yes The invasive ant Solenopsis invicta is established in Europe [1, 1 ed.]

Citation preview

ll Magazine Correspondence

The invasive ant Solenopsis invicta is established in Europe Mattia Menchetti1,5,*, Enrico Schifani1,2, Antonio Alicata3, Laura Cardador4, Elisabetta Sbrega1, Eric Toro-Delgado1, and Roger Vila1,5 The red imported fire ant (Solenopsis invicta) is classified as one of the worst invasive alien species1 and as the fifth costliest worldwide2, impacting ecosystems, agriculture and human health3. We report the establishment of S. invicta in Europe for the first time, documenting a mature population in Sicily. We use genetic analyses to assess its putative origin, as well as wind tracking and species distribution modelling to predict its potential range on the continent. We show that half of the urban areas in Europe are already suitable and that climate warming expected under current trends will favor the expansion of this invasive ant. In less than one century, the South American ant S. invicta established and spread throughout much of the United States, Mexico, the Caribbean, China, Taiwan and Australia (Figure 1A), while eradication succeeded only in New Zealand4,5. In Europe, there have been at least three documented interceptions of S. invicta, in Spain, Finland, and the Netherlands6. We documented 88 nests extending over about 4.7 ha during winter 2022/2023 in Sicily (Italy), near the city of Syracuse (Figure 1A–C and Data S1A,B). The invaded area, bordering a river estuary, is heavily disturbed, but falls within a larger regional protected site. Locals informed us of frequent ant stings in the area since at least 2019, suggesting a prolonged presence of S. invicta that is coherent with the large invaded area and high number of mature nests (Figure 1B,C). How the species reached this site is not clear, but no large landscaping or planting projects seem to have taken place over the last few years and it is highly unlikely that it represents the first arrival point and only location in the area. The proximity of one of the main R896

cargo harbors of the island, the Augusta port (~13 km northward), may be relevant for its introduction. Long-range dispersal of ant queens during nuptial flights tends to be aided by wind and follow its direction3. Locally prevailing wind directions at ground level indicate that, if arrived by flight, queens colonizing the invaded site may have come from the north-west, where further monitoring efforts should be prioritized (Figures 1A and S1). Likewise, swarming queens are likely to be directed southeast and therefore towards the sea, which may represent a limiting factor for further inland spread. Alarmingly, we observed nuptial flights even in winter (Figure 1B), well outside the spring-autumn seasonality typical of the northern hemisphere3. S. invicta colonies can be either monogynic or polygynic (hosting one or multiple queens, respectively), and the two social forms are characterized by specific alleles of the Gp-9 gene7. Polygynic colonies, which have become prevalent across much of the invaded range, adopt a supplemental short-range dispersal strategy, and may perform long-range flight dispersal less frequently3. We confirmed the polygyny of this population by detecting multiple dealate queens per nest and conducting genetic analyses on the Gp-9 gene. The history of the global invasion of S. invicta has been reconstructed through genetic data: they were first introduced from northeastern Argentina into the southern US, while all other alien populations apparently originated from at least nine distinct introductions from North America8. The Italian population presents one of the three main mitochondrial haplotypes, widely distributed across invaded areas and in Argentina, named H5 (Figure 1D)8. Among invaded areas, this haplotype is particularly frequent in the populations of southern US, mainland China and Taiwan (Figure 1E), the most likely introduction sources considering their top position in the global trade. We assessed environmental suitability for S. invicta across Europe and the Mediterranean using ensemble species distribution modeling. Under current environmental conditions, this ant may be able to establish in about 7% of the study region (Figure 1G), mainly occupying agricultural areas and, to a lesser extent, urban and protected areas (Figures 1F and S2, and Data S1C). Remarkably, half of the urban areas are recovered

Current Biology 33, R879–R897, September 11, 2023 © 2023 Elsevier Inc.

