Stacked Species Distribution Models Research Articles

Mapping multiscale breeding bird species distributions across the United States and evaluating their conservation applications.

Species distribution models are vital to management decisions that require understanding habitat use patterns, particularly for species of conservation concern. However, the production of distribution maps for individual species is often hampered by data scarcity, and existing species maps are rarely spatially validated due to limited occurrence data. Furthermore, community-level maps based on stacked species distribution models lack important community assemblage information (e.g., competitive exclusion) relevant to conservation. Thus, multispecies, guild, or community models are often used in conservation practice instead. To address these limitations, we aimed to generate fine-scale, spatially continuous, nationwide maps for species represented in the North American Breeding Bird Survey (BBS) between 1992 and 2019. We developed ensemble models for each species at three spatial resolutions-0.5, 2.5, and 5 km-across the conterminous United States. We also compared species richness patterns from stacked single-species models with those of 19 functional guilds developed using the same data to assess the similarity between predictions. We successfully modeled 192 bird species at 5-km resolution, 160 species at 2.5-km resolution, and 80 species at 0.5-km resolution. However, the species we could model represent only 28%-56% of species found in the conterminous US BBSs across resolutions owing to data limitations. We found that stacked maps and guild maps generally had high correlations across resolutions (median = 84%), but spatial agreement varied regionally by resolution and was most pronounced between the East and West at the 5-km resolution. The spatial differences between our stacked maps and guild maps illustrate the importance of spatial validation in conservation planning. Overall, our species maps are useful for single-species conservation and can support fine-scale decision-making across the United States and support community-level conservation when used in tandem with guild maps. However, there remain data scarcity issues for many species of conservation concern when using the BBS for single-species models.

Boosting freshwater fish conservation with high-resolution distribution mapping across a large territory.

The lack of high-resolution distribution maps for freshwater species across large extents fundamentally challenges biodiversity conservation worldwide. We devised a simple framework to delineate the distributions of freshwater fishes in a high-resolution drainage map based on stacked species distribution models and expert information. We applied this framework to the entire Chinese freshwater fish fauna (>1600 species) to examine high-resolution biodiversity patterns and reveal potential conflicts between freshwater biodiversity and anthropogenic disturbances. The correlations between spatial patterns of biodiversity facets (species richness, endemicity, and phylogenetic diversity) were all significant (r=0.43-0.98, p<0.001). Areas with high values of different biodiversity facets overlapped with anthropogenic disturbances. Existing protected areas (PAs), covering 22% of China's territory, protected 25-29% of fish habitats, 16-23% of species, and 30-31% of priority conservation areas. Moreover, 6-21% of the species were completely unprotected. These results suggest the need for extending the network of PAs to ensure the conservation of China's freshwater fishes and the goods and services they provide. Specifically, middle to low reaches of large rivers and their associated lakes from northeast to southwest China hosted the most diverse species assemblages and thus should be the target of future expansions of the network of PAs. More generally, our framework, which can be used to draw high-resolution freshwater biodiversity maps combining species occurrence data and expert knowledge on species distribution, provides an efficient way to design PAs regardless of the ecosystem, taxonomic group, or region considered.

Community-level predictions in a megadiverse hotspot: comparison of stacked species distribution models to forest inventory data

AbstractGiven the current scenario of climate change and anthropogenic impacts, spatial predictions of biodiversity are fundamental to support conservation and restoration actions. Here, we compared different stacked species distribution models (S-SDMs) to forest inventories to assess if S-SDMs capture emerging properties and geographic patterns of species richness and composition of local communities in a biodiversity hotspot. We generated SDMs for 1499 tree species sampled in 151 sites across the Atlantic Forest. We applied four model stacking approaches to reconstruct the plant communities: binary SDMs (bS-SDMs), binary SDMs cropped by minimum convex polygons (bS-SDMs-CROP), stacked SDMs constrained by the observed species richness (cS-SDMs) and minimum convex polygons of species occurrences (MCPs). We compared the stacking methods with local communities in terms of species richness, composition, community prediction metrics and components of beta diversity—nestedness and turnover. S-SDMs captured general patterns, with bS-SDMs-CROP being the most parsimonious approach. Species composition differed between local communities and all stacking methods, in which bS-SDMs, bS-SDMs-CROP and MCPs followed a nested pattern, whereas species turnover was most important in cS-SDMs. S-SDMs varied in terms of performance, omission and commission errors, leading to a misprediction of some vulnerable, endangered and critically endangered species. Despite differing from forest inventory data, S-SDMs captured part of the variation from local communities, representing the potential species pool. Our results support the use of S-SDMs to endorse biodiversity synthesis and conservation planning at coarse scales and warn of potential misprediction at local scales in megadiverse regions.

