Climate change alters global invasion vulnerability among ecoregions

We assess climate similarity among global freshwater and terrestrial ecoregions under historical and future climate scenarios to determine where climate change will impact the climate filter of invasion process.


| INTRODUC TI ON
Invasive species are a primary cause of global biodiversity decline (Bellard et al., 2016) and cause significant economic damages, resulting in estimated economic damages since 1960 of at least $1130.6 billion, in 2017 values (Cuthbert et al., 2022).Climate change is predicted to become a leading cause of biodiversity loss within the century (Thomas et al., 2004) and alter future biological invasion outcomes (Essl et al., 2020;Rahel & Olden, 2008).The human-mediated movement of species and resulting range expansion is also reorganizing global biogeography, with biotic homogenization likely to occur among regions similar in climate (Capinha et al., 2015) and climate change is expected to lead to homogenization of biodiversity (Saladin et al., 2019).The impacts of invasive species on biodiversity will likely result in declines in ecosystem services, indicating that proactive adaptive actions are necessary to preserve ecosystem function (Cuthbert et al., 2022).
The invasion process, beginning with the transfer of species from a source region to the release in a non-native range, followed by the survival, establishment, spread and impacts of released species, is influenced by biotic and abiotic factors (Lodge et al., 2016).These factors include the number of individuals released in a non-native region (propagule pressure; Bradie et al., 2013;Lockwood et al., 2005), the climate of the region of introduction (Bomford et al., 2009(Bomford et al., , 2010;;Forsyth et al., 2004;Howeth et al., 2016), interactions with native communities (Zenni & Nuñez, 2013), and secondary spread networks (Vander Zanden & Olden, 2008).Climate change is predicted to influence the introduction of species to non-native areas (Chan et al., 2019;Walther et al., 2009), the removal of climatic barriers to spread (Buckland et al., 2001), species survival in non-native regions (Walther et al., 2007), non-native species interactions with novel biotic communities and the magnitude of invasion impacts (Rahel & Olden, 2008;Walther et al., 2009).A key uncertainty related to survival is the regional pattern and intensity of current and future climate similarity among source and recipient ecosystems.Understanding global patterns in regional climate change-driven impacts to introduced species survival is essential because invasive species are transported globally via trade networks (Hulme, 2009;Sardain et al., 2019;Seebens et al., 2017).Freshwater and terrestrial ecoregions represent appropriate spatial units to assess geographic patterns of climate change-driven vulnerability to freshwater and terrestrial species survival, as they reflect the distinct distribution and composition of freshwater fish and terrestrial plant species respectively (Abell et al., 2008;Olson et al., 2001).Ecoregions are intended to represent the original extent of biotic communities prior to major disturbances resulting from human activities and represent distinct biotic groups and evolutionarily significant processes.Therefore, these biogeographic regions have applications for global and regional conservation-related assessments and planning.Identifying regions where climate change and invasive species management policy could be prioritized would be advantageous for informing global conservation efforts.
Climate change influences the colonization stage of the invasion process by altering the success of species survival following arrival (Walther et al., 2009).Climate matching is a tool commonly used to assess how climatic conditions of a recipient region (i.e. an invaded region) will affect the survival of potential invasive species (Lodge et al., 2016;Mandrak & Cudmore, 2015).Climate matching involves quantifying the similarity in climates between a species' source location, that is, native and introduced ranges, using a suite of bioclimatic variables, such as mean annual temperature, and has been validated for predicting invasive species survival in introduced regions (Bomford et al., 2009(Bomford et al., , 2010;;Bradie et al., 2015;Duncan et al., 2001;Forsyth et al., 2004;Howeth et al., 2016;Van Wilgen et al., 2009) and serves as a useful method for invasive species screening protocols (Richardson & Thuiller, 2007).Predictions based on climate matching will likely be disrupted by climate change due to disproportional changes in climates among ecoregions (Chapman et al., 2017), implying that climate change projections should be incorporated into analyses of biological invasions at regional scales (Britton et al., 2010;Campbell et al., 2022;Hubbard et al., 2023;Kriticos, 2012).
Our research focuses on climate similarity and species survival because global environmental gradients act as strong filters for species establishment (Rahel & Olden, 2008).Therefore, the aim of this study is to assess the impacts of climate change on patterns of climate similarity among freshwater and terrestrial ecoregions to identify where the colonization phase in the invasion process would be most affected.
