Search

The Northwest Pacific Ocean is the most productive fishing ground in the Pacific Ocean, with a continuous rise in water temperature since 1990. We developed stacked species distribution models (SSDMs) to estimate the impacts of climate change on the distribution dynamics of economically significant species under three climate change scenarios for the periods 2040-2060 and 2080-2100. Overall, water temperature is the most important factor in shaping the distribution patterns of species, followed by water depth. The predictive results indicate that all the species show a northward migration in the future, and the migration distance varies greatly among species. Most pelagic species will expand their habitats under climate change, implying their stronger adaptability than benthic species. Tropical fishes are more adaptable to climate change than species in other climate zones. Though limitations existed, our study provided baseline information for designing a climate-adaptive, dynamic fishery management strategy for maintaining sustainable fisheries. [ABSTRACT FROM AUTHOR]

Freshwater turtles are often used as terrarium pets, especially juveniles of exotic species. At the adult stage they are often released by their owners into the wild despite their high invasion potential. In Europe these thermophilic potentially invasive alien species occupy the habitats of the native European pond turtle Emys orbicularis (Linnaeus, 1758), with new records from the wild being made specifically in Eastern Europe (Latvia and Ukraine) during recent decades. Assessing the potential of alien freshwater turtles to establish in new territories is of great concern for preventing invasion risks while preserving native biodiversity in the present context of climate change. We explored this issue by identifying the present and future (by 2050) suitable habitats of the European pond turtle and several potentially invasive alien species of freshwater turtle already settled in Europe, using a geographic information system (GIS) modelling approach based on datasets from CliMond for climate, Near-global environmental information (NGEI) for freshwater ecosystems (EarthEnv) and Maxent modelling using open-access databases, data from the literature and original field data. Modelling was performed for seven species of alien freshwater turtles occurring from the extreme northern to southern borders of the European range of E. orbicularis: the pond slider Trachemys scripta (Thunberg and Schoepff, 1792), the river cooter Pseudemys concinna (Le Conte, 1830), the Florida red-bellied cooter Pseudemys nelsoni (Carr, 1938), the false map turtle Graptemys pseudogeographica (Gray, 1831), the Chinese softshell turtle Pelodiscus sinensis (Wiegmann, 1835), the Caspian turtle Mauremys caspica (Gmelin, 1774) and the Balkan terrapin Mauremys rivulata (Valenciennes, 1833). In Ukraine, the most Eastern limit of E. orbicularis distribution, were previously reported northern American originated T. scripta, M. rivulata, M. caspica, whereas in Latvia, Emys' most northern limit, were additionally reported P. concinna, P. nelsoni, G. pseudogeographica and Asia originated P. sinensis. The resulting Species Distribution Models (SDM) were of excellent performance (AUC > 0.8). Of these alien species, the most potentially successful in terms of range expansion throughout Europe were T. scripta (34.3% of potential range expansion), G. pseudogeographica (24.1%), and M. caspica (8.9%) and M. rivulata (4.3%) mainly in Eastern Europe, especially in the south of Ukraine (Odesa, Kherson, Zaporizhzhia regions, and Crimean Peninsula). Correlation between the built SDMs for the native E. orbicularis and the invasive alien T. scripta was reliably high, confirming the highly likely competition between these two species in places they cooccur. Moreover, a Multiple Regression Analysis revealed that by 2050, in most of Europe (from the western countries to Ukraine), the territory overlap between E. orbicularis and potentially invasive alien species of freshwater turtles will increase by 1.2 times, confirming higher competition in the future. Importantly, by 2050, Eastern Europe and Ukraine are predicted to be the areas with most suitable habitats for the European pond turtle yet with most limited overlap with the invasive alien species. We conclude that Eastern Europe and Ukraine are the most relevant priority conservation areas for the European pond turtle where it is now necessary to take protective measures to ensure safe habitat for this native species on the long-term. [ABSTRACT FROM AUTHOR]

The Northwest Pacific Ocean is the most productive fishing ground in the Pacific Ocean, with a continuous rise in water temperature since 1990. We developed stacked species distribution models (SSDMs) to estimate the impacts of climate change on the distribution dynamics of economically significant species under three climate change scenarios for the periods 2040-2060 and 2080-2100. Overall, water temperature is the most important factor in shaping the distribution patterns of species, followed by water depth. The predictive results indicate that all the species show a northward migration in the future, and the migration distance varies greatly among species. Most pelagic species will expand their habitats under climate change, implying their stronger adaptability than benthic species. Tropical fishes are more adaptable to climate change than species in other climate zones. Though limitations existed, our study provided baseline information for designing a climate-adaptive, dynamic fishery management strategy for maintaining sustainable fisheries.

