes on those whom the research is intended to benefit. On page 18 I reflect on the methodological limitations which the political situations in Zambia and Zimbabwe impose on data collection. However, there is a wider relationship between reflexivity and the ontological and epistemological assumptions surrounding the methods used in this study. In Chapters 5 and 6 I use a positivist approach, which makes the assumption of value neutrality when making assessments of land cover change, issues surrounding MRV and describing forest phenology. In Chapter 7 a constructivist epistemology is almost unavoidable when using qualitative methods to assess the nature of institutions and the extent to which they meet the minimum requirements for successful CPR utilisation. The institutions which mediate the relationship between people and natural resources are social constructions, compounded in their genesis by the layered values of the people who erect them (Murphree, 2009). Is it therefore valid to conclude that reflexivity is superfluous when using an objectivist approach to the acquisition of knowledge? I suggest that reflexivity is equally important in both methodologies and that the relationship between REDD+ and forest communities provides a good example of why this is the case. Several studies have shown that community monitoring of AGB by local communities can fulfil the highest standards required by the Intergovernmental Panel on Climate Change (IPCC) (Danielsen et al., 2013, 2011; Skutsch & Ba, 2010). As I will discuss later in this book, efforts continue to develop a methodology using remotely sensed data, mostly at a scale which is not appropriate for decision making and REDD+ implementation on the ground. The data is collected at huge expense, analysed by institutions which are overwhelmingly based in the global North and the results are published in ways which are largely inaccessible to local communities

M ETHODOLOGY, O NTOLOGY AND E PISTEMOLOGY FOR SD

23

Table 2.2: Attributes of Mode 1 and Mode 2 knowledge production. From (Hessels & Van Lente, 2008) Mode 1

Mode 2

Academic Context Disciplinary Homogeneity Autonomy Quality control via Peer Review

Context of Application Transdisciplinary Heterogeneity Reflexivity Novel Quality Control

(Britz et al., 2006; Ondari-Okemwa, 2004). The degree to which ecological studies have avoided taking the steps to integrate the process of evaluating anthropogenic ecological change in an interdisciplinary manner and cling to value neutrality, emerges as a theme in this study but has also been the subject of criticism by Norton & Noonan (2007). It is by reflecting on the implications of the biophysical research methodology and data that we gain insights into how these data impact on the actions needed to implement SFM or indeed, Sustainable Development. Reflexivity is a property of Mode 2 knowledge production (Gibbons et al., 1994) in which social accountability for the effects of research on wider society are explicitly taken into account. Gibbons et al. (1994) claim that Mode 2 research yields socially robust knowledge which has a different epistemological status to knowledge produced by Mode 1 science. Table 2.2 shows the main attributes of Mode 1 and Mode 2 knowledge production. Reflecting on knowledge as power is an important component of placing biophysical research in its context as received and experienced by local people (Ojha et al., 2010). Participatory approaches to SFM are made easier when power relations are aligned. The theory of how this is achieved in practice is limited, but Ojha et al. (2010) suggest a deliberative approach which involves self-reflexivity where participants reflect on their internalised cosmologies and unquestioned assumptions which in turn leads to a cognitive crisis, where these views are seen to be in mismatch, either with conditions in society or the biophysical data which are thought to describe an objective reality. Post-modern theorists look for power relations that might distort any claims for a wider relevance, particularly the claims made by researchers in the biophysical sciences. The diversity of perspectives is enhanced by questioning the universality of these claims and insisting on a more inclusive approach to knowledge (e.g. Schelhas, 2003). Reflexivity therefore, which

24

C HAPTER TWO

is traditionally restricted to researchers using qualitative methodology, has implications for all aspects of research (Sultana, 2007). The combination of research techniques required for a holistic assessment of the problems of land cover change demands a degree of integration of approach and reflexivity provides a platform for this integration.

Ontology, Epistemology and Cosmology The sustainability literature acknowledges the contested nature of SD in a more general sense by identifying philosophical issues that arise from this contestation. Questions about what is admissible as knowledge, epistemological differences between different academic disciplines, cosmologies that place human economic social and political activity separate and independent to that of natural phenomena, are all potential sources of problems when it comes to the implementation of SD projects. Ontological Dichotomies

There is a strong line of argument in the sustainability literature which highlights the limitations of the reductionist scientific paradigm in finding solutions to environmental problems which are emergent from complex interactions between systems (Mebratu, 1998). Critiques of the positivist epistemology which dominates scientific practice, both in terms of the funding it attracts and the authority it claims, are becoming more confident as the expertise associated with this research tradition fails to solve the complex problems of building a sustainable society (Brand & Karvonen, 2007). Indeed, it is the crisis that SD seeks to address that creates a sense of urgency and consequent rejection of the scientific method because it has not produced the predictive heuristic knowledge required to address environmental problems, including problems of SFM (Ojha et al., 2010). In many cases the rejection of the scientific method and of positivism may be overstating the case. Conservation practice in Zambia and other parts of Africa has not escaped the dominance of positivist technical experts in making decisions (Adams, 2003; Adams & Hutton, 2007; Anderson & Grove, 1989). Often these ecologists merely commit the error inherent in their field, which is to ignore human participation in environmental processes and to advocate an Edenic conception of nature and what needs to be preserved (Adams & McShane, 1992; Denevan, 1992). But this does not mean that ecological understanding obtained through positivist analysis is wrong. There is a need for holistic thinking which integrates the findings of reductionist research by

M ETHODOLOGY, O NTOLOGY AND E PISTEMOLOGY FOR SD

25

identifying linkages within and between the ecosystem and the human sociocultural system and understanding the feedbacks that result (Li, 2000). The value of local ecological knowledge needs to be given equal epistemological status, with the intention of not discrediting scientific understanding, but to expose the hegemony of the latter tradition and construct a more even playing field in the politics of environmental knowledge (German, 2010; Jackson, 2014; McGregor, 2005). There is no reason why a mixed methods approach cannot form the basis of this holism, or for not forging ahead with methods that work in the context of the research questions (Howe, 1988). As Johnson & Onwuegbuzie (2004) point out, the purists in both camps, (although this probably applies to positivists to a greater extent) often make the simple error of equating methodology with the logic of justification which underpins ontological and epistemological claims for a specific methodology. This logical justification may be less narrowly defined by postmodern anti-positivists, but even a cursory reading of the social science literature will reveal a wealth of discussion regarding the admissibility of certain techniques and contain detailed discussion of their contribution to quality research. Questions regarding the admissibility of observations and measurements in the positivist research literature, and the extent to which the observer exerts an influence on the outcome of these measurements are, by comparison, given little space in considerations of the quality of the research findings. There is a need for using both scientific positivist and postmodernist approaches where appropriate. Johnson & Onwuegbuzie (2004) advocate an adherence to the philosophy of pragmatism (of whom early advocates were Charles Pierce, William James and John Dewey (Menand, 2002, pp. 350-358) to provide the ontological logic of justification for different research methodologies rather than stick to dogmatic claims of ontological exclusivity (Howe, 1988). Wilson (1999) advocates a far more ambitious vision. His perspective, that all phenomena are ultimately reducible to the laws of physics and consequently transcend cultural differences, is at the heart of how many scientists are trained to think. His arguments for ‘consilience’ between the natural and the social sciences are eloquent and powerfully expressed. His ideas that human behaviour (and therefore culture) may ultimately be understood by mechanistic explanations of brain function and biochemistry are typical of an ardent proponent of scientific positivist reductionism. His view that complex systems will ultimately be understandable and predictable through increasingly sophisticated mathematical models and computer processing power is at odds with the suggestion that these systems are computationally irreducible (Wolfram, 2002). Even if Wilson (1999) is correct, the advances

26

C HAPTER TWO

in understanding human behaviour and social-ecological systems in such a detailed and mechanistic way may not arrive before climate change results in catastrophic effects on human society.

Epistemological Discord

The extent to which it is possible (or even desirable) to reconcile different epistemological standpoints presents SD with a significant challenge. The positivist epistemology is dominant in conservation practice, often to the detriment of the planning process which, for example, often fails to take into account indigenous knowledge (Lane & McDonald, 2005) because it is largely derived through inductive reasoning. In formal logic theory it can be shown that it is possible to arrive at a false conclusion through inductive reasoning even when the premises are true, but much indigenous knowledge has been tested in the fire of hard experience and would therefore have not survived this process had it not corresponded to a reality of some kind, be it independent or constructed. There is a larger area of mismatch when attempting to reconcile an empirical epistemology with a postmodernist position. However, reconciliation is not necessarily the solution. Some humility on the part of the empiricists may be enough, coupled with a clear recognition that a constructivist epistemology is more inclusive, and that a technocratic approach to SD often excludes people, especially in rural Africa (German, 2010; Larson et al., 2005). Many conservation scientists trained in the positivist tradition may now recognise the requirement for inclusivity, especially in the context of CBNRM, but often fail to grasp the process and significance of inclusivity with respect to social idea generation. The founder of modern jurisprudence, the United States Supreme Court Justice Oliver Wendell Holmes Jr, provided a powerful justification for protecting free speech in a democratic society which underpins this significance. Unimpressed with claims to intrinsic human rights to justify freedoms, he reasoned that it is in the interests of society to allow free expression, because “we need the resources of the whole group to get the ideas we need” (Menand, 2002, pg. 350). Any approach to SD which insists on a single epistemological position runs the risk of circumventing one of the central principles of democratic idea making. It is this idea that underpins the process of social experimentation that has become an important approach in dealing with environmental issues and managing natural resources (Tàbara & Pahl-Wostl, 2007). Giving the process of policy making epistemological significance allows new policy making to become part of a process which feeds back into policy formulation (Mukamuri & Manjengwa, 2009).

M ETHODOLOGY, O NTOLOGY AND E PISTEMOLOGY FOR SD

27

Brand & Karvonen (2007) advocate four different types of experts who would enhance knowledge acquisition and generation in a SD project: • the outreach expert who can communicate effectively to non-experts, • the interdisciplinary expert who understands the thinking and approach of other disciplines, • the meta-expert who acts as a mediator between differing claims of legitimacy for different groups, and • the civic expert who engages in inclusive and participatory discourse between expert and non-expert alike.

These roles are all potential solutions to the apparent impossibility of reconciling different ontological approaches to knowledge. Of course, the principle of democratic idea making and social learning must be accepted a priori in the epistemological sense, before any expert can play a constructive role in the implementation of SD projects. The approaches suggested by Brand & Karvonen (2007) nevertheless provide roles for positivist and other experts who are used to being in positions of authority, and in all likelihood will continue to be in these positions for some time to come, to continue to assume authority and yet play a more inclusive role. The approaches provide a more expansive outlet for a personal expression of their acceptance of alternative epistemologies, without having to be seen to abandon their own. In this study I use multiple methods in acknowledgement of the restrictions faced by clinging to a single methodology (Norton & Noonan, 2007). The insights I have gained have important implications for how Monitoring, Reporting and Verification (MRV) in REDD+ projects is undertaken, but does not resolve the discord between different epistemologies because the publishing, reporting and evaluation methods inherit the same prejudices from their practitioners (Poteete et al., 2010). Mismatch of Cosmology

At the heart of many of the problems with SD implementation is the idea that the biological system, the social system and the economic system are separate entities which may occasionally intersect, and that it is only at this intersection that we need to address sustainability (Mebratu, 1998). This cosmic mismatch underpins the reluctance of those ecologists and environmentalists who deny an ecological role for humans in African ecosystems where humans evolved, and have lived ever since (Gibbons, 1997). Even if ecosystems outside Africa received their human populations through immigration rather

28

C HAPTER TWO

than evolution, the role of humans and their social and economic systems that are the inevitable consequences of being human, cannot ever be (and indeed, have never been) separate from a natural universe (Mebratu, 1998). The model of cosmic interdependence in Figure 2.1 needs to be more widely accepted if we are to avoid conflict in the management of ecosystems for SD.

Figure 2.1: Cosmic perceptions that affect the conceptualisation of the relationship between human institutions and natural systems. The model of interdependence between human economic, social and non-human biological systems and the abiotic environment is more conducive to addressing problems in SD than a model which allows for only some overlap between these components. From Mebratu (1998). To a large degree, advocates of SD and the wider environmental movement lack a coherent worldview. Many environmentalists, who claim to subscribe to a cosmology of cosmic interdependence (and even those who identify with deep ecology), would place humans outside of a natural order when asked to accept as natural large scale vegetation change through human influence in the past. The full range of cosmologies to which environmentalists subscribe is governed by the dominant Western philosophical assumption of reality being underpinned by permanent entities, and change being subordinated to these entities. This classical ontology makes change accidental, or incidental, to the existence of these entities, be they forests, vegetation structures, habitats

M ETHODOLOGY, O NTOLOGY AND E PISTEMOLOGY FOR SD

29

or species. Western philosophy has long maintained a separation between humans and nature (Glacken, 1973) but it is largely the ideas of Descartes, who advocated a separation of mind from body, which have become influential in subsequent philosophical thought. Descartes made a clear separation between the mind (which was equivalent to the soul, and the essence of what it meant to be human) and the structure and functioning of the biological organism (Damasio, 1994). The consequences of this Cartesian dualism for science have been profound. Scientists have attempted to study the mind without reference to neurobiology, and medicine has approached the study and treatment of diseases without considering the consequences of psychological conflict on the body. Damasio refers to this as "Descartes’ Error" in his detailed argument for placing the mind firmly at the nexus of neuroanatomy, neurophysiology and neurochemistry. This has the effect of replacing Descartes with Darwin when conceptualising the human mind. As an environmental scientist I find this intuitively appealing and deeply satisfying. The implications for the relationship between humans and nature which emerge from this replacement are noted by Damasio: Versions of Descartes’ error obscure the roots of the human mind in a biologically complex but fragile, finite and unique organism; they obscure the tragedy implicit in the knowledge of that fragility, finiteness and uniqueness. And where humans fail to see the inherent tragedy of conscious existence, they feel far less called upon to do something about minimizing it, and may have less respect for the value of life.(1994, 251) If we replace Descartes with Darwin, we reach a more holistic and normative sense of how integral and vulnerable humans are to the reforming of the environment that is possible through climate change (Costanza, 2003). These effects are not only about food supply, biodiversity, forests and oceans, but are also about the effects on one of the most unusual and unique species that has ever evolved. Although Damasio (1994) may have been referring to the respect for the value of human life in the above quotation, disregard for the environmental effects of human activity indicates that this statement could equally apply to all life on earth. The extent to which the process philosophy of Alfred North Whitehead is able to inform a coherent cosmology for the environmental (in the widest possible sense) movement has been investigated by Griffen (2007). Throughout this book, the importance of relationships within and between systems and the understanding of how feedback interacts across scales and entities is

30

C HAPTER TWO

essential to understanding how human activity and ecosystem functions can be integrated in SD. Process philosophy (Whitehead, 1929) regards change as being the foundation of reality (Griffen, 2007). However, a detailed discussion of Whitehead’s philosophy is beyond the scope of this book. It remains to be seen whether Whitehead’s ideas (1992, chapters 3 and 4) can be formulated into a cosmology that reconciles the divisions inherent in the environmental movement, and which even more deeply separates those outside of it from the aspirations of SD. Smith (1984) has attempted to unify humanity and nature through his theory of the production of nature through human labour. He draws on Marxist ideas to propose that: "Through human labour and the production of nature at the global scale, human society has placed itself squarely at the centre of nature" (Smith, 1984, 65). This vision of nature as a social construction is appealing, but Smith’s determination to erect Marx as the unifier of humans with nature attempts to ignore Darwin. Smith was clear in his rejection of a biological role for breaking down the Cartesian separation of humans from nature: Now those using and often abusing Darwin have attempted to extend Darwin’s insights back into the social world. The latest and most sophisticated attempt at this comes from sociobiology, the authors of which claim to explain the intricacies of individual and social behaviour by reference to biology; society is become a biological artefact. (Smith, 1984, 6) This attempt to cast all biologists who would make any statement about a biological origin for human behaviour as social Darwinists is outdated, and sets up a straw man with respect to the significance of findings in contemporary neuroanatomy, neurochemistry and neurophysiology outlined by Damasio (1994). Even with the latest findings of neuroscience, there are few biologists who would attempt to explain human social behaviour using solely mechanistic explanations of biological phenomena, and Smith acknowledges this, but nevertheless rejects any role for Darwin. The argument for Marxist theory as the only explanation for how nature is a production of human political economy reinforces the division between researchers who study human society and scientists who study biophysical phenomena, and in doing so negates a key requirement of SD that requires a multidisciplinary approach to social-ecological problems. I examine Smith’s ideas with respect to the production of nature and the production of space in a discussion of scale in chapter five and attempt to synthesise constructivist views of scale with positivist ideas more extant in ecology.

M ETHODOLOGY, O NTOLOGY AND E PISTEMOLOGY FOR SD

31

In contradiction to his earlier rejection of any contribution of biology to explanations for social behaviour, Smith later invokes Darwin when explaining the evolution of modern humans from earlier ancestors to support his theory of the production of nature: “It is human productive activity, not as a general concept but as a concrete historical act designed to create means of subsistence, that differentiates human beings from animals. . . . From the start human nature was a human product, and this applies not simply to consciousness, but even to human physiology. The development of the hand from a means of locomotion to a sophisticated limb for the manipulation of tools, is accomplished gradually through thousands of years of labour.” (Smith, 1984, 37) (emphasis added) If Smith is not invoking Darwin with this statement about the evolution of the human hand, then it is difficult to understand how adaptations for tool use link humans with nature. Smith is correct to invoke Darwin, but there seems little justification for rejecting sophisticated studies of the brain or sociobiology which attempt to make the same link between social behaviour and evolution. The next step, which is to explain all social behaviour with reference to mechanistic biological processes, is a step few biologists would attempt. In Figure 2.2, I attempt to show how both Smith (1984) and Damasio (1994) invoke Darwin to unify humans and nature. This does not negate a constructivist theory for conceptualising nature, but merely places evolution (via Darwin) as a nurturing force from which physical adaptations and the mind arise. Social behaviour is an emergent (as from a complex system) characteristic of the Darwinian structures. The emergence of a construction of nature as an ontological phenomenon is congruent with the emergence of physical adaptations which both emerge from an evolutionary separation of humans from animals. The demand which SD makes for interdisciplinary research is approached by reconciling the apparently antagonistic positivist and anti-positivist epistemologies and their ontological implications for a cosmology that unifies humans with nature, and ultimately informs a more mature and cohesive environmental worldview than that which is common among environmentalists.

Conclusion Interdisciplinary research is a defining characteristic of sustainability research (Blewitt, 2008), but presents its practitioners with significant methodological

32

C HAPTER TWO

Figure 2.2: Both Damasio (1994) and Smith (1984) invoke Darwin in an attempt to end the separation of humans and nature instigated by Descartes. The rejection of biological (and by implication positivist scientific) explanations for this unification by Smith (1984), who claims primacy for a structuralist epistemology mediated by Marxist theory, is contradicted by reference to Darwin. The roots of this contradiction potentially point to a potential reconciliation between opposing arguments. There is equal validity for both positivist and constructivist explanations of humanity’s emergence from and integration with nature

M ETHODOLOGY, O NTOLOGY AND E PISTEMOLOGY FOR SD

33

challenges (Klein, 1996; Poteete et al., 2010). In this chapter I have offered insights into the importance of positionality and reflexivity. These issues take on unique significance when conducting research in the global South and in countries with authoritarian governments (Sultana, 2007). I have justified my interdisciplinary approach where I combine what are normally seen to be different and conflicting epistemological positions by explaining how these positions complement the study of a social-ecological system. I have explored the potential for consilience between different philosophical positions through a critique of the separation of humans from nature. Detailed methods are explained in each of the following chapters to enable a critique of the methodology and results, and because, in each case, innovations in methodology were developed and employed. The interdisciplinary nature of the research requires familiarity with a range of literature and in the next section I provide a review of the major themes which inform the research in this study. There is a global literature that is pertinent to the interdisciplinary nature of this research, but where possible and relevant I draw on the literature from South-Central Africa.

Chapter Three Sustainable Forest Management for Local and Global Outcomes

The history of interdisciplinary thinking in relation to environmental problems is not new (Meine, 2013). When Aldo Leopold was president of the Ecological Society of America in 1947, he called for a “land ethic” that integrated developments in ecology, history, ethics and aesthetics: “An understanding of ecology does not necessarily originate in courses bearing ecological labels; it is quite as likely to be labelled geography, botany, agronomy, history or economics.” (Leopold, 1949, 262) The holistic synthesis which I attempt in this study draws together several strands of thought that have existed on the edges of scientific endeavour for more than 60 years. The emphasis on holism by Smuts (1926), Leopold’s (1949) interdisciplinary views and his strong emphasis on ethics, and Rachel Carson’s call to action in Silent Spring (Lytle, 2007) are gradually coalescing into interdisciplinary approaches in ecology, politics and economics and the emerging field of sustainability science (Clark, 2007; Clark & Dickson, 2003; Fiksel, 2006). Prior to the emergence of this holistic approach, the only academically valid kind of systems thinking was the development of a mathematically expressed general theory of systems in which the problems of many different disciplines would be expressed using meta-language and theory rooted in mathematics (Checkland, 2000). There are now few researchers who work on problems of society and the environment that believe 35

36

C HAPTER THREE

this is possible. It is the relatively new holistic, interdisciplinary approach that underlies the breadth of this review of the published literature in this chapter. I examine the nature of sustainable development and its contested and complex status, questioning the assumptions about development and conservation that prevail in the global South in general, and Africa in particular. The science of forest carbon and the complexities of managing forests for carbon sequestration under REDD+ are discussed with particular reference to dry tropical forests in Zambia. New developments in theoretical ecology are examined in a review of resilience theory and the influence of complex systems theory on our views of social-ecological systems. The review concludes with an assessment of institutional and governance issues and particularly the theories of Elinor Ostrom (e.g. Ostrom, 1990) and how they have added to our understanding of Common-Pool Resource (CPR) management. The research approach in this study covers a wide range of disciplines and requires an equally wide ranging consideration of the literature. This review will inform the conclusions which are reached in each chapter, as well as the holistic synthesis which is attempted.

Sustainable Development The rise of the environmental movement during the twentieth century is attributed to the publication of Rachel Carson’s novel Silent Spring in 1962 (Lytle, 2007). Although the roots of environmental thought extend further back in the nineteenth and twentieth centuries to John Muir (Nash, 1982) and Aldo Leopold (Leopold, 1949), it is the demand for action which emerged from Silent Spring that led to the formation of the United States Environmental Protection Agency, and contributed to the widespread public awareness which prompted the formation of international organisations concerned with environmental issues (Lytle, 2007). The birth of the concept of sustainable development can be traced to the United Nations Conference on the Human Environment in Stockholm in 1972, which was an attempt to find a compromise between the development needs of the global South, and the conservation concerns of the global North (Robinson, 2004). The World Commission on Environment and Development (WCED) (or Brundtland) Report (Brundtland, 1987) is the most frequently cited document responsible for the widespread awareness of the concept of Sustainable Development, but the ideas embodied in this document, and around Sustainable Development (SD) in general, are a logical extension of the combined environmental and social-political concerns of Western society in the 1960s, 1970s and 1980s (Robinson, 2004). In this context, although the Brundtland

SFM FOR L OCAL AND G LOBAL O UTCOMES

37

report could be described as reformist and radical in one sense, it is very much a product of its time. The definition of SD adopted in the Brundtland report in 1987 has resulted in both widespread adoption of the term with diverse interpretations, and widespread criticism of SD as a meaningless concept, taken ownership of by individuals or organisations for the political purpose of stamping their definition on a term which has become increasingly influential (Mebratu, 1998; Rogers et al., 2008). Herman Daly, a senior economist at the World Bank, is frequently quoted as having called the term SD an “oxymoron” (Redclift, 2005), but a closer reading of his 1990 paper (Daly, 1990) and subsequent publications (Daly & Farley, 2010), shows that he was referring to “sustainable growth” as an oxymoron. Daly (1990) criticises the Brundtland report for stating the problem which SD is supposed to address in a way which conflates the terms “development” and “growth” in the context of the economy. Daly is clear that SD is both desirable and achievable (Daly, 1993). It serves little purpose (and makes for tedious reading) to review the more than 70 definitions (Holmberg & Sandbrook, 1992) (and frequent accompanying semantic debates) of SD in current use. Mebratu (1998) makes a useful division of how different groups understand and conceptualise SD by looking at the institutional, ideological and academic understandings of SD and the ideas and solutions which emerge from this conceptualisation. Institutional Understanding

Emerging from the WCED Report is an understanding of SD which is focussed on human needs and is widely adopted by international organisations which focus on poverty alleviation, as well as the World Business Council for Sustainable Development (WBCSD) which represented business interests at the Earth Summit in Rio in 1992 (Redclift, 2005). Within this definition emerges a refinement which identifies three pillars of SD: the ecological system, the economy and society (Mebratu, 1998). This so-called “Triple Bottom Line” (Elkington, 1997) focusses on economic growth as a means of alleviating poverty and delivering social justice and accepts that this must, and can, be done without destroying the environment. Ideological Understandings

fits in well with extant ideological traditions that are critical of capitalism as well as those which propose a more general green ideology. Marxism, eco-socialism (Pepper, 2002), eco-theology (Deane-Drummond, 2008), deep ecology (Naess, 1973) and the new apogee of all of these and more, eco-feminism (Buckingham, 2004; Warren, 2000), offer differing but similar radical SD

38

C HAPTER THREE

interpretations of SD. There are certainly questions about “needs,” and how development, economic growth or culture are the determinants of these needs and their changing nature, which are seldom asked outside of radical green circles (Redclift, 2005). Although the solutions for SD which emerge from these traditions (abandonment of the capitalist system, socialist egalitarianism, adoption of a nature-centred spiritualism and adoption of female centred values, respectively) may not be practical in the context of this study, they still have value in confronting problems of SD in developing countries and contributing to notions of social justice with respect to culturally differing needs and north-south conceptions of SD (Redclift, 2005).

Academic Understanding

Academic responses to the ideas which encompass SD have informed diverse disciplines such as education, management and law, but perhaps have their most wide ranging influence in reforming how economists, ecologists and sociologists have developed new bodies of theory to respond to the challenges which SD presents (Mebratu, 1998). In the following sections I will examine the influence of SD on economics and social-ecological theory. SD has important implications for economics, especially in the context of REDD+, and the relationship between society and nature has been profoundly affected by SD. Economists, most of whom advocate a neo-liberal agenda (which since the 1980s is the dominant force in international policy formulation (Mebratu, 1998), have advocated the commodification of the environment to reconcile growth with ecological limits. There are several reasons for this approach. Economists view many environmental problems as arising from the failure to cost externalities effectively so that businesses regulate their activities accordingly (Turner et al., 1994). By creating a market for environmental resources which are not usually bought or sold, scarcity can be established and the costs and benefits of using those resources can be properly assessed (Pearce & Barbier, 2000). Carbon dioxide from industrial processes, for example, has been discharged into the atmosphere without the true cost to society which climate change predicts being taken into account. The central idea behind REDD+ is to use the price of carbon (which emerges from regulations of the market in carbon) so that a value can be placed on the carbon sequestered by forests. Pearce & Barbier (2000) make the argument that the advantage of framing environmental problems in this way is that it allows these problems to be represented in the political decision making process in the same way that other economic problems are represented.

SFM FOR L OCAL AND G LOBAL O UTCOMES

39

Economists have made an important contribution to the debate around theory by framing the issue in a purely economic (albeit simplified) interpretation of human well-being-that of utility (Dietz & Neumayer, 2007). The concept of utility is seen to be embodied by four kinds of capital: Produced, Human, Natural and Social, and development is sustainable if it “maintains the capacity to provide non-declining per capita utility for infinity” (Neumayer, 2003). The extent to which natural capital can be substituted for other forms of capital, especially produced capital, defines the concepts of “weak” versus “strong” sustainability. Proponents of weak sustainability argue that natural capital is entirely substitutable by produced capital, or that technological advancement can increase the productivity of the stock of natural capital faster than it is depleted (Neumayer, 2003). By contrast proponents of strong sustainability regard natural capital as non-substitutable both in the production of goods and in utility for human well-being (Neumayer, 2003). Beyond the economic models of economists it seems that the case for weak versus strong sustainability may well depend on the different kinds of natural capital being considered and that substitutability may vary both within and between these different types of capital. The ecological conceptualisation of SD offsets the inherent sustainability and resilience of ecosystems against human requirements for a constant supply of goods and services and the resulting disruption of ecological processes. Ecology as a science has hitherto been strictly reductionist and firmly rooted in the positivist tradition (e.g. Peters, 1991). The response of ecological science to human induced ecological problems results in a so-called “shallow ecology” approach, in which an assumption is made that we can solve the immediate ecological problems without confronting the philosophical issues which underpin current social and economic actions. A “deep ecology” approach advocated by the philosopher Arne Naess (1973) advocates similar ideas around social justice and biological egalitarianism that eco-socialism or eco-theology espouse, but includes many of the principles of complexity and social-ecological systems that have recently become a part of the resilience literature in ecology (Berkes et al., 2003; Holling, 1973). Inasmuch as the resilience literature attempts to stay true to its positivist roots, Deep Ecology represents a unique nexus of spiritual and moral values and scientific principles to potentially make an important contribution to the conceptualisation of sustainable development. The assertion that biodiversity is a value in itself underpins the biocentric vision of Deep Ecology and assumes that humans have no right to reduce biodiversity except to satisfy their most basic needs (Naess, 1973). Nevertheless the resilience literature appears to represent the most imSD

40

C HAPTER THREE

portant embrace of what could be called a SD approach to solving ecological problems (Fiksel, 2006; Ludwig et al., 1997) by moving away from viewing ecological systems as entities which can be wholly understood through the reductionist modelling of environmental parameters, and including human influences and the way they affect ecological processes. Attempts are being made to incorporate social systems and their associated policy and technology responses to ecological changes into dynamic models which are aimed at the sustainable management of these complex systems (Folke, 2006; Ostrom, 2009a; Walker et al., 2006). The combined development of the concept of sustainable development, the emergence of sustainability science, and the idea of the social-ecological systems approach to ecology may represent the emergence of a holistic response by academia to the problems of sustainability. While this may be the case for natural resource management, other forms of response are required, most notably in economics where there is a requirement for alternative economic frameworks. Despite this awareness the structures of funding bodies, academic departments and academic journals have not entirely embraced this holism. Poteete et al. (2010) dedicate an entire book to the problems of interdisciplinary work and Elinor Ostrom frequently mentions the difficulty of overcoming faculty, funding bodies and journals to do interdisciplinary work. Sustainable Development (SD) to Sustainable Forest Management (SFM): From Local to Global

One of the key differences between debates about development and the environment, which emerged after the concept of SD became more widely embraced, was not only their close relationship, but also the global challenge which SD represents (Elliott, 2012). Rockström et al. (2009) introduced the concept of “planetary boundaries” which define safe biophysical limits for conditions on earth to be maintained in the state under which human civilisation has developed and thrived. Defining global limits in this way makes a direct link to local actions and their global effects by making the global effects relevant to activities which are part of normal life in the Western world. This is an extension to the foundational ideas of the Scottish biologist and sociologist Patrick Geddes (1854-1932) whose urban planning framework proposes a convergence of the disciplines of geography, economics and anthropology (or place, work, and folk as he described it) in order to build urban environments which positively influence social harmony on a larger scale. The effects of consumption have the potential to affect everyone on earth and give rise to the concept of climate equity which provides the rationale for REDD+ as a system of payment for ecosystem services. It is the

SFM FOR L OCAL AND G LOBAL O UTCOMES

41

connection between the local (at varying scales, from individuals to nation states) and the global that makes payment for CO2 sequestration logical and justifiable, even if the mechanisms for the implementation of this system are complicated. Ultimately this connection between the local and the global provides strong justification for sustainable development to be a wide ranging concept that affects economic, social and political activity across scales. The term Sustainable Forest Management (SFM) has a long history. It has had different meanings in the past, and no matter what definition is applied to it today, the term is difficult to define and reflects the values which society ascribes to forest functions at a particular point in time (Douglas & Simula, 2010c). These values change over time and therefore the definition of forest sustainability will also change. The Food and Agricultural Organisation (FAO) defines SFM as,

The stewardship and use of forests and forest lands in a way, and at a rate, that maintains their biodiversity, productivity, regeneration capacity, vitality and their potential to fulfil, now and in the future, relevant ecological, economic and social functions, at local, national, and global levels, and that does not cause damage to other ecosystems. (FAO, 2005, 6)

Like most definitions concerned with sustainability, it is more useful to adopt an operational definition of SFM. Douglas & Simula (2010c) consider that consensus among all stakeholders that a forest is being used in an appropriate way, and that its biological condition is stabilised for an extended period of time, is a sufficient definition for the achievement of SFM. The link between local actions and global consequences is behind much of the emphasis on forests and SFM, and the relationship between forests and climate change highlights this connection. The presence of large forested areas of earth is crucially important for stabilising global climate systems (Bonan, 2008). The actions which need to be taken for these forests to be preserved will benefit the entire planet, but the need to take action arises from carbon producing activities in the developed world. There is no hyperbole in making a direct connection between individual consumption in developed countries and the effects this has on climate change (e.g. Carlsson-Kanyama & González, 2009; Marlow et al., 2009). The actions taken as a result of a REDD+ programme negotiated at international climate negotiations have direct effects on the lives of forest people. The links are clear.