We compare the climate match between ecoregions under historical climates and under climate change projections.Specifically, we test whether the climate match among ecoregions globally will become more similar under climate change projections than under historical climatic conditions, indicating if climates between ecoregions will homogenize or become heterogeneous.At the biogeographic realm scale, we aim to show where climate change may lead to new potential source and recipient pairs of ecoregions likely to support survival of non-native species following arrival.Understanding species survival in recipient ecosystems can help direct conservation efforts in resource allocation and policy goals and aid in the forecasting of future climatedriven changes in the regional vulnerability of invasion.

| Overview
To develop the climate similarity data set, we calculated climate similarity among all possible pairs of freshwater and terrestrial ecoregions, with each ecoregion serving as a potential source and biogeography, biological invasions, climate matching, freshwater, terrestrial recipient of introduced species (Abell et al., 2008;Olson et al., 2001).
We used a historical climate data set and six global climate models (GCMs) for the climate data.We wrote the Climatch algorithm (Crombie et al., 2008) in R (R Core Team, 2022) to allow the versatility of incorporating climate change data and calculating climate match among source-recipient ecoregions.Finally, we analysed the climate match data and mapped results of the ecoregions to develop an understanding of the changing spatial distribution of climate similarity at a global scale.

| Ecoregions
We used the World Wildlife Foundation Ecoregions of the World biogeographic classifications for freshwater ecoregions that are based on the distribution of freshwater fish species (Abell et al., 2008) and terrestrial ecoregions that represent distinct plant assemblages (Olson et al., 2001).Both sets of ecoregions are designed to be representative of regional environmental and ecological variables providing geospatial units that encompass distinct biodiversity features and ecosystems at large scales that are regarded as useful for conservation planning (Thuiller et al., 2011).
Therefore, ecoregions offer a useful spatial delineation for analysis of the impacts of climate change on estimates of future sources of introduced species.Freshwater ecoregions were typically larger (mean unit surface area = 312,315 km 2 ; n = 426) than terrestrial ecoregions (mean unit surface area = 164,691 km 2 ; n = 821).Seven terrestrial ecoregions were removed from analyses: Climate data were not available for five small oceanic island ecoregions; the 'rock and ice' ecoregion covering inland Antarctica and Greenland; and the 'lake' ecoregion because this ecoregion is a set of large-lake systems from different locations globally, with different climates and, thus, these lakes are better represented in the freshwater ecoregion data set.

| Climate data
The spatially interpolated bioclimatic data set from global weather stations, WorldClim v2.1, was used for the historical time period 1970-2000 (Fick & Hijmans, 2017) (Fick & Hijmans, 2017).These climate projections were developed by research groups using different narratives of future human development known as shared socioeconomic pathways (SSPs), which reflect projections of future economic growth and demographics affecting the mitigation of global greenhouse gas emissions and adaptation to impacts (O'Neill et al., 2014(O'Neill et al., , 2017)).
CMIP6 has modeled multiple radiative forcing scenarios specific to each SSP (Chen et al., 2020).However, WorldClim offers a single radiative forcing scenario (e.g.8.5) for each SSP.The IPCC does not assign probabilities to predictions of SSP outcomes in the future; therefore, we conducted our analysis with three SSPs to represent a range of possible projected future climatic conditions.Scenario SSP2-4.5 represents a narrative of intermediate challenges to emissions mitigation and adaptation (Fricko et al., 2017).Scenario SSP3-7.0 represents a narrative of regional rivalries with strong challenges to mitigation and adaptation to effects of climate change.Scenario SSP5-8.5 represents a narrative of continued fossil fuel-driven economic growth through much of the 21st century (O'Neill et al., 2017).