Knowledge of the spatial patterns of biodiversity is essential for identifying conservation hotspots. Two commonly used methods of biodiversity assessments are point-to-grid maps and habitat suitability maps generated by stacked species distribution modelling (S-SDM). We compared these two types of maps using three metrics: species richness, weighted endemism (WE), and corrected weighted endemism (CWE) produced for three monocot genera (Allium, Tulipa and Eremurus) in Uzbekistan. Also, we asked how close the patterns of species richness and weighted endemism predicted by S-SDM will be if the sampling density is significantly reduced. To answer this question, we compared the maps of species richness and WE produced using the entire dataset, with one in which species occurrences were thinned to no more than ten occurrences per species. The patterns of species richness and WE in point-to-grid maps and maps generated by S-SDM were very similar for all three genera. However, the spatial distribution of CWE generated by S-SDM was dramatically different from that of point-to-grid maps, and this difference was consistent across the three genera. For species richness and WE, the maps produced by S-SDM from the reduced and complete datasets were similar and close to the analogous point-to-grid maps. These results suggest that even incomplete surveys can provide valuable data for elucidating spatial biodiversity patterns and that for conservationists, S-SDM should be an essential part of their toolbox. However, the use of CWE in S-SDM applications should be avoided. [ABSTRACT FROM AUTHOR]

Measuring species richness at varying spatial extents can be challenging, especially at large extents where exhaustive species surveys are difficult or impossible. Our work aimed at determining the reliability of species richness estimates from stacked ecological niche models at different spatial extents for taxonomic groups with vastly different environmental dependencies and interactions. To accomplish this, we generated ecological niche models for the species of Cactaceae and Pinaceae that occur within 180 published floras from North America north of Mexico. We overlaid or stacked the resulting species' potential distribution estimates over the bounding boxes representing each of the 180 floras to generate predictions of species richness. In general, our stacked models of Cactaceae and Pinaceae were poor predictors of species richness. The relationships between observed and predicted values improved noticeably with the size of spatial extents. However, the stacked models tended to overpredict the richness of Cactaceae and over‐ and underpredict the richness of Pinaceae. Cactaceae stacked models showed higher sensitivity and lower specificity than those for Pinaceae. We conclude that stacked ecological niche models may be somewhat poor predictors of species richness at smaller spatial extents and should be used with caution for this purpose. Perhaps more importantly, abilities to compensate for their limitations or apply corrections to their reliability may vary with taxonomic groups. [ABSTRACT FROM AUTHOR]

Freshwater biodiversity loss is a major concern, and global warming is already playing a significant role in species extinctions. Our main goal was to predict climate change impacts on aquatic insect species distribution and richness in Swiss running waters according to two climate change scenarios (RCP2.6 and RCP8.5), using different modeling approaches, that is, species distribution models (SDMs), stacked‐SDMs (S‐SDMs) and a macroecological model (MEM). We analyzed 10,808 reaches, used as spatial units for model predictions, for a total river network length of 20,610 km. Results were assessed at both the countrywide and the biogeographic regional scales. We used incidence data of 41 species of Ephemeroptera, Plecoptera and Trichoptera (EPT) from 259 sites distributed across Switzerland. We integrated a coupled model for hydrology and glacier retreat to simulate monthly time‐step discharge from which we derived hydrological variables. These, along with thermal, land‐cover, topographic and spatially explicit data, served as predictors for our ecological models. Predictions of occurrence probabilities and EPT richness were compared among the different regions, periods and scenarios. A Shiny web application was developed to interactively explore all the models' details, to ensure transparency and promote the sharing of information. MEM and S‐SDMs approaches consistently showed that overall, climate change is likely to reduce EPT richness. Decrease could be around 10% in the least conservative scenario, depending on the region. Global warming was shown to represent a threat to species from high elevation, but in terms of species richness, running waters from lowlands and medium elevation seemed more vulnerable. Finally, our results suggested that the effects of anthropogenic activities could overweight natural factors in shaping the future of river biodiversity. [ABSTRACT FROM AUTHOR]