42

C HAPTER THREE

Development, Community and Well-being

Development theory is in the midst of a crisis from which it has yet to emerge (Parfitt, 2002). In a disparaging summary of development theory Sachs (1997, 1) wrote: The idea of development stands like a ruin in the intellectual landscape. Delusion and disappointment, failures and crimes have been the steady companions of development and they tell a common story: it did not work. Moreover the historical conditions which catapulted the idea into prominence have vanished: development has become outdated. But above all, the hopes and desires that have made the idea fly, are now exhausted: development has grown obsolete. The concept has certainly created its share of controversy. The use of the term “development” has created a category of people who are “underdeveloped” relative to those who are “developed” and has justified the involvement of Western countries in the lives of millions of rural people (Banerjee, 2003). It is the categorisation of people outside of Western countries, who live in many different ways, as a simple mirror opposite of the developed, and as needing ‘development’ in order to reach the heights achieved by the developed, that many researchers, community workers and community leaders in those countries find problematic. The association of development with economic growth has resulted in misery for millions of people as traditional livelihoods became less meaningful in this new context and were replaced with alternatives such as growing cash crops. It is in this context that the term has become loaded with meaning (Escobar, 1995), much of which is increasingly rejected by grassroots movements which harness local knowledge to improve people’s lives in ways which are different to the way development is conceptualised by Western donors or development aid organisations. Escobar (1995, 217) states it succinctly when he suggests that, “the authors representing this trend state that they are interested not in development alternatives but rather alternatives to development, that is, the rejection of the entire paradigm altogether.” It remains to be seen whether the meaning of “development” can be reconstituted in a way which carries less of the negative associations of the widespread failure of development programmes. Despite these criticisms, failure has led to introspection and an attempt to re-evaluate the basis on which development practice and theory were founded. The wide range of issues which Chambers (1983) raises with respect to development is an attempt to change the way development is practiced and perceived by the rural poor. The challenges which Chambers (1983) presents

SFM FOR L OCAL AND G LOBAL O UTCOMES

43

for those concerned with rural poverty and rural development are personal, professional and institutional and are deemed to be largely responsible for the marginalisation of the rural poor. This marginalisation is behind many of the failures of development agencies to address the problems of the rural poor and the awareness of this shortcoming has resulted in development professionals seeking new approaches and methods in their work. Action-reflection research (Reason & Bradbury, 2001), agro-ecosystem analysis (Conway, 1985), applied social anthropology (Nadel, 2013) and rapid rural appraisal (Chambers, 1981) have led to the emergence of a family of approaches and methods in which the participation of local people in the research is central to the generation of knowledge. I discuss these methods more fully on page 51, but it is important to make the link between these participatory methods of research and the emergence of participation as an essential component of sustainable natural resource utilisation. Participation has become an important component of the rural development methodological toolbox independent of the process of introspection underway in the development community. If there can be said to be an emerging law of sustainable development in a rural context, then that law would have participation as its central tenet. The ideas of Sen (1999) go some way to redefining development as “a process of removing unfreedoms and of extending the substantive freedoms of different types that people have reason to value” (3). This removes the cultural link with freedoms that are associated with Western assertions of human rights, but which are valued differently in a particular culture. The definition also dissociates development from economic growth and yet provides the space for “entitlements” and “capabilities”, which are essentially economic terms which represent the aspirations of people without implying what they should be. It seems that this may satisfy Escobar’s (1995) requirements for an alternative to development which still allows for people to aspire to a life of well-being. The link between freedom and development however, and Sen’s (1999) clear support for democratic freedom, have worrying implications for development in Africa where governments have mostly persisted in restricting these freedoms, even in states which have free and fair national elections. Democratic freedom is more than the right to vote in an election. The existence of communities which are ill-disposed to resource extraction, for example, indicates that although a country may practice democracy on a national scale, local institutions which empower communities may be entirely lacking. Stedman (1999) provides an excellent critique of sustainable forest management which links SD and SFM through the consideration of a “sense of place” which suggests measuring the extent to which SFM answers the ques-

44

C HAPTER THREE

tion “what should forests sustain ?” Stedman (1999) defines a sense of place as the meaning which is attached to a spatial setting by a person or a group (community). By recording responses of individuals in communities to quality of life statements, descriptive beliefs or place meanings, one can assess the sense of place which pervades the community, and both identifies the cohesiveness of the community through a common sense of place, and provides a link between the achievement of SFM in sustaining the community. It is important that this variation be taken into account when implementing any programme of SFM, and that the scales of community affiliation, from the village to the state, be considered when planning development programmes which affect large numbers of individuals. The lines which partition communities are seldom obvious to the uninitiated researcher, but are clearer to those who comprise and claim affiliation to different communities, despite the potential for differences between communities at different scales. When assessing a sense of place that SFM is intended to sustain, it is important for the concept of community to be clear. However, “community” is a contested concept which has different meanings in different contexts and different literatures (Gusfield, 1975; McMillan & Chavis, 1986; Peterson et al., 2008; Wenger & Snyder, 2000; White & Harder, 2013). Preservationist Ideals

There is a tension between the concept of SD and the concerns of a large and vocal preservationist movement which has a long tradition among environmentalists and conservationists (Robinson, 2004). The emphasis of SD on human needs and what is required to satisfy these does not support the preservationist ideal. Preservationists will argue for the role of wilderness (areas devoid of human presence) in maintaining ecological integrity, or for fulfilling spiritual needs, but the evidence for the wilderness ideal in an ecological sense is thin, and fulfilling spiritual needs contrasts poorly with the requirement of moving large numbers of people out of areas in which they live to make these wildernesses. To be sure, the general sustainability literature ranges widely with respect to moral and ethical issues and draws on Eastern philosophy to support a sustainable lifestyle less linked to material wealth (Feng, 2009), and also places emphasis on human mental and spiritual well-being (Blewitt, 2008). However, there is a considerable difference between the sustainability literature and the widely accepted interpretation of the concept of SD, even if we accept Robinson’s (2004) advice and embrace the concept’s “constructive ambiguity.” It is difficult to see how the term “development” can be reconciled with the preservationist sine qua non which is the absence of humans from vast landscapes. In an ecological sense, the

SFM FOR L OCAL AND G LOBAL O UTCOMES

45

preservation of some large wilderness areas free of human influence may be justified in the context of ecosystem function, especially when these areas are unsuitable for human habitation and where human influence would have a dramatic effect on ecosystem function. Wetland systems are often sensitive to human disturbance or nutrient enrichment and probably function better with minimal human presence. However, the designation of many areas of wilderness, justified as the preservation of a pristine, ecologically ideal state, often does not withstand scrutiny (Adams & McShane, 1992; Denevan, 2011). Traces of human influence are often found to have had a profound influence in the past and the historical record shows clearly that sanitisation is what happened in many of the areas in Africa we think of as pristine today (Adams & McShane, 1992). This has important consequences for the implementation of SD in eastern and southern Africa. The foundation of conservation in Africa can be traced back to preservationist European ideas of a natural habitat teeming with wildlife. It is not widely appreciated that concerns about artificially induced climate change and species extinctions had their genesis in the colonial experience generally and the experience at the Cape in South Africa was particularly influential (Grove, 1990). Indeed it was in South Africa, Australia and North America that governments were first urged to make the natural environment and the threats posed by human induced degradation a concern and a responsibility, long before this was possible in Europe (Anderson & Grove, 1989). The problem with preservationist ideas is that they exclude human influence as an integral and important part of shaping the environment. Anderson and Grove (1989, 4) summarise the issue by stating that, Much of the emotional as distinct from the economic investment which Europe made in Africa has manifested itself in a wish to protect the natural environment as a special kind of “Eden”, for the purposes of the European psyche, rather than a complex and changing environment in which people have actually had to live. The desire to maintain ‘Eden’ has been particularly pronounced in eastern and southern Africa, where European ambitions have extended to permanent settlement. Many of the first national parks in colonial Africa were proclaimed in this spirit of preservation. The preservationist movement in the United States, and the formation of Yellowstone National Park in 1872, inspired a worldwide movement which was enthusiastically copied in southern and east Africa (Adams & Hutton, 2007). The formation of Kruger National Park in South

46

C HAPTER THREE

Africa and Albert (now Virunga) National Park in what was the Belgian Congo were directly inspired by the Yellowstone phenomenon (Nash, 1982). Declarations of other national parks were soon to follow. Paradoxically, nearly all of these areas required the movement of people who had been living in this environment for long periods of time. In 1934, Pitman (1934) urged the colonial government of Northern Rhodesia to form new reserves and expand existing ones, and recommended that “the native population should be removed from the game reserves as soon as possible”. This had disastrous consequences for people in eastern Zambia, many of which were still having an impact decades later (Vail, 1977). The formation of national parks in Tanzania resulted in large scale movement of people out of these areas in order to enforce the preservationist view of a pristine wilderness (Nelson, 2003). Humans, and their influence on the habitat, had no place in the preservation of wild Africa. These preservationist ideas have a powerful influence on how African countries practice conservation and implement SD with respect to conservation (Adams & McShane, 1992). The institutions involved with conservation in Africa today are by and large the product of international conservation organisations and conservation professionals trained in ecology and conservation science. Although these institutions are part of, or closely aligned to, state governments, most of the funding comes from overseas development organisations. This is not only a problem with respect to SD, but represents a fundamental difference between the objectives of conservation on the part of these organisations (who are largely biocentric) and the concerns and objectives of rural Africans which are more instrumental and economic (Barrow & Murphree, 2001). Critics of this statement would point to the changes in the conservation paradigm during the 1980s, which was one of including people rather than excluding them from conservation programmes (Adams & Hutton, 2007). The community-based approach to conservation now dominates the practice of conservation, and Zimbabwe’s Communal Areas Management Programme for Indigenous Resources (CAMPFIRE) was an innovative regional pioneer of CBNRM (Bond, 2001; Duffy, 2000; Frost & Bond, 2008; Murombedzi, 1999). Despite these efforts, however, the dominant paradigm remains focused on sharing the revenue of sustainable use (mainly of large mammals, but also timber in some cases) in the protected areas rather than an integrated view of the role humans may have had in shaping the ecosystems we see today. The contentious relationship between conservation authorities in Zimbabwe and the Convention on Trade in Endangered Species of Wild Flora and Fauna (CITES) which centres around the sustainable use of elephants (Mofson, 1997), is an example of how different views of wild Africa come into conflict,

SFM FOR L OCAL AND G LOBAL O UTCOMES

47

despite elephant utilisation being focussed on improving the livelihoods of rural people. Zambia has a poor record of involving communities in conservation, and therefore stands out regionally as one of the countries where communities have little role in benefit sharing (Jones, 2007), even if they were viewed as part of the ecosystem rather than as intruders into an Edenic preconception of pristine wilderness. This Edenic vision is not restricted to the formation and management of national parks. The study of ecology in Africa (and elsewhere, most notably North America) has been almost wilfully ignorant of the evidence which shows human influence on habitats over long periods of time. Descriptions of vegetation structure still use terms linked to the concept of succession theory to describe vegetation, despite the evidence that humans appear to have been using fire to change habitats in Africa for about four hundred thousand years (Bird & Cali, 1998). The existence of forest patches in many west African countries is often due to the presence and practice of the inhabitants in areas which would otherwise be open savanna (Fairhead & Leach, 1998). Similarly but conversely, areas of the Serengeti-Mara ecosystem were forested as late as the 1950s before livestock levels recovered from the rinderpest epidemic, and the area appears to have undergone several cycles of vegetation change over the last hundred years (Dublin, 1991). Elephants, fire and humans have played an integrated role in shaping this “pristine savanna” habitat. Chidumayo & Gumbo (2010) give several examples of how the dry forests of Africa may have been influenced by human activities over thousands of years. They claim that the current disturbance regime, characterised in the past by charcoal manufacture for iron smelting and low levels of slash-and-burn agriculture, has not changed much but merely increased in intensity with increased population density. This means that the vegetation structure we see today in south western Zambia is to a greater or lesser extent the product of thousands of years of human disturbance. Nevertheless, human influence does result in habitat degradation to the point where ecosystem function is altered in some circumstances. The most dramatic effects include changes to ecosystem resilience that result in systems moving to an alternative stable state. However, not all change is movement away from a pristine condition (Beinart, 2008). The extent to which human influences degrade or enhance (or even simply create) the ecosystems we see today in Africa has not been sufficiently accounted for in theory or management practice. This is problematic for SD and SFM in Zambia and southern and east Africa. The maintenance or return of the vegetation to a “natural” state is seen by state authorities and development NGOs as a goal which must

48

C HAPTER THREE

be achieved if one is to claim success (Adams, 2003). This is directly in contrast to whatever definition or conceptualisation of SD or SFM one wishes to advocate. Without humans as an integral part of the environment, there is no SD, only greater or lesser forms of destruction and interference, albeit possibly sustainable. This tension can only serve to create conflict as the state and other organisations use their entrenched views of “Africa as Eden” to manage resources which evidence increasingly shows should be managed as if Africa were home. The preservationists who perpetuate an idea of Africa as Eden are not solely responsible for placing humans outside of the ecosystem. The role of historians of environmental history has not been conducive to establishing a role for humans in African ecosystems because: they have generally focussed on individual historical figures and political events. Despite their discussion of “the economic factor” or various aspects of “social change”, few have sought to recast their narrative to depict human beings in Africa as one element in complex and evolving ecosystems. (Weiskel, 1998, 142). There is, nevertheless, a growing acknowledgement that landscapes have been influenced to greater or lesser degrees by humans. The works of Fairhead & Leach (1995, 1997, 1998); Leach & Fairhead (2000); Leach & Mearns (1996) in west Africa, Williams (2000) in Europe, Denevan (1992, 2011) in North America and Denevan (2007) and Peters (2000) in the Neotropics, indicate that human activity in the past had a profound effect on the landscapes of the present, in most areas of the world. This needs to be increasingly acknowledged if the practice of managing of natural resources is to fully embrace the concept of SD. Sustainability Research and Sustainability Science

In her review of sustainability research, (White, 2013) takes a broad approach to defining the domain of sustainability research (“relevant to all disciplines”) and its attributes. Although interdisciplinarity has previously been identified as almost defining sustainability research (Franklin & Blyton, 2011), there are certainly other attributes which are important. The consideration of different knowledge forms, the explicit linking of theory with practice, the recognition of the value of participatory approaches to knowledge generation and a reflective process of self-assessment on the part of the researcher are all potential attributes of sustainability research (White, 2013). The aim of sustainability research and its subset, sustainability science, is to gain

SFM FOR L OCAL AND G LOBAL O UTCOMES

49

an integrated understanding of the dynamic interactions between nature and society (Clark & Dickson, 2003) so that resources can be utilised in perpetuity whilst still preserving their ecological services. Exactly how the process of interdisciplinary research is to be conducted remains open to debate. Sustainability science is defined by the problems it addresses rather than the disciplines it employs (Clark, 2007), and since there are few, if any, problems in SD which are restricted to a single discipline, finding a way to conduct interdisciplinary research is as important as the research itself. Ostrom (2005) makes an important contribution to the process of conducting interdisciplinary research with the development of the Institutional Analysis and Development (IAD) framework. Approaching the problem of understanding sustainable interactions between society and the environment by the analysis of institutional diversity and complexity, and a clear recognition that there are diverse and complex arrangements which work well for sustainably governing and managing natural resources, has led to the development of the IAD framework. The underlying assumption of the IAD framework is that there are regularities in human behaviour which are amenable to the development of a multilevel taxonomy of rules which describes the fundamental building blocks of organised human behaviour (Ostrom, 2005). The strength of the conceptual approach which the IAD uses is that its nested scaling of institutions in a hierarchical system facilitates dialogue between natural and social scientists. The scaling of ecological processes and the importance of measurements at the appropriate scale in order to inform understanding of these processes are well understood in the natural sciences. However, considerations of scale have been less explicit, less precise and more variable in the social sciences and have more to do with conceptual constructions of scale than with measurement using statistical methods (Gibson et al., 2000b). I discuss the IAD in more detail on page 89 and expand on the definitions of key terms related to the concept of scaling, along with the implications of an explicit consideration of scale for institutional analysis and how this may be used to link with appropriate scales in ecological systems thus facilitating interdisciplinary research. Kates & Dasgupta (2007) present a rather more ambitious challenge for SD research methodology by claiming that African poverty, and by extension African exceptionalism (sub-Saharan Africa’s unique failure to show marked improvements in indicators of poverty over the last twenty years), represent a “grand challenge for sustainability science”. There are several problems with presenting the challenge of understanding the interaction between African ecosystems and African societies in this way. Attempts to generalise about African development represent an oversimplification of the problem and

50

C HAPTER THREE

indeed a paradox with respect to a generally accepted principle of sustainability research: the solutions are often found in the place-based details of the interactions and their outcomes. Therefore, these details must be studied at a scale which is appropriate for finding a local solution (Clark & Dickson, 2003; Lane & McDonald, 2005), which is most definitely not at the scale of a continent larger than the United States, China, India and western Europe combined. Nevertheless, attempts at the integration of different research disciplines remain difficult, or even actively resisted by some sectors of the research community. SD research (at least with respect to its application in the context of environmental decline) still faces the challenges which Aldo Leopold noted in ecology in 1935 when he wrote: One of the anomalies of modern ecology is that it is the creation of two groups, each of which seems barely aware of the existence of the other. The one studies the human community almost as if it were a separate entity, and calls its findings sociology, economics and history. The other studies the plant and animal community and comfortably relegates the hodge-podge of politics to the liberal arts. The inevitable fusion of the two lines of thought will, perhaps, constitute the outstanding advance of the present century. (Meine & Knight, 1999, 272). To some extent these problems in ecology are being addressed in studies of ecological resilience (Gunderson & Holling, 2002; Gunderson, 2000; Holling, 1973; Ludwig et al., 1997). Does the formation of ideas in ecology and resilience studies have anything to contribute to SD research and its defining requirement, that of being interdisciplinary or ultimately transdisciplinary? The idea of a social-ecological system with its associated feedbacks has important implications for SD research and provides inspiration for the conceptual tools which have emerged that may serve as analogues in conceptualising SD research (Anderies et al., 2006; Walker et al., 2006; Walker & Salt, 2006). In addition there is much to learn from the insights into governance which common-pool resource theory offers (Ostrom, 1990, 1999, 2007b) and the emerging field of political ecology (Adams & Hutton, 2007; Dobson & Eckersley, 2006; Zimmerer & Bassett, 2003) appears to represent a useful synthesis, although it is yet to rise to the heights suggested by Aldo Leopold’s observation. Later sections (Page 89) present a review of the major themes from these current research agendas and how they may be useful in SD research.

SFM FOR L OCAL AND G LOBAL O UTCOMES

51

To a large degree, sustainability research has been informed by the work of development professionals working in developing countries over the last 40 years. Significant problems with this research have emerged as development programmes fail to address the issues they were designed to resolve and in some cases have exacerbated rural poverty rather than ameliorated it. In the next section we look at the response of the development community to these failures. Participatory Approaches

In recognition of the global disparities between and within countries and the global deterioration of ecosystems, the 1992 United Nations Conference on Environment and Development in Rio de Janeiro established a document called Agenda 21, which is a commitment by most of the world’s governments to SD (Nations, 1992). This document sets out a dynamic programme of action intended to address the problems of development. It is a non-binding agreement and is intended to be carried out by various actors according to the different capacities and priorities of the signatory country. Agenda 21 has an institutional focus and links institutions to tasks which need to be initiated or completed. Importantly, participatory principles are linked to actions by institutions on local and national scales so that participation becomes an integral part of SD implementation. Evidences for the success of participatory approaches to environmental management include the Kenya Greenbelt Movement (www.greenbeltmovement.org) which was formed to respond to the needs of rural Kenyan women (Kapoor, 2001). The extent to which Agenda 21 principles have been adopted by local authorities and institutions varies widely. There has been widespread adoption in Africa and Europe, but some areas in the United States have actively legislated against the proposals, citing concerns about the restriction of individual liberties (Spangenberg et al., 2002). The success of Agenda 21 implementation varies widely, but there is nevertheless widespread consensus that the principles are an important blueprint for SD implementation. The range of personal, professional and institutional barriers to the application of development theory has been discussed in detail by Chambers (1983). Solutions to these problems more fully explored in Chambers (1997). Participatory approaches to research in sustainability have been rising to prominence, particularly in the last 10 to 15 years. Although the foundational work of Robert Chambers has been the subject of a wide ranging critique (e.g. Kapoor, 2002; Kindon et al., 2007; Parfitt, 2004) there is generally rising popularity and credibility of participatory approaches. The motivation behind this credibility are the development failures which characterise much of the

52

C HAPTER THREE

work done in the global South by professionals and development agencies in the global North. Chambers (1997) indicates that the source of these failures include: • The development profession constructing its own reality with respect to the causes of rural poverty and the economic and social interventions required for addressing poverty. Epistemological rigidity has contributed to this construction. • The arrangement of knowledge and power has been skewed towards development professionals and institutions in the global North. These professions claim a monopoly on reality which they largely construct, but which does not match the reality of the development environment. • The contradictory nature of constructing reality and claiming objectivity within the same discipline. • The simple failure to take into account the knowledge which nonliterate people have about their environment. This often emerges as development projects fail and the rural poor become more militant and assertive as a result of these failures. The response of the development community has been the emergence of a family of research methodologies collectively known as Participatory Rural Appraisal (PRA). Figure 3.1 shows some of the research methods which have contributed to the concept, although the approach has been considerably expanded to include urban areas. Its roots in addressing problems of rural poverty are now less exclusive than the acronym implies. PRA is broadly defined as a family of approaches and methods to enable rural people to share, enhance, and analyse their knowledge of life and conditions, to plan and to act and to monitor and evaluate (Chambers, 1997). PRA has ideas in common with reflexivity (page 22) and much of the methodology has been developed from practice rather than a priori principles. Figure 3.2 shows the three pillars of PRA namely: • the behaviours and attitudes of outsiders, who facilitate the dialogue and do not dominate • the methods require that the discussion be shifted from verbal to visual, individual to group, from measuring to comparing • sharing of experience and knowledge between individuals perceived as outsiders and those perceived as insiders and between organisations

SFM FOR L OCAL AND G LOBAL O UTCOMES

53

Figure 3.1: This diagram shows some of the research methodologies which have contributed to the evolution of participatory approaches to learning which have emerged as a family of techniques known as Participatory Rural Appraisal (PRA). From Chambers (1997).

54

C HAPTER THREE

Figure 3.2: The three pillars of Participatory Rural Appraisal (PRA). The diagram shows how the methodology has evolved over time. Some of the data which may be gathered from a PRA approach is indicated by the list of methodology indicated in the lower left. From Chambers (1997)

SFM FOR L OCAL AND G LOBAL O UTCOMES

55

PRA techniques have become widely used in southern Africa, especially in the context of natural resource use (e.g. Mombeshora et al., 2009). The technique has contributed to the success of some programmes, but there are indications that the opportunity to further advance development theory, or at least the opportunity to implement it, may be nearing an end in some areas of Africa. Kragelund (2014) has suggested that the international development space is undergoing major changes in the geographies of power in Zambia. The emergence of China, India and Brazil as economic powers and their increasing influence in Africa has allowed the Zambian state to pursue a sovereign frontier with respect to its development agenda which lessens the Zambian government’s dependence on traditional donors. Development as currently practised may die the death it was inevitably destined for as countries emerge from poverty, although the speed and manner of its passing may be wholly unexpected.

Summary

In this section I have introduced some of the multiple meanings of SD and how the concept has had a wide influence in shaping our understanding of what needs to be done to address the many, complex problems that emerge from the interaction between environmental degradation and society. Forests and SFM are important components of addressing issues of climate change and the concept of REDD+ embraces many of the concepts of SD. In the next section I discuss how forests interact with the atmosphere to accumulate carbon, present an assessment of the status of REDD+ and in later sections I consider how resilience theory and Common-Pool Resource (CPR) theory may contribute to a holistic synthesis of the problems around Sustainable Forest Management.

The Science of Forest Carbon Forests accumulate carbon in a dynamic process which includes gross photosynthesis, plant respiration, the transfer of carbon to the soil through litter accumulation and the gradual release of this carbon to the atmosphere through processes of decomposition and respiration by the soil microbial community. However, not all forested regions contribute equally to this process. There is surprisingly little known about the net effect on atmospheric temperatures of evaporative cooling by tropical forests, versus the masking of increased snow induced albedo that occurs when boreal forests expand over formerly open ground (Bonan, 2008). Betts (2000) suggests that the effects of decreased

56

C HAPTER THREE

Table 3.1: Measures of Above Ground Biomass (AGB) for different dry forests in Africa. The measures for the Congo Basin forest and African closed canopy tropical forests (mean values) are the highest in Africa and are included for comparison Forest Type Congo Basin forests African closed canopy forests Sub-humid dry forests Zambezian Woodlands Sudanian Woodlands Kalahari Scrubland Somali-Masai Bushland

AGB (t

ha−1 )

429 395.7 160-209 88-97 56-78 22-34 13-18

Reference Lewis et al. (2013) Lewis et al. (2013) Timberlake et al. (2010) Timberlake et al. (2010) Timberlake et al. (2010) Timberlake et al. (2010) Timberlake et al. (2010)

albedo as a result of expanding boreal forests might offset any positive benefits of carbon sequestration. This further reinforces the climate benefits of tropical forests. Fire plays an important part in the carbon cycle of dry tropical forests and is responsible for large, episodic releases of carbon from the system (Figure 3.3). The study of the carbon dynamics of tropical forests is still in its infancy and there are few tropical forest locations where we can account for the carbon dynamics in detail (Malhi & Grace, 2000). Keith et al. (2009) show that forests that occur in areas with cool temperatures and moderately high rainfall are some of the most carbon dense in the world, with above ground carbon values of about 1800 t ha−1 (living plus dead material). The cool temperatures and high precipitation produce fast growth rates with relatively slow decomposition. Decomposition in tropical forests would be faster due to the high temperatures. The AGB of African tropical forests varies considerably, with large differences between moist tropical forests and dry deciduous tropical forests. Within deciduous tropical forest and woodland there is also considerable variation, as shown in Table 3.1, which also includes maximum values from central African Congo forests which represent the highest AGB of any forests in Africa. In a study of Amazonian rainforest, Malhi & Grace (2000) estimated that carbon has a residence time of about 16 years in above ground biomass and 13 years in soils, to give a total of 29 years before it is returned to the atmosphere. There are few studies of carbon residence times in dry tropical forests, but it would be expected that the residency of carbon in the soil may be longer, given that decomposition takes place slower in a dry environment, or potentially much shorter if fires are frequent (Chidumayo

SFM FOR L OCAL AND G LOBAL O UTCOMES

57

Figure 3.3: The Forest Carbon cycle. In accounting for net CO2 sequestration by photosynthesis in forests, it is necessary to factor in CO2 production via litter and soil respiration, as well as from fire, tree mortality and harvesting. Fire is an important factor in the study area. Since black carbon can have long periods of residence in the soil, accounting for the role of fire in CO2 emission should consider the residence time of black carbon in the soil. (Author’s own diagram)

58

C HAPTER THREE

& Gumbo, 2010). Taking into account these potential differences in carbon flux on a year to year basis in order to account for carbon in a REDD+ project will be an extremely challenging task in any area. In western Zambia, which is remote and difficult to access, remotely sensed imagery, coupled with targeted local studies, will be the only practical way of monitoring carbon flux. In attempting to quantify the amount of carbon stored through net primary production, Malhi & Grace (2000) make the important point that most current estimates of this value ignore or underestimate the role of the roots and below ground carbon of living trees. I examine this problem with reference to the the dry tropical forests of Africa on page 58. Tree growth rates are an important component of carbon sequestration and growth rates are determined by a combination of genetic factors, climate, soil type and levels of disturbance, of which fire is the most important in dry tropical forest. The relationship between age of regrowth and AGB appears to be linear in dry miombo woodland (Figure 3.4), although the coefficient of determination (r2 = 0.49) indicates that further data are required to confirm this relationship with more confidence. Increments in the diameter of tree species vary widely in the southern African region (Chidumayo & Gumbo, 2010), and it seems there is little point in using these figures to infer growth rates unless restricting these observations to a particular location. Jeffers & Boaler (1966) found that humidity and minimum temperature were the most important factors determining the growth rate of Pterocarpus angolensis which varied between 0.8 mm and 4.8 mm per year. This emphasises the need for local measurements of growth rates (and indeed climatic conditions) so that regional variation in climate is accounted for when measuring carbon sequestration. Roots, Below Ground Carbon and Climate

In forests which are adapted to survive long dry seasons, some trees have deep root systems in order to reach underground water reservoirs. Acacia erioloba (which in some areas occurs in close association with Baikiaea plurijuga) has roots which penetrate to 60m in the central Kalahari (Canadell et al., 1996). Baikiaea plurijuga itself produces a long tap root soon after germination and has been shown to have a cone of root exploitation about 9m deep and 12m in diameter for trees of 40cm DBH and 18m in height (Calvert, 1986a). This root system represents significant below ground biomass. Rooting depth in plants in semi-arid areas has been successfully modelled using a combination of plant allometry and ecohydrological optimisation models which take into account soil type, the amount and timing of rainfall and the effect of temperature and humidity on potential evapotranspiration

SFM FOR L OCAL AND G LOBAL O UTCOMES

59

Figure 3.4: Woody Above Ground Biomass (AGB) accumulation in Kalahari and Somali-Masai woodlands and wet miombo, dry miombo and Sudanian woodland (Redrawn from Chidumayo & Gumbo (2010))

60

C HAPTER THREE

(Collins & Bras, 2007; Schenk & Jackson, 2002b). Optimal rooting is deepest in sandy soils and where large amounts of rain are deposited in infrequent storms (Collins & Bras, 2007) but the depth of rooting is ultimately determined by the interaction of these factors with the depth to which water penetrates and the timing and magnitude of water input and evaporative demand (van Wijk, 2011). Many plants in seasonally dry habitats develop underground storage organs. Laden & Wrangham (2005) compared the biomass of edible underground storage organs (rhizomes, tubers and corms) eaten by the Hadza people in the savanna and dry forests of Tanzania to that available to forest people in the Ituri Forest in the Central African Republic. The Tanzanian habitat contained greater than 400 times the biomass of edible underground storage organs compared with the rainforest of central Africa. Underground storage organs form an important food source for the San people of the Kalahari, and the utilisation of this resource may have formed an important step in human evolution in east and southern Africa (Laden & Wrangham, 2005). Further evidence for the abundance of this resource comes from examining the distribution of the five genera of Batherygid mole rats, which feed exclusively on underground storage organs (Burda, 2001). They are restricted to non-rainforest regions of central, east and southern Africa and have never been recorded from the rainforest regions. Although some studies have identified the plant species which have large underground storage organs in the Kalahari region (Hargreaves, 1996), there has not been an adequate investigation of this carbon store to begin an assessment of its importance in African dry tropical forests. Estimating the scale and importance of below ground biomass has hitherto not been an active area of research relative to the considerable effort undertaken to measure above ground biomass. However, an increasing body of work indicates the potential importance of below ground biomass as a carbon store in tropical dry forests because of the higher root biomass relative to tropical rainforest (Canadell et al., 1996; Collins & Bras, 2007; Jackson et al., 1996; Schenk & Jackson, 2002a, 2005; van Wijk, 2011). Robinson (2007) estimates that the global root carbon pool is about 68% larger than originally thought and the geographical distribution of this carbon pool is heavily skewed towards the areas where dry tropical forests are found. Since dry forests account for some 42% of the world’s tropical and subtropical forests, and 70%-80% of the forested area of Africa (Murphy & Lugo, 1986), the measurement of below ground carbon should be attracting more attention than it does at present.

SFM FOR L OCAL AND G LOBAL O UTCOMES

61

The importance of large swathes of deep rooted forests to global atmospheric circulation has recently received attention by some researchers. Using models of general atmospheric circulation, Kleidon & Heimann (2000) have suggested that deep rooted forests are important to the maintenance of tropical circulation patterns. The mechanism by which this occurs in shown in Figure 3.5.

 

           



  

      

 

        

 



    

Figure 3.5: This diagram shows the mechanism by which deep rooted trees influence atmospheric circulation. The trees have access to groundwater during the dry season and are able to continue transpiration well after the rains have stopped. This results in local cooling and more moisture (and therefore energy) being transported to the intertropical convergence zone resulting in enhanced precipitation in the wet season hemisphere. This process results in more convection and invigorates the general circulation (Authors own diagram) The process will be accentuated where deep rooted species are evergreen. Although Baikiaea plurijuga is deciduous, the trees maintain their leaves for longer than any other species in the region, and are the first to come into leaf before the rains start, and in some areas many trees never completely lose their leaves (Childes, 1988; Martin, 1940). However, the extent to which Baikiaea plurijuga forests play a part in regional atmospheric circulation is speculative at present, and none of the work on deep rooted forests has

62

C HAPTER THREE

examined forests in Africa. Currently fewer than 20% of plant ecology studies consider roots (Wilson, 2014) which makes it unlikely that REDD+ projects will be able to draw on the basic science which is necessary to be able to account for below ground biomass. The issue of below ground biomass is one of many challenges faced in trying to establish a global system of payments for carbon sequestration. In the next section I consider some of the political, scientific and economic problems faced by REDD+ and place these in the context of carbon sequestration in the dry tropical forests of western Zambia.

Reduced Emissions from Deforestation and forest Degradation (REDD), Carbon and Climate Change To a large degree, the global public goods that come from tropical forests have not been translated into economic value either to governments of forested countries or to people who depend on forests for their livelihoods (Douglas & Simula, 2010b). As a result SFM has not been implemented successfully on a scale large enough to have an impact on reducing CO2 emissions. The potential solution to this problem are the proposals put forward under the UN-REDD programme which propose to pay for carbon sequestered by or stored in tropical forest biomass. Reduced Emissions from Deforestation and forest Degradation (REDD) is not a clearly defined set of methodological guidelines, but rather a broad set of approaches, actions and proposals that aim to reduce emissions from deforestation and forest degradation (Verchot & Petkova, 2009). In a formal sense it is an international financing mechanism for reducing carbon emissions which was negotiated under the UNFCCC, but the widespread use of the acronym REDD as a proper noun belies any apparent single interpretation of the approach. In fact, many of the central issues concerning implementation are highly contested. Nevertheless the core concept is simple: given that deforestation and forest degradation account for around 20% of carbon emissions worldwide (Stern, 2006), payment for reducing these emissions will lead to a reduction in emissions from this sector. REDD can be broadly conceptualised as a global system for making payment for ecosystem services. Payments will be made by buyers to providers of an environmental service, the main component of which will be carbon sequestration from the atmosphere by trees, although REDD+ includes many other criteria, most notably social benefits which must accompany the provision of this core service. The increasingly widespread use of the acronym REDD+ is intended to bring attention to a more holistic vision of REDD which includes conservation

SFM FOR L OCAL AND G LOBAL O UTCOMES

63

of biodiversity, sustainable forest management practices and the equitable treatment of indigenous people and communities in the process of reducing emissions from deforestation and forest degradation. I use the term REDD+ because this is becoming the most widely accepted vision for what REDD needs to achieve, although the acronym which describes the official UNFCCC programme does not use the extended version REDD+. History and Design Issues for REDD+

The proposal for REDD+ to be included as one of the strategies for atmospheric carbon reduction has its genesis in global climate negotiations and particularly the Kyoto Protocol. At the UNFCCC Conference of Parties (COP) 13 meeting in Bali in 2007, a decision was made “affirming the urgent need to take further meaningful action to reduce the emissions from deforestation and forest degradation in developing countries” (United Nations, 2007). This led to discussions on REDD+ being included in subsequent meetings with a deadline set for the UNFCCC meeting in Copenhagen in 2009 for reaching an agreement on how REDD+ was to be implemented in the short to medium term. Since the well publicised failure of REDD+ to be formally included in the Kyoto Protocol during the UNFCCC COP 15 Copenhagen talks, the details about how REDD+ is to be implemented have gathered controversy and become the object of intense negotiation in order to seek a compromise. The main points of agreement include (Verchot & Petkova, 2009): 1. All parties agree that REDD+ can form an important part of the mitigation efforts of developing countries 2. There is agreement on the conceptual level with respect to REDD+ in that any implementation must include social and other co-benefits, broad participation, sustainable forest management and that the issues of permanence and leakage must be addressed 3. There is agreement that REDD+ policy must be based on measurable and verifiable carbon emissions reductions 4. Parties agree that the process be implemented at a national level rather than at subnational levels.