as suitable. This is concerning because most suitable urban areas are coastal Mediterranean cities highly connected by seaports, potentially favoring the spread of the species. Worryingly, future projections depict a far worse scenario, in which the suitable range of S. invicta largely increases (Figures 1H,I and S2, and Data S1D). Coordinated efforts for early detection and action in the region are key for successfully managing this new threat. Citizen science may play a key role in the detection of S. invicta considering that it is frequently encountered in urban and peri-urban areas, and due to its painful stings able to cause anaphylactic shocks9 and the characteristic large nest mounds. Monitoring efforts should be extended to a larger geographic scale, given the species dispersal capability and the presumed existence of an unknown first site of introduction. The establishment of effective detection strategies for alien ant species appears particularly important considering their rising numbers at the European and global scales5,10. SUPPLEMENTAL INFORMATION Supplemental information including two figures, methods, data file and inclusion and diversity statement can be found with this article online at https://doi.org/10.1016/j.cub.2023.07.036. ACKNOWLEDGMENTS We wish to thank Francesco Petralia for the early report that allowed the detection of the species. Support for this research was provided by “la Caixa” Foundation (ID 100010434) to M.M. (grant LCF/BQ/ DR20/11790020), by a Beatriu de Pinós fellowship (funded by the Catalan Government and EU COFUND program nº 801370) to L.C., and by the Secretaria d’Universitats i Recerca (Departament de Recerca i Universitats, Generalitat de Catalunya) with a Joan Oró predoctoral program grant and the European Social Fund Plus, EU (grant 2023 FI-1 00556) to E.T.-D. We acknowledge the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT model. DECLARATION OF INTERESTS The authors declare no competing interests. REFERENCES 1. Luque, G.M., Bellard, C., Cleo, B., Bonnaud, E., Genovesi, P., Simberloff, D., and Courchamp, F. (2014). The 100th of the world’s worst

ll Magazine A

D H36

Wind tracking Backward Forward

48 23

N

13

33

H5

H22

s rie

Sicily

1 0.75 0.5 0.25

W

(%)

Main hubs Cargo ports Airports

15

16

Tra jec to 25 samples 5 samples 1 sample

E

n

Missing haplotype

NATIVE A Argentina

Brazil

ALIEN Paraguay

South US

1939

Trinidad

1991

California

1998

Australia New Zealand

2001

Taiwan

China

2003

2004

2001

Italy

? - 2022

ARRIVAL DATE

E ALIEN ALIEN NATIVE A

50 km

B

C

100 m

S 51

Nest Nest + winged sexuals

277

568

133

7

H5 haplotype frequency 7

64 107

107

n

5 Eradicated

n total samples Image Landsat / Copernicus. Data SIO, NOAA, U.S. Navy, NGA, GEBCO

F

Land use categories suitable for S. invicta

G Current

H Future (2050)

I

35

Suitable area (%)

30

Suitable fraction of the total land use category

SUITABILITY 0-200

23.2%

URBAN 46.9%

200-400 400-600 600-800 800-1000

20.7%

PROTECTED

25 20 15 10 5

6.7%

90

70

20

20

nt re

20

ur C

AGRICULTURAL

50

0 42.4% 9.2%

Figure 1. Location, genetic analysis, and modeling of the potential spread of the new alien population of S. invicta. (A) The invaded area in Sicily is marked with a star. The directions of wind trajectories starting from (forward) and arriving at (backward) the study area are indicated as percentages over the total time frame. Main commercial hubs on the island are highlighted. Inset map summarizes the records retrieved for the species alien and native ranges. (B) Nuptial flight recorded in January 2023. (C) Satellite view of the study area (37.055N, 15.267E) and ant nest positions (Data S1A). (D) Haplotype network of mitochondrial sequences. The three main haplotypes are annotated. Colors indicate the sample origin and the sizes of the circles represent the number of samples. (E) Worldwide frequency of the H5 haplotype recorded in Italy, highlighting possible introduction sources. (F) Bars represent suitable area estimated for the species partitioned by land use category (green: % of total area of Europe; blue: % of total area of that category). (G) Ensemble model map prediction under current and future (H) environmental conditions. (I) Future trends of predicted suitable area (% of total area).

2.

3. 4.

5.

invasive alien species. Biol. Invasions 16, 981–985. https://doi.org/10.1007/s10530-0130561-5. Diagne, C., Leroy, B., Vaissière, A.C., Gozlan, R.E., Roiz, D., Jaric´, I., Salles, J.-M., Bradshaw, C.J.A., and Courchamp, F. (2021). High and rising economic costs of biological invasions worldwide. Nature 592, 571–576. https://doi.org/10.1038/ s41586-021-03405-6. Tschinkel, W.R. (2013). The Fire Ants (Cambridge, MA: Belknap Press). Morrison, L.W., Porter, S.D., Daniels, E., and Korzukhin, M.D. (2004). Potential global range expansion of the invasive fire ant, Solenopsis invicta. Biol. Invasions 6, 183–191. https://doi. org/10.1023/B:BINV. 0000022135.96042.90. Wetterer, J.K. (2013). Exotic spread of Solenopsis invicta Buren (Hymenoptera: Formicidae) beyond North America. Sociobiology 60, 50–55. https://doi. org/10.13102/sociobiology.v60i1.50-55.