Stacked species distribution models (S‐SDM) provide a tool to make spatial predictions about communities by first modelling individual species and then stacking the modelled predictions to form assemblages. The evaluation of the predictive performance is usually based on a comparison of the observed and predicted community properties (e.g. species richness, composition). However, the most available and widely used evaluation metrics require the thresholding of single species' predicted probabilities of occurrence to obtain binary outcomes (i.e. presence/absence). This binarization can introduce unnecessary bias and error.Herein, we present and demonstrate the use of several groups of new or rarely used evaluation approaches and metrics for both species richness and community composition that do not require thresholding but instead directly compare the predicted probabilities of occurrences of species to the presence/absence observations in the assemblages.Community AUC, which is based on traditional AUC, measures the ability of a model to differentiate between species presences or absences at a given site according to their predicted probabilities of occurrence. Summing the probabilities gives the expected species richness and allows the estimation of the probability that the observed species richness is not different from the expected species richness based on the species' probabilities of occurrence. The traditional Sørensen and Jaccard similarity indices (which are based on presences/absences) were adapted to maxSørensen and maxJaccard and to probSørensen and probJaccard (which use probabilities directly). A further approach (improvement over null models) compares the predictions based on S‐SDMs with the expectations from the null models to estimate the improvement in both species richness and composition predictions. Additionally, all metrics can be described against the environmental conditions of sites (e.g. elevation) to highlight the abilities of models to detect the variation in the strength of the community assembly processes in different environments.These metrics offer an unbiased view of the performance of community predictions compared to metrics that requiring thresholding. As such, they allow more straightforward comparisons of model performance among studies (i.e. they are not influenced by any subjective thresholding decisions). [ABSTRACT FROM AUTHOR]

Understanding the importance of climate in determining species distribution and how it might change as a function of spatial grain size is a vital issue for species distribution modeling (SDM), yet it is often not accounted for in models and has not been extensively addressed in under sampled areas in tropical forests. Using extensive field sampled vegetation plots data on species occurrences and current climate conditions we modeled 150 vascular plant species in the Okavango River Basin, to map their current projected suitable climate space at 2, 5, 10, 20, and 50 km2 pixel resolution. Relationships between the variable importance scores and variable identity and their interaction with predictor spatial grain were investigated using Generalized Linear Models and post-hoc analysis. We found variation in the relative influence of temperature and precipitation variables across the spatial grains. The importance of the determinants of species distribution may change between species but such changes are less determined by the predictor's spatial grain. Potential evapotranspiration consistently exhibited the greatest influence in determining species and richness distribution across fine to coarse spatial grains. We found that the spatial grain of predictors had no effect on the model predictive power and that varying predictor spatial grains had only negligible effects on the model performance measured by AUC and Kappa statistics. The spatial grains of the climatic predictors used showed no effect on species richness pattern either. Our results indicate that in areas with relatively low topographic variation, modeling at coarse spatial grain for conservation purposes can be acceptable. Moreover, we show that in tropical areas that have comparatively homogeneous climatic conditions along large spatial extents the variable importance is not influenced by predictor spatial grain. For projections of contemporary species suitable climate space in relatively flat and topographically homogeneous areas which often have a climatically homogenous landscape, more attention must be given to the identity of the selected predictor variables for modeling the species distributions than to their spatial grain size. We suggest that in species distribution modeling for conservation planning, assessment of the input datasets spatial grain should be informed and guided by knowledge of the landscape level topographic conditions, as protocol.

Issue: Approaches to predicting species assemblages through stacking individual niche‐based species distribution models (S‐SDMs) need to account for community processes other than abiotic filtering. Such constraints have been introduced by implementing ecological assembly rules (EARs) into S‐SDMs, and can be based on patterns of functional traits in communities. Despite being logically valid, this approach has led to a limited improvement in prediction, possibly because of mismatches between the scales of measurement of niche and trait data. Evidence: S‐SDM studies have often related single values of a species' traits to environmental niches that are captured by abiotic conditions measured at a much finer spatial scale, without accounting for intraspecific trait variation along environmental gradients. Many pieces of evidence show that omitting intraspecific trait variation can hinder the proper inference of EARs from trait patterns, and we further argue that it can therefore also affect our capacity to spatially predict functional properties of communities. In addition, estimates of environmental niches and trait envelopes may vary depending on the scale at which environmental and trait measurements are made. Conclusion: We suggest that to overcome these limitations, surveys sampling both niche and trait measurements should be conducted at fine scales along wide environmental gradients, and integrated at the same scale to test and improve a new generation of spatial community models and their functional properties. [ABSTRACT FROM AUTHOR]