Several groups of climatic variables are typically used as inputs for species invasiveness and climate match analyses, including select variables from a suite of 19 bioclimatic variables (Bomford et al., 2010;Britton et al., 2010;Howeth et al., 2016).For freshwater animals, Bradie et al. (2015) demonstrated that annual mean temperature, minimum temperature of coldest month, mean temperature of driest quarter and mean temperature of coldest quarter provided the most accurate predictions of non-native species presences; therefore, we used this set of variables for the freshwater ecoregions.For terrestrial plants and animals, Bradie et al. (2015) demonstrated that annual mean temperature, minimum temperature of coldest month, mean temperature of wettest quarter, mean temperature of coldest quarter best predicted non-native species establishment; therefore, we used this set of variables for the terrestrial ecoregions.For freshwater ecoregions, air temperature data were used as a substitute for water temperature because bioclimatic variables are commonly used in freshwater assessments (Bomford et al., 2010;Bradie et al., 2015) and surface freshwater systems readily exchange heat with the atmosphere and stream and river water temperatures have strong positive correlation to air temperature (Caissie, 2006;Mohensi & Stefan, 1999).In lakes, temperature dynamics are complicated by mixing or stratification and ice-over periods; however, even in large lakes, surface temperatures remain strongly correlated with air temperature (Trumpickas et al., 2009).Furthermore, climate change has led to increases in lake surface temperatures (Dobiesz & Lester, 2009) and significant warming of deep water in non-stratified lakes, although summer hypolimnion temperatures in stratified lakes are unlikely to change (Robertson & Ragotzkie, 1990).

| Climate similarity
The Euclidean climate distance measuring algorithm, Climatch (Crombie et al., 2008), calculates the standardized distance between a source region i and a recipient region j as: .
Here, the floor function rounds the match score down to the nearest integer, y represents the grid cell and k represents the climate variable.This algorithm calculates a 'climatch' score from 0 to 10 for each grid cell within a recipient region, representing the closest match that each grid cell in a recipient region has to any grid cell in a source region.A climatch score of 0 represents a very poor match and 10 a perfect match.Sum Climate scores are the number of climate data points within a recipient region that have a match at and above a given climatch score (Bomford et al., 2010).This value is then typically expressed as a percentage of the total number of climate grid cells within the receiving region, which we refer to as 'climate match' here.Bomford et al. (2010)  score for all possible source-recipient pairs of each ecoregion set.
Climate match between two regions can be asymmetrical and differs depending on whether a region is considered as a source or recipient region.This is the result of differences in number of grid cells and variation in climatic conditions between the regions.
For example, if region i is large and has a heterogeneous climate, whereas region j is smaller and has a fairly homogenous climate, it is possible for region i to be a strong match as a source to region j, but for j to be a weaker match as a source to region i.We used climate data of each grid cell within each ecoregion rather than an average or centroid measurement to maintain the variation across the region, as it better reflects the climate match between regions.We computed the climate match between all possible pairs of ecoregions with each ecoregion as a potential recipient (j) from each ecoregion as a source (i).Climate matches were computed for the historical climate period and for each individual GCM, then climate matches of the GCMs were averaged across all GCMs for each projected climate change scenario (Sofaer et al., 2017).The change in mean climate match for a given ecoregion was the number of climate match percentage points increasing or decreasing between the historical and projected climate scenarios.For example, if the historical mean climate match of an ecoregion was 10% and the mean climate match under the SSP2-4.5 scenario was 12%, the change in mean climate match was 2.

| Analyses
Friedman's non-parametric two-way analysis of variance with repeated measures by ranks (Legendre & Legendre, 2012) was used to test for an effect of climate scenario on mean climate match of all ecoregions of each set (freshwater and terrestrial).Pairwise Wilcoxon rank sum with paired samples post hoc tests were used to examine differences between each scenario (Wilcoxon, 1945).
Non-parametric tests were used because the residuals from the analysis of variance models violated the assumption of normality.An alpha of 0.05 was used to determine significance for all statistical analyses where relevant.Climate similarities were displayed visually for each ecoregion set (e.g.change in mean climate match between historical and future climate conditions).All statistical analyses, modelling and mapping were performed using the statistical programming software R v.4.3.0 (R Core Team, 2022).Maps were made using the 'tmap' package (Tennekes, 2018) in the Robinson projection and using the World Geodetic System (WGS84).
Circular plots were used to display where potential sourcerecipient pairs of ecoregions were projected to increase in climate similarity above thresholds of 70%, 80% and 90%, among biogeographic realms.In these plots, each line originates from the biogeographic realm of the source ecoregion of the pair and is represented with a smooth end, which then connects with another or the same biogeographic realm with a pointed end indicating the realm of the recipient ecoregion in the pairing.The thickness of the lines and tick marks along the outside of the plots indicates the number of pairs of source-recipient ecoregions that were projected to rise above the given threshold.