6. Wong, M.K., Economo, E.P., and Guénard, B. (2023). The global spread and invasion capacities of alien ants. Curr. Biol. 33, 566–571.e3. https:// doi.org/10.1016/j.cub.2022.12.020. 7. Valles, S.M., and Porter, S.D. (2003). Identification of polygyne and monogyne fire ant colonies (Solenopsis invicta) by multiplex PCR of Gp-9 alleles. Insectes Soc. 50, 199–200. https://doi. org/10.1007/s00040-003-0662-8. 8. Ascunce, M.S., Yang, C.C., Oakey, J., Calcaterra, L., Wu, W.J., Shih, C.-J., Goudet, J., Ross, K.G., Shoemaker, D. (2011). Global invasion history of the fire ant Solenopsis invicta. Science 331, 1066–1068. https://doi. org/10.1126/science.1198734. 9. Kemp, S.F., DeShazo, R.D., Moffitt, J.E., Williams, D.F., and Buhner II, W.A. (2000). Expanding habitat of the imported fire ant (Solenopsis invicta): a public health concern. J. Allergy Clin. Immunol. 105, 683–691. https://doi.org/10.1067/ mai.2000.105707.

10. Schifani, E. (2019). Exotic ants (Hymenoptera, Formicidae) invading Mediterranean Europe: a brief summary over about 200 years of documented introductions. Sociobiology 66, 198–208. https://doi.org/10.13102/sociobiology. v66i2.4331. 1

Institut de Biologia Evolutiva (CSIC-Univ. Pompeu Fabra), Barcelona 08003, Spain. Department of Chemistry, Life Sciences and Environmental Sustainability, University of Parma, Parma 43124, Italy. 3Department of Biological, Geological and Environmental Sciences, University of Catania, Catania 95124, Italy. 4CREAF, Cerdanyola del Vallès 08193, Spain. 5Twitter: @MattiaMenchetti (M.M.); @RogerVila_Lab (R.V.) *E-mail: [email protected]

2

Current Biology 33, R879–R897, September 11, 2023

R897

Figure S1. Wind trajectory analyses. (A-D) Raster maps of trajectory density for the trajectories starting from the point of observation of S. invicta at ground level (i.e. forward trajectories) and (E-H) for the trajectories arriving at the point of observation of S. invicta (i.e. backward trajectories), computed at a resolution of one trajectory per hour from 11:00 to 15:00 CET for all days from April 1st to October 31st of either the past three (A, B, E, F; N = 3,210) or five years (C, D, G, H; N = 5,350). The plots are shown for trajectories lasting 5 hours (A, C, E, G), at a spatial resolution of 1Km, and lasting 24 hours, at a spatial resolution of 10Km (B, D, F, H). Each cell is colored according to the proportion over the total number of computed trajectories that pass within it. The black cell indicates the point of observation of S. invicta, at which all trajectories arrive (and therefore all trajectories pass within it). (I-L) Circular histograms representing the directions of wind trajectories starting at the point of observation of S. invicta (i.e. forward trajectories) and (M-P) arriving at the point of observation of S. invicta (i.e. backward trajectories), computed at a resolution of one trajectory per hour from 11:00 to 15:00 CET for all days from April 1st to October 31st of either the past three (I, J, M, N; N = 3,210) or five years (K, L, O, P, D; N = 5,350). The direction is indicated considering the angle between the starting coordinates and those of the trajectory endpoint after running for five hours (I, K, M, O) or 24 hours (J, L, N, P). The plot highlighted in red (K, O) corresponds to the same one shown in Fig. 1A of the main text.

Figure S2. Species distribution modeling. (A) Map showing occurrence data (red dots) and background (blue areas) used for species distribution models of S. invicta. (B) Ensemble model predictions under current environmental conditions at a global scale. (C) Ensemble model predictions for future environmental scenarios at a global scale for the year 2050. (D) Ensemble model predictions for future environmental scenarios at a global scale for the year 2070. (E) Ensemble model predictions for future environmental scenarios at a global scale for the year 2090. (F) Ensemble model predictions for future environmental scenarios at European and Mediterranean scale for the year 2070 and (G) for the year 2090. (H) Binary ensemble model predictions under current environmental conditions and (I, J, K) for future environmental scenarios at European and Mediterranean scale.