Yunnan's biodiversity is under considerable pressure and subtropical evergreen broad-leaved forests in this area have become increasingly fragmented through agriculture, logging, planting of economic plants, mining activities and changing environment. The aims of the study are to investigate climate change-induced changes of subtropical evergreen broad-leaved forests in Yunnan and identify areas of current species richness centers for conservation preparation. Stacked species distribution models were created to generate ensemble forecasting of species distributions, alpha diversity and beta diversity for Yunnan's subtropical evergreen broad-leaved forests in both current and future climate scenarios. Under stacked species distribution models in rapid climate changes scenarios, changes of water-energy dynamics may possibly reduce beta diversity and increase alpha diversity. This point provides insight for future conservation of evergreen broad-leaved forest in Yunnan, highlighting the need to fully consider the problem of vegetation homogenization caused by transformation of water-energy dynamics.

Abstract Correlative species distribution models (SDMs) are widely used to predict species distributions and assemblages, with many fundamental and applied uses. Different factors were shown to affect SDM prediction accuracy. However, real data cannot give unambiguous answers on these issues, and for this reason, artificial data have been increasingly used in recent years. Here, we move one step further by assessing how different factors can affect the prediction accuracy of virtual assemblages obtained by stacking individual SDM predictions (stacked SDMs, S-SDM). We modelled 100 virtual species in a real study area, testing five different factors: sample size (200-800-3200), sampling method (nested, non-nested), sampling prevalence (25%, 50%, 75% and species true prevalence), modelling technique (GAM, GLM, BRT and RF) and thresholding method (ROC, MaxTSS, and MaxKappa). We showed that the accuracy of S-SDM predictions is mostly affected by modelling technique followed by sample size. Models fitted by GAM/GLM had a higher accuracy and lower variance than BRT/RF. Model accuracy increased with sample size and a sampling strategy reflecting the true prevalence of the species was most successful. However, even with sample sizes as high as >3000 sites, residual uncertainty remained in the predictions, potentially reflecting a bias introduced by creating and/or resampling the virtual species. Therefore, when evaluating the accuracy of predictions from S-SDMs fitted with real field data, one can hardly expect reaching perfect accuracy, and reasonably high values of similarity or predictive success can already be seen as valuable predictions. We recommend the use of a 'plot-like' sampling method (best approximation of the species' true prevalence) and not simply increasing the number of presences-absences of species. As presented here, virtual simulations might be used more systematically in future studies to inform about the best accuracy level that one could expect given the characteristics of the data and the methods used to fit and stack SDMs. Highlights • Virtual species were used to test factors affecting spatial assemblage predictions. • Modelling technique and sample size proved to be the most important factors. • Sampling procedures caused uncertainty in predictions even under a "known" truth. • A 'plot-like' sampling method ensures accurate predictions even at small samples. • Confirmed the importance of evaluating relative effects of factors affecting SDMs. [ABSTRACT FROM AUTHOR]

In this study, we compare two community modelling approaches to determine their ability to predict the taxonomic, functional and phylogenetic properties of plant assemblages along a broad elevation gradient and at a fine resolution. The first method is the standard stacking individual species distribution modelling ( SSDM) approach, which applies a simple environmental filter to predict species assemblages. The second method couples the SSDM and macroecological modelling ( MEM- SSDM- MEM) approaches to impose a limit on the number of species co-occurring at each site. Because the detection of diversity patterns can be influenced by different levels of phylogenetic or functional trees, we also examine whether performing our analyses from broad to more exact structures in the trees influences the performance of the two modelling approaches when calculating diversity indices., We found that coupling the SSDM with the MEM improves the overall predictions for the three diversity facets compared with those of the SSDM alone. The accuracy of the SSDM predictions for the diversity indices varied greatly along the elevation gradient, and when considering broad to more exact structure in the functional and phylogenetic trees, the SSDM- MEM predictions were more stable., SSDM- MEM moderately but significantly improved the prediction of taxonomic diversity, which was mainly driven by the corrected number of predicted species. The performance of both modelling frameworks increased when predicting the functional and phylogenetic diversity indices. In particular, fair predictions of the taxonomic composition by SSDM- MEM led to increasingly accurate predictions of the functional and phylogenetic indices, suggesting that the compositional errors were associated with species that were functionally or phylogenetically close to the correct ones; however, this did not always hold for the SSDM predictions., Synthesis. In this study, we tested the use of a recently published approach that couples species distribution and macroecological models to provide the first predictions of the distribution of multiple facets of plant diversity: taxonomic, functional and phylogenetic. Moderate but significant improvements were obtained; thus, our results open promising avenues for improving our ability to predict the different facets of biodiversity in space and time across broad environmental gradients when functional and phylogenetic information is available. [ABSTRACT FROM AUTHOR]