| RE SULTS
Mean climate match of all ecoregions of each freshwater and terrestrial set globally were significantly different between historical and projected future climate scenarios for freshwater ecoregions (n = 426) as recipients (Friedman X 2 = 168.73,df = 3, p = <.0001,small effect size: 0.132) and sources (Friedman X 2 = 134.37,df = 3, p = <.0001,small effect size: 0.105), and for terrestrial ecoregions (n = 821) as recipients (Friedman X 2 = 811.83,df = 3, p = <.0001,moderate effect size: 0.330) and sources (Friedman X 2 = 637.36,df = 3, p = <.0001,small effect size: 0.259).There was a general trend of increased mean climate match with higher emissions scenarios although no differences between emissions-scenario projections were found for freshwater ecoregions as sources (Table 1 and We identified 10 recipient freshwater and terrestrial ecoregions with the greatest increases and decreases in mean climate match between the historical climatic conditions and for the projected climate of 2090 SSP5-8.5.The 10 recipient freshwater and terrestrial ecoregions with the greatest changes were in five biogeographic realms, although primarily in the Nearctic and Palearctic realms (Appendix Table S1 and Table S2).The freshwater and terrestrial ecoregions as sources with the greatest changes in mean climate match between historical climatic conditions and 2090 SSP5-8.5 were dispersed across six biogeographic realms and most commonly in the Nearctic and Palearctic realms (Appendix Table S3 and Table S4).

| DISCUSS ION
Our study found that, under climate change, many ecoregions in North America and Eurasia, particularly in the Arctic, will have TA B L E 1 Results of Wilcoxon post hoc test of the differences in freshwater and terrestrial ecoregion mean climate match between the historical  and projected 2090 (2081-2100) climates for the shared socioeconomic pathways SSP2-4.5,SSP3-7.0,SSP5-8.5.
Adjusted p-values were done using Bonferroni adjustment.historical and projected climate scenarios implies that, at status-quo levels of species transport between regions, invasion rates could increase because overall, climates of introduced regions may be more suitable for introduced species.The results of this study build upon previous research that predicted that climate change will drive biotic homogenization of mammals (Hidasi-Neto et al., 2019) and terrestrial plants (Saladin et al., 2019) due to environmental niche shifts and that homogenization of biodiversity is more likely to occur among regions similar in climate (Capinha et al., 2015).If historically biogeographically distinct regions, which reflect evolutionary processes shaped by their climatic conditions (Abell et al., 2008), become more similar in climates to each other, homogenization is likely to be a consequence.
Historical records of international species flows indicate that invasive species are transported at a global scale and between all continents and biogeographic realms, although unevenly (Capinha et al., 2017;Hudgins et al., 2023).The introduction of species is driven by patterns of economic trade flows (Chapman et al., 2017;Seebens et al., 2015) that can be modulated by bridgehead effects (Bertelsmeier & Ollier, 2021).Climate similarity between source and recipient regions, the focus of our research, is also a significant predictor of invasions (Capinha et al., 2023;Sofaer & Jarnevich, 2017).Records of species flows from 1960 to 2020 from analysis of the InvaCost project show the greatest source regions to be from the Nearctic (mostly North America), Palearctic and Indo-Malay realms, with the greatest donor of cost flows being China to the recipient USA (Hudgins et al., 2023).The interbiogeographic realm analysis (Figure 5) indicated that many ecore- The climates of many ecoregions of different biogeographic realms were predicted to become more similar under climate change conditions, which could lead to increased biotic homogenization among these regions, and possibly contribute to the reshaping of biogeography, a process that has begun due to historical invasions (Capinha et al., 2015).Here, we used several thresholds in our biogeographic realm analyses to forecast how climates of ecoregions among biogeographic realms could become more similar (i.e. Figure 5).We used several thresholds because survival probability likely has a non-linear relationship to climate match, such that a change in climate match may not result in a proportional change in survival probability (Bomford et al., 2010).