Disagreements nevertheless continue to create obstacles for REDD+ to be included into a legally binding global climate agreement after the Kyoto Protocol expired at the end of 2012. Many of the issues are related to the real difficulty and complexity of implementing REDD+ on a global scale, while other disagreements stem from ideological and political differences with respect to financing REDD+ and governance and tenure issues.

64

C HAPTER THREE

There are high financial expectations in many forested countries from the transfer of funds as part of the REDD+ process (Phelps et al., 2011). The rejection of the project-based approach on the part of developing countries in favour of government-based national implementation at the COP 16 meeting in Cancun in November 2010 (Bosetti & Rose, 2011; Karsenty et al., 2014), led to questions about the extent to which some countries are negotiating to maximise future revenue for government departments rather than being realistic about what is achievable. This is especially true for countries which lack capacity in the short to medium term. Zambia has neither the technical capacity to implement a countrywide REDD+ programme (Kowero & Spilsbury, 1997), nor has the government demonstrated the political will to address the changes in forest tenure and community participation in forest management that are essential to the success of REDD+ (Leventon et al., 2014). As an example of this lack of political will, The Forest Act (1999) which addresses some of these issues, allowing for Joint Forest Management and sharing of revenue between government and communities, has been waiting for final parliamentary approval for over ten years (Chundama, 2009). At its heart, REDD+ is an incentive based mechanism which draws on the marginal theory of value to represent the conservation of forests as a rational choice which is made relative to a choice between other income generating uses of forested land, and solely dependent on their relative value (Turner et al., 1994). It is suggested that governments of developing countries would pay an opportunity cost to forego the utilisation of their forests and therefore REDD+ programmes must provide a payment which is higher than the foregone opportunity cost of cutting down forests (Gregersen et al., 2010). There is a growing body of criticism of this economic approach. Karsenty & Ongolo (2012) question the assumption that governments, and particularly governments of fragile or failing states, are economic agents who behave rationally. The idea that governments which have systemic problems with corruption and a poor record of transparency and service delivery in general, would (a) make a decision about a major development agenda based only on a cost-benefit analysis and (b) deliver a reduction in carbon emissions from deforestation based on this decision, is optimistic at best. Much of the disagreement around how to fund REDD+ arises as a result of assumptions under marginal economic theory about how choices are made by these governments. In many cases it seems it is this disparity which is the “elephant in the room” when disputes about fundamental issues arise, such as whether a market-based or fund-based approach to funding REDD+ is preferable. The ranking of Zambia by various international organisations in measures of transparency, governance and corruption is presented in Table

SFM FOR L OCAL AND G LOBAL O UTCOMES

65

Table 3.2: Measures of competitiveness, corruption and development for Zambia Indicator

Index/Rank

Source

Global Competitiveness Index Institutional Competitiveness Infrastructure Competitiveness

108 (out of 144) 56 (out of 144) 111 (out of 144)

Schwab, et al. (2012)

Corruption Index (0 = most corrupt; 100 = least corrupt)

37

Transparency International, (2012)

Life Expectancy Gross National Income per capita

49 years USD1490

World Bank, (2012)

Human Development Index

164 (out of 187)

UNDP, (2012)

3.2. The possibility of the Zambian government acting as a rational economic agent, with respect to the decisions which need to be made to reverse one of the highest deforestation rates in the world, seems unlikely when these indicators show that few other development goals have been met. The World Bank assessment for 2012 (World Bank, 2012b) indicates that: “Despite strong economic growth in the last decade, Zambia has made very little progress in reducing poverty and providing basic opportunities for children still remains a challenge.” Government decisions are often self serving and are based on many factors of which a cost benefit analysis may be only one. Large amounts of money flowing into government departments will not necessarily deliver the reductions in carbon emissions which are required under REDD+. According to the International Budget Partnership (2012) report on Zambia, which examines budget processes and budget policies: “Zambia is among the worst performers in budget oversight and engagement in southern Africa” which does not bode well for the establishment of a fund-based REDD+ programme. Nevertheless these factors do not mean that project-based REDD+ programmes are not feasible, and in fact there is a precedent for project-based conservation initiatives working successfully in Zambia (for example Kasanka Trust Zambia (2013) and Frankfurt Zoological Society (2013)). A related problem arises from attempting to place the correct cost on the

66

C HAPTER THREE

opportunity which is lost through foregoing deforestation. Gregersen et al. (2010) highlight an issue which is especially relevant with respect to Zambia, by indicating that it is the perceived opportunity cost by the recipient that is important when providing an incentive not to cut down a forest. If a farmer needs to plant crops for the forthcoming year the opportunity cost of not doing so is very high when weighed against payments from a government which has seldom, if ever, provided him with anything. The difference between this perceived cost and the cost calculated by an economist who assumes a particular government will make rational economic decisions would lead to the economist undervaluing the perceived opportunity cost to the farmer responsible for cutting down the trees. The solution proposed by Karsenty & Ongolo (2012) is not to abandon incentive mechanisms, but to operate them at the scale at which they were identified in marginal economic theory and thereby avoid some of the problems which arise when working at a government level. This can be achieved through smaller scale project-based REDD+ initiatives which deliver real benefits to local communities. The question of whether REDD+ is going to be fund-based or market based is a complex one, and it will more than likely emerge as a mixture of both, especially in the short to medium term (Clements, 2010), and will differ for different countries. Capacity building of forest institutions and policy changes are amenable to being funded by development aid, or a fund-based approach, while Clements (2010) is of the opinion that delivering the long term sustained flow of incentives at the scale at which they are required can only be achieved by a compliance-based market approach. However, this is not necessarily the case in every situation and there are possibly circumstances where a fund could pay for the establishment of a project on behalf of a community which may take carbon to market once they have the capacity to do so. Sandbrook et al. (2010) discuss a more fundamental problem which appears to make the market based funding of REDD+ through global carbon markets incompatible with allowing more meaningful local participation and devolution of forest tenure to local communities. The substantial inflows of money which are proposed under REDD+ (which in many cases, and almost certainly in the case of Zambia, exceed the entire budget of the Forestry Department), if not conditional upon attending to local governance issues, will result in an increased tendency for governments to centralise control. An increasing accumulation of evidence (Dietz et al., 2003; Ostrom, 2009a) shows that local governance is essential for the sustainable management of common property such as forests. The potential paradox is that if carbon markets have the effect of undermining local governance, preventing changes

SFM FOR L OCAL AND G LOBAL O UTCOMES

67

in forest tenure and investments in local capacity and slow governments changing legislation which gives clear definitions of property rights, then they will fail to produce the conditions that lead to emissions reductions. This is another reason that a nested approach would work better than a national approach to REDD+ implementation. Projects within-country require these changes to be put in place before they can be successful and before investors will commit substantial funding. Therefore, governments will be compelled to enact changes before projects can begin and the delivery of emissions reductions can start. Currently the voluntary market has a diverse, but nevertheless well established set of verification criteria which include meeting standards for social and governance measures and these could easily be included in a REDD+ verification framework which is implemented for individual projects. Therefore the apparent paradox which Sandbrook et al. (2010) highlight can be resolved with the correct structuring of REDD+ implementation. However, these authors are correct to be concerned about the impact of carbon markets on REDD+ in the context of the UNFCCC negotiations in which participants appear to emphasise national programmes in preference to a nested or project-based approach (Bosetti & Rose, 2011). The question of MRV is immensely complicated and as such it will probably be implemented differently for different countries which have different capacities for MRV and may require different methods for different forest types (Peskett & Brockhaus, 2009). The level at which MRV should take place is under discussion by parties to the UNFCCC. One of the issues under discussion is whether MRV should be undertaken by a national organisation or should be the responsibility of an international body such as the UNFCCC. The question of who performs this process is largely a political one, but there are real methodological issues with respect to the methods used for making an inventory of carbon (Herold et al., 2015). This is further complicated by the question of which carbon pools to include and the extent to which this can be practically undertaken in different forest types (Verchot & Petkova, 2009). Measurement of above ground carbon, below ground carbon, soil carbon, dead wood and litter will require an extraordinary effort on the scale at which REDD+ needs to be implemented. Few of these issues have been resolved, and I hope that this study will highlight what is possible and practical in south western Zambia. The biomass inventory method which uses standard forest enumeration techniques, measuring Diameter at Breast Height (DBH) and converting these measurements to estimates of living above ground biomass using allometric relationships, is a simple technique with relatively low cost that can be used to monitor changes in carbon stocks over wide areas (Malhi et al.,

68

C HAPTER THREE

2002). Used in conjunction with remotely sensed land use mapping and Geographic Information System (GIS) it becomes a powerful technique to account for changes in standing carbon over time, and will more than likely be the technique which is used in many African countries. The method needs some research to perfect-for example, allometric equations still need to be developed for many species-but this is routine forest enumeration work and represents quite modest research ambitions for national Forestry Departments if REDD+ implementation is to be seriously implemented. The potential of Light Detection and Ranging (LiDAR) technology to provide a quicker and more accurate method to model forest structure has been reviewed by Dubayah & Drake (2000) and recently by Akay et al. (2009) who all conclude that the technique has wide application for forest management and biomass assessment. The recent map of global canopy height produced using the Geoscience Laser Altimeter System (GLAS) aboard ICESat (Ice, Cloud and land Elevation Satellite) is an impressive demonstration of what is possible using LiDAR from space (Simard et al., 2011) but the technique has limitations of spatial resolution (1 km) and the range of sampling and prediction errors with respect to the area and forest types under consideration in this study. A key question for the study area, and for dry tropical forests in general, is the importance of below ground carbon. There are important potential implications for expanding the scope of MRV to include below ground carbon in this region. Measurements of above ground biomass in this habitat are considerably lower than those for tropical rainforest, but may be underestimating relative below ground biomass by two or three orders of magnitude (Laden & Wrangham, 2005). Nevertheless, despite the evidence for the importance of below ground carbon, the inclusion of this carbon pool in any REDD+ accounting scheme for Zambia would not be practical until further research is able to conclusively establish its importance. It is likely to vary substantially across different forest types and different soil types and represents a complicated carbon pool to assess. Soil carbon and its association with emissions related to agriculture would be an important carbon pool to measure in Zambia, especially where rural agriculture is an important source of deforestation and land use change. Several authors have measured a significant and sustained reduction in soil organic matter and nitrogen where deforestation followed cultivation in Miombo woodland in Zambia (Chidumayo & Kwibisa, 2003) and Malawi (Walker & Desanker, 2004). Good soil maps are available for most of Zambia, and it would be relatively simple to measure soil carbon using representative site measurements which are then extrapolated and mapped using GIS in conjunction with soil maps and regularly updated land use maps to monitor

SFM FOR L OCAL AND G LOBAL O UTCOMES

69

changes in soil carbon. Mather (1992) first highlighted the relationship between the area of a country under forest coverage and human population growth rates. As populations increase, forest cover declines rapidly until a point where it starts increasing again in response to slowing population growth rates and changed attitudes to forest conservation. What has become known as Forest Transition Theory (FTT) is often used to depict deforestation patterns in developing countries and illustrate the problem of setting baselines for different countries. The procedures for setting baselines from which emissions are measured is contested, but nevertheless is one of the most important aspects of implementing a REDD+ programme. The baseline is a benchmark which determines whether projects or countries get rewarded for reducing emissions or not rewarded for increasing emissions (Angelsen, 2008b). The main aim of forest researchers in a REDD+ programme is to determine the BusinessAs-Usual (BAU) scenario to predict the rate of decline of forests over time (Angelsen, 2008b). Given the rate of deforestation which is predicted in the BAU scenario, there are various proposals for calculating a baseline, above which countries would be rewarded for decreasing emissions. Using national historical deforestation rates to set the baseline is a popular proposal (Angelsen, 2008b). However, the Zambezi Teak forests are an example of where project specific baselines may be more appropriate. Since the process of deforestation started in the study area long before deforestation reached its current high levels in Zambia, setting a national baseline based on deforestation rates for the last ten years, for example, would ignore the fact that significant deforestation of the Zambezi Teak forests had already taken place by 1996 (JICA, 1996), the date from which widespread deforestation is usually measured in Zambia (Stringer et al., 2012). This is illustrated schematically in Figure 3.6. The setting of a national baseline from a fixed date in the past does not take into account regions where deforestation occurred before that date. On a global scale, Angelsen (2009) shows how different countries will inevitably occupy different positions on the forest transition curve predicted by FTT (Figure 3.7). The problem of setting baselines based on FTT is that countries which have large areas of undisturbed forest and occupy the early stages of the forest transition will receive less compensation than countries which have higher rates of deforestation, despite incurring a higher opportunity cost by reducing rates of deforestation. FTT has some similarities to the Environmental Kuznets Curve (EKC) which postulates a similar relationship between the rise of environmental pollutants and per capita income (Dinda, 2004). The theories are similar in

70

C HAPTER THREE

Figure 3.6: The forest transition curve of Baikiaea plurijuga forests relative to other forested areas in Zambia is shown schematically. The effect of using the average national deforestation rate for the last ten years to set a baseline for Zambia ignores the spatial variation in deforestation patterns between different woodland and forest types. (Authors own diagram)

SFM FOR L OCAL AND G LOBAL O UTCOMES

71

Figure 3.7: Country A has a lower historical rate of deforestation than country B, and a higher percentage forest cover, yet would benefit less from Reduced Emissions from Deforestation and forest Degradation plus (REDD+) payments if baseline emissions are set based on the forest transition curve. Redrawn from Angelsen (2008b)

72

C HAPTER THREE

that they suggest that environmental pressure rises as population growth rate (FTT) or income (EKC) rises and then decreases as population growth and per capita income decreases and increases respectively. The problem with both theories is that deforestation and the increase in environmental pollutants are both complex processes which may not necessarily be attributable to one factor. With respect to deforestation in particular, the process is not necessarily gradual and predictable, but can be erratic based on commodity prices, market demand, war or political upheaval (Angelsen, 2008b). Both theories imply that deforestation or a rise in environmental pollutants is the inevitable price to pay for population growth and economic progress, and that all countries will follow the path from a largely agrarian economy to that of an industrial economy with higher per capita income and lower growth rates. Most disturbingly, especially in the context of SD, the theories seem to suggest that we are powerless to influence this process. Both FTT and EKC are valuable conceptual tools for thinking about the relationship between population growth and economic development and their influence on environmental degradation, but appear to be too general to apply to specific situations. The fact is that the Zambezi Teak forests are at a different position on the forest transition curve to other areas of Zambia, and the drivers of this process have a different history from that which explains the more general high rate of deforestation in Zambia. The unique characteristics of these forests with respect to carbon sequestration, growth rates and drivers of deforestation will be discussed in more detail in later chapters, but they nevertheless provide a clear example of where the setting of a national baseline based on national forest transition curves may not be appropriate. In this case FTT does not provide anything other than a large scale conceptual model, but is not able to meaningfully account for spatial variation in deforestation within a particular country at a national level. It is precisely this complex and variable nature of global deforestation that has led to different proposals for calculating baselines (Sathaye & Andrasko, 2007). One of these is that the countries with a deforestation rate half of the global average deforestation rate use this as a baseline, and those countries with deforestation rates higher than the global average use a national historic baseline. However, these proposals, including the suggestion that the poorest countries receive higher baselines, is more about attempting to solve the political problem concerning the inclusion of REDD+ in an international agreement than establishing a workable system of reward for real reductions in emissions. The debate appears to have similarities with respect to issues around payment for carbon and whether it should be project-based

SFM FOR L OCAL AND G LOBAL O UTCOMES

73

or fund based. Ultimately, only solutions which are able to demonstrate real reductions in emissions are likely to be successful in the long term. The Forests Act (1999)

The first statement of the Forests Act (1999) (Government of Zambia, 1999) in Part I, specifies that The ownership of all trees standing on, and all forest produce derived from, customary areas, National Forests, Local Forests, State Lands and open areas is vested in the President on behalf of the Republic, until lawfully transferred or assigned under this Act or any other written law. (Part I, 3) There are important implications of this statement for REDD+ and successful CBNRM or SFM projects. It places ultimate authority in the hands of the state and defines a narrow view of property rights when there is a need for clarity on different kinds of ownership rights and a clearer establishment of community rights. Rights of ownership are almost never absolute and almost always involve an interplay between rights and responsibilities (Mukamuri, 2009). What is needed is a move away from ownership schemes which emphasise the responsibilities of communities (such as to control or reduce burning) without devolving any rights (Barrow & Murphree, 2001). This stamp of state authority reflects past and, it seems, current thinking on the part of the Zambian Government on the role of the state as a dispenser of privileges to communities in the form of revenue sharing, rather than legally empowering communities with clear ownership rights which have meaningful authority (Leventon et al., 2014). This is at odds with both the requirements for REDD+ implementation and the lessons from nearly 30 years of CBNRM implementation in southern Africa. A major problem is that, at the time of writing, the act has not formally been passed by parliament despite being written up in draft form since 1999. Although the government is capable of issuing Statutory Instruments to similar effect, the lack of a clear mandate to establish Joint Forest Management Areas (JFMAs), for example, without the Act being passed, makes it difficult to implement policy, as the grounds on which actions are established are highly uncertain and leave little room for legal challenge by communities (Leventon et al., 2014). The current licensing system as laid out in the Zambia Forests Act, 1999 is confusing and contradictory, does not support forestry policy and does not result in enough revenue collection for the Forestry Department to function effectively (Whiteman, 2013).

74

C HAPTER THREE

Currently the licensing arrangements are not contained within a single document, but spread across the Forests Act, statutory instruments, licence agreements and policy statements. This results in confusion for both timber concessionaires with respect to which licence they need, and the amount they need to pay. There are contradictions between documents (and sometimes within the same document) and in many cases no guidance is given at all to some categories of use. There is no definition of “personal use”, for example, and leaving this interpretation open to Forestry Department officials creates potential for corruption. The entire licensing system is, through the confusion it generates, open to corruption and is sometimes used to the advantage of the official concerned, who claims to be able to apply certain charges but then accepts a bribe to turn a blind eye to them (pers. obs). More importantly, with respect to the viability of the forests, the current emphasis on volumes of production rather than area of forest under concession is leading to a gradual deterioration of the forest reserve (Whiteman, 2013). The five year time period for which concessions are awarded is too short for the concessionaire to establish any infrastructure and the emphasis is to extract as much value as possible from the area before expiry of the concession licence. The current situation, where forestry tries to measure every tree that is cut in the field, and calculates royalties on the basis of timber volumes, is an expensive way to collect revenue. Charges on the basis of area awarded to a concessionaire would be much easier to collect and require less expenditure on the part of the Forestry Department. However, this requires careful forest enumeration before areas are designated and charges can be set, and the Forestry Department lacks the capacity and expertise to conduct these surveys (pers. obs.). Where volume charges could become applicable is in adjustments to the current category of Conveyance License. Once logs are loaded and ready for transport it is easier and cheaper to calculate the volume at designated check points, and high penalties for illegally conveying logs would serve as a deterrent to illegal transport. The use of so-called ‘shadow permits’ to circumvent rules requiring more stringent environmental compliance for exported timber has recently been highlighted (Global Witness, 2013). Although the report only looks at the problem in The Democratic Republic of Congo, Cameroon, Ghana and Liberia, a similar issue exists with respect to the Pitsaw License in Zambia (pers. obs.). As the name implies, this license category is intended to enable small scale producers to cut timber using a pitsaw. The nature of a pitsaw means that it is difficult to process large amounts of timber and therefore the license has no requirements for an ecological impact assessment and charges

SFM FOR L OCAL AND G LOBAL O UTCOMES

75

are supposed to be minimal. The interpretation of this license category has become progressively more expanded to the point where medium size saw milling operations, capable of producing between 10m3 and 50m3 of sawn timber per month, are operating under a pitsaw licence. Some of these producers are exporting timber. One of the largest furniture manufacturers in Livingstone buys timber from small suppliers who cut under a Pitsaw Licence, all of which is perfectly legal. Since Joint Forest Management (JFM) has many similarities with CBNRM, the success of JFM depends on adhering to the principles of CBNRM which have proved successful. After thirty years of CBNRM work in southern Africa, there is a high degree of certainty about how to implement successful programmes, despite local variations in governance traditions, culture or resource type (Dalal-Clayton et al., 2003). One of the major failings of the Forests Act, 1999 is the lack of clear fiscal devolution to the Forest Management Committee. Dalal-Clayton et al. (2003), working in the Luangwa Valley, found that fiscal devolution to the level of the Village Area Group (VAG) resulted in more efficient implementation of the project and that less than 1% of money was unaccounted for, compared with around 40% loss of revenue at the higher level of the Area Development Committee. The authors introduce the concept of “scale laziness” (the idea that higher administrative scales are less efficient than lower ones) which is emerging as a universal principle of CBNRM in southern Africa, and which appears to be the most important failing in CBNRM projects in Zimbabwe. The requirement for devolution of fiscal responsibility is a recurring theme in regional reviews of CBNRM policy (Jones, 2007). Campbell & Shackleton (2001) state the case more plainly, saying that: “The greater the authority village organisations receive the more likely they are to succeed.” Despite the widespread tendency in southern Africa to decentralise natural resource management (Campbell & Shackleton, 2001), Zambia has largely been the exception to this trend. There has been some success in the Luangwa valley with the Luangwa Integrated Rural Development Programme (LIRDP) (Dalal-Clayton et al., 2003), but even this project has had a tortuous journey during its more than 20 year history. Although there is no panacea when it comes to natural resource management (Ostrom, 2007a; Ostrom & Cox, 2010), SFM and JFM policy in Zambia needs to draw from the many lessons which come from case studies of CBNRM projects in the region (Campbell & Shackleton, 2001; Mukamuri & Manjengwa, 2009). The Forestry Department in Zambia has explicitly rejected an 80:20 revenue split with communities and unless this policy is changed, with clear and explicit laws on fiscal devolution, SFM or JFM in Zambia, it does not stand much

76

C HAPTER THREE

chance of success (Jones, 2007). The current focus on replanting indigenous forests, as outlined in a ministerial statement by the Minister of Lands, Natural Resources and Environmental Protection (Ministerial Statement on the National Tree Planting Programme, 2012), represents a misguided vision of what needs to be done about deforestation, and particularly deforestation in Zambezi Teak forests. Unfortunately the Forestry Department have taken this statement as a policy directive and are actively encouraging timber concessionaires to replant trees. Given the scale of the deforestation problem in Zambia, and the clearly demonstrated futility of trying to replant Zambezi Teak (Calvert, 1986a), it exposes the ignorance of anyone who proposes such a policy as a solution to deforestation in Zambia. To date, the government’s plan to “raise 17,500,000 seedlings in 11 nurseries country-wide” and to “create six thousand jobs” has not materialised. Should there be any doubt about the scale intended for this programme the Minister concludes his statement by saying that: It is my sincere belief that once this programme is implemented, the country will have contributed its share in the reduction of green house gases which are responsible for global warming and climate change with its adverse effects to mother earth [sic]. (Ministerial Statement on the National Tree Planting Programme, 2012) Unless a concerted effort is made to increase technical capacity and address the basic legal issues, SFM or JFM in Zambia faces insurmountable difficulties with respect to implementation of basic forest management and enumeration (Jones, 2007). Climate Change in Southern Africa

The increased release of CO2 , CH4 , NO2 and other greenhouse gasses into the atmosphere since the start of the industrial revolution has contributed to a rise of globally averaged land and sea surface temperatures of 0.85 ◦ C (0.65 − 1.06 ◦ C) over the period 1880 to 2012 (IPCC, 2013). The increase in CO2 is largely due to the burning of fossil fuels, while the increase in CH4 and NO2 is largely due to changes in land use and deforestation. The changes in climate that will result from this increase in average temperature vary in magnitude as well as geographically. Africa is expected to experience a higher degree of climate change than the global average (Collier et al., 2008). While the IPCC (2013) report indicates with “medium confidence” that soil moisture drying on a global

SFM FOR L OCAL AND G LOBAL O UTCOMES

77

scale in dry regions will get worse and increase the chance of agricultural drought, this change is predicted with “high confidence” in the southern African region. Globally, warmer and/or fewer cold days and nights are “virtually certain” in the late twenty-first century and it is “very likely” that these are as a result of human induced changes. Similarly, the report indicates that warmer and/or more frequent hot days are “virtually certain” in the late twenty-first century. From these predictions we can assume, with a high degree of certainty, that the study area will experience drier conditions with decreasing soil moisture, fewer cold nights and more hot days with a general trend for the cold nights to be warmer and the hot days hotter in the future than they have been in the past. Collier et al. (2008) indicates that a 10%20% decrease in rainfall can be expected and that average temperatures will increase by 4 ◦ C in southern Africa, and that this will have severe effects in the Zambezi drainage basin. The climate of the African continent, and of southern Africa in particular, is highly variable, particularly with respect to rainfall (Hulme et al., 2001). A key challenge for climate scientists is to isolate natural variability from human induced greenhouse gas increases in order to predict the effects of climate change on the African continent (Hulme et al., 2001). Key drivers of local variability are the El Niño Southern Oscillation (ENSO) and land cover change. It appears that Africa is unique in the effect that land cover change has on local climate. In a General Circulation Model (GCM) Xue (1997) linked land cover change to local processes which interact with the upper troposphere, and was able to account for rainfall anomalies in the Sahel and further south in the October-November-December rainy season over 40 years. The removal of large amounts of vegetation, and the accompanying increase in the surface air temperature, results in reductions of rainfall, runoff and soil moisture, and an increase in the albedo, or reflection coefficient of the land surface. There is a feedback process in which the reduction of total diabatic heating rate of the air (heating which is caused by a warm earth surface or the sun), caused by the increase in albedo, results in an increase in the subsidence of air in the upper troposphere which is consistent with changes in rainfall. Feedback occurs when decreases in rainfall result in decreases in vegetation cover. The effect of the El Niño phase of the ENSO is to cause a decrease in precipitation and an increase in temperature in southern Africa (Nicholson & Entekhabi, 1986). Nevertheless the magnitude, consistency, seasonal timing and duration of the rainfall response to ENSO varies regionally (Nicholson & Kim, 1997). Predictive models of climate change remain uncertain because of the influence of the above mentioned factors in determining climate in

78

C HAPTER THREE

Africa. Nevertheless, this does not negate the usefulness of these models, as modelling is as much about quantifying the influence of different inputs as it is about predictive accuracy. A more serious concern is the difficulty of downscaling GCMs to spatial and temporal scales which are relevant for management at the landscape scale (Hulme et al., 2001). Downscaling the model does not escape the fact that local climates are still dependent on the large scale processes included in the GCMs and present only one of a range of possible scenarios. The influence of climate change on Zambezi Teak forests must be considered with these caveats in mind. Nevertheless there have been major advances in climate modelling over the last 20 years and current models are a reliable guide to future climate change, at least at the continental scale. In this study I examine the effect of these continental or regional scale changes in climate on the vegetation of south western Zambia in general and the Zambezi Teak forests in particular. The effects of climate change on Africa are likely to be severe because of the high reliance on subsistence agriculture among the populations of the affected regions and the limited human and institutional resources which provide the capacity to adapt (Collier et al., 2008). On a macroeconomic scale Zambia, like many African countries, is still dependent on a narrow range of exports (mainly copper) which limits the opportunities for economic adaptation in the face of climate change. In Zambia generally, and in the study area in particular, soils are relatively infertile and rural agriculture is already vulnerable to the existing variation in rainfall. In addition to economic weakness, the capacity to adapt is restricted by weak regional integration and the failure by governments to build institutions which study, predict and assess the risks of climate change and design plans for the future (Collier et al., 2008). Maize is the most important crop grown in the region for feeding both humans and livestock. Jones & Thornton (2003) have modelled the effects of climate change on maize production in southern and eastern Africa and predict average decreases in yields due to climate change of about 10% by 2055, although there is large spatial variability to this decrease. Collier et al. (2008) indicates that maize production may not be possible in large areas of Zimbabwe and South Africa once climate change starts affecting the region. Lobell et al. (2008) show that southern Africa is a “climate risk hotspot” with respect to crop production and that maize, wheat and sugarcane are all likely to experience losses in production by 2030 unless there is investment in adaptation through development of different crop varieties or expansion of irrigation. In the context of SD neither of these options are attractive. Zambia does not currently allow the import or growing of genetically modified maize

SFM FOR L OCAL AND G LOBAL O UTCOMES

79

and the expansion of irrigation is likely to contribute to higher CO2 emissions in the long term, further exacerbating the effects of climate change. Rural communities often depend on NTFP to supplement income and food supply in times of scarcity and these products act both as a safety net and as an alternative source of cash if there is no food surplus to sell (Shackleton & Shackleton, 2004; Shackleton et al., 2011). Profits from the trade of NTFP can commonly comprise 40%, and sometimes as much as 80%, of household income (Barany et al., 2001; Eastaugh, 2010). The supplementation of the diet with protein from bushmeat and insects, fruits which provide vitamins and minerals and medicinal plants which cure disease or relieve its symptoms, makes an important contribution to community health. It is against this background that the effects of climate change on forests become important. The extent to which rural African populations are dependent on forest resources makes them vulnerable to changes which affect the ability of a forest to supply this environmental service (Eastaugh, 2010). Where NTFP are harvested for local use the demand is unlikely to pose a threat to the resource. Commercial trade in NTFP however, can result in over harvesting and unsustainable use of the resource. The recent commercial harvesting of Devil’s Claw (Harpagophytum spp.) in south western Zambia is an example of unsustainable quantities of a resource being extracted for trade, resulting in the Zambia Forestry Department banning the export and extraction until a regulatory framework can be set up to monitor its use (Stewart & Cole, 2005). The effects of climate change on forests and the provisioning of NTFP are therefore linked to the changes in ecosystem function that climate change will bring and the effect this has on forest survival. Poorer crop yields will require more land to be cleared for agriculture, a drier climate will result in more frequent fires which result in conversion of forest to grassland, and reduced rainfall may see some forest types which are at the southern edge of their distribution, or locally adapted to withstand current conditions (e.g. Zambezi Teak forests), disappear completely thus reducing ecosystem biodiversity. Although dry deciduous forests are fire adapted systems, increases in fire frequency and intensity are implicated in feedbacks between vegetation and climate and are a proximate cause of the conversion of tropical wooded savanna to grassland (Bond & Keeley, 2005; Hoffmann & Jackson, 2000). Understanding the effects of climate change, even at a broad spatial scale, is important for planning possible adaptation to future conditions. In this study I examine the possible effects of climate change on forests in the region with particular reference to dry deciduous forests. Leaf phenology varies widely in these forests and is a response to drought (Singh & Kushwaha, 2005). Local

80

C HAPTER THREE

soil conditions, species specific variation in other adaptations to drought and variation in rainfall across the study area result in a variable pattern of leaf senescence and emergence across the region. Given that predictions for climate change are for higher temperatures and lower rainfall, there are clearly implications for leaf phenology with respect to climate change. There are indications from theory and practice that building resilience into human and ecological systems enhances their ability to cope with climate change (Tompkins & Adger, 2004; Walker et al., 2006). The adaptive capacity of a community is dependent on social capital which is developed through participation in natural resource management. The social experimentation and granting of epistemological status to policy mentioned on page 26 form part of a larger process of democratic idea making that has a distinguished history (Menand, 2002). Building community resilience is achieved through expanding networks of dependence and engagement, and this is achieved through building institutions which are participatory and facilitate social learning. Normal governance procedures tend to follow more rigid decision-making processes and the erection of institutions which incorporate principles of SD are more conducive to the group learning which can be incorporated into management processes (Tompkins & Adger, 2004). Wider participation also improves the management of Common-Pool Resources by extending networks which improves monitoring (Steins & Edwards, 1999). The existence of civil society institutions plays a role in enhancing adaptive capacity, but in many African countries (and Zambia especially) these organisations are viewed with suspicion by governments because they require and demand changes in governance which frequently translates into a devolution of power from the centre to the periphery (Brown et al., 2010). The ability of people who depend on natural resources for their livelihoods to adapt to climate change therefore, depends not only on their own considerable suite of skills and capacity for survival, but on wider governance and institutional arrangements (Jones, 2007). In the case of Zambia they are not able to participate in forest governance or in structuring institutions. This is a concern for the adaptive capacity of forest communities. It is likely that adaptive capacity in the past was stronger and has been eroded by the impact of the HIV/AIDS epidemic, financial pressures to pay for the education of children and the erosion of traditional authority which, although not democratic in a Western sense, sometimes has effective institutions for allowing people to participate in governance. I examine the implications of resilience theory on ecological and social systems in the next section and outline some of the implications for forests in particular. Vincent (2007) makes particular reference to the importance of scale when making assessments of uncertainty in adaptive

SFM FOR L OCAL AND G LOBAL O UTCOMES

81

capacity and the scale-based issues which emerge repeatedly in this study are brought together in the discussion in chapter eight.