Predicting spatial distribution of flowering forage availability is critical for guiding migratory beekeeping decisions. Species distribution modelling (SDM) is widely used to predict the geographic distribution or species ranges. Stacked distributions of multiple species (S-SDM) have been used in predicting species richness or assemblages. Here, we present a method for stacking SDMs based on a temporal element, the flowering phenology of melliferous flora species. First, we used presence-only data for thirty key forage species used for honey production in Western Australia, combined with environmental variables for predicting the geographic distribution of species, using MaxEnt software. The output distribution grids were then stacked based on monthly flowering times of each species to develop grids representing the richness of flowering species by grid cell. While designed for modelling flowering forage availability for a migratory beekeeping system, the approach can be used for predicting temporal forage availability for a range of different fauna that rely on melliferous flora. •How to use temporally stacked species distribution modelling for generic distribution of flowering availability using presence-only data.•A procedure for developing flowering richness and availability grids., Competing Interests: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (© 2023 The Authors. Published by Elsevier B.V.)

Aim We assess main ecological determinants affecting the comparative performance of macroecological models ( MEMs) that model species richness directly, and stacked species distribution models based on stacking probabilities ( pS- SDMs) and binary predictions ( bS- SDMs). Specifically, we aimed to understand how statistical effects such as prevalence and environmental heterogeneity are entangled with species' ecology in Swiss avian assemblages. Location Switzerland. Methods We tested for statistical and ecological effects on overprediction and underprediction by regressing species richness residuals against community-averaged values of prevalence, functional traits and functional dissimilarity. Further, we defined bird functional groups through hierarchical clustering and compared accuracy of species richness predictions between groups to understand the differences between model types and ecological determinants thereof. Last, we tested how accuracy of species assemblages constructed from bS- SDMs relates to species' functional characteristics. Results Underprediction of high diversity sites by pS- SDMs and MEMs was mainly explained by prevalence, whereas overprediction of low diversity sites was strongly affected by diet and habitat traits, and increased with functional dissimilarity. Model performances varied strongly between functional groups with more accurate and less biased predictions for generalist species groups. Critically, overprediction bias in richness predictions by bS- SDMs was uncorrelated with assemblage prediction success. Main conclusions The reliability of all community models tested here strongly depended on functional species' characteristics related mainly to diet, foraging and breeding habitat. This underlines the need to incorporate all relevant and species-specific or functional group-specific ecological filters in the models. Improved prediction accuracy of species richness will require finer-resolved environmental predictors that better describe available niche space and account for specific spatial and resource requirements of different species. More research is needed to understand the relationship between accuracy of species richness and species assemblage predictions in bS- SDMs as well as the role of biotic interactions. [ABSTRACT FROM AUTHOR]

Aim The aim of this study was to test different modelling approaches, including a new framework, for predicting the spatial distribution of richness and composition of two insect groups. Location The western Swiss Alps. Methods We compared two community modelling approaches: the classical method of stacking binary prediction obtained from individual species distribution models (binary stacked species distribution models, bS-SDMs), and various implementations of a recent framework (spatially explicit species assemblage modelling, SESAM) based on four steps that integrate the different drivers of the assembly process in a unique modelling procedure. We used: (1) five methods to create bS-SDM predictions; (2) two approaches for predicting species richness, by summing individual SDM probabilities or by modelling the number of species (i.e. richness) directly; and (3) five different biotic rules based either on ranking probabilities from SDMs or on community co-occurrence patterns. Combining these various options resulted in 47 implementations for each taxon. Results Species richness of the two taxonomic groups was predicted with good accuracy overall, and in most cases bS-SDM did not produce a biased prediction exceeding the actual number of species in each unit. In the prediction of community composition bS-SDM often also yielded the best evaluation score. In the case of poor performance of bS-SDM (i.e. when bS-SDM overestimated the prediction of richness) the SESAM framework improved predictions of species composition. Main conclusions Our results differed from previous findings using community-level models. First, we show that overprediction of richness by bS-SDM is not a general rule, thus highlighting the relevance of producing good individual SDMs to capture the ecological filters that are important for the assembly process. Second, we confirm the potential of SESAM when richness is overpredicted by bS-SDM; limiting the number of species for each unit and applying biotic rules (here using the ranking of SDM probabilities) can improve predictions of species composition. [ABSTRACT FROM AUTHOR]