Climate change will redefine taxonomic, functional, and phylogenetic diversity of Odonata in space and time

Climate change is rearranging the mosaic of biodiversity worldwide. These broad-scale species re-distributions affect the structure and composition of communities with a ripple effect on multiple biodiversity facets. Using European Odonata, we asked: i) how climate change will redefine taxonomic, phylogenetic, and functional diversity at European scales; ii) which traits will mediate species’ response to global change; iii) whether this response will be phylogenetically conserved. Using stacked species distribution models, we forecast widespread latitudinal and altitudinal rearrangements in Odonata community composition determining broad turnovers in traits and evolutionary lineages. According to our phylogenetic regression models, only body size and flight period can be partly correlated with observed range shifts. In considering all primary facets of biodiversity, our results support the design of inclusive conservation strategies able to account for the diversity of species, the ecosystem services they provide, and the phylogenetic heritage they carry in a target ecosystem.

Water–energy, climate, and habitat heterogeneity mutually drives spatial pattern of tree species richness in the Indian Western Himalaya

Analyzing plant species richness across a broad geographic gradient is critical for understanding the patterns and processes of biodiversity. In view of this, a species richness map was developed by stacking the ranges of 51 tree species along an elevational gradient in the Western Himalaya using stacked species distribution models (SSDMs). Among modeling algorithms available in SSDMs, random forest and artificial neural networks exhibited the best performance (r = 0.81, p &amp;lt; 0.001). The predicted tree species richness distribution pattern revealed a mid-elevation peak at around 2,000 m asl, which is in concordance with the observed richness pattern (R2 = 0.94, p &amp;lt; 0.001). Additionally, structural equation models (SEMs) were used to confirm the key factors that influence tree richness. The results based on SEMs confirm that the elevational pattern of predicted tree species richness is explained by mutual effects of water–energy availability, climate, and habitat heterogeneity. This study also validates that the impact of moisture on tree species richness coincides geographically with climate factors. The results have revealed that water–energy-related variables are likely to impact the species richness directly at higher elevations, whereas the effect is more likely to be tied to moisture at lower elevations. SSDMs provide a good tool to predict a species richness pattern and could help in the conservation and management of high biodiverse areas at different spatial scales. However, more investigation is needed to validate the SSDMs in other parts of the Himalayan region to provide a comprehensive synoptic perspective of Himalayan biodiversity at a larger scale.

Diversity and conservation of endemic true bugs for four family groups in China

AbstractAimProtected areas (PAs) in China have typically been designed by considering one or several focal taxa (e.g., relict plant and vertebrate species), but the effectiveness of these “protective umbrellas” in safeguarding insects remains unclear. In this study, we aim to investigate the distribution and diversity of endemic true bugs for four family groups (i.e., Miridae, Lygaeoidea, Pentatomidae, and Reduviidae) in China, the environmental factors shaping these diversity patterns, and the effectiveness of China's PAs in safeguarding their diversity.LocationChina.MethodsWe mapped the spatial distributions of 1028 endemic true bugs in China using stacked species distribution models and pixel grids of observed points. The environmental factors that influence diversity patterns across China were explored by generalized linear models and random forest models. The effectiveness of PAs in safeguarding species distributions was evaluated by the Zonation platform.Main resultsWe found that most (84.9%) diversity hotspots for total species were located in mountainous areas of southern China. Additionally, the diversity patterns are shaped by the combined effects of different environmental factors, with the minimum temperature of the coldest month, annual precipitation, and elevation showing the strongest effects. On average, Chinese PAs covered 10.2% of the distribution areas of these endemic true bugs.Main conclusionsIn general, our research identifies the important roles of mountainous areas in southern China in maintaining species diversity. Moreover, our research also suggests that PAs in China could not provide sufficient protection for this diversity. Protecting these endemic true bugs diversity should be valued more in mountainous areas of southern China.