Resilience, Ecology and Society The insights which resilience theory have provided in ecology represent some of the most significant advances in ecological theory since papers by Clements (1916, 1936) on succession and climax, and the Lotka–Volterra model and its associated differential equations used to describe predatorprey systems (Lotka, 1920). The theories of Clements and Lotka–Volterra underpin a conceptual view of changes in ecological systems as deviations from an equilibrium state represented by a climax. Much of this ecological theory has been drawn from ideas originating in theoretical and applied physics with a view of ecosystems which is more analogous to a simple chemical reaction than a complex system such as a forest, savanna or tundra. However, in 1973 C.S. Holling, an ecologist who had already done seminal work on the use of the Lotka-Volterra equations to show the functional response of prey to predation by predators (Holling, 1959), articulated an alternative view of ecosystem function in a landmark paper published in the journal Annual Review of Ecology and Systematics (Holling, 1973). He proposed a shift from an equilibrium centred view, to an emphasis on the conditions required for persistence, and identified resilience as a property of ecosystems. He defined it as the capacity of an ecosystem to return to its original state following a perturbation. In returning to this original state, the ecosystem should maintain its essential function and structure as well as its taxonomic composition and ecological processes. These ideas have been extensively developed by Holling himself (Holling & Gunderson, 2002) and other authors (Folke et al., 2004; Gunderson & Holling, 2002; Peterson et al., 1998; Walker et al., 2002; Walker & Salt, 2006). They incorporate ideas from complex systems theory (Cumming & Collier, 2005; Holland, 2006) and have led to the long overdue (especially with respect to African ecosystems) emergence of the idea of social-ecological systems which include human influences as an integral part of ecosystem function (Folke, 2006; Holling, 2001; Ostrom, 2009a). Ecology has not been the only field to change as a result of the insights provided by resilience thinking, but as Kinzig et al. (2006) state: “The last three or four decades have fostered a revolution in the way scientists think about the world: instead of orderly and well behaved, they now view it as complex and uncertain” (Kinzig et al., 2006, 1). In the next two sections I examine some of the implications of resilience thinking for ecosystems in general and forests in particular.

82

C HAPTER THREE

Complexity and Regime Shifts

Complexity and the study of complex systems are central to ideas which encompass resilience thinking. A complex system, by its nature is difficult to define. Although we can list the characters of complex systems, some systems may show some of these more prominently than others (Cilliers, 2000). Nevertheless some of the properties which characterise complex systems (Cumming & Collier, 2005; Liu et al., 2007), are listed below: • Complex systems comprise a large number of simple elements which interact dynamically. • The effects of these dynamic interactions are propagated throughout the system and are nonlinear, which means that small causes can have large effects, and vice-versa. • There are many direct and indirect feedback loops resulting from interactions which include both positive and negative feedbacks. • Complex systems have a memory which is distributed through the system and which is rooted in its history and therefore history is important to the behaviour of the system. • The behaviour of the system is determined by the nature of the interactions, and because these interactions are dynamic, propagate feedbacks, and are nonlinear, the results of these interactions cannot be predicted from the study of the components. The property of emergence, itself a characteristic of complex systems results from these interactions. • Complex systems are self organising and adaptive in the way that they react to external phenomena. • The issue of scale is central to complex systems. This is particularly true of ecological systems, which are hierarchic and exhibit different emergent properties at different scales. An explicit consideration of different scales is essential when analysing complex ecological systems

It is the complex nature of ecosystems which means that attempts to simulate interactions by means of differential equations, which assume linear relationships between dependent and independent variables, are not only impractical but also do not enhance our understanding of the system (Cilliers, 1998). So if ecosystems behave as complex systems how do we study them? With increasing human impact in all ecosystems at a local and landscape scales, and with the potentially widespread nature of this impact through

SFM FOR L OCAL AND G LOBAL O UTCOMES

83

climate change, it is important we attempt to understand these systems and how they respond to perturbation. A property of all complex systems studied is the existence of lever points which comprise a simple intervention that creates a lasting effect (Holland, 2006). In ecological systems an alternative state, for example when a forest changes into a grassland, is induced by biotic or abiotic phenomena which have the effect of producing new types of feedback relationships which drive the system towards an alternative domain of attraction. A different example might be the way a vaccine produces a lasting change in the human immune system. A small intervention produces a lasting and permanent change in the system. Unfortunately there are no rules for discovering what these lever points are in an ecosystem (Holland, 2006). In the case of Zambezi Teak I use a combination of long term studies of ecosystem function, the effects of past human intervention and the results of stochastic events to infer both the nature of alternative stable states and the lever points which result in the ecosystem changing state.

Forests and Resilience

Forests may be usefully conceptualised as complex, self organising systems, with internal and external feedbacks which determine the state in which the system exists (Messier et al., 2013). Resilience theory provides insights into how these systems behave as well as the attributes which confer resilience on the system (Hirota et al., 2011). These include, but are not necessarily limited to, diversity, system memory, the ability to self organise, forest hierarchical structure and the nonlinear processes which maintain their state (Cumming et al., 2013). Central to this view of ecology is the recognition that biotic and abiotic processes can develop mutually reinforcing relationships over a range of scales (Holling, 2001). Complex systems may exist in one or several alternative stable states, and in ecosystems, and forests in particular, these alternative states result in major changes in ecosystem function and identity and are difficult to reverse (Hirota et al., 2011). Primary forests are highly resilient habitats, able to withstand multiple disturbances and adapt to a range of perturbations whilst maintaining their essential function and identity (Cumming & Collier, 2005). However, it is possible that under conditions of a predictable climate, together with variation in local soil conditions and the influence of topography on the microclimate, a system could emerge which exists quite close to the limits of its domain of attraction with respect to an alternative stable state. If the perturbation was such that the adaptive capacity of the main forest species was exceeded, we may expect that a change of state

84

C HAPTER THREE

would result (Folke et al., 2004). The Zambezi Teak forests are potential candidates for such a system and I investigate this possibility in chapter eight.

Common-Pool Resources, Institutions and Governance The developments in Common-Pool Resource (CPR) theory over the last 20 years have produced a dramatic change in how we understand the formation of rules for governing CPR resources. The work of Elinor Ostrom has been influential (Ostrom, 1990, 1999), and this reformulation of CPR theory has drawn on many studies of different CPRs and the communities that govern these in a sustainable manner without the disastrous results predicted by Hardin’s (1968) “tragedy of the commons”. Hardin’s (1968) theory of resource depletion has had an enormous effect on the governance of natural resources, since the only conclusion which emerges from this theory is that state or private ownership is necessary to prevent natural resource depletion (Dietz et al., 2003). Hardin’s (1968) “tragedy” was not referring to the disastrous outcome which he deemed to be the end result of uncontrolled communal use of CPRs, but to the sense of “the remorseless working of things” or the “remorseless inevitableness” of a tragic drama described by the philosopher Alfred North Whitehead (Whitehead, 1926). Elinor Ostrom’s insight is that CPRs are not necessarily depleted in a tragic drama in which humans are swept along, out of control of their own destiny. Communities can come together to formulate rules which restrict and monitor resource use, censor those who break the rules and prevent outsiders from exploiting resources. To be fair, Hardin (1998) admitted that rules which regulated extraction had the potential to prevent overuse of resources. His idealised, theoretical commons would end in ruin, but the mistake made by those who subscribed to the theory of the tragedy of the commons was to assume that this idealised commons was representative of CPR use everywhere. Elinor Ostrom’s insight was to highlight the fact that this sort of commons did not exist and that there were always rules regulating extraction where communities were able to formulate them. Ostrom’s (1990) key insight was to develop a set of eight design principles which characterise institutions which do not deplete communally used CPRs over long periods of time. The design principles which characterise the institutions responsible for long-lasting CPR use are shown on page85. Some 20 years after the publication of these design principles, Cox et al. (2010) reviewed 91 studies that evaluate the design principles and found that they are well supported empirically. Elinor Ostrom’s work was recognised by the award of the 2009 Nobel Prize in Economic Sciences.

SFM FOR L OCAL AND G LOBAL O UTCOMES

85

The criticism that Ostrom’s (1990) design principles are not universally applicable across different resources types, and that as a general principle it is not valid to treat the participants or the resources as homogenous and rational, or without heterogeneity, respectively, has largely been addressed (Ostrom, 2007a; Ostrom & Cox, 2010). There has been a considerable amount of work in developing a diagnostic approach to social-ecological problems (Anderies et al., 2004; Crawford & Ostrom, 1995; McGinnis, 2011; Ostrom, 2009a) so that variation in these factors can be accounted for. The criticism that these general design principles constitute a panacea or a prescriptive approach to solving CPR problems is disingenuous and as Agrawal (2002) explains, Ostrom’s (1990) design principles, “are expressed as general features of long-lived, successful commons management rather than as relationships between characteristics of the constituent analytical units or as factors that depend for their efficacy on the presence (or absence) of other variables.” (Agrawal, 2002, 49). There is clear awareness of the need to assess local variation in social, ecological and institutional factors in order to understand why national or international legislation or other prescriptive approaches may not be suitable for a particular resource type, social system or institutional structure (Gibson et al., 2000a). In some respects the phrase CPR is not sufficiently well defined in most usage, especially with respect to the cumulative hierarchy of rights which the concept of property encompasses. This issue is clarified later, although the scale and level of detail at which CPR theory and problems are discussed justify some loose use of the phrase without implying uniformity of CPRs or the institutional arrangements that govern their use. Ostrom’s (1990) Design Principles for Sustainable Common-pool Resource Management are as follows: 1. Clearly defined boundaries - both the boundaries of the resource, and the individuals, households or villages who are allowed to utilise the resource, must be clearly defined. 2. Congruence between appropriation and provision rules and local conditions - these requirements are diverse and resource specific. They could, for example, involve matching rules for resource use with local cultural traditions, or restricting traditional use of NTFP to the use of traditional collection methods rather than the use of industrial machinery. 3. Collective-choice Arrangements - people who are affected by the operational rules must be involved in modifying them. This participation has become an important component of implementing CBNRM projects and is a widely accepted principle.

86

C HAPTER THREE

4. Monitoring - monitors who police compliance with the rules of resource use, and who assess the resource condition, must be accountable to the people who are using the resource or, ideally, be drawn from the same community. 5. Graduated sanctions - members of the community who are the users of the resource are punished for breaking the rules, depending on the severity and context of the offence. This is done by other members of the community or by officials who are accountable to the community. 6. Conflict-resolution mechanisms - conflict resolution is low-cost, quick and resolves disputes effectively. If disputes always have to revert to the court system, and become drawn out over long periods of time, it is not conducive to effective CPR utilisation. 7. Minimal recognition of rights to organise - the rights of communities to devise their own rules and create their own institutions are not challenged by central government. 8. Nested enterprises - all of the activities which involve administering the sustainable utilisation of a CPR are nested in multiple layers. This principle provides a way of addressing the issue of scale with respect to governance of a CPR. Horizontal linkages between community user groups and vertical linkages between user groups and different levels of government are equally important for success. Social Capital

Some of the criticisms of Ostrom’s (1990) work make the elementary mistake of confusing correlation with causation. The fact that Ostrom (1990) is able to find a general relationship between successful commons management and eight broadly defined criteria says nothing about how this success is achieved. Singleton & Taylor’s (1992) view is that a community either has the capacity to utilise a CPR according to Ostrom’s (1990) principles or does not, depending on the conditions prevailing in the community relating to trust, transparency and what would more generally be called social capital (Coleman, 1988). This begs the question of whether one needs to create the conditions which foster the building of social capital before communities are able to govern the sustainable use of CPRs, or whether creating conditions under which communities are able formulate rules to govern a CPR would result in the building of social capital. The building of community resilience and adaptive capacity (see page 80) to climate change is enhanced by the building of social capital (Tompkins & Adger, 2004) and it could be expected that the participatory nature of formulating rules for governing a CPR would result in the building of social capital. Singleton & Taylor’s (1992) views

SFM FOR L OCAL AND G LOBAL O UTCOMES

87

may be correct, but if the rules and institutions for successful commons management are envisaged as an outcome, then the process of building these institutions and formulating the rules can contribute to building the social capital that will underpin their future successful implementation. In chapter eight I will examine this issue in more detail with respect to western Zambia.

Tenure, SFM and Property Rights

The involvement of communities in forest management has not only emerged as an important principle of CBNRM, but in western Zambia in particular, it will be a sine qua non for the successful implementation of a REDD+ project (Cliggett, 2014; Siangulube, 2007). The success of community participation in SFM is ultimately linked to the question of forest tenure (Simbizi et al., 2014). The lessons learned from more than thirty years of CBNRM in southern Africa are directly applicable to the question of tenure in forests. These lessons are clear: fiscal responsibility for revenue generated from natural resources and the decisions about harvesting these resources must be devolved to the village level in order for CBNRM or SFM or, ultimately, REDD+ to be successful (Bond, 2001; Mukamuri & Manjengwa, 2009; Murombedzi, 1999). In western Zambia the issue of tenure rights is of central importance with respect to REDD+ implementation because of the potential conflict between managing forests for carbon sequestration, and the importance of shifting cultivation to food production in the region. When pitched against the state’s demands for forest conservation, the demand of poor communities for land on which to grow crops has important political implications and therefore often prevails, resulting in forest destruction (Malambo, 2014). The concern that the devolution of tenure rights will result in forest degradation is therefore not a valid reason for governments to resist the process and ignore the considerable skill with which people in western Zambia have conducted their affairs in other areas of community development and traditional administration (see page 115). Siangulube (2007) has shown that the communities of western Zambia have well established traditional conservation practices, the most effective of which is adherence to directives from the Barotse Royal Establishment. In most cases of successful CBNRM the state has a role to play in assessing the viability of the resources and allocating quotas (van Putten et al., 2014). This service not only provides a check on unsustainable exploitation which should address the concerns of government in this regard, but also provides the community with a service which they would almost certainly lack the capacity to provide themselves.

88

C HAPTER THREE

Property Rights

Many of the disputes and the confusion surrounding allocation of property or tenure rights to communities stem from a simplistic view of property and the failure to recognise that control over resources is not an all-or-nothing concept. Property is never solely held by an individual person or institution, nor is it ever held completely in common ownership (Fennell, 2011). Schlager & Ostrom (1992) have provided a taxonomy of property rights that recognises five resource control rights: 1. Access - the right to enter a designated physical property (a forest for example). 2. Withdrawal - the right to obtain the products of a resource (harvest timber, honey, medicinal plants). 3. Management - the right to regulate internal use patterns and to make decisions to improve the resource (control burning, size limits for logging). 4. Exclusion - the right to determine who has access rights and how these may be transferred. 5. Alienation - the right to sell or lease management and/or exclusion rights.

When a purely dichotomous view of tenure or ownership prevails, the institution being called upon to devolve rights (in the case of western Zambia it is the Zambian government) takes the view that it must devolve all rights and therefore loses all control over the resource. This ignores the different scale at which these rights need to be devolved to allow communities to manage resources sustainably. The ability to determine access and withdrawal rights, for example, is essential for a community to draw clearly defined boundaries which is the first of Ostrom’s (1990) design principles for the long term success of CPR institutions. The rights of the community to define boundaries will legitimise monitoring and any graduated sanctions which are imposed on offenders. This does not mean that the government is permanently abrogating the rights to management of the resource (which may be held jointly in the case of a forest), or the alienation rights, which, by virtue of government’s position of ultimate authority, it will almost always hold. The unique position of forestry, particularly in this case Zambezi Teak, is that its harvesting is not amenable to small scale logging or use by individuals (Lescuyer et al., 2013). Successful harvesting requires large scale capital investment in machinery and a marketing and sales capacity which is beyond the reach of the communities living near the forests (Godwin, 2014). Nevertheless, devolving a right to withdrawal for the proceeds from licensed timber operators to communities is no different in principle from the

SFM FOR L OCAL AND G LOBAL O UTCOMES

89

current situation that only allows withdrawal rights for products harvested by individuals (honey, food, medicinal plants). The main difference is that once communities start commercialising the withdrawal of products, an administrative structure is required to regulate this withdrawal and to equitably distribute the benefits. Governance and Institutions

Governance issues are at the heart of successful CPR utilisation. The success or failure of a CPR utilisation programme will depend on the formation of institutions which deliver real benefits to the village level and yet which do not present a threat to the status quo with respect to established government institutions so that they are actually implemented. The two countries in the region which have most successfully introduced national policies for CBNRM, Namibia and Zimbabwe, share a similar history. Their shared history has been important in getting governments to devolve ownership of natural resources (mostly wildlife, but also timber in some cases) to local communities (Jones & Murphree, 2001). The main factor which resulted in the devolution of tenure to communities was the success of the devolution of ownership of wildlife to white commercial landowners (Jones & Murphree, 2001). In the 1970s both countries introduced legislation which allowed landowners to take ownership of the wildlife on their land. Wildlife was formerly the property of the state and the devolution of ownership had a dramatic effect on the viability of wildlife. The vibrant commercial game ranching sector which emerged placed thousands of hectares of land under wildlife use. Before this change in legislation, wildlife had been shot to make a place for cattle ranching. After independence in Zimbabwe in 1980 and Namibia in 1990, proponents of CBNRM used the example of the successful devolution of wildlife ownership to white commercial farmers and persuaded the governments to extend this ownership to rural communities outside the ranching areas. There are many lessons to be learned from these two experiences and there is an extensive literature on Zimbabwe’s CAMPFIRE programme (e.g. Frost & Bond, 2008; Taylor, 2009), and a growing literature on Namibia’s success in establishing community conservancies (Barnes et al., 2002; Jones & Murphree, 2001; Jones & Weaver, 2009). At first consideration the fortuitous opportunity which history provided in Namibia and Zimbabwe, as a lever for changes in the ownership of natural resources, may be disheartening in those countries, like Zambia, where history has not provided this opportunity. However, this need not necessarily be the case. If policy making is seen as a dynamic process, and more

90

C HAPTER THREE

especially if the process of policy making is given epistemological status, changes become more acceptable and even essential as a part of the policy formulation process (Jones & Murphree, 2001). Nevertheless governments often do not see policy making in this context. The desire to entrench power and control is a strong instinct, even in governments with a long history of democratic conviction. For the greater part of Zambia’s independence, the government has subscribed to command and control policies which still have a powerful influence on the style of government Zambia practices today. The Zambian government is reluctant to let go of power (Jones, 2007). There is a rhetoric of community empowerment in legislation, but this is not matched with a political will to solve the many difficulties and iterations of policy making which devolution of ownership of natural resources entails. Given the wealth of information available from CBNRM studies in southern Africa that inform institutional change (e.g. Bond, 2001; Mukamuri & Manjengwa, 2009; Murphree, 2009), as well as the important work that provides the tools to analyse and design institutions (McGinnis, 2011; Ostrom, 2008; Ostrom & Basurto, 2011) there may be the possibility that a policy making experiment that builds new institutions could create a way for both parties to come together for the benefit of the people who live near the Zambezi Teak forests. This will require compromise on both sides, as is often the case when resolving disputes. On page 209 I make a link between forest condition and institutional structure and governance which shows how important it is to structure the institutions with communities at the forefront of SFM.

Fragmented Nature of Theory and Environmental Change The challenge of producing a holistic synthesis is partly due to the fragmented nature of theory on environmental change (Barbier et al., 2010; Lambin et al., 2003, 2001; Walker, 2004). VanWey et al. (2005, 61) summarise the confusing situation regarding theories of environmental change: Research teams composed of scholars from multiple social and biophysical disciplines face the problem of having a plethora of theories that can inform their work. Further, some of the theories accepted in one discipline have been challenged and rejected in others. Which theories should form the foundation for broad, multidisciplinary, cumulative research is a puzzling, and at times, contestable issue.

SFM FOR L OCAL AND G LOBAL O UTCOMES

91

Despite the absence of a coherent body of theory which informs deforestation and environmental change, the role of forests in SD agendas at the national and local level is becoming increasingly important, particularly in the Miombo biome of South-Central Africa (Campbell et al., 2007; Dewees et al., 2010). The populations of these countries are increasing and rural people are still largely dependent on subsistence agriculture and NTFPs for survival (Chidumayo & Gumbo, 2010; Nair & Tieguhong, 2004; Shackleton & Shackleton, 2004). Poor urban populations are still almost solely dependent on forests for fuel, particularly in Zambia (Malimbwe et al., 2010; Nair & Tieguhong, 2004). There have been attempts by development agencies to bridge the wide division between ecologists and social scientists, and the farmers, fishermen and forest people who manage and harvest these resources for a living (Gibbs, 2000). Despite their efforts, and a symptom of the theoretical entropy which permeates the field, the outcome has been a tangled web of catchphrases and an alphabet soup of acronyms which describe conceptual approaches to environmental change (Sayer & Campbell, 2004). Few of these have been successful at managing natural resources in a way which serves the needs of local people, while giving confidence that ecosystem function is preserved on a local and global scale in perpetuity (e.g. Kumar & Corbridge, 2002). Sayer & Campbell (2004) provide several pages of criticism of the development industry’s role in managing natural resources and development under the disparaging heading “Dysfunctional Development Assistance Projects”. I outline three concepts which contribute to the fragmented nature of theory and which result in the failure to recognise and address ultimate causation with respect to land-use change. Neo-Malthusian Assumptions

At the heart of the failure to produce a clear theory of land cover change are problems with the received wisdom concerning the nature of forest use and expanding populations (Leach & Mearns, 1996). Theories which propose population and poverty as major sources of environmental change have seldom survived under the scrutiny of detailed empirical research (VanWey et al., 2005). Assumptions about environmental degradation often rest on neoMalthusian ideas about the relationship between society and environmental change. These ideas are deeply engrained in Western thought and, together with the Edenic conceptualisation of African environments discussed on page 36, form a formidable body of ideas based on the misconception and failure to correctly identify the correct cause for a particular effect. In a wide ranging review of some 140 models of deforestation Angelsen & Kaimowitz (1999)

92

C HAPTER THREE

reach the conclusion that significant doubts exist to support the hypothesis that increasing human populations are an ultimate cause of deforestation. The models which claim to show that human population increase leads to deforestation mostly propose a deterministic relationship between a set of independent variables (e.g. population, poverty) and deforestation. Leach & Mearns’s (1996) book provides several examples where the received wisdom about forest degradation is either wrong, or considerably more complex than initially thought.

Application of Common Pool Resource Theory

The application of Hardin’s (1968) theory of the “tragedy of the commons” has provided the justification for central government control of CPRs for a generation of central planners and resulted in the disempowerment of millions of people (Ostrom et al., 1999). While it is probably true that Garret Hardin has been misinterpreted (Bajema, 1991; Lozano, 2007) there is no doubt about the effect this has had across the developing world. In Zambia this disempowerment is still a source of conflict between rural communities and central government. The work of Elinor Ostrom (Dietz et al., 2003; Gibson et al., 2000a; Ostrom, 1990) has provided some clarity with respect to the sustainable use of natural resources. Ostrom’s recasting of CPR theory is both wide ranging (over 1,000 case studies), draws on a large geographical range, and examines institutions which have been in existence for as long as one hundred years. Ostrom’s eight design principles have been scrutinised by over 90 different studies and have been found to be well supported empirically (Cox et al., 2010). The key difference between Ostrom’s work and the economic models which attempt to reduce deforestation or similar resource use failures to a deterministic relationship between a few dependent variables, is that Ostrom assigns agency to individual players and uses models of human decision making to inform the set of rules by which CPRs can be sustainably utilised (Ostrom & Ostrom, 1999). The assignation of individual agency to problems of CPR use is both radical and innovative and perhaps not well accepted or understood when assessing problems of land-use change. It offers a challenge to authoritarian governments when considering how to devolve ownership over natural resources. In this study I use Elinor Ostrom’s theories to explore the outcomes of different governance structures on forest condition, but ultimately it is the effect on individual agency against which sets of rules for forest use must be judged (Milward & Provan, 2000; Ostrom & Ostrom, 1999).

SFM FOR L OCAL AND G LOBAL O UTCOMES

93

Context and Scale

The importance of context and scale is becoming increasingly recognised by researchers studying deforestation (VanWey et al., 2005). An underlying factor which may contribute to the failure of current theory, is that most theories have been developed in very few countries. Brazil, Cameroon, Costa Rica, Indonesia, Mexico and Thailand are where the bulk of the studies originate from (Angelsen & Kaimowitz, 1999). Ecuador, the Philippines and Tanzania contribute slightly fewer studies (Angelsen & Kaimowitz, 1999). Of these countries, only Tanzania is similar to Zambia, and this depends on the scale at which one compares the characteristics. At smaller scales, the differences are magnified. In their review of what is becoming called “sustainability science” Clark & Dickson (2003, 8059) quote the conclusion reached by a joint workshop of the International Council for Sciences, the Third World Academy of Science and the Initiative on Science and Technology for Sustainability (www.icsu.org): agenda setting at the global, continental, and even the national scale will miss a lot of the most important needs. . . The transcendent challenge is to help promote the relatively ‘local’ (place or enterprise-based) dialogues from which meaningful priorities can emerge, and to put in place the local support systems that will allow those priorities to be implemented. Clark & Dickson (2003) go on to say that: “Where such systems exist, the production of usable, place-based knowledge for promoting sustainability has been impressive indeed.” This provides strong support for a local focus for this study, and an explicit consideration of context and scale in all landuse studies. The fragmentation of theory around landuse change can be linked to the attempt to apply theory too widely and to ignore context and scale. One may argue that Ostrom (1990) does exactly this by her advocacy of the eight design principles for CPR use. However, there is a subtle distinction in calling these criteria which have emerged from an inductive process of case study review and research “design principles,” and advocating these ideas as a panacea for structuring CPR institutions. Ostrom is very clear on the limitations of her design principles (Ostrom, 2007a; Ostrom & Cox, 2010). Their application in a local context and at an appropriate scale will result in differing degrees of adherence to these principles according to a set of local norms and principles (Ostrom, 2007a).

94

C HAPTER THREE

The Landscape Approach: Ten Principles

Although Ostrom (2007a) and Ostrom & Cox (2010) warn about the attempts to find universal solutions to environmental problems, some general guidelines seem reasonable, especially if these allow the complexity and diversity of interactions in social-ecological systems to emerge. Sayer et al. (2013) outline ten principles which are useful in guiding a landscape approach to achieve social, economic and environmental objectives and promoting a holistic synthesis which is context based and scale appropriate. I discuss these ten principles below in the context of the theoretical framework for this study presented in the preceding sections. 1. Continual learning and adaptive management Landscapes change over time and this change comes from both human and nonhuman causes. In measuring and documenting this change the requirements for management interventions become clear. In this study I document changes in land cover over a period of 30 years so that the nature of the changes can be used to inform management at the landscape scale. These changes will continue, hence the need for an effective monitoring regime which accurately depicts the changes at a scale which is relevant for management. 2. Common concern entry point Participatory negotiation of outcomes builds trust and facilitates the achievement of objectives. The structure of institutions and the scaling of solutions which are relevant to local communities are important to building these relationships. The work of Ostrom (1990) and Ostrom (2009c) emphasise the importance of inclusive approaches to rule formulation which build trust. 3. Multiple scales Issues of scale are central to successful natural resource management. This study examines mismatches of scale in measurements of land cover change and AGB and emphasises the importance of matching scales. Ostrom (1990) makes the scaling of institutions one of her eight rules for successful Common-Pool Resource use. 4. Multifunctionality Landscapes have multiple legitimate uses and the proponents of conservation and REDD need to accept this and allow for these uses in planning. To some extent this is addressed in the expanded concept of REDD+. 5. Multiple Stakeholders All stakeholders need to be recognised, with their respective claims given equal importance inasmuch as these claims allow for the existence of diversity

SFM FOR L OCAL AND G LOBAL O UTCOMES

95

in stakeholder claims. Different stakeholders engage with a landscape scale process in different ways and the decision making process needs to take this into account. Effective methods of communication need to be established to allow for different stakeholders to have their views expressed. 6. Negotiated and transparent change logic Transparency and good governance help to build trust and are essential components to a programme of landscape scale management. Change can be traumatic and controversial and negotiating this change requires transparent decision making and a logic of change which is justified and legitimate. 7. Clarification of rights and responsibilities Rules of use and for resource access are essential as a basis for good management. Ostrom’s (1990) rules encompass many different aspects of this principle including the requirement for access to cheap and quick methods of conflict resolution. 8. Participatory and user friendly monitoring The mismatch of scale with respect to remotely sensed measures of land cover change is problematic, but given that communities are capable of measuring and monitoring Above Ground Biomass to the highest standards required by the IPCC (Danielsen et al., 2013, 2011; Skutsch & Ba, 2010), the continued assumption that remotely sensed measures are the most cost-effective methods of monitoring represents a lost opportunity to make the entire REDD+ process more participatory. Grounding the monitoring process in the community promotes ownership and acceptance of the project and is one of Ostrom’s (1990) eight design principles for successful CPR utilisation. 9. Resilience A focus on maintaining system resilience ensures long term goals of delivering ecosystem services can be met while accommodating different uses of the landscape. Resilience theory provides a strong justification for social and ecological systems to be given equal importance in epistemological terms, and demands that a systems approach looks at the interaction between these two domains. 10. Strengthened stakeholder capacity Participation in a landscape scale planning process is a learning exercise. This learning, as well the learning outcomes which follow from implementation, give stakeholders in the community an improved capacity for participation in further and ongoing development. Shared experience between sites further strengthens the capacity for problem solving and fosters innovative solutions, which increases the likelihood of successful project delivery.

96

C HAPTER THREE

Many of Sayer et al.’s (2013) ten principles match the issues which have been discussed earlier in this chapter and could form the basis of a management plan as they incorporate many of the requirements of SFM.

Conclusion There are a wide range of issues which are relevant to the problem of SD and SFM in the dry deciduous forests of western Zambia. The complexity becomes apparent from the large number of factors which interact in this socialecological system. Although Stern (2006) considers forest conservation one of the most cost effective methods of climate change amelioration, there is potential for one of Conklin’s (2006) “wicked problems” to emerge from the process of implementing a REDD+ programme. The most important theme which emerges from this review is the requirement for interdisciplinary research as essential for taking into account historical, ecological, political, institutional and governance issues which affect SFM. The need for a holistic synthesis is clear. There have been calls for such research for over 60 years (e.g. Leopold, 1949) and it is an approach which is becoming more widely acceptable in the research community with the development of sustainability science (Clark, 2007).

Chapter Four Towards a Holistic Synthesis of a Social-Ecological System

Resource management is still largely viewed through a positivist biophysical lens (Ojha et al., 2010). This chapter emerges from the deficiency of the purely biophysical approach to problems of SFM and the failure of epistemological assumptions, which were supposed to deliver predictive generalisations, but fail to address the multidimensional complexity of social reality (Fischer, 1998). My approach is not to disregard the insights which the biophysical sciences afford, but to widen the scope of knowledge included in an analysis. Later in this chapter I argue for the use of a case study approach which allows one to demonstrate how each locality and context is different and provide some of the detailed context which supports later empirical chapters. There is a detailed overview of the study area which describes some of its biophysical characteristics and socio-economic parameters and its socio-political history. I attempt to reveal complexities of scale and present the socioecological framing which is important because there have been few studies of any kind in this remote part of western Zambia. The most recent botanical literature is more than 40 years old (Fanshaw, 1968) and Colin Trapnell did his original surveys more than 70 years ago (Trapnell, 1937; Trapnell & Clothier, 1937). The last vegetation map of the area was published in 1976, but was based on Trapnell & Clothier’s (1937) original report (Edmonds, 1976). I describe some aspects of the social-ecological system for the first time and explore how different components of the system may be affected 97

98

C HAPTER FOUR

by external drivers such as climate change or financial incentives. The examination of phenology provides a detailed analysis of potential impacts of climate change on leaf phenology which is important for the resilience of the forests. The livelihoods analysis examines the extent to which communities are vulnerable to changes which come about through climate change and the impact of REDD+ projects on their livelihoods. The major research approach to presenting this chapter is to investigate how a case study approach can be used to produce a holistic synthesis of the major factors which govern the functioning of the social-ecological system and highlight issues which are relevant for Sustainable Development, Sustainable Forest Management and REDD+. Some of what I include in this chapter is based on personal observations over eight years of working in the area and personal communication with forestry and wildlife officials, area councillors, members of both royal households under whose jurisdiction the area falls (HRH Senior Chief Inyambo Yeta and HRH Chief Moomba), businessmen, timber merchants and district commissioners. These observations are acknowledged as such. The conclusion draws this information together in a synthesis of how the social-ecological system functions in the context of possible REDD+ project implementation.

The Case Study Approach to Researching a Social-Ecological System Local context and considerations of scale have emerged as central to SFM (VanWey et al., 2005) and studies of land-use change and deforestation (Agrawal, 2003; Lambin et al., 2001; VanWey et al., 2005). To date, the UN-REDD framework has not shown that it is capable of taking into account the diverse perspectives, interactions and uncertainties facing forest communities (Evans et al., 2014). This is despite the recognition that local people are integral to the success of the UN-REDD programme, a recognition made explicit in the seven REDD+ Safeguards, (also known as the “Cancun Safeguards” as they were adopted during COP16 in Cancun, Mexico) which have been agreed by governments participating in the United Nations Framework Convention on Climate Change (UNFCCC) (World Bank, 2012a).These policies entrench the rights of local people in the REDD+ process and effectively make most of the co-benefits implied in REDD+ part of the UN-REDD process. A purely biophysical approach to SFM or REDD+ is no longer tenable and therefore researchers need to consider their positionality and be reflexive about

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

99

their role in the process of project implementation, knowledge generation and decision making. This places local context and detailed case study approaches at the heart of the UN-REDD process. In the absence of a general theory to predict land use change (Geist & Lambin, 2002; Lambin et al., 2001; VanWey et al., 2005) a detailed, contextual understanding of REDD+ implementation is necessary. If there are no panaceas for successful CPR use, as Ostrom & Cox (2010) suggest, then local context must assume the importance for developing local solutions to local problems rather than overly simplified institutional prescriptions which are recommended as “the way” to solve problems of environmental conservation. In chapter five, I examined some of the problems which emerge when scale is not explicitly part of the REDD+ process, but some of these problems also apply when local context is not part of it. Local context is both important in its own right and affects the extent to which scale is important for both ecological and social phenomena under investigation, but also because each locality is unique. This is important for management because it requires unique solutions (Humphreys, 2008; Perz, 2007), but also because it illustrates the complexity of social-ecological interactions and the need for adaptability or resilience rather than one solution for every situation (Walker et al., 2004). This aligns with the epistemological position of recognising multiple realities, whether these are value-based social constructions, traditional knowledge or the findings of qualitative social science research. Most studies of deforestation in Zambia date from 1990 and are at the continental scale (FAO, 2010). This limits the consideration of local context which is essential to understanding land cover change in south western Zambia. With an 80 year history of timber exploitation, a low population density relative to the rest of the country and a socio-cultural history that is unique, the region demands a treatment which is different to the major narrative of landuse change in Zambia as outlined by international development agencies such as the FAO. The synthesis which emerges will be important for placing the UN-REDD programme with respect to REDD+ implementation in different areas of Zambia. A “one size fits all” approach to REDD+ is not going to be possible if Zambia is to avoid the development aid failures of the past (Dalal-Clayton et al., 2003; Wainwright & Wehrmeyer, 1998).