It is likely that there are thresholds of climate similarity that, once surpassed, would allow for species survival.et al., 2016;Lovell et al., 2006;Vander Zanden et al., 2010).Risk assessment tools, such as the freshwater Fish Invasiveness Scoring Kit (FISK) (Copp, 2013) based on the weed risk assessment tool (Pheloung et al., 1999) The climate matching layer that we provide at the ecoregion scale could be combined with other information on propagule pressure and species invasions elsewhere or other data used within these frameworks to conduct species-or pathway-specific assessments.
Supporting future research and management, particularly that of pre-invasion, is vital because the high cost of biological invasions has not been met with sufficient proactive management (Cuthbert et al., 2022).
There are several limitations of our study.Here, we examine the survival stage of the invasion process in relation to climate change using climate matching, yet there are other important factors such as propagule pressure (Lockwood et al., 2005), biotic interactions of introduced species with native communities (Theoharides & Dukes, 2007), habitat suitability (Lodge et al., 2016) and secondary transport pathways following introduction (Johansson et al., 2018) that affect invasion success.Some studies have found economic trade to be a more significant driver of invasions than climate match or environmental distance measures (Dawson et al., 2017;Pyšek et al., 2010;Sardain et al., 2019), limiting the scope of this study.However, the combination of climate matching and connectivity through trade networks has been shown to be more predictive than trade alone (Chapman et al., 2017;Seebens et al., 2015), and research that accounts for biases in sampling effort has found climate similarity to be of greater importance than trade (Sofaer & Jarnevich, 2017), underscoring the utility of our analysis.Here, we do not aim to predict future invasion rates in ecoregions, but to show how climate change will modify the survival stage of the invasion process among ecoregions and provide an assessment and data that can underpin rigorous risk assessments in these regions.Furthermore, our results should be interpreted in the context of freshwater animals, and terrestrial plants and mammals because we use the bioclimatic variables predictive of each taxonomic group's survival in introduced regions (Bradie et al., 2015).Three of the four bioclimatic variables of freshwater animals are shared with aquatic plants (Bradie et al., 2015); therefore, this, along with the consistency between the analyses for freshwater and terrestrial ecoregions, indicates that results may be similar for other taxonomic groups.Future directions of research could include other taxonomic groups, such as marine species and their ecoregions (Spalding et al., 2007).
In using ecoregions as our spatial unit, not species-specific (e.g. native and non-native ranges) spatial data to address ecosystem vulnerability to biological invasions, we do not estimate survival probabilities for individual species.The scale and configuration of spatial units used in our analyses could lead to potential bias in aggregated results.Potential for bias is ubiquitous in ecological data and may not be resolved in all cases (Wakefield & Lyons, 2010).Therefore, . Projected climate change data were accessed from the WorldClim platform (WorldClim, 2020), which provided global downscaled and bias-corrected bioclimatic data for averages of the period 2081-2100 (hereafter 2090), produced from the Intergovernmental Panel on Climate Change (IPCC) Coupled Model Intercomparison Project 6 (CMIP6) of six GCMs (BCC-CSM2-MR, CNRM-ESM2-1, CanESM5, IPSL-CM6A-LR, MIROC-ES2L, MRI-ESM2-0) at a resolution of 10 arc minutes

Figure 1 )
Figure 1).There was strong variation in change in mean climate match between historical and the highest emissions scenario (Figure 2), and the variance of the change in mean climate match of ecoregions was higher under higher emissions scenarios.The variance of change in mean climate match of freshwater ecoregions as recipients was SSP2-4.5 = 3.77, SSP3-7.0 = 7.64, SSP5-8.5 = 10.65, and as sources was SSP2-4.5 = 5.16, SSP3-7.0 = 9.28, SSP5-8.5 = 13.35.The variance of change in mean climate match of terrestrial ecoregions as recipients was SSP2-4.5 = 3.29, SSP3-7.0 = 6.28,SSP5-8.5 = 8.82, and as sources: SSP2-4.5 = 4.89, SSP3-7.0 = 8.90, SSP5-8.5 = 12.67.Mean climate match of freshwater and terrestrial