Aim Species distribution models ( SDMs) are common tools in biogeography and conservation ecology. It has been repeatedly claimed that aggregated (stacked) SDMs ( S- SDMs) will overestimate species richness. One recently suggested solution to this problem is to use macroecological models of species richness to constrain S- SDMs. Here, we examine current practice in the development of S- SDMs to identify methodological problems, provide tools to overcome these issues, and quantify the performance of correctly stacked S- SDMs alongside macroecological models. Locations Barents Sea, Europe and Dutch Wadden Sea. Methods We present formal mathematical arguments demonstrating how S- SDMs should and should not be stacked. We then compare the performance of macroecological models and correctly stacked S- SDMs on the same data to determine if the former can be used to constrain the latter. Next, we develop a maximum-likelihood approach to adjusting S- SDMs and discuss how it could potentially be used in combination with macroecological models. Finally, we use this tool to quantify how S- SDMs deviate from observed richness in four very different case studies. Results We demonstrate that stacking methods based on thresholding site-level occurrence probabilities will almost always be biased, and that these biases will tend toward systematic overprediction of richness. Next, we show that correctly stacked S- SDMs perform very similarly to macroecological models in that they both have a tendency to overpredict richness in species-poor sites and underpredict it in species-rich sites. Main conclusions Our results suggest that the perception that S- SDMs consistently overpredict richness is driven largely by incorrect stacking methods. With these biases removed, S- SDMs perform similarly to macroecological models, suggesting that combining the two model classes will not offer much improvement. However, if situations where coupling S- SDMs and macroecological models would be beneficial are subsequently identified, the tools we develop would facilitate such a synthesis. [ABSTRACT FROM AUTHOR]

• Habitat saturation impacts predictions from Species Distribution Models (SDM) • Habitat saturation biases stacked SDMs (S-SDMs) predictions. • Probability-based richness predicts local SR without bias whatever habitat saturation. • Comparing different S-SDMs predictions can shed light on community assembly processes. Several studies have proposed to predict Species Richness (SR) by combining the predictions of independent Species Distributions Models (SDMs) (the predict first-assemble later strategy). Alternative methods propose to combine outputs from SDMs differently, by either summing predicted presence probabilities at each location, or summing binary presence predictions after thresholding the probabilities. Species can occupy various proportions of their suitable habitats (i.e, have various levels of habitat saturation), which can cause discrepancy when predicting their presences through SDMs. Furthermore, these discrepancies can be increased when combining the predictions of individual SDMs to predict SR. In this article, we performed simulations of species distributions with varying habitat saturation (i.e., the amount of suitable habitat occupied by a species), and we compared observed richness with that predicted by the alternative approaches. We found that probability-based richness is not biased by the level of habitat saturation, while threshold-based richness over-predicts richness at low habitat saturation and under-predicts it as high habitat saturation. Probability-based richness should thus be used in priority when predicting species richness locally. Nonetheless, threshold-based richness represents species richness constrained by environmental filtering only and thus is a useful indicator of potential species richness when species fully saturate their habitats. Thus the systematic comparison of probability-based and threshold-based richness predictions can reveal the importance of habitat saturation and can thus help identify community assembly mechanisms at play. [ABSTRACT FROM AUTHOR]

Yunnan's biodiversity is under considerable pressure and subtropical evergreen broad-leaved forests in this area have become increasingly fragmented through agriculture, logging, planting of economic plants, mining activities and changing environment. The aims of the study are to investigate climate change-induced changes of subtropical evergreen broad-leaved forests in Yunnan and identify areas of current species richness centers for conservation preparation. Stacked species distribution models were created to generate ensemble forecasting of species distributions, alpha diversity and beta diversity for Yunnan's subtropical evergreen broad-leaved forests in both current and future climate scenarios. Under stacked species distribution models in rapid climate changes scenarios, changes of water-energy dynamics may possibly reduce beta diversity and increase alpha diversity. This point provides insight for future conservation of evergreen broad-leaved forest in Yunnan, highlighting the need to fully consider the problem of vegetation homogenization caused by transformation of water-energy dynamics.