100

C HAPTER FOUR

The Study Area: History, Geomorphology, Culture and Ecology The study is located in south western Zambia and covers an area of some 23,500 km2 , including parts of both the Western and Southern Provinces. The explorer Dr David Livingstone was the first European person to document the area, although his travels were concentrated along the Zambezi river, and he makes no mention of the Zambezi Teak forests (Livingstone, 1857; Livingstone & Livingstone, 1866). On his journey by canoe along the Zambezi in 1895, Gibbons (1897) makes mention of the “high banks covered with forest and teeming with tsetse fly” which he encountered about 35 miles upstream of present day Mwandi. This would place him in the heart of the present distribution of the Zambezi Teak forests. Western Province was known as Barotseland during colonial rule of Northern Rhodesia and the name was changed when Northern Rhodesia became the independent country of Zambia. Over the last 80 years, the timber from these forests has been extensively harvested. The building of Cecil John Rhodes’s Cape to Cairo railway line required thousands of railway sleepers, and the Zambezi Teak forests of the then new territory of Northern Rhodesia were a primary source of this timber. Logging for railway sleepers finally ceased in 1972 (Calvert, 2005). Limited harvesting for furniture and flooring timber continues to the present day (pers. obs.). The scale at which many of these problems are examined in this study is that of the landscape. Zambezi Teak forests are spread out in a discontinuous distribution over 23,500 km2 (about the size of Belgium or Lesotho). Regional conservation planning (such as the recent formation of the KavangoZambezi Transfrontier Conservation Area by five neighbouring countries (Zambia Wildlife Authority, 2008)) would initially be better informed by studies at the landscape scale which may form the basis of more detailed studies in the future. The role of local communities in managing the forests and sharing in their wealth has been a problem ever since the Barotseland Agreement of 1964 was revoked by a constitutional amendment enacted by the government of newly independent Zambia in 1969 (Caplan, 1968). The agreement preserved specific powers for the Litunga (the official title of the King of Barotseland) to “make laws for Barotseland in relation to the following matters, that is to say; land, forests and control of bush fires” among others (Sandys, 1964). The abrogation of this agreement ended local responsibility for fire prevention and forests and is a source of resentment among the traditional leadership

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

101

Figure 4.1: Map of Africa showing Zambia and neighbouring countries

102

C HAPTER FOUR

21°0'0"E

23°0'0"E

25°0'0"E

27°0'0"E

29°0'0"E

Zambia Zam b

ez i

Angola 15°0'0"S

15°0'0"S

an Cu do

Kafu e

O ka v

an

n ya Lin

obe ti Ch

go

Zimbabwe 20°0'0"S

Namibia

´

21°0'0"E

20°0'0"S

Botswana

0

23°0'0"E

110

220

25°0'0"E

440 Km

27°0'0"E

29°0'0"E

Figure 4.2: Map of Zambia showing the location of the study area in the southwest of the country

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

103

Figure 4.3: A mature closed canopy Zambezi Teak forest is densely shaded with a sparse understory of shrubs. Fire is not able to penetrate the forest unless it is opened through logging which allows light to penetrate and encourages a grass layer to grow

and population of Barotseland to this day (pers. comms.). Before these powers were appropriated by the Zambian state the Barotse Native Forest Service, which was formed in 1935, was funded from royalties on timber (Mubita, 1986). The service comprised the local Induna Silalo (a senior headman) as Forest Induna, under whom a number of Range Induna‘s were appointed to monitor fires, clear brush and organise fire fighting teams (Mubita, 1986). Despite the success of this system, ownership of natural resources has been in the hands of the state for most of Zambian history. The study area has a unique history and a unique mix of forest and woodland. Together with a long record of environmental exploitation and conflict over resources, the need for a strong contextual analysis seems clear. To a certain extent this context dictates the scale at which the problems are examined. A more detailed account of the study area is given in the sections which follow.

104

C HAPTER FOUR

Prehistory and Geomorphology

The area between the Zambezi river and its main westerly flowing tributaries to the east, the Njoko, the Machile, and the Ngwezi rivers, has been inhabited by agricultural societies for more than a thousand years (Vogel, 1986). Iron age pottery, from near the Machile river, and from the banks of the Zambezi river 120 kilometres to the west, has been dated using radiocarbon dating of the associated charcoal deposits (Clark & Fagan, 1965). Further centres of population have been found about 70 kilometres west of the Victoria Falls near Bovu vlei, and close to the Falls itself near the Maramba river (Vogel, 1986). This archaeological evidence corresponds to the widespread dispersal of Iron Age food producers throughout the savanna of South-Central Africa between A.D.400 and A.D.800. The most striking geomorphological feature of the landscape is the predominance of Kalahari sands throughout the region. These sands determine much of the ecological characteristics of the region and have influenced the nature of the archaeological record of Iron Age human habitation, which is relatively poor compared to other areas in Zambia. Phillipson (1974) suggests that this may be because wooden vessels were used in preference to clay vessels. Clay deposits are rare in the region, which is overlain by aeolian sands. Nevertheless, the evidence of an ancient food producing culture in immediate proximity to the Zambezi Teak forests has profound implications for this study. It indicates that humans have been living near Zambezi Teak forests for a considerable period of time. It is possible that these forests are anthropogenic structures, whose distribution is partially determined by a combination of naturally occurring differences in soil depth, soil moisture and thousands of years of human use of fire in the landscape. If “landscapes are our unwitting cultural autobiographies” (Lewis, 1979, 12), and fire is an important factor in controlling Zambezi Teak forest growth (Calvert, 1986a; Gambiza et al., 2008), then the existence of discrete forest patches with evidence of ancient human presence nearby lends support to this interpretation of forest distribution. At the subcontinental scale, the geomorphology of southern Africa is characterised by two major features. The first of these is the presence of what Thomas & Shaw (2010) call the “Great Escarpment” around the rim of the subcontinent. The second feature is the existence of a depression in the interior called the Kalahari basin. The formation of both structures is linked to rifting associated with the breakup of Gondwanaland and the resultant subsidence and uplift. Approaching the interior of southern Africa across the coastal plain from any direction, one meets with an escarpment at a mean distance of between

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

105

150 to 200 kilometres from the sea. The lateral rifting processes which produced this escarpment are not well understood, but are thought to be associated with thermal expansion of the crust during rifting which results in subsidence and uplift at different contact points (Thomas & Shaw, 2010). A simple analogy to describe this geomorphology is that of an upturned saucer, with a narrow coastal plain rising to a ridge, which once surmounted, gives way to a broad and shallow depression which has gradually become filled with sediment by the erosion of endoreic river systems. The Kalahari sediments were initially derived from erosion of the uplifted Great Escarpment by rivers flowing towards the interior. This infill and sedimentation are largely responsible for the relatively flat topography of the Kalahari basin (Thomas & Shaw, 2010). Subsequent erosion and wind action during dry periods in the Quaternary resulted in the formation of dunes. There seems to be a consensus that the dunes were in existence prior to 32000 yr B.P., but that later formation of dunes occurred at intervals around 19000 to 17000, 10000 to 6000 and 4000 to 3000 yr B.P (Lancaster, 1989). The timing of dune formation in the northern Kalahari is complex (Thomas et al., 2000) and it is difficult to pinpoint a period of dune formation in western Zambia, as there may well have been several episodes. Nevertheless ancient linear dune systems are visible in western Zambia to a greater or lesser degree. The most prominent of these occurring in an area between the Kwando and Zambezi rivers and are clearly visible on satellite images. The ridges of these dunes support dense forests of Zambezi Teak and other species commonly associated with this forest type (Thomas, 1984). The pattern of vegetation growth across the ancient dune formations can be disrupted by fire, timber logging and agricultural clearing, but is nevertheless clearly defined in undisturbed areas. These dune formations are not clearly visible in the study area and appear to terminate at the Zambezi river. This would appear to indicate that the Zambezi formed a barrier to aeolian sands from the west. However, there are two arguments suggested by Thomas (1984) that may explain the lack of distinct linear dune formations east of the Zambezi river. Dune orientation is almost certainly a proxy for wind direction and the dominant winds in this area are related to the South African anticyclone and come from the east. In addition, the pattern of drainage in the study area indicates that it was formed in ancient dune troughs because the streams are orientated in the same direction as the dunes to the west of the Zambezi. The lack of distinct ridges is possibly due to degradation through erosion by the greater density of drainage on the east side of the river (Thomas, 1984). Nevertheless, observations in the field show that despite the apparent flat topography between Sesheke and Mulobezi, there are areas of higher

106

C HAPTER FOUR

ground and usually it is on these patches of higher ground that Baikiaea plurijuga grows in tall, closed canopy stands (Huckabay, 1986a). Ultimately the dune formations of western Zambia have received little paleoclimatic investigation (Thomas & Shaw, 2002) and there may well be relic but eroded dune structures in the areas where Zambezi Teak now grows in dense stands. This would show a continuum with the pattern of vegetation distribution seen elsewhere where the dune ridges are more distinct. At a landscape scale, patterns of vegetation structure were last mapped in 1972 (Edmonds, 1976).

Vegetation and Classification

The first mapping and description of the vegetation of Zambia started with the pioneering work of Colin Trapnell from 1937 to 1943 (Smith, 2002; Trapnell, 1937; Trapnell & Clothier, 1937). His system of classification was, and in some ways still is, unique, although subsequent work in other areas has examined ethnographical classification systems in some detail (Ellen, 2006). His combination of soil type and explicit consideration of the indigenous crops grown in the different soils was used to identify and classify vegetation categories. Later maps of vegetation were compiled using aerial photographs, field observation and forestry measurements and follow the classification proposed for “African Vegetation Types” proposed at the 1957 Yangambi Conference (Aubreville, 1957). In many areas of Zambia no further vegetation surveys have been undertaken following Trapnell’s work. There are relatively few publications and maps describing vegetation for south western Zambia. Other than Trapnell’s work there is only Fanshaw’s (1968) paper and a description of the Zambezi Teak forests by Martin (1940), whilst the map drawn by Edmonds (1976) remains the only detailed vegetation map of the area. Fanshaw (1968) provides the only detailed description of the vegetation in the area and lists some 550 woody species which occur in seven main vegetation types. The majority of vegetation in the study area comprises a large swathe of Kalahari Woodland, in which patches of Baikiaea plurijuga forest occur. Although Fanshaw (1968) recognises some variation in Kalahari Woodland, for the purpose of this study it is considered as a single type in which Julbernadia paniculata is one of the most common large trees, but is found in a heterogeneous mixture with Burkea africana, Erythrophleum africanum, Guibortia coleosperma, Pterocarpus angolensis and Schinziophyton rautenenii. There are some areas where Brachystegia species occur, making these communities of Kalahari Woodland more similar in species composition than the Miombo Woodland to the east, but the areas are not large and do not

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

107

warrant classification as a separate vegetation type in a study of this scale (pers. obs.).

Figure 4.4: Kalahari Woodland is open, with a grass layer that supports regular fires. Many species are adapted to withstand fire and have thick, corky barks that smoulder but do not burn easily (Dantas & Pausas, 2013; Gignoux et al., 1997). The two dark trunks (left foreground) are Pterocarpus angolensis which show signs of a fire from the previous season. The most common tree is Julbernadia paniculata, known locally as Lutondo, and hence the Kalahari Woodland is known as Mutondo, or the place where Lutondo trees grow The river courses, which are seasonally flooded, are bounded by bands of grassland (known locally as “dambo”) of varying width and which comprise a small, but important habitat. Reedbuck (Redunca arundinum) require tall grass cover and a constant water supply and are found almost exclusively on the dambos, while oribi (Ourebia ourebi) are found exclusively in habitats with a mixture of short and tall grasses and rely on the heterogeneity of the dambo habitat for food and shelter (Skinner & Smithers, 1990). Both sable antelope (Hippotragus niger) and roan antelope (Hippotragus equinus) feed on dambo grasses and the dambos are used by all grazing mammals in the region to varying degrees (pers. obs.). In the south east there is a large area of woodland dominated almost exclusively by Colophospermum mopane, a vegetation type known locally as Mopane woodland. In this area the occasional Baobab (Adansonia digitata)

108

C HAPTER FOUR

is found, as well as some Acacia nigrescens and the semi-succulent Commiphora spp. which frequent hot, dry habitats (Coates Palgrave, 1977). Some patches of Zambezi Teak forest occur within the Mopane woodland where sandy soils occur, but the area is largely interspersed with grasslands. The soils in this region are predominantly expansive montmorillonite clays on which Colophospermum mopane is adapted to grow (Coates Palgrave, 1977). Few other species can tolerate the alternation between seasonal flooding and drying which makes the clays expand and contract and the only other woody species diversity occurs on the large termitaria which form slightly elevated mounds above the surroundings (pers. obs.).

Figure 4.5: Field camp in Mopane Woodland. This vegetation type is dominated by Colophospermum mopane, distributed in patches interspersed with open grassland. Large areas are flooded in the rainy season, and the black clay soil dries hard during the dry season. Few species are able to survive these extremes Outside the Mopane Woodland areas, soils are predominantly Kalahari sand, which is a product of aeolian weathering of weak Upper Karoo sandstones (Thomas & Shaw, 2010). The sands are highly permeable but due to the low relief and an unknown less permeable substratum, the macro-drainage is poor and there are many seasonal pans, flooded plains and watershed grassland plains. The sands vary from pale to orange in colour and contain from 3% to 12% clay, and 15% to 60% fine sand (0.02mm to 0.2mm) with particles more than 0.5mm in diameter rarely forming more than 5% of the

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

109

total (Thomas & Shaw, 2010). The sands are not fertile, with nitrogen content being particularly low where the humus content has been disturbed by cultivation (Thomas & Shaw, 2010). Zambezi Teak forest occurs almost exclusively on deep, well drained sands in areas of rainfall from 650mm to 1200mm (Huckabay, 1986a; Martin, 1940). Most of the study area has an annual rainfall at the lower end of this range, although there is a north-south gradient with higher rainfall in the north (Hijmans et al., 2005). The sands are moderately acid, and it has been suggested that the composition of the sands is fine enough to keep almost the entire annual rainfall available for use by the deep rooted Zambezi Teak (Calvert, 1986a). In areas where drainage is restricted and there is a chance of the soil becoming waterlogged, the species will not grow. Of minor interest with respect to forest types in the study area is the mention by Fanshaw (1968) of the occurrence of closed canopy Cryptosepalum exfoliatum forests. During extensive field expeditions over the last nine years I have not encountered any stands of Cryptosepalum exfoliatum which fit Fanshaw’s (1968) description and it seems these forests may be restricted to the extreme north east of the study area, are at the southern end of their range with respect to rainfall levels and almost certainly do not occupy a substantial enough area to be considered in this study. Chidumayo (1997) indicates that these forests may have been destroyed by cultivation and fire and that only small, disturbed fragments remain. Baikiaea plurijuga Forest Ecology

The five other members of the genus Baikiaea are tropical rainforest species. Baikiaea plurijuga is the only species which occurs in dry tropical forest, and indications are that it is at the edge of its range (Huckabay, 1986a). Baikiaea plurijuga occurs further south than any of the other Baikiaea species, and grows in areas which receive between one third to one half the mean annual precipitation (Huckabay, 1986a). The species which are associated with Baikiaea plurijuga are more characteristic of the thorn woodlands and dry deciduous forests of the Kalahari basin, and it is not clear how the species came to be distributed through its current geographical extent. It appears as a climax vegetation type under specific conditions, the most important of which are the presence of deep, well drained sands and the reduced frequency of fire (Martin, 1940). However, Zambezi Teak forests only cover a small percentage of the area. Kalahari woodland is by far the dominant vegetation class, with many typical miombo species (Julbernadia paniculata, Burkea africana, Pterocarpus angolensis and some Brachystegia) and an open structure which allows the growth of grass.

110

C HAPTER FOUR

Comparisons of modern pollen assemblages in the Congo with forest inventory data from the same sites show that although the physiognomy of the vegetation is well represented by the pollen rain (the pollen data clearly indicate a habitat dominated by trees rather than herbaceous plants), the precise species composition is difficult to determine from the pollen record (Elenga et al., 2000). Different plants produce different quantities of pollen and dispersal of this pollen varies between species. It is therefore unlikely that any fossil pollen record would be able to show direct evidence of Baikiaea plurijuga invasion from the north. Nevertheless the accuracy with which vegetation types can be reconstructed is improving in Central Africa as the concept of Plant Functional Types (PFTs) is refined for the region (Lebamba et al., 2009; Vincens et al., 2006). PFTs are an ecological classification which is an alternative to traditional taxonomic categories and is aimed at simplifying floristic complexity and improving the ecological information in a classification through a consideration of traits which have ecological relevance (Díaz & Cabido, 1997; Prentice et al., 1992). The late Quaternary climate in South-Central Africa appears to show extensive fluctuations between dry and wet periods which have influenced vegetation distribution (Vincens, 1991). Given that Baikiaea plurijuga is related to other Baikiaea species which favour a moist climate, we assume that a moist period at some stage during the Quaternary would have enabled the species to migrate south in the ensuing changes in vegetation distribution. Using pollen cores from southern Lake Tanganyika, Vincens (1991) shows that from about 12,000 yr B.P. Zambezian woodlands with a species composition similar to the modern vegetation started to replace the vegetation which had previously shown significant montane species as well as numerous Ericacea. There appears to have been a cool, dry period from about 22,000 through to 15,000 yr B.P., with a transitional period between 15,000 and 12,000 yr B.P., followed by a great increase in rainfall and temperature from about 12,000 yr B.P. Most significantly, Vincens (1991) documents the occurrence of arboreal taxa in South-Central Africa which have more affinities with West and Central Africa flora than with the Zambezian flora from about 12,000 yr B.P. We may speculate that Baikiaea plurijuga was among these taxa. Although attempts have been made to model Quaternary vegetation distribution in Africa using pollen and plant macrofossil data (Jolly et al., 1998b) and compare these distributions to those predicted by climate models (Jolly et al., 1998a), the continental scale at which these predictions are made is not useful for explaining the southward migration and establishment of Baikiaea plurijuga forests in south western Zambia. The focus on forests as systems which accumulate carbon produced by

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

111

human industrial activity ignores their extraordinary diversity (Gibson et al., 2011; Thompson et al., 2009). Whilst the dry tropical forests of Africa do not harbour the number of species which is normally associated with tropical rainforest, the study area nevertheless contains a higher number of tree species than the whole of western Europe (Fanshaw, 1968). White (1983) and Linder et al. (2005) both designate the Zambezian region as a centre of endemism, with the highest floristic diversity of dry forests and woodlands in Africa. Baikiaea plurijuga and Entandrophragma caudatum (which occurs on the periphery of some Baikiaea plurijuga forests) are endemics to the region and are currently listed on CITES Appendix II, indicating that the species are not necessarily threatened with extinction, but may become so unless trade is closely controlled. The forests also harbour extraordinarily large mammalian diversity. Although numbers are much reduced there are still herds of elephant (Loxodonta africana), African buffalo (Syncerus caffer), sable antelope (Hippotragus niger), roan antelope (Hippotragus equinus) and Lichtenstein’s hartebeest (Alcelaphus lichtensteinii) which use the Zambezi Teak forests and Kalahari Woodland (pers. obs.). The dambos (seasonally flooded grasslands) are a permanent habitat for grassland specialists such as southern reedbuck (Redunca arundinum), oribi (Ourebia ourebi) and Defassa waterbuck (Kobus ellipsiprymnus, ssp. defassa). The Mopane forests are flooded during the rainy season and largely inaccessible, but during the dry season they provide a habitat for elephants, zebras (Equus quagga) and greater kudus (Tragelaphus strepsiceros) (pers. obs.). The complexity of the system quickly becomes apparent when one considers the potential role of keystone herbivores such as elephant in affecting habitat change and maintaining patch diversity (Owen-Smith, 1989, 1992). The role of large herbivores in distributing nutrients in ecosystems was recently modelled by Wolf et al. (2013) and the effects of the extinction of large herbivores in Amazonia were examined by Doughty et al. (2013). The authors conclude that the Amazononian forest vegetation is still adjusting to the effects of large herbivore extinction which decreased the amount of phosphorus in the system which was transported in an artery-like fashion in the dung and bodies of these animals. Large herbivores (>100kg) in particular play a disproportionately large role relative to their population density in dispersing nutrients throughout an ecosystem (Owen-Smith, 1992), and may be particularly important in environments with nutrient poor soils as are found in the study area. The role of large mammals such as elephants, sable antelopes and African buffaloes, which use Zambezi Teak forests as shade areas during the heat of the day, in transporting nutrients to these locations is an intriguing possible source of nutrient enrichment.

112

C HAPTER FOUR

Given a habitat as complex as a forest with a large number of plant species, the concept of Plant Functional Type (PFT) represents a way of understanding the role of plants in ecosystem function. PFTs are generally positively related to the number of species, but lists of species do little to enhance our understanding of their role in ecosystem function. Even in a general sense, Diaz & Cabido (2001, 647) state that “Rates and magnitudes of ecosystem processes have been found to be more consistently associated with functional composition (presence of certain plant functional types or traits) and functional richness (number of different plant functional types) than with species richness.” The evidence for a few key PFTs playing an important role in ecosystem function and ecosystem resource dynamics is accumulating. The evidence for the mass ratio hypothesis, in which it is proposed that the importance of a species in ecosystem function can be directly estimated from its total biomass within the system, is supported by an increasing body of work (Grime, 2002, 1997; Huston et al., 2000; Tilman et al., 1997). A detailed review by Huston (1997) makes a strong case for studies which report effects of declining species number on ecosystem function being fundamentally flawed, and it is clear that the loss of a few dominant species which have a large effect on ecosystem processes has a greater effect than declining species numbers per se. Field observations and a few published studies (Calvert, 1986a; Huckabay, 1986a; JICA, 1996; Rudel, 1989) indicate that Baikiaea plurijuga has functional traits which allow it to contribute to ecosystem resource dynamics and functioning in an important way. From this it can be argued that the loss of Baikiaea plurijuga forests represents the loss not only of a single species, but also of a functional type for which no other species is able to substitute. There is certainly no other species which forms closed canopy forest patches in the study area which is overwhelmingly characterised by the open Kalahari woodland vegetation type normally associated with a climate that has a long dry season and highly variable rainfall (JICA, 1996; Martin, 1940). Baikiaea plurijuga has long, deep roots with a highly developed early rooting ability, nitrogen fixing root bacteria and explosive seed dispersal (Calvert, 1986a; Martin, 1940). The loss of this functional type may have a larger impact than the loss of several other species whose functional role is duplicated, or at least closely matched by other species. For example Pterocarpus angolensis is a common fire tolerant species in Kalahari Woodland and is under threat because of harvesting for timber (Whiteman, 2013). However, other common species (Burkea africana, Erythrophleum africanum, Bobgun-

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

113

nia madagascariensis) have similar fire tolerant characteristics (Dantas & Pausas, 2013; Gignoux et al., 1997) and the loss of Pterocarpus angolensis would not mean the loss of a species of this functional type. History of Baikiaea plurijuga exploitation

The exploitation of Baikiaea plurijuga forests started with the development of railways and mines in Zambia. Cecil John Rhodes died in 1902, three years before his Cape to Cairo railway line crossed the Zambezi below the Victoria Falls. His instructions to the engineers were to “build the bridge across the Zambezi where the trains, as they pass, will catch the spray of the Falls” (Calvert, 2005). The bridge was built and completed accordingly and linked the south and north banks of the Zambezi for the first time. Professor Sir George Darwin, the son of Charles Darwin, and the President of the British Association (now the Royal Society) officially opened the bridge on September 12, 1905 (Davis, 1911). The town of Livingstone, 10 kilometres away, was established as a railhead and much of the initial development of the town was connected with the railway industry (Calvert, 2005). The first record of logging Baikiaea plurijuga forests using machinery is from a front page advertisement in the Livingstone Mail of November 11, 1911 announcing that Messrs. Trombas and Jacobs were “now prepared to accept orders for all kinds of Native Timber” at their Dimitra Saw Mill in Livingstone (Calvert, 2005). There were extensive Baikiaea plurijuga forests to the west of Livingstone, but Pterocarpus angolensis and other timber species were all plentiful in close proximity to the new town. In early 1915 the Bulawayo based concern, A.F. Philip and Company, became interested in investing in the operation. The result of the negotiations was the formation of Zambesi [sic] Saw Mills Ltd in 1916 (Calvert, 1986b). This company was to become the most important player in the exploitation of the Zambezi Teak forests as it expanded over the following decades. The new company established a saw mill near Bovu Vlei, 45 kilometres west of Livingstone (Calvert, 1986b). Calvert (2005) describes how the opening of the copper mines in the north led to increased demands on the rail network. To date, pressed steel sleepers had been used to support the railway lines, but these were not strong enough to withstand the heavier carriages carrying coal to the north and copper ore to the south. The concave pressed steel sleepers would gradually flatten down under a heavy load. Wooden sleepers, and particularly those made of Zambezi Teak, were suddenly in great demand so that the line from Livingstone to the Copperbelt could be upgraded to withstand the heavier loads. In 1918 Rhodesia Railways placed an order for 800,000 wooden

114

C HAPTER FOUR

sleepers from Zambesi Saw Mills. The relatively small size of Zambesi Saw Mills, which had mostly been cutting timber for furniture and construction in Livingstone, meant that they were unable to meet this order. Rhodesia Railways therefore advanced £10,000 for the expansion of the mill and acquired shares in the company in exchange for supplying the capital for expansion of the business (Calvert, 2005). A two-foot gauge tram line of sawn Zambezi Teak was laid from the mill at Mapanda forest back to Livingstone in 1918. Eight hundred sleepers a week were hauled on 10 ton bogie wagons, drawn by spans of 18 oxen. However, by 1924 the wooden tram line was no longer sufficient, and Rhodesia Railways took the decision to lay a standard Cape gauge railway powered by steam locomotives and using standard rolling stock. The line was extended as required by the demand for timber, and by 1934 the line had reached Mulobezi where it terminates today. An excellent history of the Zambezi Saw Mills Railway has been written by Calvert (2005), but for the purposes of understanding the history of exploitation, this early pattern of logging by railway was more or less maintained until the Mulobezi mill was closed down. By the 1950s the high quality timber close to the line of rail had been exploited and the company was experiencing problems maintaining profitability. It had now become the second largest employer in Northern Rhodesia after the copper mines and the decreasing supply of timber was starting to have severe economic effects in Barotseland, where Zambesi Saw Mills (1948) Limited was the largest industry. After independence in 1964 the Zambian government nationalised all major industries. In 1968 the Zambian government acquired 51% of the shares of the company, but it never regained profitability and in 1972 it was closed down. During the period of operation, increasing pressure on Zambesi Saw Mills (1948) to maintain high levels of production led to recommendations by forest officers being ignored by the government. Repeated recommendations were made by forest officers to reduce logging and maintain minimum size classes (Calvert, 1986a). Despite this, minimum size classes have been lowered repeatedly in an attempt to open up more timber to logging. Although forestry operations continued into the 1980s, the forests were beyond large scale economic exploitation after 1972. Small scale cutting of Zambezi Teak continues, but volumes rarely exceed 3000 m3 per year, compared to the 30,000 to 40,000 m3 per year which were cut in the 1940s (Calvert, 1986b; Martin, 1940). In the last 10 years, attempts have been made to establish viable businesses using mobile saw milling equipment and modern drying kilns, but due to a number of factors which we will examine later, these

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

115

Figure 4.6: Felling a large Zambezi Teak tree in mature closed canopy Zambezi Teak forest. This process opens the forest to fire, preventing young trees from becoming established. Without fire management the forests never recover

businesses have failed, or are barely profitable (pers. obs.). Socio-political History

The exploitation of Zambezi Teak is closely connected with the political history of Zambia and of Western Province, or Barotseland, in particular. The origin of the Lozi people in Barotseland has never been resolved (Mainga, 1966; Muuka, 1966). There is an official version of events recounted by the Lozi royal clan, but like many myths of origin it is linked to attempts at establishing primacy and uniqueness in order to support claims to territory and the right to exercise authority (Mainga, 1966). The importance of these claims and rights is no less important today than it was 150 years ago when the first explorers and missionaries arrived in the area and started to document the history of the Lozi people (pers. obs.). The official story denies the presence of any other people in the area and claims that the Lozi kings are descended from a unitary god called Nyambe who spent time on earth and from whom divine authority to rule is derived (Mainga, 1966). The Lozi

116

C HAPTER FOUR

Figure 4.7: Abandoned steam engine and other rolling stock from the heyday of Zambesi Saw Mills. These are now under the nominal care of the Natural Heritage Conservation Commission of Zambia, but are regularly vandalised by scrap metal speculators

Figure 4.8: Map of tribal and linguistic groups in Zambia. These groups represent a diverse cultural and traditional landscape which transcends the national and regional boundaries of the state and which are seldom considered in development projects (Government of Zambia, 1965)

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM 117

118

C HAPTER FOUR

state, with its divine monarchy centred on a sacred king, is similar to two other well documented states in the region which preceded it by centuries, and from which it is thought to have been an offshoot (Mainga, 1966). The Luba-Lunda kingdom in the Congo and the Karanga-Rozwi state in what is now Zimbabwe are well documented by Arab and Portuguese sources as having been in existence since the eighth and ninth centuries and AD 900 respectively (Mainga, 1966). These two kingdoms have similar structures with respect to leadership and royal institutions (Mainga, 1966), and are the major regional nuclei for other states which can be clearly demonstrated to have derived from one of these centres. The Lobedu Kingdom in northern South Africa, for example, was founded in the seventeenth century as a breakaway group from Monomotapa’s empire in Zimbabwe (Mainga, 1966). Although Lozi oral tradition supports evidence for both origins, and indeed both Lunda and Rozwi oral traditions claim credit for the origin of the Lozi state, there are several indications which suggest that a Congo origin is more likely. The regalia of kingship, the elevation of the royal family to a position of earthly and spiritual power, the circumscription of the movements and public appearances of the king, and the significance of the royal graves in cultural traditions all have similarities to Lunda culture (Muuka, 1966). Caplan (1970) considers a split from the Lunda-Luba kingdom to be the most persuasive theory for the origin of the Lozi people, who arrived on the Zambezi floodplain during the latter half of the seventeenth century. The most important event in the history of Barotseland was the arrival of the Kololo in 1840 from an area near present day Lesotho and the Free State in South Africa (Caplan, 1970). Displaced by the turmoil of regional Zulu aggression, this group of Sotho speaking people under the leadership of Chief Sebituane quickly forced the Lozi elite into retreat and exile. Within a short period they had established hegemony over most of Barotseland (Caplan, 1970). Prior to the Kalolo invasion, the Lozi political system was distinguished by the division of the system into silalo and makolo, which was unique to societies in southern Africa. The silalo is a simple territorial division with no administrative function. The makolo is a political division which does not map to any territorial entity but is instead represented by a figure with an important title at the capital (Caplan, 1970). People in a single village (several of which may fall under the territory of a local headman called the Induna Silalo) may be members of different political sectors (makolo). The consequence of this is that it was difficult to organise a localised block of rebellion against the king (Caplan, 1970). When it comes to structuring institutions in Barotseland which are responsible for implementing the sustainable utilisation of natural resources

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

119

(and forests in particular), these need to be designed to take into account the traditional administrative and political structures. The reality of Barotseland today is that there is considerable tension between the Zambian state structures and the structures of what is now known as the Barotse Royal Establishment (BRE) (McLoughlin, 2014). The BRE expects and demands that their traditional structures are respected and taken into account when any programme of development or sustainable utilisation of natural resources takes place. These demands are largely supported at the lowest level of traditional administration and indeed by the people themselves (pers. obs.). Although the tensions between the BRE and the Zambian government tend to be exaggerated and taken advantage of by extreme groups (there is a small but vocal secessionist movement) by and large what the BRE expect is similar to what Welsh and Scottish governments have been granted over the last 10 years in the United Kingdom: more control over local decisions and decentralisation of control over local issues.