ecoregions as recipients increased under climate change largely in the Northern Hemisphere, within the Nearctic and Palearctic biogeographic realms and particularly in the Arctic and Subarctic (Figure3).Freshwater and terrestrial ecoregions decreased in mean climate match as recipients, particularly in Africa and South America, the Aftrotropics, greater climate similarity to other ecoregions, thereby increasing the overall likelihood of survival for species introduced to these regions.Other areas, particularly in Africa and South America, showed lower climate similarity under climate change.Results also indicated that there were increases in the change in average climate similarity of ecoregions as potential sources of biological invaders, meaning that the organisms within these ecoregions under future climatic conditions would, on average, have greater potential to survive in other ecoregions should they be introduced.Our assessment at the ecoregion scale shows variation in mean climate match among ecoregions nested within countries, particularly in larger countries (e.g.Canada, Russia and the USA).Climate matching at the ecoregion scale can provide insights that may not be found if climate matches are combined or averaged across entire countries.Homogenization of climates among ecoregions may lead to a homogenization of biodiversity(Rahel & Olden, 2008) due to a reduction of abiotic filters (climate) on the survival of introduced propagules.With climates becoming more similar among ecoregions, species adapted to those increasingly similar climatic conditions are more likely to survive following introduction.Invasive generalist species are already replacing more specialized native species due to global changes like climate disturbances and habitat loss(Clavel et al., 2011).Furthermore, the finding of global mean climate match of freshwater and terrestrial ecoregions increasing between F I G U R E 1 The mean climate match (%) of freshwater (left panels) and terrestrial (right panels) ecoregions as recipients (top) and sources (bottom) globally for historical (1970-2000) climate and projected climates of 2090 (2081-2100) for SSP2-4.5,SSP3-7.0 and SSP5-8.5.Friedman effect size on the right side of each panel.**** indicates significant difference with an adjusted p-value < .0001,** indicates an adjusted p-value < .001.
gions would become more similar in climate, occurring between all biogeographic realms and that the Palearctic and Nearctic are projected to have the greatest numbers of potential sourcerecipient pairs of ecoregions increasing in climate similarity.Ecoregions in the USA and Canada are projected to increase in mean climate match as recipients to all other ecoregions affecting the success of arriving species.Furthermore, ecoregions in the Arctic and Subarctic were predicted to become, on average, more similar in climate to potential global source regions, indicating that a changing climate will affect the number and locations of potential F I G U R E 2 The mean climate match of each freshwater (left) and terrestrial (right) ecoregion as a recipient (top) and as a source (bottom), using historical climate data versus data projected for 2090 under the SSP5-8.5.A 1:1 line is shown in red.Points above the 1:1 line indicate ecoregions predicted to increase in climate match and those below the line to decrease.sources of biological invasions in those recipient areas and possibly become a major driver of future invasions, an existing concern of experts in this field (Essl et al., 2020) and supporting current hypotheses about whole system changes in the Arctic (Heino et al., 2020; Ravolainen et al., 2020).Because species flows are global and climate similarity among ecoregions is projected to increase globally, albeit with heterogeneity in the mean climate match between ecoregions, future research and risk assessment of managed regions should consider the impacts of climate change, and this study provides a basis for formal assessment that would incorporate other important factors of the invasion process (e.g.propagule pressure) at an ecologically relevant scale.The location of these changes is where conservation or restoration resources could be most effectively used.These areas, and individual ecoregions with the greatest increases in mean climate match, represent priority targets for detailed assessment of biological invasion risk and detection of new invaders and future research into the impacts of climate change.