Supplemental Experimental Procedures: RESOURCE AVAILABILITY: DNA sequences are deposited in GenBank (accession numbers OQ974928 and OQ974929). Datasets and scripts are deposited in the Open Science FrameworkS1 at https://doi.org/10.17605/OSF.IO/Y8AMD. dataset_nests.txt. Table containing georeferenced nests with information about the presence of winged sexuals. Fields: “decimalLatitude", "decimalLongitude", “measurementAccuracy” (following Darwin Core Archive format), “wingedSexuals” (Y = winged sexuals observed; N = winged sexuals not observed). dataset_occurrences.txt. Occurrence dataset used in the species distribution models. Fields: “type” (occurrence type, i.e. preserved specimen, literature, human observation, record presented in this study), “db” (source database, i.e. GBIF, iNaturalist, GABI, downloaded from literature, record presented in this study), “id” (ids of the occurrence in their original source), “observed_on” (observation date reported), “year” (year reported), “year2” (year used for calculating the background in the sdm analysis), “decimalLatitude”, “decimalLongitude”, “measurementAccuracy” (following Darwin Core Archive format), “locality”, “county”, “state”, “country”, “source” (exact source, e.g. literature references). dataset_sequences.txt. Genetic dataset used in the haplotype network analysis. Fields: “an” (GenBank accession number); “species_name”; “country”; “country_label” (country labels used for haplotype network); “source” (references, when available); “nucleotides” (untrimmed sequence). data_preparation.R. R script used for data preparation for the species distribution modeling. sdm_modelling.R. R script used for species distribution modeling. future_model_prediction.R. R script used future prediction in the species distribution modeling. extent_suitable_areas_calculations.R. R script used for the calculation of the extent of suitable areas. Solenopsis_opentraj_commands_all_altitudes.sh. Command file used to carry the wind trajectories analysis. METHOD & RESULTS DETAILS: 1. Wind trajectories We reconstructed wind trajectories using the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) dispersion model from National Oceanic and Atmospheric Administration (NOAA) Air Resources LaboratoryS2. Analyses were based on the Reanalysis databaseS3. We used a locally installed version of HYSPLITS3 and a custom R pipelineS4 to automate the execution of a large number of runs of the dispersion model. This pipeline parses a set of input dates using the “lubridate” packageS5 to generate the complete list of dates and times for which trajectories need to be computed, and then downloads the necessary files from the database and sends the calls to HYSPLIT using the “splitr”S6 and “opentraj”S7 packages. Analyses were conducted using R version 4.1.2. Plots were later edited with Adobe Illustrator. We computed both forward and backward trajectories for the preceding years to detect the predominant winds starting from and arriving at the locality of observation of S. invicta. In all cases, trajectories were computed for all days from April 1st to October 31st and between 11:00 and 15:00 CET, as late morning to early afternoon of the warm season is the typical timeframe of the species’ nuptial flightsS8. Since the amount of time the nests have been at the studied locality is unknown, we computed trajectories considering the previous five as well as the previous three years; the same overall patterns were detected. Trajectories were initialized at ground level and were computed for a duration of 5 hours as S. invicta tends to land shortly after mating and flights are not expected to last very longS8, but also for a duration of 24 hours. Different altitude settings (100m and 500m) produced very similar patterns and can be inspected through the script we provided. The trajectories were plotted in a raster of trajectory density with the “sp” package S9,S10 to detect potential regions to which dispersal could be more likely; the same patterns were observed for either 3 (20192021) or 5 years (2017-2021) and for 5 or 24 hours of duration (Figures S1A, S1B, S1C, S1D, S1E, S1F, S1G, and S1H). We could not compute the analysis for the year 2022 due to the lack of meteorological data. Given the similarities between the patterns observed using 3 and 5 years, we do not expect that this may alter the analysis. In addition, the azimuth was computed for each trajectory using the “raster”S11 and “geosphere”S12 packages, considering the angle between the point of origin and its coordinates after the total duration of the trajectory (i.e., 5 and 24 hours). Azimuths were plotted as circular histograms using “ggplot2”S13 to visualize the predominant directions of dispersal (settings using 5 years and 5 hours are in Figure 1A, all combinations of settings Figures S1I, S1J, S1K, S1L, S1M, S1N, S1O, and S1P). 2. Genetic analyses Mitochondrial sequences. Ant specimens collected during the field surveys in November 2022 and January 2023 were identified under a stereomicroscope. Two specimens from different colonies were selected for COI analysis. Due to the low