Climate Change Africa is predicted to experience climate change more severely than other continents, but the effects will vary regionally. There are no continent-wide changes which are likely to occur (Collier et al., 2008). Climate change predictions for Zambia indicate that the south western part of the country, in the area where Zambezi Teak Forests are located, will be most severely affected. Data on ClimateWizard (www.climatewizard.org), which use data from Mitchell et al. (2004), show that mean temperature in the study area is set to rise by between 2 − 4 ◦ C over the next 80 years and rainfall is predicted to decrease. In dry deciduous forests, rainfall is thought to be the main cue for seasonal phenological events (Childes, 1988; Singh & Kushwaha, 2005) and the combination of changes in temperature and lower rainfall may result in the adaptive capacity of some species to survive the dry season being exceeded. The dry areas of southern and eastern Africa experience a short rainy season of four to five months, a long dry season of seven to eight months, large fluctuations between day and night temperatures in winter (with occasional frosts), and hot diurnal temperatures in summer (Thomas & Shaw, 2010). Trees have adapted to cope with these conditions using a wide range of structural and physiological adaptations. Variations in phenology are an important adaptation of dry tropical forest species to seasonal drought (Singh & Kushwaha, 2005). Trees vary in the timing of leaf senescence, the extent of deciduousness and the response to factors which trigger leaf fall (Singh &

120

C HAPTER FOUR

Kushwaha, 2005). Understanding the factors which determine leaf emergence and senescence is important for predicting the effects of climate change. The timing of leaf phenology is of particular importance to dry tropical forest species (Poorter et al., 2014; Xiao et al., 2006). The extent to which people rely on forests for building materials, food and traditional medicine makes changes in forest species distribution and forest structure of direct relevance to human survival. Vegetation mapping in Africa often fails to account for the diversity of forest types (Gond et al., 2013). Predicting responses to climate change requires a good understanding of the spatial distribution of these forest types and the characteristics which determine their possible response to change (Gond et al., 2013). Poor knowledge of the physiological basis for differences in the leaf phenology of forest types makes it difficult to predict what effects climate change will have on tree survival and more research is urgently needed in this area. Being able to predict the effects of climate change on forests takes on more relevance than in temperate forests because of the degree to which populations in this part of Africa depend on the environmental services which forests provide (Collier et al., 2008). The chitemene agricultural system, which is widely practiced in the study area, depends on forest regrowth for soil regeneration during the rotation (Trapnell & Clothier, 1937). If the rate or extent of forest regrowth is affected by climate change it will have a significant impact on soil fertility and therefore on food security. Non Timber Forest Products (NTFPs) provide an important source of nourishment during lean years, often provide the only source of cash income, and are central to traditional medical practices and cultural ceremonies (Shackleton et al., 2011). The extent to which climate change affects forests will determine whether these products are available for local populations to use in the future. Being able to anticipate the effects of climate change on forests will add considerably to building community resilience and adaptive capacity. The effects of climate change on forests in south western Zambia will occur at the scale of individual tree responses to predicted changes in temperature or rainfall. It is only through understanding individual species’ responses, and how site related factors contribute to these responses that will allow the scaling up to regional or continental scale predictions for responses to climate change. The reverse also applies with the ability to relate the predictions of the General Circulation Model (GCM) to local conditions that are meaningful for making assessments of climate change effects on household level resilience. In temperate forests phenological events are usually linked to seasonal

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

121

changes in temperature and changing day length, with a particular emphasis on a period in late winter (Clark et al., 2013). In tropical deciduous forests the phenology is largely determined by seasonal variation in water availability (Childes, 1988; Singh & Kushwaha, 2005). However, despite water availability being the major cue for triggering leaf flush or leaf fall, there is a large degree of variation in the timing of these phenological events. The control of vegetative phenology in tropical trees is not well understood (Kushwaha & Singh, 2005), but is thought to be due to a range of adaptations which determine the water relations of a species (Singh & Kushwaha, 2005), as well as local variation in soil moisture conditions. Borchert (1994) found that the variation in stem water status was the principal determinant of phenology in dry tropical forests in Costa Rica. The state of stem water storage depends on a number of factors including the individual water storage capacity of a species, which was strongly correlated with wood density. Species with lower wood densities were able to retain higher levels of stem water status by the end of the dry season relative to species with higher wood density (Borchert, 1994). However, rooting habit and local variation in soil water were also factors which influenced the variation in stem water status within species in Borchert’s (1994) study. In many seasonally dry tropical forests one of the functional traits which varies between tree species is the degree of deciduousness (Kushwaha & Singh, 2005). Leaf senescence reduces water loss and the period during which trees are leafless confers different degrees of drought tolerance. Due to the variation in factors which determine stem water status, however, there is a high degree of variation in deciduousness observed in dry tropical forests. For this reason, it is difficult to generalise using field based observations of phenological changes when mapping leaf phenology over large geographical areas. By the same logic, characterising trees or forests by plant functional type over large areas may also be problematic given the variation in vegetative phenology (Kushwaha & Singh, 2005). The contribution of multiple factors to phenological events results in a high level of variation being observed in the field in the timing of leaf phenology. A patch of forest in one area may have lost its leaves by the middle of the dry season, while a forest patch of the same species in a different area will retain leaves until a few weeks before the first rains (pers. obs. and Figure 4.9). Understanding how climate change may affect forests is made more difficult because of this variation in phenological timing. An increase in the severity of drought and heat stress due to climate change, has the potential to be more severe than trees can survive, leading to widespread tree mortality and altering the structure and composition of

122

C HAPTER FOUR

Figure 4.9: Aerial view of differences in leaf phenology. The forest in the foreground without leaves is a patch of Zambezi Teak forest. The Kalahari Woodland along its border and further towards the horizon has already flushed. The photograph was taken on the November 19, 2009, a few days after the first rains, and clearly shows how leaf phenology varies between different vegetation types in the same location

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

123

forests (Allen et al., 2010). Ultimately there are three possible fates for tree populations in the face of rapid climate change: migration to track the ecological niche to which they are adapted, adaptation to new conditions, or death (Aitken et al., 2008). The effect of climate change on plant phenology has the potential to be of immense significance (Figure 4.10), but the severity of the effect will depend on the nature of existing adaptations. Mapping differences in leaf phenology using remote sensing will allow for clearer prediction of the effects of climate change on forested areas.

Figure 4.10: The effects of climate change on tree phenology. The direction of the arrows is from the affecting to the affected factor. From Singh & Kushwaha (2005)

124

C HAPTER FOUR

Climate Change in Zambia

Climate maps for Zambia were generated using the online tool developed by The Nature Conservancy called ClimateWizard (www.climatewizard.org). The data are from Mitchell et al. (2004) and include past climate trends and future predicted changes. The downscaled 50km data are generated by a GCM according to the following method described on the ClimateWizard website: “Global climate model output from the World Climate Research Programmes (WCRPs) Coupled Model Intercomparison Project phase 3 (CMIP3) multi-model dataset (Meehl et al., 2007), were downscaled as described by Maurer et al. (2009) using the biascorrection/spatial downscaling method (Wood et al., 2004) to a 0.5 degree grid, based on the 1950-1999 gridded observations of Adam & Lettenmaier (2003).” (www.climatewizard.org) It is this data and the output from the CMIP3 GCM that are referenced in the IPCC Fourth Assessment Report (Pachauri, 2008) which is used in examining climate predictions for Zambia. The maps selected from the ClimateWizard website used Emission Scenario High A2 which makes the following assumptions about the global factors which determine future emissions (Nakicenovic & Swart, 2000): • Relatively slow demographic transition and relatively slow convergence in regional fertility patterns. • Relatively slow convergence in inter-regional Gross Domestic Product (GDP) per capita difference. • Relatively slow end-use and supply-side energy efficiency improvements (compared to other storylines). • Delayed development of renewable energy. • No barriers to the use of nuclear energy.

These assumptions represent a relatively high global emissions scenario compared with an alternative medium or low emissions scenario upon which climate change can be modelled. Although the emissions scenario is high, each scenario (high, medium or low) develops three models which represent the extremes of the predictions within that scenario. The output of the high emissions scenario model depicted in Figure 4.11 shows changes predicted by the middle model so that half the models in the high emissions scenario predict a greater amount of change, and half the models predict a lesser amount of change.

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

125

The maps in Figure 4.11 show that Zambia has experienced an increase in mean annual temperature since the 1950s, a trend which is predicted to continue through to 2099. Although the level of warming appears to slow during the first half of the century, by 2070 most of the country starts to experience mean annual temperature increases of 3 − 4 ◦ C. Changes in the south western part of Zambia in the area where the study area is located, appear to be the most extreme of the entire country.

Figure 4.11: The map on the top left shows the historical rate of change for annual temperature. Statistical confidence for these data is high, between 95% and 99%. The graph shows the same data over the years 1951-2002. The map on the lower left shows predicted temperature change for 20402069 and the map on the lower right shows predicted temperature change for 2070-2099, both compared to the 1961-1990 baseline average. Generated from data at www.climatewizard.org There is a general trend of decreasing rainfall since the 1950s through to 2002. Climate change predictions are for this trend to reverse and show a slight increase relative to a 1961-1990 baseline average. However, Hulme et al. (2001) indicate that rainfall predictions of General Circulation Model (GCM) models in southern Africa are not well defined due to the response of the region to ENSO events, and given the small increases predicted through to

126

C HAPTER FOUR

2099 it is important to treat these data with caution. The data from Mitchell et al. (2004) show that mean temperatures in July have been rising since the 1950s and this trend is predicted to continue through to 2099. Although this model does not predict minimum temperatures, data from the WORLDCLIM model (Hijmans et al., 2005) indicate a current average monthly minimum temperature of 3.4 ◦ C in the south western corner of the study area. Observations from this area from 2011, as well oral records from residents, show that occasional frosts at this time of year result in widespread crop damage and the deaths of young trees. Occasional frosts occur throughout the study area and are a source of mortality for young Baikiaea plurijuga trees (Calvert, 1986a). The effect of rising temperatures on the crucial pre-rain period in October was also examined using the ClimateWizard tool. Changes to climate in October appear to show increasing temperatures and decreasing rainfall (Mitchell et al., 2004). This has potential consequences for Zambezi Teak forests as well as other species in the south east of the study area which produce a pre-rain flush. These species will experience considerable stress from a combination of higher temperatures and lower precipitation, whilst the survival of Zambezi Teak seedlings may also be affected (Calvert, 1986a). Most of the population of Zambia (and especially south western Zambia) rely on subsistence agriculture. The start of the rainy season in November/December is a critical time for the establishment of crops and any variation in these conditions has the potential to affect crop yields and result in widespread crop failure. Zambia, and southern African countries in general, are poor with deficient infrastructure and rapidly growing populations. This makes the region extremely vulnerable to any change in climate which affects the hydrological cycle (Schulze et al., 2001). Climate Change and Forests

The maps of predicted climate change in Zambia show clearly that these changes will be most extreme in the south west of the country, in exactly the area where Zambezi Teak forests are found. Zambezi Teak forests and Mopane woodland show similar phenological characteristics, at least in the south east of the study area. Leaf flush starts increasing in December, peaks in February and March and starts decreasing again by April. This relatively short growing season will be more vulnerable to change than in species or areas that have longer growing seasons. The long period of leaflessness confers a degree of drought resistance, but it is possible that the forests are vulnerable if this resistance is at the limit of the physiological adaptations which allow the forests to survive in this area.

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

127

Although the predictions for temperature increase are alarming, and those for rainfall increase not well defined, mean annual figures are often misleading when it comes to their effect on critical phases of the lifecycle of a species. In the study area the start of the wet season in October/November is a critical time for many plants. It is the hottest time of year and comes after five to six months of dry weather. Some tree species produce an early pre-rain flush (Ochna pulchra, Terminalia sericea) which is sustained through the use of limited water reserves stored in the plant tissue (Childes, 1988). The ability to flush early provides a competitive advantage to these species, but any changes to temperature or changes to the timing and volume of rain at this time of year may have a deleterious effect on them. Zambezi Teak seedlings are evergreen and are able to survive in dry soil through the dry season. This gives them a competitive advantage relative to other deciduous tree species, as well as the herbaceous and grass layer, as they are able to start growing as soon as the first rains fall (Childes, 1988). Nevertheless these seedlings will have limits to their survival and increases in temperature during the critical months in October/November may affect their viability. Baikiaea plurijuga and Bauhinia petersiana bear pods which desiccate progressively during the dry season until they twist and explode, dispersing seeds several metres from the parent tree. Warmer temperatures in the dry season and the accompanying lower humidity may result in the pods drying faster and dispersing seeds earlier, thereby exposing them to predation by rodents for longer periods of time. There are numerous examples of different species in the study area having evolved a response to the variable nature of the climate and the timing of rainfall, with different timings of fruiting, fruit maturation, flowering and leaf emergence and senescence. Phenology is an important adaptation, enabling large numbers of species to coexist (Cleland et al., 2007). Changes to this regime will potentially have an effect on the structure and composition of the vegetation. Although there are few studies from Africa, several metaanalyses have looked at identifying the ‘fingerprint’ of climate change by studying how species respond to it (Root et al., 2003). These include diverse changes in the spring phenology in plants (Cleland et al., 2007; Parmesan, 2006), an earlier start to the growing season in European trees (Chmielewski & Rötzer, 2001) and the shifting distributions of mammals, molluscs, grasses and trees (Root et al., 2003; Walther et al., 2002). There is ample evidence for biological communities responding to climate change and it is against this background that we attempt to infer responses of Baikiaea plurijuga forests to predicted climate changes over the next 80 years. Since Zambezi Teak is known to be frost sensitive the model predictions

128

C HAPTER FOUR

for the coldest month of July have potential implications for Zambezi Teak forests. With predicted mean temperatures set to rise by between 2 − 4 ◦ C over the next 80 years, this may have the effect of reducing the frequency and intensity of frosts which would be a favourable development for Zambezi Teak forests. However, lower rainfall and increased evapotranspiration could lead to the depletion of deep water reserves upon which Zambezi Teak forests depend. The human induced changes in forest structure over the past 50 years have increased significance in the light of predictions for climate change in the area and the future resilience of the forests. While it is difficult to be conclusive about the effects of climate change on forests, the effects on rural agriculture are more predictable. Jones & Thornton (2003) predict an average decrease in maize production in southern Africa of 10% by 2055. However, this may be larger in the study area due to more extreme changes predicted. Lobell et al. (2008) anticipate that southern Africa is a “climate risk hotspot” with respect to crop production and the indirect effects of lower crop production are increased reliance on the “safety net” of NTFP (Shackleton & Shackleton, 2004). Although population density in the areas is low, increased reliance on NTFP is likely to place more pressure on forests. Lower crop productivity leads to increasing demand for land and results in larger areas of forest being cleared for agriculture (Shackleton & Shackleton, 2004). In the next section I examine how communities in the study area utilise forest products and examine the contribution this makes to overall livelihood strategies. Inasmuch as climate change affects forest species distribution and forest structure, it is the details of forest use that will determine how rural communities are affected.

Sustainable Livelihood Analysis Globally, the main cause of deforestation is the clearing of land for commercial or subsistence agriculture (Angelsen & Kaimowitz, 1999; Hosonuma et al., 2012). There is no unifying theory of land use change which enables us to predict how land use will change given a set of environmental and social conditions (Geist & Lambin, 2002; Lambin et al., 2001; VanWey et al., 2005). The drivers of behaviour that lead to forest clearing are complex and have been found to be context dependent (Babigumira et al., 2014). To explain the decisions which determine land use change some authors have used a sustainable livelihoods framework (e.g. Bebbington, 1999; Campbell et al., 2002). At its simplest, the concept of a sustainable livelihoods framework is an integrated measure of the circumstances which enable an individual to

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

129

earn a living (Chambers & Conway, 1992). The core idea behind using a sustainable livelihoods framework is that a measure of poverty, together with a set of contextual factors, enables an understanding of livelihood decisions which determine patterns of resource use (Babigumira et al., 2014; Campbell et al., 2002). Poverty is measured using five categories of ‘capital assets’ (Figure 4.12) which communities possess in different amounts depending on their circumstances. These assets are natural, physical, human, social and financial capital and are the building blocks upon which households engage in markets (Bebbington, 1999; Chambers & Conway, 1992). It is the local context that influences the choices made by individuals who improvise livelihoods depending on the social, economic and ecological circumstances in which they find themselves (Chambers & Conway, 1992). The livelihoods framework is used in this study as a general method for thinking about multiple interacting factors and the effect they have on livelihoods and hence on resource use decisions (Babigumira et al., 2014; Babulo et al., 2008). In a general sense the impacts of SFM and REDD+ can be assessed in terms of their impact on livelihoods and the decisions people will make when a REDD+ project impacts their lives. The livelihoods approach to analysing the different elements that shape people’s livelihoods, the factors that influence them, and the linkages between these factors are shown in Figure 4.14.

Participatory Rural Appraisal

The three small villages of Kapanza, Moyandulwa and Simbeza are typical of the rural lifestyle of the people who live around the Zambezi Teak forests. Each village has a headman and the 50 households live a subsistence lifestyle in a remote part of the Mulobezi Game Management Area (GMA). Using techniques of Participatory Rural Appraisal (PRA) (participatory mapping, transect walks, timelines and change analysis, seasonal calendars, livelihood analysis and group discussions) (Chambers, 1994), assessments were made of local people’s perceptions of institutions in which they participated and the effectiveness with which these institutions addressed their needs. Participatory mapping produced a measure of local infrastructure and livelihood analysis enabled us to produce a list of income and expenditure ranked in order of importance. Seasonal calendars indicated times of shortage where livelihood decisions and resource use decisions would be determined by the straightened circumstances. The seasonal calendars showed when times of plenty occurred and how these influenced decisions about resource use. A database of forest products and their relative importance was compiled, and

130

C HAPTER FOUR

Figure 4.12: The five capital assets in a livelihood framework (adapted from Bebbington (1999) and Campbell et al. (2002)

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

131

Figure 4.13: Drums are an important part of daily life. They are used to communicate across the villages to call meetings, to announce that traditional beer is ready to drink or that a visitor has arrived, and during ceremonial events

132

C HAPTER FOUR

Figure 4.14: Conceptual framework for livelihoods approach. From Babulo et al. (2008)

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

133

a measure of their financial capital was made, and the threats which wildlife or the environment posed to it were assessed.

Figure 4.15: Participatory mapping and the final results of the process. The process offers the community a chance to communicate their understanding of the landscape and what is important to them. Inasmuch as landscapes are social constructions of space and place and are “our unwitting cultural autobiographies” (Lewis, 1979, 12), the process of participatory mapping allows the researcher to gain an insight into the cultural significance of the landscape

Livelihood Analysis

For the livelihoods analysis I grouped 22 variables which contribute to livelihood decisions in the study area under five categories of capital: Natural, Social, Human, Physical and Financial (Figure 4.12). Variables were selected based on information which emerged from the PRA exercises and adapted from the literature(Babigumira et al., 2014; Bebbington, 1999; Chambers & Conway, 1992; Chhatre & Agrawal, 2009; Hahn et al., 2009; Sayer & Campbell, 2004; Suich, 2010). Each of these variables was assigned two criteria which would determine if it was scored low (minimum score of 0) or high (maximum score of 10) (see Table 4.1). If the variable did not fully meet the scoring criteria, a score between 0 and 10 was assigned.

134

C HAPTER FOUR

Figure 4.16: Ranking the relative importance of income sources from forest products

No livestock, no draught power Isolated, no roads or communications

Range of livestock to supply draught power, food Roads, mobile phones

Zero compliance Community members do not trust each other No leadership/disputes over leadership Self interest dominates the community Weak kin and ethnic (tribal) networks No social organisations

Struggle to buy basic household assets No agricultural implements

Full compliance with traditional and local rules Strong relationships of trust Wise and well established leaders Community members value mutual interest Strong kin and ethnic (tribal) networks Well run social organisations

Social Capital Adherence to rules Relationships of trust Leadership Mutuality of interest Kin and ethnic networks Social organisations

No bank account or cash saving Never receive payments from CRB

Criteria for low score (0)

Full range of cooking pots, blankets, clothes Ploughs, hoes, knives

An amount of savings in cash Regular payments in accord with the legislation

Financial Capital Savings Revenue from CRB

Physical Capital Household assets Agricultural implements Livestock Infrastructure

Criteria for high score (10)

Type of Capital

Table 4.1: Livelihood variable scoring criteria H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM 135

Fertile soils, with fertility maintained Clean water, easily accessible Forests intact, provide materials needed Grazing land available Plenty of land available

Completed Secondary school Knows tree names, uses traditional materials Community members with special skills Medical / clinic available Plenty of labour for essential tasks

Human Capital Western knowledge Traditional knowledge Skills Health Labour availability

Natural Capital Soil fertility Water Resources Forest Resources Grazing resources Land quantity and quality

Criteria for high score (10)

Type of Capital

Table 4.1 continued . . .

Infertile soils, fertility declining No reliable water source or far away Deforested, no materials for community Grazing land degraded Land overcrowded and in high demand

No formal schooling, not literate No traditional knowledge No community members have special skills No medical facilities available No young, fit people to perform manual labour

Criteria for low score (0)

136 C HAPTER FOUR

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

137

Forest Products

The community relies on forests for food, medicine, fuel, household items and products to sell to generate cash. Different products are obtained from each habitat (Dambo (D), Kalahari Woodland (KW) (Lutondo), Baikiaea plurijuga forest (BPF) (Mukusi) and Chundu, an ecotone habitat containing a mix of species) and in some cases there is clear separation between men and women with respect to the different products harvested. The majority of products are harvested by the entire household with no separation between men and women. Honey gathering is done by men, and is a source of widespread fires. Men set fires to clear the forest of wild animals and prepare an easier path for a possible return visit. Observations and interviews with honey collectors returning from their expeditions indicate that men also gain status from the difficulty and distance they have to travel to collect the honey, and may set a fire to mark a spot on the horizon they can point to when they return to the village, so they can show other community members the distance.

Figure 4.17: A traditional bellows (or Mvubu) used to fire the furnace for manufacturing knives, axe heads and hoes. The blacksmith is an important part of the community and a source of income through his employment of several assistants

138

C HAPTER FOUR

Figure 4.18: Bwili (Dioscorea spp.) is a tuber which is boiled and eaten like a potato. The plant is a climbing vine and the tubers are located by following the dry tendrils towards the ground and carefully digging them out. These are used in the household or sold to generate income Income and Expenditure

Sources of income for the community at Mabwe are mostly natural products which are collected from the surrounding forests. Fish and sweet potatoes are the exception to this, with fish being highly seasonal and sweet potatoes requiring a considerable amount of labour to cultivate. The relatively low rank of employment as an income generator indicates the extent to which the community is reliant on natural products for generating income. Table 4.2 shows different sources of income with their importance (a variable which combines the degree of effort required to collect or produce and the temporal availability of the item) and the amount of income they generate as a proportion of annual income. Expenses for community members are shown in Table 4.3. These expenses are mostly for manufactured goods which the community are not able to make themselves. School fees are the exception to this and are the only service which the community pays for in cash.

Tonga name

Munkoyo

Buchi

Kusambala

Bwizu

Bwili

Chimbwale

Inchili

Kufula (blacksmith)

Name

Tuber

Honey

Fish

Grass

Tuber

Sweet potatoes

Pounding Mortar

Blacksmith & other

8

7

6

5

4

3

2

1

7 5

Only one individual is blacksmith but he employs assistants. Some income from safari company employment

3

6

5

For sale to surrounding community

Cultivated for sale locally and in towns

For sale locally and in towns

For sale to areas with less grass and safari camps

3

1

For sale, brewing alcoholic drink. Some people keep beehives Both men and women catch for selling and eating

2

Income ranking

Eminia holubii collected, beaten flat and dried for sale locally and in towns

Importance Information

Table 4.2: Sources of income for Mabwe community H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM 139

140

C HAPTER FOUR

Figure 4.19: Collecting and processing Munkoyo for sale in nearby towns. From top left: the plant is located among grass stems, the tubers are dug up and beaten flat, after which they are left to dry on the roofs of houses before being bundled up for sale. Munkoyo is an important source of income for rural communities, but large quantities are sold in towns and villages and the sustainability of the resource needs urgent research

Livelihood Analysis

The results of scoring the variables in the livelihoods framework are presented in Table 4.4. There are different numbers of variables in each category which makes a direct comparison of the relative importance of type of capital difficult. To overcome this the scores were transformed by calculating the natural logarithm of each category so their relative importance could be ranked. Natural and Social capital equally contribute the most to the livelihood assessment. The community has slightly lower levels of Human capital, whilst the relatively low levels of Physical and Financial capital make a small contribution to the overall score. The score of 150 out of a possible 220 indicates that the community has a livelihoods index of 0.68 (Chhatre & Agrawal, 2009; Persha et al., 2010). A livelihoods index of 1 would indicate that all livelihoods needs are satisfied under the different categories of capital considered in this assessment.

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

141

Table 4.3: Expenses requiring cash payment for Mabwe community members, arranged in order of importance. These are for goods which they cannot manufacture themselves with the exception of school fees which is the only service they pay for Product

Importance

Salt Soap Cooking oil Blankets Pots and plates School fees Clothes

1 2 3 4 5 6 7

Approximate amount in Zambian Kwacha (GBP1 = ZMK8) ZMK10/month ZMK50/month ZMK50/month ZMK150 each: 2- 4 per year ZMK120 per year 3-5 children: ZMK60 per term per child ZMK300 per year

142

C HAPTER FOUR

Discussion

Rural communities in western Zambia are completely dependent on forest resources for every aspect of their daily lives. One of the major omissions from this study is the reliance of the community on traditional medicines. The practices of the traditional healer are not shared openly and the time spent with the community at Mabwe was not sufficient to build up measures of trust so that the extent of this practice could be assessed. Previous studies have shown that up to 95% of rural and urban dwellers in southern Africa use traditional medicines (Mukamuri, 2009) and the importance of this resource should therefore not be overlooked. Indications of this knowledge being more widespread emerged from how local research assistants identified plants. If they could not remember the name of a plant, they first tried to remember what it was used for. This frequently included a medicinal use. Sometimes wives were consulted as the plant in question was something which women used more than men. Once a use was established this jogged the memory to produce a name, which I then was able to look up in Fowler’s (2007) dictionary of vernacular names. The link between plants and their utility value (medicinal or otherwise) shows the importance of forest resources to the local community. Names of plants and trees may be forgotten, but remembering their use is the key to survival.

Figure 4.20: Maize and sorghum on a drying rack. Sorghum was traditionally more widely grown as a staple food, but is now only grown for brewing beer The implications of these findings for REDD+ are important. Planning to

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

Table 4.4: Livelihood variable scores Type of Capital

Mabwe Score

ln

Financial Capital Savings Revenue from CRB Subtotal

1 3 4

1.39

Social Capital Adherence to rules Relationships of trust Leadership Mutuality of interest Kin and ethnic networks Social organisations Subtotal

8 5 8 7 10 9 47

3.85

Physical Capital Household assets Agricultural implements Livestock Infrastructure Subtotal

7 7 4 3 21

3.04

Human Capital Western knowledge Traditional knowledge Skills Health Labour availability Subtotal

4 10 10 2 7 33

3.50

Natural Capital Soil fertility Water Resources Forest Resources Grazing resources Land quantity and quality Subtotal Total

6 9 10 10 10 45 150

3.81

143

144

C HAPTER FOUR

manage forests as stores of carbon is an oversimplification of their utility to local communities and will result in an oversimplification of SFM practices unless a detailed local understanding of forest dependence by local communities is taken into account. SFM policy may have to accept that community usage of forests will involve occasional fires which set the process of CO2 sequestration back in the areas where they burn. Eliminating fire from the system would not be ecologically viable, but it would also not be viable if people are going to go on living their lives in these forests as they currently do. The livelihood analysis is not able to establish clear links between livelihood variables and forest use. In studies which attempt to make these causal connections, the relationships are seldom clear (e.g. Babigumira et al., 2014). One reason for this is that livelihood decisions are dependant on local context as well seasonality and variation between years (Ellis, 2000). Different years may bring drought, floods, insect outbreaks or shocks such as elephant damage to crops or lion predation of livestock. At Mabwe I was told of a single incident when elephants damaged crops which affected the ability of the community to produce a surplus for sale that year, which prevented the Parent Teachers Association (PTA) progressing with construction of their school. The data is useful in a more general sense to assess the extent to which the community in Mabwe is resilient in the face of challenges they experience in making a living. High levels of social capital facilitate cooperation, build relationships of trust and are essential for long-term resource management and have been shown to be important for sustainability (Coleman, 1988; Pretty, 2003). High levels of natural capital support a wide range of livelihood choices and strategies (Chaminuka et al., 2014). The high levels of natural and social capital underpin the resilience of the community at Mabwe. In a global analysis of forest clearing and rural livelihoods Babigumira et al. (2014) found little evidence for forest clearing being driven by extreme asset poverty despite the widely accepted theory that poverty alleviation is an essential step to preventing deforestation (VanWey et al., 2005). In the Mabwe community the categories of physical and financial capital are the lowest, indicating that the households are asset poor. The findings at Mabwe support those of Babigumira et al. (2014) with little evidence of widespread deforestation in the study area (chapter six). More disturbingly, Babigumira et al. (2014) suggest that it is households with moderate or high asset holdings that are likely to cut forests as they have the means to expand their agricultural activities to improve their livelihoods. The implications for REDD+ are disturbing. REDD+ projects may encourage deforestation if

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

145

income from the sales of carbon results in an increase in asset holdings and eliminates the capital constraints that have kept communities from clearing forests.

Conclusion: A Holistic Social-Ecological Synthesis The lack of an overarching theory of land use change (Lambin et al., 2001) and Ostrom & Cox’s (2010) caution to avoid the “panacea problem” make a case study approach to land use change and SFM a requirement for understanding the potential impact of REDD+. A number of issues emerge from the recognition that each locality is unique. The most pertinent of these is that identifying the factors which confer both social and ecological resilience (Folke, 2006), is more important than advocating a solution to SD issues which arise at one particular time and space, or which advocate panaceatype solutions. Development projects, including REDD+, should prioritise the preservation of factors which confer resilience. The south western part of Zambia is predicted to experience more severe climate change than other areas of the country (Mitchell et al., 2004). In a climate which is already quite extreme this could have important ecological effects on the forests and woodlands which underpin the ability of people to survive in the area. There is an urgent need for basic research into forest ecology to understand how these changes will affect forest survival. Phenological adaptations are an important evolved response to long dry seasons and phenological events are likely to be affected by climate change (Singh & Kushwaha, 2005; van Schaik et al., 1993). With current knowledge it is not possible to predict how these changes will affect the forests of south western Zambia. Urgent research into these basic ecological responses is required in order to understand how resilient the forests are in the face of changes in temperature, timing and amount of rainfall. Understanding the capacity of these forests to adapt to climate change will enable an understanding of how communities who live in the forests will be affected. A clearer understanding of climate change effects on phenology requires a better understanding of the environmental cues which trigger phenological events in dry tropical forest species. There is a requirement for detailed, long term phenological observations over a wide geographical area in the dry deciduous forests of southern Africa. These observations are easy to make in the field and the data are comparatively inexpensive to collect for those Forestry Department or Zambia Wildlife Authority (ZAWA) employees who are resident in these areas. Coupled with climatic and weather data this could reveal correlations between local climate events and phenology which

146

C HAPTER FOUR

would guide later experimental approaches to make the link between cause and effect. The study area has a hot climate, seasonal rainfall which is spatially and temporally variable and relatively infertile soils which make south western Zambia less than ideal for growing crops. This has implications for the livelihoods of the people who live in the area and their subsistence lifestyle. Growing crops for food is only one of many activities which are required for making a living. Development aid projects that propose improving crop yields (e.g. Almekinders et al., 1994; Kwesiga et al., 1999) may have benefits, but they are not able to change the relatively infertile soils and erratic rainfall. More importantly, they do not address the capacity to withstand occasional catastrophic events such as insect outbreaks or elephants’ raiding of crops. It is in the many different livelihood activities, food sources and traditional skills, together with the abundant natural and social capital that the community at Mabwe possesses, that a capacity for resilience or adaptability is found. There are several implications in this holistic synthesis for REDD+ implementation. Governance issues are central to REDD+ (Secco et al., 2013; Springate-Baginski & Wollenberg, 2010) and unique governance issues emerge from this case study. They will need to be addressed if REDD+ is going to be successful. The resentment brought about by the abrogation of the Barotseland Agreement (Sandys, 1964) persists to this day (pers. comms. HRH Chief Inyambo Yeta) and needs to be addressed. Together with this must be a recognition of the capacity for governance which is demonstrated daily at the Kuta (traditional court) at Mwandi, and at the smaller Nkoshi meetings where local headmen judge minor cases. Siangulube (2007) has shown that people in Barotseland have more respect for directives governing the conservation of forests which come from the Barotse Royal Establishment (BRE) than those from central government. These traditional governance arrangements are an untapped resource for the governance of natural resource use in south western Zambia. The evidence for increases in material wealth resulting in increases in forest cutting (Babigumira et al., 2014) has important implications for REDD+. Poverty is difficult to define (Agola & Awange, 2014; Laderchi et al., 2003), but with a livelihood index of 0.68, the people of Mabwe cannot be said to be suffering from poverty. They experience uncertainty and unpredictability, which is partly the nature of the environment in which they live, but one does not observe extreme deprivation in the villages of Mabwe. Cultural ceremonies are regularly conducted, celebrations are held, leadership is strong and there are social institutions addressing health, education and

H OLISTIC S YNTHESIS OF A S OCIAL -E COLOGICAL S YSTEM

147

environmental issues. There are lean times and there are times of plenty. The possibility of REDD+ increasing the financial and physical capital of the community at the expense of natural or even social capital must be considered before any projects go ahead. There is a risk that the capacity for resilience which is essential for the survival of the community will be undermined if small increases in material prosperity result in large changes in natural or social capital. This synthesis has drawn together the many unique characteristics of the study area in an attempt to make a holistic assessment of the implications of future REDD+ projects in the region. In the following chapters I quantify land cover change in the study area over the 30 year period between 1975 and 2005 and then go on to compare the effects of different governance arrangements in Zambezi Teak forests in Zambia and Zimbabwe before concluding with a general discussion in chapter eight.

Chapter Five Scale Mismatches and Implications for SD and SFM

The concept of scale and the challenges it presents for management and research in ecology is well documented (e.g. Cumming et al., 2006; du Toit, 2010; Meentemeyer, 1989). However, this does not mean that adequate consideration is given to scale-based issues which emerge from these studies. The scale-based implications of research findings, particularly with respect to issues around REDD+ implementation and climate change, are often overlooked, but the theoretical basis for a treatment of scale in ecology is well enough developed (Levin, 1992; Turner, 1989). Efforts to implement REDD+ involve developing a methodology and a policy framework for the largest global programme of payment for ecosystem services that has ever been attempted (Angelsen et al., 2009). In this chapter I argue that there are scale-based implications across the entire range of actions which are required for any REDD+ programme to be successful. There is currently a deficit in the consideration of scale in the formulation of policy with respect to SFM and REDD+, and in the research aimed at improving Monitoring, Reporting and Verification (MRV) of Above Ground Biomass (AGB). I will examine some examples of scale mismatch and argue that the matching of different types of scale can contribute to REDD+ projects being implemented successfully. I will examine some of the issues that emerge when scale is made an explicit criterion which is addressed in an assessment of Reduced Emissions from Deforestation and forest Degradation plus (REDD+). 149

150

C HAPTER FIVE

Despite the acknowledgement of the coupled nature of social-ecological systems (Holling, 2001; Westley et al., 2002), few attempts have been made to reconcile the important conceptual differences between how social scientists and ecologists theorise scale. At the heart of these different approaches are differences of ontology and epistemology between the different academic traditions as discussed on page 24. The demands of Sustainable Development (SD) for interdisciplinary, holistic synthesis require that differing conceptualisations of scale are at least discussed in the same space, even if reconciliation is challenging. The demands for Sustainable Forest Management (SFM) are even more immediate. Unruh (2011, 185) argues that “the continued inattention regarding the human realities of putting and retaining trees on tropical landscapes has resulted in proposals that are neither feasible nor realistic” and highlights the extensive attention given to ecological research on carbon storage in tropical forests, compared to attempts to address the sociopolitical problems of achieving levels of carbon sequestration that will make a difference to atmospheric levels of CO2 . Part of addressing this issue is developing a cohesive body of theory that can inform a common understanding about scaling issues (Gibson et al., 2000b; Unruh, 2011). In this chapter I address the question of how scale and sampling methods affect the implementation of REDD+ in the deciduous forests of south western Zambia. The spatial implications for sampling across vast geographical areas, which is at the heart of the measurement of AGB, are also examined. I highlight the economic assumptions which underpin REDD+ and show how a mismatch of analytical scale in the economic sphere results in policy deadlock between negotiating parties, and argue that this reverberates as political disputes in climate negotiations. Explicit consideration of scale can help to avoid some of these problems. I also explore how different conceptualisations of scale can be reconciled between social scientists who subscribe to a constructivist notion of space and scale and ecologists who have more positivist conceptualisations of scale and space.