Howeth et al. (2016) F I G U R E 3 Change in mean climate match of ecoregions as recipients between the historical climate (1970-2000) and projected climates of 2090 (2081-2100) for SSP2-4.5,SSP3-7.0 and SSP5-8.5.(a) Freshwater ecoregions (n = 426).(b) Terrestrial ecoregions (n = 821, Antarctic regions not shown).Mean climate match is represented with each ecoregion as a potential recipient of invasive species from every other ecoregion.Points represent the centroids of ecoregions of the Oceanic biogeographic realm for visibility.found that, for the Laurentian Great Lakes basin recipient region and source regions of the native ranges of introduced freshwater fishes, a climate match of 71.7% or greater was predictive of survival in the Great Lakes.While this threshold may be suitable for that particular region, climate match thresholds that are predictive of species survival likely vary across taxonomic groups and ecoregions of different sizes.Therefore, increases in the climate match of source-recipient ecoregion pairs among and within biogeographic realms, at multiple levels, indicate new opportunities for species survival in recipient regions under climate change.Results indicated that broad trends in changes in climate similarity among ecoregions nested within biogeographic realms were similar at several levels.The Palearctic realm had the most withinrealm terrestrial ecoregions to increase in climate match.This realm includes Europe and much of Asia, continents that have had the largest transfers of introduced vascular plants over the last 60 years (Seebens et al., 2015).The prevalence of species transfers and the increases in climate match of ecoregions in the realm could make the Palearctic susceptible to biotic homogenization under climate change.Climate matching is often an important component of broader frameworks that aim to prevent, eradicate or minimize the spread and impacts of invasive species.Horizon scanning can use climate matching and an assessment of species invasiveness elsewhere, to predict if species will become invasive and cause potential environmental and socio-economic impacts(Bayón & Vilà, 2019;Matthews et al., 2017).The aim of horizon scanning is often to identify potential invaders prior to arrival for the prevention of introduction, a priority of conservation organizations worldwide and the preferred strategy over eradication or adaptation(Leung et al., 2002;  Lodge F I G U R E 4 Change in mean climate match for ecoregions as sources between the historical climate (1970-2000) and projected climates of 2090 (2081-2100) for SSP2-4.5,SSP3-7.0 and SSP5-8.5.(a) Freshwater ecoregions (n = 426).(b) Terrestrial ecoregions (n = 821, Antarctic regions not shown).Mean climate match is represented with each ecoregion as a potential source of invasive species to all other ecoregions.Points represent the centroids of ecoregions of the Oceanic biogeographic realm for visibility.
of potential bias with supplementation of climate matches of each individual ecoregion is necessary for future analyses.Future research may refine the potential source regions through incorporating data of specific invasion pathways (e.g.aquarium trade for freshwater fishes, ornamental trade for terrestrial plant species) that connect sources of introduced species to particular recipient regions.Furthermore, microclimates within recipient ecoregions may support invasive species where overall climate match appears low, which is recognized in the climate match percent score used in this study.Future research directions could include an examination of the influence of microclimates and the spatial scale used in climate matching analyses.Overall, our results support previous predictions of climate change affecting patterns of the survival of biological invaders(Essl et al., 2020;Rahel & Olden, 2008) and, by the end of this century, climate change will influence patterns in climate match between ecoregions.Climate similarity among ecoregions was predicted to increase overall, and between individual pairs of ecoregions, impacting the climatic filter to species invasions.Regional heterogeneity of the change in mean climate match supports the importance of regionally focused analyses(Hulme, 2017).We identify biogeographic regions for prioritized research and management efforts, including climate-integrated risk assessments and invasive species monitoring, where novel technology could improve species detection (e.g.eDNA in freshwater systems; Rees et al., 2014).Monitoring is of particular importance because invasive species whose survival and establishment have previously been limited by climate suitability are likely to be among the first to benefit from climate change.Further analysis of trade connectivity and potential biodiversity loss in the identified regions will be necessary to slow biotic homogenization and meet the goals of the Kunming-Montreal Global Biodiversity Framework (GBF) of the UN Convention of Biodiversity (CBD, 2022).Our results serve F I G U R E 5 Circular plots of the number of new source-recipient pairs of ecoregions to rise above a climate match of 70%, 80% and 90% between the historical climate (1970-2000) and future climate of 2090 (2081-2100) SSP5-8.5 within and among biogeographic realms.Freshwater ecoregions (n = 426) are shown in the left panels; terrestrial ecoregions (n = 817) are shown in the right panels.Pointed ends of lines indicate the recipient ecoregion in each pairing.Source regions nested in biogeographic realms are indicated by their colour in the circular margin.Map of ecoregions by biogeographic realm (bottom), following the colour scheme of the circular plots.Antarctica is not included in freshwater ecoregions and has been removed from the terrestrial ecoregions due to having <5 new matches.
modelled the results of 1687 introduction events of 280 non-indigenous species in 10 countries and found that Sum Climate scores 5-8, expressed as a percentage of the total number climate stations in each country, were predictive of invasion success.However, the study found that a Sum Climate score of 6 indicates suitable habitat and provided the widest distribution of climate match (Bomford et al., 2010)was best for differentiating between regions that have suitable and unsuitable climate within a recipient region(Bomford et al., 2010).We based our climate match values on Sum Climate 6