urbanization and the lack of significant commercial hubs in the area, it is highly unlikely that the population originated from multiple independent sources of introduction. DNA was extracted following the protocol by Schär et al.S14. We amplified 920 bp of the mitochondrial gene cytochrome c oxidase including both the subunits I and II using the primers C1-J-2195 (5’TTGATTTTTTGGTCATCCAGAAGT) and DDS-COII-4 (5’TAAGATGGTTAATGAAGAGTAG)S15. The following components were added to each PCR tube: 2 μl of genomic DNA, 0.1 μl of GoTaq® G2 Flexi DNA Polymerase from Promega, 5 μl of 5X Green GoTaq® Flexi Buffer, 2 μl of MgCl2 25 mM, 0.5 μl of dNTPs 10 mM, 0.5 μl of each primer 10 mM, mili-q® H2O up to a final volume of 25 μl. A negative control without genomic DNA was used. The PCR cycling conditions were: 1 cycle of 1 min. at 94ºC, followed by 35 cycles of 30 sec. at 94ºC, 1 min. at 48ºC, 120 sec. at 68ºC, followed by a final cycle of 5 min. at 72ºC. The PCR products were visualized by gel electrophoresis with a 1% agarose and sent to Macrogen Europe for Sanger sequencing in both directions. Sequences are available on GenBank under the accession number OQ974928 and OQ974929. A total of 1,413 mitochondrial sequences covering the worldwide range of the species were retrieved from GenBank and multiple literature sourcesS1. If the number of specimens sharing a given haplotype in a particular location was reported, but instead of sequences only haplotypes were uploaded to GenBank, we duplicated the haplotype sequences in order to obtain data at the specimen level. Sequences were aligned and trimmed to 714 bp using Geneious 2020.2.4 (www.geneious.com). A haplotype network (Figure 1D) of S. invicta sequences was built with the program TCS 1.21S16 and later edited with tcsBUS17 and Adobe Illustrator CC 2022. The mitochondrial haplotype present in the Italian population is also present in Argentina, Trinidad, California, Southern US, China, Taiwan, Australia, and New Zealand (Figures 1D and 1E). The New Zealand haplotype refers to an intercepted colony, while a formerly established population with a different haplotype has been eradicated. GP-9 allele. DNA was extracted from four specimens from different colonies following the protocol by Schär et al.S14. We amplified the Gp-9 gene using the primers 24bS (5’TGGAGCTGATTATGATGAAGAGAAAATA) and 25bAS (5’GCTGTTTTTAATTGCATTTCTTATGCAG)S18. The following components were added to each PCR tube: 2 μl of genomic DNA, 0.1 μl of GoTaq® G2 Flexi DNA Polymerase from Promega, 5 μl of 5X Green GoTaq® Flexi Buffer, 2 μl of MgCl2 25 mM, 0.5 μl of dNTPs 10 mM, 0.5 μl of each primer 10 mM, mili-q® H2O up to a final volume of 25 μl. A negative control without genomic DNA was used. The PCR cycling conditions were: 1 cycle of 1 min. at 92ºC, followed by 35 cycles of 15 sec. at 92ºC, 45 sec. at 55ºC, 150 sec. at 68ºC, followed by a final cycle of 7 min. at 68ºC. The PCR products were separated with a 2% agarose gel, prepared by adding 0.2 g of agarose for every 10 ml of Tris buffer – Boric acid – EDTA (TBE) 1X and by adding 7.5 μl of SYBR Safe DNA Gel Stain from ThermoFisher Scientific for every 60 ml of solution. The size of the fragment was checked by loading in the gel 5 μl of 100bp DNA Ladder 50ug (1.0 ug/ μl) from Invitrogen. The presence of the b allele of the Gp-9 gene was determined by the approximate length of 422 bp (Valles & Porter, 2003). For confirmation, one of the products was sent to Macrogen Europe to be sequenced in both directions and the sequence was identical to AF427898.1. 3. Species distribution modeling Occurrence and environmental data. To model the distribution of the species we compiled a worldwide dataset of S. invicta occurrences by gathering museum specimen dataS19,S20, citizen science data from iNaturalist.org, and from literature sourcesS1. Identification of all iNaturalist.org records were double-checked by authors (MM and ESc) through photo examination, except for those from the United StatesS21, where the species is very well known by the public. We retained only records with an error accuracy below 10 km. A total of 6,888 occurrences of S. invicta were retrievedS1. Duplicate records at the 5-arcmin resolution were handled as only one occurrence, with a total of 2,410 occurrences available for analyses (Figure S2A), compared to the 88 of Bertelsmeier et al.S22, 827 of Sung et al. S23, and 1,610 of Chen et al. S24. We also compiled 19 global climate variables from the WORDLCLIM v2S25 at 5-arcmin resolution. To avoid multicollinearity issues, we conducted a principal component analysis (PCA) to identify clusters of collinear variables and retained for analyses the variable within each cluster with larger biological meaning: namely, mean diurnal range (bio2), temperature seasonality (bio4), mean temperature of the coldest quarter (bio11), and precipitation of driest quarter (bio17). We also inspected them by applying the variance inflation factor (VIF) using the R package “usdm”S26. All VIF values were