Concepts of Scale Scientists who study scale-based phenomena accept that explanatory variables for a phenomenon in question change as the scale of analysis changes (Gibson et al., 2000b). In searching for commonalities between ecological and social concepts of scale, this characteristic is common to both disciplinary areas, which struggle to formulate generalisations which are applicable across scales (Brenner, 2001; Gibson et al., 2000b; Marston, 2000; Turner et al., 1989). Figure 5.1 shows the way scale relates to different phenomena in

S CALE M ISMATCHES AND I MPLICATIONS FOR SD AND SFM

151

ecological conceptions of the concept and the conceptualisation of social scientists. The idea of a hierarchy of scale which is applicable in both concepts is shown as an overarching concept because of its potential as a tool for mapping scales of ecological phenomena to appropriate scales of social phenomena. This mapping is likely to be highly variable depending on both the ecology and the society under examination and there is unlikely to be a uniform mapping of scale in different contexts. Nevertheless, the arrangement of these scales in an explicit and hierarchical structure is designed to facilitate mapping their interaction. I will show an application of this mapping on page 162.

Figure 5.1: Diagram showing how different concepts of scale apply to ecological conceptualisations of scale and those extant in sociological conceptions of scale. Cartographic scale is included for the sake of completeness but has little role to play in ecological research. Both conceptualisations of scale have properties which are common. Size (or dimension), level (locations along a measurement dimension of a scale), grain (precision of the measurement) and extent (size of the analytical dimensions of a scale) are all common properties of scale in both disciplines. I make hierarchy an overarching property because of the value it has in structuring the many objects and processes of scale which emerge from studying an actual social-ecological system. Those who support constructivist notions of space and scale are likely to reject the application of hierarchy if it is understood to be only either inclusive (with lower categories being subsets of higher categories, as in classical reductionist science) or exclusive (with lower categories not contained within higher categories). The addition of the possibility of a constitutive hierarchy, in which new objects, structures or processes are emergent from the combination of existing ones, adds to the complexity of hierarchy as a concept (author’s own diagram)

152

C HAPTER FIVE

Scale in Ecology

The consideration of scale has a long history in ecology. Clements (1916) described his successional theories using a temporal scale, but did not consider explaining patterns of vegetation growth on a spatial scale. The development of landscape ecology is an attempt to explain the influences of spatial heterogeneity on biotic and abiotic processes and emphasises large spatial scales and the ecological effects of the spatial patterning of ecosystems (Turner, 1989). Although ecologists have long understood the significance of scale, they have largely misunderstood its importance with respect to policy (Norton, 1998). Awareness of this deficit in linking scales across ecology and society is starting to emerge (e.g. Cumming et al., 2006), but as I show on page 155, ignoring the policy and management implications of ecological research aimed at improving methods of Monitoring, Reporting and Verification (MRV) for REDD+ continues in even the most recent work. The term ‘scale’ refers to the spatial, temporal, analytical or quantitative dimensions used by researchers to measure and study objects or processes (Gibson et al., 2000b). I use the term ‘small scale’ to refer to smaller areas or objects under investigation and ‘large scale’ to refer to large areas or quantities, which is opposite to the cartographic use of the term which refers to the size of the scale on a map. Different positions along a measurement dimension are referred to as levels of that scale. Levels of a temporal scale would include short or long time periods, while levels of spatial scale would include phenomena which are small or large in size. Deforestation by rural farmers, for example, takes place over a smaller area (and level) than clear felling of forests for timber. Scale affects the ecological phenomena under investigation in different ways (Wiens, 1989). 1. Changing scale results in changing relationships between ecological entities and the physical or biological linkages which are important for determining ecological patterns. Relationships between climate and vegetation show more clearly at large spatial scales and may disappear at small spatial scales where competition and other biological processes are more important. 2. Ecosystem transfers of species or nutrients in the system, or measures of how closed or open an ecosystem is, depend on the scale at which the system is examined. 3. When larger scales are examined (for example a continental scale biomass assessment), measurements use a larger grain size (lower resolution) than when smaller quantities (for example a ≈ 10,000 ha REDD+ project) are being studied where a small grain size, or higher resolution, is more appropriate. The

S CALE M ISMATCHES AND I MPLICATIONS FOR SD AND SFM

153

consequences of changing scale are to lower the resolution and this results in a scale mismatch when trying to explain phenomena which occur at a higher resolution. There are few, if any, scale-independent processes and the loss of explanatory power as grain size increases is unavoidable. 4. The spatial and temporal variance of a variable changes when the scale of measurement is changed. As grain size increases, the ability to discern spatial variability decreases.

Scale has a powerful influence on the ability to make statements about ecological process and pattern and this makes it essential to limit the reporting of these processes to the scale at which they have explanatory power. On page 155 I examine how ignoring the effect of scale in ecological research leads to incorrect assumptions around data gathered from remote sensing studies of AGB.

Scale in Human Geography Studies of space and scale in human geography have a long history, but recently it is possible to distinguish two distinct paths in the consideration of spatial concepts and issues of scale. Quantitative geographical analysis has a long history and continues to be an active area of research, using GIS and increasingly sophisticated analytical methods to identify local indicators of spatial association (Fotheringham, 1997). An important step in clarifying analytical techniques with respect to the effects of scale was the identification of the ecological fallacy, which was the observation that the correlation of two variables increases for spatial units of larger resolution, making it incorrect to infer relationships between individuals from results of studies at a larger scale (Robinson, 1950). This work laid the foundation for the identification of the Modifiable Areal Unit Problem (MAUP) (Openshaw & Taylor, 1979) which highlights the effects which grain (resolution) can have on the findings of geographic analysis. The MAUP is applicable across all areas of scientific equity that examine spatial phenomena, and has important implications because it makes it difficult to identify the resolution which is best for the investigation of both social and ecological phenomena. However, the idea that space and scale in the social arena can be adequately described in a conventional sense, by Cartesian coordinates, has come under question in the last two decades (Sheppard & McMaster, 2008). Researchers have queried what space and scale mean in the social arena. Sheppard & McMaster (2008, 15) summarise the central concerns of scale theorists:

154

C HAPTER FIVE

Their analysis begins with two claims: First, if space is socially constructed, the same must also be true of scale, so we need to think about how scales come into existence. Second, if scale is socially constructed, then we cannot simply take for granted the existence and importance of the geographic scales usually invoked in human geographic writing: Neighbourhood, city, regional, national, global. Rather, we need to understand not only why their relative importance may vary over space and time, but also whether these are even the right scales to be thinking about

The idea that space and scale are socially produced and contingent on political struggle arises from the seminal work of Henri Lefebvre (Lefebvre & Nicholson-Smith, 1974) and Neil Smith (1984). In Smith’s theory, the constructivist approach to the production of space and scale has been informed by Marxist theory and its critique of the capitalist system and the conditions it imposes on labour and the value of goods. Both Smith’s and Lefebvre’s works are informed by Marxist theory (Marston, 2000). Various authors have critiqued and attempted to refine these theories by looking at how this social construction takes place (Brenner, 2001; Marston, 2000) but have not significantly altered the views of an active and prominent (and I would argue, dominant) group of social scientists who support constructivist theories of space and scale. There are fundamental differences between the ontological conceptualisation of space and scale which ecologists support, which are in line with a positivist tradition, and of those human geographers and social scientists who ascribe the existence and structure of space and scale in the social domain to social constructions predominantly arising from the capitalist system and its role in structuring human society. These are difficult views to reconcile. However, the interaction between social institutions and ecological systems, and the demand for interdisciplinary work that is required for solving problems which emerge from this interaction, require not only that there be a more explicit recognition of scale in ecological, economic, social and institutional factors, but also that there be an accommodation of different ontological positions. How can this be achieved? In the next section I discuss the potential for hierarchy theory to be a conceptual bridge for linking ideas of ecological space and social space and suggest that ontological differences need not be a barrier to avoiding scale mismatches between the different systems.

S CALE M ISMATCHES AND I MPLICATIONS FOR SD AND SFM

155

Scale and Hierarchy

Scales are closely linked to the concept of hierarchy, and it is the hierarchical arrangement of knowledge that contributes to coherence and congruence in the natural sciences (Gibson et al., 2000b; Wilson, 1999). Gibson et al. (2000b) claim that the social sciences lack a clear concept of hierarchy, and I suggest this may be due to the undertones of dominance and control which the idea of a hierarchy in social systems implies. However, there are several concepts of hierarchy which are different to that commonly understood in the natural sciences. Biologists understand a hierarchy across scales to involve causally linked phenomena along an analytical scale. Theory at larger scales of the hierarchy is informed as phenomena at the smaller scales are better understood and contribute to the development of theory. This hierarchy is at the heart of the reductionist method, but the study of complex systems in particular requires a consideration of hierarchy that is more nuanced and has more in common with a concept of hierarchy that would be application to social phenomena (e.g. Holling, 2001). Simon (1962) was one of the earliest proponents of hierarchy theory which he developed in order to more easily understand complex systems. Hierarchy theory lacks clearly defined terms and although it has had a strong influence in ecology, leading to the ecological paradigm that centres on the study of pattern, process and scale, the theory does not have a well developed methodology and its application to different fields is in its infancy (Wu, 2013). The application of different forms of hierarchy to the study of social institutions has the possibility to yield insights, especially when considering how these institutions interact with respect to managing natural resources. One of Ostrom’s (1990) eight design principles for successful CPR management is the existence of a nested hierarchy of enterprises which are involved in administering the sustainable utilisation of a CPR (see page 85). The weakness of vertical linkages between user groups and different levels of government is the source of potential scale mismatches when it comes to conceptualising institutional arrangements and the policy which informs them. To a lesser degree than in ecology, there is no one correct level at which to investigate social phenomena, and a multi-level approach to studying institutions is required because of the interactions which occur between levels and how interactions affect the sustainable utilisation of natural resources (Gibson et al., 2000b). Social phenomena reverberate across hierarchies in social institutions over a shorter time scale than they do in ecological systems, and in a way that may be socially constructed, but the role of hierarchy, even if it does not have ontological significance, is equally important for understanding the complexity of these interactions of social constructs. Later

156

C HAPTER FIVE

in this chapter I suggest a role for different forms of hierarchy so that social institutions can be arranged in a hierarchy alongside ecological systems making mismatches of scale easier to visualise. Ultimately I suggest that the use of hierarchy facilitates removing the epistemological barriers and allows for the accommodation of different forms of knowledge in assessments of social-ecological systems.

Effects of Changing Scale

Ecologists working on scale related phenomena such as vegetation change continue to look for ways to extrapolate information between spatial and temporal scales, but to a large extent any predictions are scale dependent and mechanisms that are found to operate at one scale rarely explain patterns found at different scales (Gibson et al., 2000b). The extent to which samples of data from a population are adequate to make statements and claims about that population are at the heart of statistical methodology. When these samples are used to make generalisations across a wide geographical area, the assumptions and analytical methodology become affected by issues of scale. This is because the characteristics of biological populations vary across geographical space. This variation is the basis of evolution and natural selection. Samples of a population at a particular location will be less representative of a population further away than they are of a population in the near vicinity. Despite this, allometric equations which are used to approximate the biomass of trees are frequently applied to vast geographic areas outside of the area where the allometric relationships were calculated. Archibald & Bond’s (2003) study of tree allometry in Acacia karroo in South Africa shows that branching structure and stem architecture differ significantly in different environments. The application of allometric equations across large areas of Africa, without consideration of the possible variation in growth forms in different environments, is an example of scale issues not being taken into account in an analytical study. In this chapter I assess the implications for sampling of scaling up in space, which is what happens when allometric equations developed at the microlevel are widely applied at the mesoscale or macroscale in the calculation of AGB in dry deciduous forests of Africa.

S CALE M ISMATCHES AND I MPLICATIONS FOR SD AND SFM

157

Scale Mismatches Scale Mismatch in Ecology

Several authors have called for a more appropriate consideration of scale in global climate change research (Kates & Wilbanks, 2003; Wilbanks & Kates, 1999). Climate change may be a global phenomenon, but its effects, and the actions needed to ameliorate these effects (and this includes REDD+) are taken at a local scale. There is a need to improve the understanding of how local scale phenomena are linked to regional or global scale phenomena, but more importantly there is a need for researchers to restrict claims about the findings of macro-scale studies to the scale at which these finding are relevant or enhance understanding of the process. When individuals, communities and the institutions in which they participate make decisions about resource utilisation, it is at a scale which differs from that of the macro-scale. The diagrams and discussion in Peterson et al. (1998) depict time and space scales in boreal forests and their relationships to some of the processes that structure the forest, indicating the disparity between the scale of processes in different domains. These processes will be different in dry deciduous African forests, but the concept of scaling will be similar, and the scale at which satellite imagery measures structure will determine which processes that can be informed by these structural measurements. Many studies have used low resolution imagery (≈1km) to examine land use change at the continental or subcontinental scales (e.g. Achard & Blasco, 1990; Brink & Eva, 2009; Cabral et al., 2006). This appears to be a particularly favoured approach in Africa, presumably because of the difficulty of accessing some areas and the lack of data from ground based research. However, the results from these analyses (e.g. Baccini et al., 2008; Saatchi et al., 2011) are not suitable for mapping deforestation at smaller administrative scales which are relevant for SFM by communities. Where pixels represent >1km2 many of these pixels will cover an area with more than one land use category and therefore cannot be accurately represented in a classification by one land cover class. This may lead to errors in estimating forest cover at a scale which is relevant to local communities (Boyd & Danson, 2005). This is a mismatch of analytical scale which appears to be an all too common error among natural scientists who are trying to develop improved methods of MRV for REDD+. In many parts of Africa deforestation is related to local factors such as land tenure, agricultural expansion and government policies (Hamandawana et al., 2005). For this reason the results of these continental scale studies are applicable only in a general sense to local communities. They may be important for conceptualising issues

158

C HAPTER FIVE

relevant to land use change, but they do not provide the information required by local government or communities to manage their local forest resources. However, the data can be useful when used at the appropriate scale. As a consequence of Saatchi et al.’s (2011) work many countries now have AGB data on a national scale for use at international negotiating forums where the scale of the problems under discussion matches the scale at which the data are informative. Duveiller et al. (2008) recognise the problems associated with analytical scale in the Congo River Basin. They used high resolution imagery to map deforestation in the central African forests, but augmented this with a sampling based approach used in combination with object oriented methods. It remains to be seen whether this data can be used to inform community level actions which need to be taken for SFM, but it is encouraging to see an explicit recognition of the scale mismatch inherent in using low resolution imagery, and an attempt to address this issue. Hudak & Wessman (2000) use an approach which combines the advantages of both high and low resolution data to assess deforestation in Malawi, but their approach is not widely used in assessing deforestation in other areas in Africa. Nevertheless Hudak & Wessman’s (2000) study is an example of an analytical approach which matches the scale of the analysis with the scale of the problem. Another excellent example of using remote sensing data at the appropriate scale is the work by Kamusoko & Aniya (2007) in the Bindura district in Zimbabwe. Landsat data is used in an analysis to produce a model of land use change through to 2030 (Kamusoko et al., 2009) and thus yield information relevant for planning and management at the appropriate scale. Although natural scientists are tempted to produce analytical tools with wide application, the questions of scale which are inherent in any spatial analytical methodology cannot be ignored. To some extent the analytical techniques are limited by the resolution of the data which is available, or by the computer processing power required to increase the extent of a study without lowering the resolution. If these limitations are present they should limit the scale at which the analytical process can be informative and the temptation to make claims for the utility of an analytical product should not ignore the scale at which that product may be useful. There are few allometric equations which have been developed for subSaharan Africa and the equations in use therefore have a wide application (Henry et al., 2010). In tropical forests globally, errors in biomass estimates may contribute as much to differing estimates of carbon emissions as uncertainties in deforestation rates (Houghton, 2005). The most important predictors of tree AGB, in order of decreasing importance, are trunk diameter,

S CALE M ISMATCHES AND I MPLICATIONS FOR SD AND SFM

159

wood density, tree height, and forest type (Chave et al., 2005). The fact that trunk diameter, wood density and height are good predictors of AGB, potentially allows for the development of accurate allometric equations. However, there are a number of factors which determine the accuracy of an allometric equation. Among the sources of error of equations in common use is the failure to take into account the variation in wood density between species (Chave et al., 2004). Wood density can vary tenfold in tropical forests and has an important effect on estimates of AGB (Chave et al., 2004). Allometric equations constructed using sample sizes of less than fifty trees are also an important source of error and Chave et al. (2004) recommend not using allometric equations where a sample size of less than 100 individual trees have been used to calculate the relationship. Many studies fail to achieve this level of sampling (e.g. Henry et al., 2010). Allometric equations are often applied to large trees (≥70cm DBH) which are outside of the diameter range for which these equations are accurate. Large trees can make up a disproportionate amount of the biomass and have different scaling relationships to smaller diameter trees. Ideally, the most accurate allometric equations would be those which are constructed for individual tropical species, but the scale of this work makes it impractical. However, the growth form of trees in the study area, and generally in southern Africa, shows wide variation due to edaphic factors, browsing by large herbivores and the influence of fire in different habitats (Archibald & Bond, 2003; Styles & Skinner, 2000; Veenendaal et al., 2008). Given this spatial variation in growth form it is prudent to construct allometric relationships which are at least area specific, if not species specific, so that the source of error due to the use of allometric equation is reduced in AGB estimates. The use of allometric relationships developed in one habitat for wide application across swathes of the African continent is an example of the scale based mismatch which is common in the application of analytical techniques to forest enumeration. Without a consideration of scale based variation, impractically large sample sizes are needed in order to be confident of not committing a Type I error when testing for differences in the results of different allometric equations. There seems little point in persisting with trying to refine these relationships for wide application when applying these equations over a wide geographic area will introduce more error than any refinement will serve to remove. The effects of a mismatch of scale are seen clearly in applying analytic techniques which may work well in a particular environment, or for a specific growth form or species, but which contribute large sources of error when used on a larger scale.

160

C HAPTER FIVE

Scale Mismatch in Social Systems

International climate negotiations which seek to establish a system of payments for forest conservation are motivated by two main factors: deforestation accounts for between 12 and 18% of global carbon emissions, and paying for carbon sequestration by forests is a low-cost solution for reducing greenhouse gas emissions (Gregersen et al., 2010). However, the economic arguments which underpin the use of forests to sequester CO2 frequently ignore scale when discussing implementation. Specifically, there is a mismatch of scale between the assumptions of marginal economic theory and the role of the state in receiving and distributing payment for ecosystem services, which ultimately must be delivered by individuals who live in and around the forests. It is widely recognised that the opportunity cost which is forgone by a country when not using the products from its forests, must be compensated for if forested countries are to be persuaded to stop deforestation. In a free market economy, however, this opportunity cost only has implications for the state indirectly through a loss of taxation. In Africa especially, it is individuals or local communities that make the decisions which lead to deforestation despite the legal protection which forests may or may not receive from a weak state. It is individuals living in the forests, or those who are directly dependent on forest products or the land on which forests grow, who will be required to bear the opportunity cost involved with REDD+ implementation. The application of the opportunity cost approach to avoiding deforestation is therefore correct, but unless payments are made to those who bear this cost then it is unlikely that they will forgo the opportunity to cut down the forests and utilise their products for short term economic gain (Karsenty & Ongolo, 2012). The application of opportunity cost approaches to forested states, as if they were decision making entities analogous to individuals, is an example of a scale mismatch between economic theory about how individuals make economic decisions and the role of governments in delivering the CO2 reductions. Since the early 1990s international forest negotiations have become increasingly linked with demands by developing countries for a global redistribution of economic resources in order to address global inequalities (Humphreys, 2008). Currently there are disagreements about whether funding for avoided deforestation to forested countries will take the form of a market-based, project approach or a development aid model which uses a fund-based approach where payments for REDD+ are administered by the state. Countries which are making payments for CO2 reductions through REDD+ favour a project based approach where accountability can be monitored and .

S CALE M ISMATCHES AND I MPLICATIONS FOR SD AND SFM

161

can ensure that reductions in CO2 emissions are being achieved. This has resulted in the problem being polarised in a political sense so that it becomes an “us and them" or “developing versus developed” problem with associated accusations of neopatrimonialism (Erdmann & Engel, 2007), or neo-colonialism and the threats to sovereignty that this entails. Some of these differences could be resolved by applying marginal economic theory at the scale at which it was intended to be applied and not at the scale of the nation state, which does not make major decisions about a development objective based on an opportunity cost approach as mentioned on page 158. However, there is also a mismatch of scale in the perceived power relationships of the different parties. Democratic Western governments, many of which are dealing with an economic crisis, are increasingly accountable to taxpayers for the money they spend. Payment for reductions in CO2 emissions are difficult to justify when domestic conditions require austerity measures to be implemented, but when these payments are made directly to governments which have high levels of corruption, poor track records of accountability to their own peoples, and histories of resisting the decentralisation of ownership of natural resources, it is unlikely that these payments will be politically viable. Forested countries are not negotiating a deal with autocrats; they are negotiating an agreement to reduce global CO2 emissions with the people of democratic countries. It is in the interest of both parties to make sure this deal ensures a long term commitment to payments. The scale mismatch is in the perception of forested countries that they are negotiating with a monolithic entity rather than representatives of people who will make their views known at the next election if they think their governments have signed an agreement which is not in their (or the global) interest. An agreement which clearly benefits local forest communities is more likely to be politically justifiable, even under conditions of economic hardship in CO2 producing countries. Taxpayers are more likely to accept these payments if they can see a benefit for people in other countries and achieve the objective of reducing CO2 emissions. The scale of the payment (by taxpayers of CO2 producing countries) needs to match the scale of the opportunity cost (to people who live in or near forests) to achieve global CO2 reduction in emissions from land use change. In this example the acceptance of the scale related issues could remove some of the obstacles caused by the adoption of a position based on political principles and ideas of global climate equity and redistribution of economic resources. Kok et al. (2007) highlights the importance of scale-based issues in scenario planning, which is an increasingly recognised tool for exploring change in social-ecological systems (Carpenter et al., 2006). In southern Africa, MRV

162

C HAPTER FIVE

scale-based considerations with respect to scenario planning around natural resources frequently map onto governance structures and the associated political connections and disputes. Kok et al. (2007) make an explicit call for making sure that issues explored at each scale be relevant and credible to stakeholders at that scale, but recognise that the process requires a substantial investment of time and resources in order to achieve the top-down, bottom-up iterative cycle that is required for developing scenarios. Scale Mismatch between Ecological and Social Systems

Mismatches of scale between ecological systems and social systems are often related to the differences in the scaling of time of different processes. In Figure 5.2 social institutions are plotted on a scale of turnover time (time to reform or adapt) and the number of people involved in the institution. The mismatch of scale occurs when scaling issues between ecological processes and social processes are not considered, and this can involve both time and other factors (number of people in this case) which emerge from the social construction of institutions. Other issues such as gender, poverty and relative power of elites may emerge as socially constructed scales in socially constructed space. The literature on political ecology in southern Africa has emerged from the innovative application of political scale to the problems of sustainable natural resource use. The CAMPFIRE programme in Zimbabwe (Frost & Bond, 2008; Taylor, 2009) and the CBNRM programs in Namibia (Naidoo et al., 2011; Suich, 2010), the Luangwa Valley (Dalal-Clayton et al., 2003) and Botswana (Jones, 2007) have contributed to a lively and ongoing analysis of the factors which contribute to the success and failure of sustainable use programmes. Considerations of scale could be more comprehensively assessed in these documents, and particularly problematic is that the literature has under theorised the treatment of scale and fallen into what Brown & Purcell (2005) calls the “scalar trap” - whereby it is assumed that a particular political or institutional scale has ontological significance. In this context the scalar trap takes on the form of a “local trap” in which it is assumed that the devolution of power to local communities is the sine qua non for achieving sustainable resource use, justice and democracy (Brown & Purcell, 2005). There is widespread, if tentative, agreement among geographers that there is no ontological significance to scale (Marston, 2000), and that: There is nothing ontologically given about the traditional division between home and locality, urban and regional, national and global scales . . . The differentiation of geographical scales

S CALE M ISMATCHES AND I MPLICATIONS FOR SD AND SFM

163

Figure 5.2: By replacing a spatial scale on the X-axis with the number of people affected by a set of rules or an institution, this diagram goes some way towards depicting how institutions are socially constructed and how they may vary across scales which do not comprise a conventional Cartesian concept of space. Placing different values on the X-axis may result in a different placement of institutions on the graph. Time on the Y -axis depicts the turnover time for institutions to change or reform. Redrawn from Westley et al. (2002).

164

C HAPTER FIVE

establishes and is established through the geographical structure of social interactions (Smith, 1984, 73). The implications of falling into the “local trap” include the potential for generating scale mismatches between ecological systems and social systems by insisting on a local scale for forest monitoring, decision making and management, when these functions may be better performed by other institutions at a larger scale. There is strong evidence for devolution being successful with respect to the sustainable utilisation of natural resources in south-central Africa (Jones & Weaver, 2009; Maveneke, 1998), but this does not mean that scale has ontological significance over process. The problems experienced by the CAMPFIRE programme and the Luangwa Integrated Rural Development Programme (LIRDP) over funds being inappropriately spent by Rural District Council (RDC) and local authorities respectively, have more to do with process than scale. Dalal-Clayton et al. (2003, 293) introduce the concept of “scale-laziness” to describe the propensity for governance at lower levels of participation (higher scale) to be less efficient at project management and governance than when projects are implemented with fully participatory governance by the community. The social pressures exerted by the direct involvement of the community members provide the primary controlling mechanism for the checks and balances needed to control funds and implement the program. In this example the authors provide strong evidence for the scale dependent nature of CBNRM in this particular context, but it is important to be aware of avoiding the “local trap” and attaching ontological significance to these observations. Many of the examples used in southern Africa come from efforts to manage large wildlife species in a sustainable way, and in different contexts (Sustainable Forest Management for example) large scale organisations may be more appropriate managers of certain aspects of sustainable resource use.

Towards a Solution It is difficult to select an appropriate scale of enquiry in an a priori manner and be confident that this will enable one to generalise about processes that are affecting a phenomenon such as deforestation. In addition to this methodological issue, some studies attempt such a grand sweep of scale that it becomes impossible to verify data on the ground and the mistakes which emerge from the analysis are unacceptable. Although global maps of biomass are likely to improve with technology such as LiDAR and other high resolution techniques, the restrictions of scale in establishing causative links between

S CALE M ISMATCHES AND I MPLICATIONS FOR SD AND SFM

165

phenomena will remain. They are not necessarily a barrier to research, but explicit consideration of the effects of scale on studying landscape scale processes needs more attention among ecologists and landscape scientists. Reconciling Epistemology

Some constructivist notions of scale draw on a Marxist critique of the capitalist system, no doubt highlighting valid points about inequality, the exploitative nature of the relationship between capital and labour and the effects of using an exchange-value versus a use-value for the products which humans need from the ecological system (Marston, 2000; Smith, 1984). However, the difference between social scientists who subscribe to these ideas and reject the positivist and reductionist scientific paradigm, and ecologists who may not be especially concerned with arguments around political economy, and may reject constructivist arguments out of hand, serves to polarise the debate in two ways: 1. Some social scientists attempt to reject positivist epistemology entirely and there is strong vein of thought among social scientists which equates a positivist, reductionist epistemology with the sociological positivism of Auguste Comte (Comte, 1868; Gartrell & Gartrell, 1996). A more constructive approach could be to restrict criticism of positivist epistemology when it is directly antagonistic to the constructivist epistemology which informs the arena of social institutions and social phenomena which mediate the relationship between humans and ecological systems. 2. Ecologists would do well to engage in the debate around political economy because it has valid contributions to make concerning the exploitative nature of the capitalist system and how the system structures space. This is particularly relevant in western Zambia where traditional institutions face profound challenges to their role and structure from the forces of globalisation and global capital. Rejection of a constructivist epistemology is widespread among ecologists (McLaughlin, 2001), although the development of more socially aware ecological theory may have softened this attitude. Greater awareness of why sociologists reject sociological positivism and have a different epistemological approach to theorising social institutions would lead to greater acceptance of different forms of knowledge which is a central criterion for interdisciplinary interaction around issues of SD.

Greider & Garkovich (1994) show how landscapes can be socially constructed to reflect the self-definition of people in a particular cultural context. This has direct impact on how people use the environment, how debates about the environment are conducted at international meetings and how changes in

166

C HAPTER FIVE

these social constructions will ultimately determine whether landscapes can be utilised for sequestering carbon on a vast global scale. However, the scale at which human institutions are constructed is not matched to ecological scales and is not, and more than likely never will be, erected solely on the criterion that they match the scale of ecological processes. Being social constructions, they are erected according to the multiple ambitions and aspirations of humanity. Whether they are proscribed by a neo-liberal capitalist system or a socialist ideology, both are unlikely to be attuned to ecological scales. The impending demands of climate change may force these institutions to become more aligned with the scales of ecological processes and rates of growth and carbon sequestration, but the crises that result in this reconstruction have yet to exert a powerful enough influence to make this a priority among the many other priorities that determine the social construction of space and scale. When cities start flooding, food supplies becomes critical and there are significant impacts of climate change on the standard of living of the majority of the people who participate in the social construction of space and scale, perhaps this will change. Until it does, the call for change from the ecologists and the socialists is likely to have little effect. It is revealing that neither the positivist nor the constructivist ideas of scale claim any ontological significance for the concept of scale (Brown & Purcell, 2005) which means that they should at least use this as a starting point for accommodating each other’s ideas, and for drawing the boundaries at where epistemological arguments end and the matching of scales can begin. I have illustrated how this may be achieved by the adoption of a conceptual approach illustrated in Figure 5.3. Ecologists will be comfortable enough with the arrangement of ecological entities along a scale (although more detail could be included for specific systems). The apparent rigidity of the arrangement of human institutions may disturb social scientists, but the inclusion of three different concepts of hierarchy serves to add a fluidity to the arrangement which allows the social construction of institutions to be part of an inclusive, exclusive or constitutive hierarchy (see Figure 5.1 for an explanation of these terms). The epistemological differences are reduced by including social constructs of space and scale as emergent from the biological complex system. The biological determinism at small scales gives way to socially constructed reality at higher scales as an outcome of consciousness. I choose to leave the debate about the nature of consciousness and its relationship with the formation of society as a grey area with respect to arguments about epistemology. In Figure 5.4 I attempt to illustrate how the conceptual framework in

S CALE M ISMATCHES AND I MPLICATIONS FOR SD AND SFM

167

Figure 5.3: The arrangement of ecological scale and operational scale (which includes the social constructs of space and scale) can reconcile epistemological viewpoints. I depict both biological phenomena studied using a positivist, reductionist approach, and constructivist theories of how social space and scale are formed, as emergent from the biological phenomena. Lefebvre & Nicholson-Smith’s (1974) and Smith’s (1984) attempts to replace Darwin with Marx confuse proximate with ultimate causation. This does not undermine the constructivist position on socially created space, but allows the possibility of other proximate causes to influence this construction. Hierarchy is an essential framework for structuring the complex factors which influence environmental impacts (author’s own diagram)

168

C HAPTER FIVE

Figure 5.3 could be applied to the social-ecological system in western Zambia. This diagram would need more research and stakeholder involvement to refine, but nevertheless it is a scale-aware depiction of interactions which facilitates the structuring of ecological interventions and institutional modifications that are scale-matched.

Figure 5.4: Diagram showing the scale at which social institutions in western Zambia interact with governance institutions through an ecological entity. This is an application of the hierarchical structuring of both ecological entities and social entities which makes no assumption about epistemological conceptualisations of space and scale. There are no doubt multiple versions of this diagram that would emerge from a participatory process of structuring the relationship between the entities. Ecological and social entities could be more finely divided. More detailed ecological processes could be included in the interaction and more detailed constructions of institutional space would emerge, but nevertheless the value of an explicit consideration of scale and of a hierarchical structuring of entities remains (author’s own diagram)

Conclusion Making causal statements about observed patterns is a central activity of science. However, making these links between observation and causation always invokes questions of scale and level. To a large extent the power of the scientific method is in its ability to develop theory at one level or

S CALE M ISMATCHES AND I MPLICATIONS FOR SD AND SFM

169

scale and use this theory to predict similar phenomena at different levels or scales. However making claims for the utility of observations at a large scale for measuring phenomena at a small scale commits a fallacy which at best highlights the ignorance of the claimant of those processes which operate at the smaller scale. Maps of global tropical biomass are a good example of where a mismatch of scale results in such a fallacy being committed. In the area examined in this study, Zambezi Teak forests have never occupied more than 15% of an area covering 23,500 km2 and are hardly visible on maps of global tropical biomass. Nevertheless they form an important part of the ecology and history of the area, and have contributed more to the local economy than any other land cover type in the area. If REDD+ projects become a reality in south western Zambia, it is in the Zambezi Teak forests that they will operate because these forests are the most threatened and they accumulate biomass of up to three times that of the surrounding woodlands. It is only when examined at the appropriate scale that this becomes apparent. Errors in the application of methodology are not uncommon in science (see Foody’s (2002) statement about the misapplication of classification accuracy assessment methods in remote sensing on page 204), and there is often temptation to make wider claims for the application of a particular method than it is capable of delivering. However, questions of scale are arguably more important in a field which involves making statements about phenomena across wide geographic ranges and should therefore be more closely considered. The principle of spatial autocorrelation needs to be considered when developing and applying allometric equations for measuring AGB. The development of these relationships is routine forest enumeration work and in the process of building capacity for REDD+ the UN-REDD programme should include the development of locally appropriate allometric equations in accordance with best international practice. The variation in African dry deciduous forests is such that a more local focus is likely to produce more accurate results for the important process of Monitoring, Reporting and Verification (MRV). Economics is a field well acquainted with scale and this is reflected in studies of microeconomics and macroeconomics (Gibson et al., 2000b). It is surprising therefore that mismatches of scale in the application of economic theory have become part of the debate about REDD+. Karsenty & Ongolo (2012) have addressed this issue and point out that appropriate scaling of the economic theory of incentives to local economic agents involved in REDD+ (companies, rural households or communities) would be a more appropriate application of theory than suggesting that fragile or failing states can, or

170

C HAPTER FIVE

would, make a decision about a major development programme such as REDD+ on the basis of a cost benefit analysis. The mismatch of scale in economic theory is mirrored by a mismatch in scale between perceptions of the parties at international climate negotiations regarding the political accountability for making the payments for reducing CO2 that REDD+ entails. Negotiations should be less about states and historical claims of inequality than they are about payments by individuals in CO2 producing countries to individuals in forested countries. A more insidious trend in the increasing number of publications which attempt to map AGB on a global scale, are the undertones of power and domination which are reminiscent of the role which imperial science played in the formation of a sociology which underpinned European colonial domination (Gilmartin, 1994). To map the AGB of the entire tropics is to command the high ground of scientific knowledge, which would be a slightly less questionable aim if the maps and the data were usable at a local scale which is relevant to REDD+ implementation and biomass mapping. Mapping at a large scale which ignores the smaller scale at which people on the ground perceive environmental issues has the potential to create mistrust and perpetuates the skewed power relationships between the global North and the global South when these maps are used to infer ecological processes. Consideration of scale needs to be at the heart of the entire REDD+ process. Almost every aspect of the way forest carbon is measured, mapped, paid for or discussed involves a consideration of scale. The scale which is used needs to be appropriate so that measuring and mapping are relevant for describing processes which lead to deforestation or reforestation on the ground, that the correct economic incentives are in place so that they influence decision making by organisations or individuals responsible for deforestation, and so that scale based misconceptions do not result in the politicisation of issues which result in negotiating deadlock. Scenario planning and stakeholder engagement which is sensitive to scale may be time consuming and labour intensive, but no more so than the considerable amount of time and resources that are being invested in producing global biomass maps which are of little relevance to local communities and which are not capable of taking into account local context or scale. The possibility of reconciling differing epistemological approaches to describing social-ecological systems through a hierarchical application of scale is exciting and will no doubt generate considerable debate. Reconciliation is important for the inclusion of different forms of knowledge and this is at the heart of requirements for interdisciplinary research and the requirements and Sustainable Development. The process of social learning which emerges from project implementa-

S CALE M ISMATCHES AND I MPLICATIONS FOR SD AND SFM

171

tion has the potential to inform scale related mismatches. By learning about how institutions interact and highlighting the problems in an iterative way, scales which were previously mismatched can be brought into alignment. Viewing policy as epistemology can make important contributions to the learning process so that scale mismatches are corrected without being ossified in policy which does not adapt to scale mismatches as globalisation and the flow of capital changes the space in which policy needs to be effective.

Chapter Six Land Cover Change and Above Ground Biomass

The fastest land cover change in human history is occurring in tropical forests (Achard & Blasco, 1990), and this has important implications for the global climate (Malhi et al., 2008; Miles & Kapos, 2008), changes in ecosystem function and resilience (Bradshaw et al., 2007; Foley et al., 2005; Lambin et al., 2003; Messier et al., 2013) and the stability of social-ecological systems (Kamusoko et al., 2009). In terms of their contribution to global carbon sequestration, African dry forests and woodlands store relatively low amounts of AGB per hectare compared to the rainforests of central and west Africa. Figures from the literature indicate a mean AGB of 395.7 tons C ha−1 for tropical rainforests in Africa (Lewis et al., 2013) versus 88-97 tons C ha−1 for Zambezian woodlands (Timberlake et al., 2010) and 40 tons C ha−1 for Miombo woodland (Chidumayo, 1990). However, dry forests comprise about 70%-80% of the forested area of Africa (Mitchard & Flintrop, 2013; Murphy & Lugo, 1986), and therefore represent huge potential as a carbon sink, provided the correct management strategies are adopted (Marunda & Bouda, 2010). Unfortunately, poor forest management practices are widespread, with the result that most deforestation in Africa is taking place in dry forests (Brink & Eva, 2009). Changes in land use in tropical forests are currently being addressed globally under the UN-REDD programme, with Zambia as one of the nine partner countries receiving support for National Programmes. There is a diverse range of problems with respect to the implementation of the UN-REDD programme at a global scale (Sanz-Sanchez et al., 2013), but 173

174

C HAPTER SIX

the MRV of emissions from land cover change is central to the success of any future REDD+ programmes. To date, most information on AGB for large areas of Africa comes from maps of global AGB using remotely sensed data at a scale of ≈1km, or from a limited number of AGB inventory plots. Both techniques have their problems, among which is the poor precision of the measurements at all scales (Hill et al., 2013). Increasing methodological sophistication in the form of remote sensing techniques does not result in the improved precision of AGB estimates (Hill et al., 2013). Danielsen et al. (2013) have shown that measurements of biomass in forests by members of the local community who receive modest training and support attain the same levels of accuracy as those achieved by professional foresters and can fulfil IPCC Tier 3 standards. In most cases remotely sensed global biomass maps only meet the criteria for IPCC Tier 1. Where detailed project scale (≈10,000 ha) monitoring of AGB is required, questions of scale have increasingly important implications for the utility of remotely sensed global biomass maps. The vegetation biomes of Africa are poorly characterised in current maps (White, 1983), and in many areas show considerable variation in structure and species composition (Privette et al., 2004; Scholes et al., 2002). At current scales of measurement in remote sensing (≈1km) much of this spatial variation may not be recorded at all, or alternatively may be well recorded if the pixel size corresponds to scales of variation on the ground. A more explicit consideration of this spatial variation will be important for accurately reporting changes in AGB in any future MRV programme under the UN-REDD programme, especially at the project scale. The setting of a baseline or a starting point against which performance of a REDD+ project is assessed is a key decision for project implementation. Past land cover change informs the rate of deforestation in the future which is used to calculate the BAU scenario versus the proposed biomass accumulations due to the REDD+ project intervention. Where deforestation has a long history it is important to fully document the start of the deforestation process (if the data is available) to properly establish the deforestation rate. Deforestation rates slow as a forest becomes depleted and measuring these rates from the incorrect starting point will make projects appear marginal, when in fact BAU deforestation rates are much higher. In addition, past land cover change can be highly heterogeneous, making the setting of a single baseline for all forest types invalid. If no change has taken place in a particular forest type, then no accumulation of biomass can be claimed for this forest type in the project accounting. We calculate the historical rate of deforestation for different forest types so that this can be used in the setting of baselines in the study area.

L AND C OVER C HANGE AND AGB

175

Although AGB comprises only a portion of total biomass, the measurement of total biomass presents many problems. Estimating below ground biomass has hitherto not been an active area of research relative to the considerable effort undertaken to measure AGB, although Saatchi et al. (2011) have produced a map of below ground biomass which is a first attempt at measuring this biomass pool globally based on allometric relationships of trees which are adapted to grow in dry ecosystems. An increasing body of work indicates the potential importance of below ground biomass as a carbon store in dry tropical forests because of the higher root biomass relative to tropical rainforests (Canadell et al., 1996; Collins & Bras, 2007; Jackson et al., 1996; Schenk & Jackson, 2002a; van Wijk, 2011). Robinson (2007) estimates that the global root carbon pool is about 68% larger than originally thought and the geographical distribution of this carbon pool is heavily skewed towards the areas where dry tropical forests are found. Soils contain an important carbon pool which is easily lost during burning and conversion of land to agriculture. Some studies have measured changes in soil carbon in Africa (Walker & Desanker, 2004) but on the whole this is a far more complex and variable component of the carbon pool to measure and map accurately. In this chapter I discuss only AGB, except where stated. Recent attempts to map the AGB of Africa have been published by Baccini et al. (2008), while Saatchi et al. (2011) have produced a map of global AGB for 75 countries in tropical regions. Baccini et al. (2008) use a combination of spectral values derived from MODIS imagery calibrated with data from field-based inventory studies. The problem with using spectral values in deciduous forests is that these values vary widely at different times of the year, and between vegetation types. Variation in phenology is one of a range of adaptations of dry tropical forest species which enable them to survive a long dry season (Childes, 1988). As space-borne technology improves there appears to be more potential for mapping AGB using direct measures of forest structure. The Geoscience Laser Altimeter System (GLAS) onboard the Ice, Cloud and land Elevation Satellite (ICESat) measures global forest height and is the main source of primary data for the global biomass maps of Saatchi et al. (2011). Remote sensing is likely to be the method of choice for measuring AGB across the vast swathes of forest that will have to be monitored in a global REDD+ programme (Herold & Johns, 2007), but it is important to assess whether it is up to the task. There are few allometric equations which have been developed for biomass inventory calculations of AGB for sub-Saharan Africa (Henry et al., 2010), and even fewer for dry, deciduous forests. There are some exceptions (e.g. Abbot et al., 1997; Chidumayo, 1997; Frost, 1996) but these equations

176

C HAPTER SIX

establish general relationships which are not necessarily transferable to different habitats, even within a vegetation type like miombo which varies widely throughout the region in both species composition and vegetation structure (Linder et al., 2005). Given this variation, area specific allometric equations may be no more or less appropriate than the more general relationships established for dry tropical forests (e.g. Brown et al., 1989; Chave et al., 2005). The lack of detailed vegetation maps also presents problems for the biomass inventory method. It is difficult to accurately depict the spatial variation in AGB across a landscape when the heterogeneity of the vegetation is inadequately mapped. Increasing the vegetation mapping capabilities of the Zambia Forestry Department in the field would assist in the production of more detailed maps. Detailed remote sensing studies using high resolution data of smaller areas would also assist in developing more detailed land cover maps. Errors in estimating AGB are not a problem if these errors are systematic. It is the uncertainty associated with the source of the error, and as a consequence the inability to distinguish random errors from systematic errors will lead to poor implementation of REDD+. Well managed REDD+ projects require accurate baselines of past biomass change as well as recent changes in biomass. These demands for data highlight several limitations with the current suite of observation approaches: 1. land cover and land cover change vary on a wide range of spatiotemporal scales that cannot be consistently captured by a single observation approach 2. field surveys are labour intensive and are not seen as a practical approach to informing the wide scale implementation of REDD+ 3. typically recent remote sensing studies have the benefit of extensive coverage, but provide products that are focused on timescales too short for the generation of baselines (i.e. < decadal timescales), and at scales that are too coarse to detect all forms of biomass loss.

This study set out to map changes in land cover in south western Zambia over a period of 30 years from 1975-2005 and generate local estimates of AGB using field biomass inventories. The accuracy of land cover change maps is assessed and the implications with respect to potential changes in ecosystem services delivery for the people who depend on the forests for their livelihoods are discussed. In particular I examine the uncertainties associated with using different AGB allometric equations commonly used for dry, deciduous forests (e.g. Williams et al., 2008). A general assessment is made of these methods for the purpose of MRV for REDD+ projects in African dry deciduous forests. The wider implications for global REDD+

L AND C OVER C HANGE AND AGB

177

implementation are discussed with a view to underscoring the enormity of the task that is being proposed with respect to the MRV of a global REDD+ programme.

Landsat Image Analysis The study area (Figure 4.2) covers some 23,500 km2 of south western Zambia. Rain falls between November and March and ranges from around 950 mm per year in the north to around 650 mm per year in the south. The area is almost completely overlain by Kalahari sand (Shugart et al., 2004), with the exception of the south east where clay rich soils dominate. The density of tree canopy cover is strongly correlated with precipitation (Privette et al., 2004), although the Zambezi Teak forests are an exception to this trend, forming pure closed canopy stands in areas where the sand is deep. White (1983) and Linder et al. (2005) both designate the Zambezian region as a centre of endemism, with the highest floristic diversity of all the dry forests and woodlands of Africa. Kalahari Woodland occupies the largest area of the landscape and contains some typical miombo species (although Brachystegia species are largely absent or occur in localised belts) with a significant grass layer. The decline of the Zambezi Teak forests has been a source of concern since the 1980s and has prompted one dedicated international conference (Piearce, 1986) and one detailed study (JICA, 1996). A combination of logging and subsequent frequent wildfires has resulted in a change to a more open, degraded Zambezi Teak forest, invaded by fire-tolerant Kalahari Woodland species or with a dense shrubby undergrowth of thicket (usually Combretum spp.) known locally as “mutemwa”. In the south east large areas of Colophospermum mopane woodland dominate the landscape where the soils contain a high proportion of clay and are waterlogged for most of the rainy season. In a significant development with respect to long term monitoring of environmental resources, the United States Geological Survey (USGS) made a change in data policy in 2008 which meant that all new and archived Landsat data would be made freely available over the internet to any user (Wulder et al., 2012). The data are especially useful for studying tropical forests. The temporal coverage of the Landsat archive corresponds to the decades during which the rate of land use change started increasing. The wide geographic coverage, the long record of consistent and regular coverage and the fact that the data have been geometrically and radiometrically corrected, makes Landsat imagery unique with respect to other data which could be used to assess the decline of tropical forests (Cabral et al., 2011; Wulder et al., 2012).

178

C HAPTER SIX

Data was used from the first Landsat satellite, Landsat 1, which was launched in 1972 and carried the Multispectral Scanner (MSS), and also from Landsat 5 which launched in 1984 and carried the MSS and Thematic Mapper (TM) scanners. The MSS data were recorded in four spectral bands, a green, a red and two infrared bands and it is this data that provides the first images from 1973 and 1975. Although Landsat 5 carried a MSS, it was turned off in the 1990s and the most recent image available is a composite TM image from 2005. This image forms part of a collection assembled for the Global Land Survey (Global Land Survey, 2013) which uses Landsat 5 TM images and gap filled Scan Line Corrector-off Landsat 7 Enhanced Thematic Mapper (ETM+) images at selected sites. TM images are multispectral but collect data at slightly different wavelengths from MSS images. In both cases the adjoining images were concatenated using a pixel-wise column and row concatenation of the images based on geographic location. A sub-image was subtracted from the concatenated image which produced an image corresponding to the polygon outline which describes the study area (Figure 4.2). The study area polygon was selected manually so that most of the economically important Zambezi Teak forests east of the Zambezi river were included in the analysis. These forests also represent patches of high AGB relative to the surrounding woodlands and have seen substantial impacts in the last 30 years. The north eastern limit is defined by the boundary of the Kafue National Park. The southern boundary was selected to exclude a vegetation type which occurs in a band along the Zambezi river known as Munga Woodland (White, 1983). This comprises a mixture of Acacia species but Faidherbia albida is the dominant tree species. Although important because it is being subject to dramatic change due to the demands of rural agriculture, it occupies a relatively small area. Preliminary analysis showed that it introduced inaccuracy in the classification results and was excluded in order to reduce the complexity of spectral responses which the classification algorithm was required to take into account. This chapter examines potential changes in forest cover over a large area. Whilst Zambezi Teak forest does not always occur in monospecific stands and often includes a significant shrub layer, these differences in forest structure are not so dramatic as to alter the ecological role or the economic potential represented by what I classify as Baikiaea plurijuga forest. The differences between Zambezi Teak forest and other vegetation types in the study area are spectrally distinct and therefore the resolution of the Landsat imagery, as well as the spectral limitation of the sensors, do not appear to have any significant drawbacks with respect to the assessment of Zambezi Teak forest coverage. In fact, numerous studies have shown that classification error of

L AND C OVER C HANGE AND AGB

179

forests reduces as spectral resolution increases because of the reduced noise to signal ratio that results (Cushnie, 1987; Hsieh et al., 2001; Marceau et al., 1994a,b). The fact that a forest, by definition, occupies a large area plays an important role in the detection process. The advantages Landsat presents with respect to forest classification have some disadvantages with respect to other categories of land use in the study area. Agricultural Fields (AF) and Dambo (D) are often in close proximity and have similar spectral values throughout the year as the wet season brings a green flush in both areas and the dry season leads to a dieback of grasses and simultaneously brings an end to the agricultural season. This results in difficulties with respect to classification and the small size of agricultural fields means that mixed pixels play an important role in contributing to the error associated with distinguishing between AF and D. Agricultural fields are detected with some success, but there is almost certainly a degree of error associated with this classification. A prerequisite for restricting the selection of images to a time of year where differences in phenology will maximise the ability to distinguish different forest types is the existence of a suitable number of cloud free images at these times of the year. For classification purposes seven land cover categories were selected: Baikiaea plurijuga forest (BPF), degraded Baikiaea plurijuga forest (DBPF), Kalahari Woodland (KW), Colophospermum mopane Woodland (MW), Dambo (D) (a local name for seasonally flooded grasslands), Agricultural Fields (AF) and Fire Scars (F). These categories are a simplification of vegetation classes across a landscape which is almost certainly more complex than they describe. It is likely therefore, that the distribution of objects in the N -dimensional parameter space (which varies because the number of bands and their frequency are different both within and between image types) is complex. However, any classification scheme must always reduce complexity, and vegetation data in this part of Africa are not detailed enough to produce an ad hoc classification scheme which is an optimal representation of the spectral responses of vegetation on the ground. A technique which is able to deal with this complexity is better suited for this analysis than supervised parametric statistical techniques (such as Maximum Likelihood, Parallelepiped, Minimum Distance) which are generally suited to classifying data with better defined categories. These would correspond to a particular probability distribution and consequently have less complex variation in their spectral signatures. Image Classification using Neural Networks

There are several reasons for choosing to use neural networks over other classification techniques. In general, neural networks make no assumptions

180

C HAPTER SIX

about the underlying distribution of the input variables, which contrasts them with parametric techniques which assume a certain form of the probability distribution (Egmont-Petersen et al., 2002). Neural networks are good at classifying problems with many parameters. The disadvantage is that it is difficult to determine how the network is making decisions. The “black-box problem” which describes the inability of the neural network to produce an indication of how reliable the resulting classification is, or how the decision was reached, has been a source of concern to those who use these techniques (Egmont-Petersen et al., 2002). The problem arises from a view of neural networks as an all-encompassing technique, and is worse where images are classified in an unsupervised manner without prior knowledge about what the groups of pixels represent. The idea that a machine can be trained to recognise objects in an image is intuitively attractive. However, these methods of analysis yield their best results when used together with other classification and verification techniques. Egmont-Petersen et al. (2002, 2293) summarise their review of image processing using neural networks with a similar caveat The conclusion must be that ANNs [Artificial Neural Networks] can play a role in image processing, although it might be a role as a supporting tool rather than a major one. ANNs are useful in image processing as either non-parametric classifiers, non-linear regression functions, or for (un)supervised feature extraction. In this study the accuracy of the final classification is assessed using other techniques and this assessment is enhanced by detailed knowledge of the study area and the processes through which vegetation change takes place. The long documented history of forest exploitation and the resulting natural experiments which have taken place further enhance this understanding. The extent to which different classifiers are suitable to use is largely dependent on the features which are being classified and the experience of the person who is doing the analysis. I used the Kohonen Self Organising Map (SOM) (Kohonen, 1990) which is a supervised neural network type classifier. Ji (2000) tested this technique using Landsat TM images and found that the SOM achieves a higher classification accuracy than the maximumlikelihood classifier. This method allows the input of a vector file that designates categories based on expert knowledge of vegetation on the ground, thus providing training data which, to some extent, describes pixel variability within each class. The SOM also assigns weights based on the proximity of winning pixels to each other and therefore incorporates a consideration of spatial awareness and the likelihood that pixels near to each other are

L AND C OVER C HANGE AND AGB

181

more likely to be similar than those further away. In the learning process the SOM adjusts weights so that nearby pixels are more likely to be classified in the same category as the pixels near to them than a category further away. This takes into account one of the fundamental principles of geographical distribution, that of spatial autocorrelation, and in the case of Baikiaea plurijuga forests the principle is particularly relevant.

182

C HAPTER SIX

Figure 6.1: Flowchart showing Landsat image selection process and analysis procedure

Accuracy Assessment

The purpose of an accuracy assessment is to convey the accuracy with which the classification technique is able to distinguish clearly bounded features

L AND C OVER C HANGE AND AGB

183

from the surrounding pixels. Baikiaea plurijuga forests are discrete and clearly identifiable features on the raw satellite image. It was not only a spectral classification which determined the outcome of the analysis, but the extent to which the SOM assigns weights based on the proximity of winning pixels to each other and therefore incorporates a consideration of spatial awareness and the likelihood that pixels near to each other are more likely to be similar than those further away. Verifying the accuracy of a map which is generated from a classifier algorithm requires some evidence of what conditions were like on the ground at the time the image was collected. A measure of accuracy is important because it provides a quantification of the error which is inevitably associated with any classification technique. When using images from remote areas in Africa from the past, it is often difficult to source data which could be used for ground truthing a map. In this case it is fortunate that we have a limited amount of data that can be used for ground truthing. These sources of data are discussed below. For both the 1975 image and the 2005 image, independent assessments of accuracy using measures outlined by Foody (2002) and Gao (2008) were used, as well as some comparisons with ancillary data which assist in determining accuracy. The reference data comprised a number of different sources. Original Forest maps

Historical maps of Zambezi Teak forest patches were available. These were drawn by the management of Zambesi Saw Mills at roughly the same time as the first satellite images were being captured. For the 1975 map a forest exploitation plan for Bombwe Forest from August 1970 drawn by J.P. Rault (Figure 6.3) was used in an accuracy assessment. The outline of Bombwe Forest was digitised and, although no geographical information is depicted on the map, the Mulobezi railway line, the Bombwe river and the Bombwe bridge are all still present in the same location and provided coordinates for georeferencing the map in ArcMap 10. The resulting vector file was converted to a raster file and the same area on the 1975 classified map was extracted for direct comparison once the images were converted to the same resolution. Figure 6.2 shows the two maps which were used to test for accuracy. Landscape features from Google Earth

The 2013 Google Earth images of the study area are from GeoEye and have a resolution of 0.41m making it possible to identify

184

C HAPTER SIX

Figure 6.2: Maps of Bombwe Forest from August 1970 and the classified 1975 map used to measure accuracy.

Figure 6.3: Map of Bombwe Forest drawn in 1970 from the Zambesi [sic] Saw Mills archive

L AND C OVER C HANGE AND AGB 185

186

C HAPTER SIX

features on the ground. The resolution is such that it is possible to use the images in a similar way to how aerial photographs are used to identify landscape features. I have detailed knowledge of the area from more than eight years of field work and research and this expert knowledge was used when interpreting the images. Nyangombe Forest in the far north east of the study area was selected for use in map accuracy testing of the 1975 map (Figure 6.4). I assumed that its area and shape have not changed substantially since 1975. The assumption is that the slow growth of Zambezi Teak (increase in DBH of between 1mm and 2mm per year), the short distance (< 30m) that seeds are dispersed from the parent tree through explosive dehiscence, the lack of any known seed dispersers, the ability of mature forest to exclude fire (Martin, 1940) and the lack of nearby human habitation and hence reduced incidence of fire would mean that forest patches such as Nyangombe should persist over a considerable period. It has also never been logged. Field sampling in Nyangombe forest in 2009 showed that it contains mature trees, has a closed canopy in most places and a relatively open understory, all signs of mature forest. The outline of Nyangombe Forest was digitised from Google Earth to use for accuracy analysis. Some distinctively shaped features among the now degraded BPF near Sesheke are the areas of slightly lower elevation devoid of trees but which show no sign of water at any time of the year (Figure 6.5). They are depicted as D on the 1975 and 2005 map in the extreme south western corner, but are in fact frost hollows, where cold air collects in the evening when temperatures are at their lowest. Winter temperatures can drop several degrees below freezing at night. Most of the trees in the region are intolerant of frost to a greater or lesser degree and therefore the area remains free of trees and shrubs, no doubt enforced by regular fires in the dry season. These are features which do not change substantially over the time period of this study and we use their distinctive shape as a means of verifying the accuracy of the 1975 map. It has been suggested by Geoff Calvert, a retired ecologist from the Zambia Forestry Department (pers. comm.) that these were former dambos which held water in the rainy season when the area was thickly forested with Zambezi Teak. The complete destruction of the forests has resulted in their ecological function being altered and they now function only as low lying depressions. The lowest temperatures of the entire study area occur in the south west corner where these frost hollows are found. Although the minimum temperature in July is 3.4 ◦ C, these represent mean temperatures for the month generated by the WORLDCLIM model (Hijmans et al., 2005). It would only take occasional but regular frosts, where temperatures dip below 0 ◦ Cto exclude trees and shrubs from these localised depressions. In

L AND C OVER C HANGE AND AGB

187

Figure 6.4: Nyangombe Forest as seen from Google Earth image in 2013 c c (Image Google Earth and CNES/SPOT)

188

C HAPTER SIX

2011 a widespread frost in the area of Sesheke and Katima Mulilo resulted in extensive leaf dieback of trees some 5km either side of the low point along the Zambezi river, and the almost total destruction of pumpkins and sweet potatoes grown as a winter crop in this area by rural farmers (pers. obs.). Young trees were also badly affected. Observations in the field indicate that these conditions occur regularly in these frost hollows.

Figure 6.5: Frost hollows near Sesheke in the south west corner of the study c c area (Image Google Earth and GeoEye) The frost hollows which bisect Simungoma West forest were digitised from Google Earth Geoeye images and the resulting Keyhole Markup Language (KML) files were converted to vector shape files in ArcMap 10 and then to raster files. In a similar manner to that described for Bombe forest, we extracted the corresponding area from the 1975 classified map and used the two maps to calculate measures of accuracy. The maps are shown in Figure 6.6. Edmonds’s 1976 Vegetation Map

The definitive vegetation map of the area, compiled by A.C.R. Edmonds for the Forestry Department, Government of Zambia in 1976, was not used for ground truthing BPF. A detailed visual examination of this map, and comparison with a false colour composite of the concatenated raw MSS data from 1973-1975 (referred to as the 1975 image or map), showed a high degree of inaccuracy in the depiction of BPF stands. Many of these forest patches show a highly distinctive shape on the raw Landsat image, easily

L AND C OVER C HANGE AND AGB

189

Figure 6.6: Maps used to measure accuracy for forest hollows bisecting Simungoma West forest near Sesheke

190

C HAPTER SIX

distinguishable without processing, and which is simply not reproduced on the Edmonds map. A number of the more distinctive forests and other landscape features were compared with a digitised version of Edmonds’s 1976 map for more accurate comparison with the 1975 Landsat image. They were found to be highly inaccurate and the map was therefore rejected as a source of ground truth information for Zambezi Teak forests. The inaccuracies may be in part due to the scale of the map (1:500,000), and the difficulty of depicting smaller forests on a map of this scale.

Figure 6.7: Flowchart showing ground truth data used for different maps and vegetation categories and analysis procedures

Edmonds’s map was, however, used to measure the accuracy of classification of the other major land cover classes, namely, KW, MW and D. These are far larger features and would be easier to depict on a map of this scale. A stratified random selection of 3,269 points was generated and used to sample both Edmonds’s map and the 1975 classified image. These data were collated in a confusion matrix and measures of user accuracy, producer accuracy and total accuracy were calculated for these classes.

L AND C OVER C HANGE AND AGB

191

Aerial and ground-based geolocated photographs

A database of over 8,000 photographs taken between 2009 and 2012 was used for accuracy assessment. These were taken both on the ground and from the air and geolocated. In areas other than those affected by fire, the changes between 2005 and 2009-2012 were not deemed large enough to substantially affect the assessment of accuracy. A total of 666 points comprising a combination of these images and data from Google Earth 2013 GeoEye images were used to construct a ground truth confusion matrix. Once each of the points was labelled with a vegetation type corresponding to what was captured in the photograph or identified from Google Earth, an accuracy assessment was calculated using the points representing the photograph/Google Earth locations and the land cover category at the corresponding point on the 2005 classified image.

Above Ground Biomass To measure AGB, 4 plots of 40m x 40m were sampled in each of the four land cover categories which are dominated by tree cover (BPF, DBPF, KW and MW). Sample sites were selected in a semi-random manner, taking into account proximity to roads (which affects ability to access the area) and the representativeness of the forest at the sample point (Figure 6.8). A wide geographical spacing between sample sites was used so that regional variations in forest characteristics were taken into account. All living stems >5cm DBH were measured and the species name recorded. The DBH of standing dead trees and dead trees on the ground was measured. The AGB of living woody species was calculated using four different published allometric equations which use a wide range of data from global relationships developed from sampling in dry forests to local relationships developed in Zambia (Table 6.1). One of the allometric equations used to calculate AGB requires the use of wood density data. The main source of this information was Zanne et al. (2009) but in a minority of cases the wood densities were not available for the species which were identified in the plots. In this case the wood densities of all the tropical African species in the same genus were sourced from Zanne et al. (2009) and the genus level mean used to arrive at an estimate of wood density for the species for which there was no data. In a small minority of cases no data were available for Africa for a particular genus and densities from the corresponding South American tropical genera were used. This technique is justified by Chave et al. (2006) who have shown

192

C HAPTER SIX

24°21'0"E

25°22'0"E

15°47'0"S

15°47'0"S

DBPF2

!

KW3

!

BPF4 KW4 KW1 DBPF4 BPF1BPF3 !

!

!

!

!

16°48'0"S

! !

!

DBPF1

16°48'0"S

KW2

!

DBPF3

!

!

!

MW4

MW2 !

MW1

MW3

!

17°49'0"S

´

24°21'0"E

0

30

60

120 Km

17°49'0"S

25°22'0"E

Figure 6.8: Map showing location of sample plots for measuring Above Ground Biomass (AGB)

Chidumayo & Gumbo (2010)

Brown et al. (1989)

Chave et al. (2005)

B = 0.702BA − 281.484

B = 34.47 − 8.067D + 0.659D2

B = 0.112 × (pD2 H)0.916

African Dry tropical forest

Dry tropical forest

Dry tropical forest

Zambia Miombo Woodland

B = 3.01D − 7.48

Chidumayo (1997) B = 20.02D − 203.37

Area

Equation(s)

Author Trees 0.1 m DBH

Notes

Table 6.1: Allometric equations used to calculate above ground biomass in this study. B = Biomass (t), p = Wood density (t/m3 ), D = diameter at breast height (DBH) (m) and BA = Basal Area (cm2 ) L AND C OVER C HANGE AND AGB 193

194

C HAPTER SIX

in an analysis of wood densities from 2546 tree species that wood density is reliably approximated by genus level means. In the few cases where wood density could not be found in the literature, I used the median wood density value for the plot where the species was recorded. Dead trees that could not be identified from bark remains were also assigned this median value. The sole exception to this was for assigning a density value to the vine Fockea multiflora (Figure 6.9) which is common in Mopane Woodland. On inspection the wood of this vine proved to be soft and pithy and I assigned it the same value as the wood of the baobab (Adansonia digitata) which is similar in texture and occurs in the same habitat.

Figure 6.9: The climbing vine Fockea multiflora, known in SiLozi as Cimulumina. In the study area it is only found in Mopane Woodland, usually climbing up a Colophospermum mopane tree as shown here

L AND C OVER C HANGE AND AGB

195

Table 6.2: Changes in forest/woodland area between 1975 and 2005. Although Baikiaea plurijuga forest (BPF) show a decline in area, vast swathes of Kalahari Woodland (KW) show a slight increase in land cover change. Where BPF has disappeared it has largely been replaced with degraded Baikiaea plurijuga forest (DBPF). The accuracy with which Colophospermum mopane Woodland (MW) was able to be classified makes the changes in area depicted here less reliable than other vegetation types

BPF DBPF KW MW D AF Fire

Area 1975 (Km2 )

Area 2005 (Km2 )

Area change (Km2 )

2051.45 1137.02 12846.43 4320.08 1778.24 648.21 728.30

1101.81 1884.48 14537.69 2226.00 2510.81 873.93 373.00

−949.64 +747.46 +1691.26 −2094.08 +732.57 +225.72 N/A

Changes in Land Cover Over Time The study area represents an extent of some 23,500 km2 and the area occupied in each land cover category in 1975 and 2005 is shown in Table 6.2. These figures are presented as percentages in Table 6.3. There is a 54% decrease in the area covered by BPF and a 60% increase in the area of DBPF. The 13% increase in Kalahari Woodland may be as a result of this vegetation type replacing BPF, especially in the south west in the area of Simungoma East where BPF losses have been widespread (Plate 1 and Plate 2). The changes in burned area are not relevant since these areas represent what was visible on the day the satellite images were captured and do not represent consistent changes over time due to ecological processes. The annual rate of land use change for BPF is 1.54% and for MW is 1.62%, while KW and DBPF are increasing at a rate of 0.44% and 2.19% respectively.

Map Accuracy 1975 map

The maps of Bombwe forest, Nyangombe forest and the frost hollows in the south west corner of the study area were used as reference images in a

196

C HAPTER SIX

Table 6.3: Percentage change in area (Km2 ) between 1975 and 2005 Vegetation Category BPF DBPF KW MW D AF

Fire

% Area 1975

% Area 2005

% change

8.73 4.84 54.64 18.38 7.56 2.76 3.10

4.69 8.02 61.84 9.47 10.68 3.72 1.59

−53.71 +60.33 +13.17 −51.52 +41.19 +34.72 N/A

measure of the spatial accuracy and classification accuracy of the Kohonen SOM neural network classification. Both the reference images and the section of the 1975 map used for testing comprised binary data. In the case of Bombwe forest it was BPF versus MW, Nyangombe forest used BPF versus KW and for the frost hollows I compared BPF versus D (representing the frost hollows). One of the results of the CROSSTAB operation in IDRISI Selva is the production of maps showing areas of agreement, areas of commission and areas of omission of the two different maps. These are illustrated in Figure 6.10 for the three different maps used for accuracy assessment. Map accuracy is calculated using the formula, M ap Accuracy =

Pa Pa + P c + P o

(6.1)

where Pa =number of pixels in the map image in agreement with reference data, Pc =number of pixels determined to be errors of commission and Po =number of pixels determined to be errors of omission. Table 6.4 shows errors of omission, errors of commission and percentage agreement between the reference images and the 1975 map. The preference for advanced statistical measures, such as Cramer’s V and the Kappa Index of Agreement (KIA) in accuracy analysis, is to take into account the element of chance. A simple percent agreement calculated from the ratio between areas of agreement, errors of commission and errors of omission does not take into account chance. For this reason a stratified random selection of 250 points was selected from both the reference images and the 1975 test image in order to calculate these measures of accuracy. The results of this quantitative assessment map accuracy are presented in Table

L AND C OVER C HANGE AND AGB

197

Figure 6.10: Maps showing areas of agreement, errors of commission and errors of omission in the three maps used for accuracy assessment of the 1975 map

198

C HAPTER SIX

Table 6.4: Percentage of agreement Pa , percentage of errors of commission Pc , percentage of errors of omission Po and percentage map accuracy for 1975 map Bombwe Forest

Nyangombe Forest

Frost Hollows

Pa Pc Po

66.60% 20.49% 12.91%

53.56% 3.58% 42.86%

73.13% 21.01% 5.86%

Map Accuracy

66.60%

53.56%

73.13%

Table 6.5: Measures of map accuracy for the 1975 classified vegetation map

χ2 df P-level Cramer’s V Overall Kappa Aspatial accuracy

Bombwe Forest

Nyangombe Forest

Frost Hollows

118810.63 1