Influence of different data cleaning solutions of point-occurrence records on downstream macroecological diversity models.

Digital point‐occurrence records from the Global Biodiversity Information Facility (GBIF) and other data providers enable a wide range of research in macroecology and biogeography. However, data errors may hamper immediate use. Manual data cleaning is time‐consuming and often unfeasible, given that the databases may contain thousands or millions of records. Automated data cleaning pipelines are therefore of high importance. Taking North American Ephedra as a model, we examined how different data cleaning pipelines (using, e.g., the GBIF web application, and four different R packages) affect downstream species distribution models (SDMs). We also assessed how data differed from expert data. From 13,889 North American Ephedra observations in GBIF, the pipelines removed 31.7% to 62.7% false positives, invalid coordinates, and duplicates, leading to datasets between 9484 (GBIF application) and 5196 records (manual‐guided filtering). The expert data consisted of 704 records, comparable to data from field studies. Although differences in the absolute numbers of records were relatively large, species richness models based on stacked SDMs (S‐SDM) from pipeline and expert data were strongly correlated (mean Pearson's r across the pipelines: .9986, vs. the expert data: .9173). Our results suggest that all R package‐based pipelines reliably identified invalid coordinates. In contrast, the GBIF‐filtered data still contained both spatial and taxonomic errors. Major drawbacks emerge from the fact that no pipeline fully discovered misidentified specimens without the assistance of taxonomic expert knowledge. We conclude that application‐filtered GBIF data will still need additional review to achieve higher spatial data quality. Achieving high‐quality taxonomic data will require extra effort, probably by thoroughly analyzing the data for misidentified taxa, supported by experts.

Improving the predictability and interpretability of co‐occurrence modelling through feature‐based joint species distribution ensembles

Abstract Species Distribution Models (SDMs) are vital tools for predicting species occurrences and are used in many practical tasks including conservation and biodiversity management. However, the expanding minefield of SDM methodologies makes it difficult to select the most reliable method for large co‐occurrence datasets, particularly when time constraints make designing a bespoke model challenging. To facilitate model selection for practical out‐of‐sample prediction, we consider three major challenges: (a) the difficulty of incorporating multiple functional forms for species associations; (b) the limited knowledge on how characteristics of co‐occurrence data impact model performance; and (c) whether individual model predictions could be combined to obtain optimised community predictions without the need for bespoke models. To address these gaps, we propose an ensemble method that uses descriptive features of binary co‐occurrence datasets to predict model weightings for a set of candidate SDMs. We demonstrate how this method may be applied through a simple case study that uses five independent Joint Species Distribution Models (JSDMs) and Stacked Species Distribution Models (SSDMs) to predict out‐of‐sample observations for a diversity of co‐occurrence datasets. Moreover, we introduce a novel SSDM that offers the potential to include multiple functional forms for each species while delivering robust community predictions. Our case study highlights two major findings. First, the ability for the feature‐based ensemble to offer more robust species co‐occurrence predictions compared to other candidate SDMs while providing insights into the data features that impact model performance. Second, the competitiveness of the novel SSDM method for forecasting species co‐occurrences, even when using a simple univariate generalised linear model (GLM) as the base model prior to stacking. We conclude that feature‐based ensembles can provide ecologists with a useful tool for generating species distribution predictions in a way that is reliable and informative. Moreover, the flexibility of the ensemble and the novel SSDM method both offer exciting prospects for incorporating a diversity of functional forms while prioritising out‐of‐sample prediction.

Predicting wildlife corridors for multiple species in an East African ungulate community.

Wildlife corridors are typically designed for single species, yet holistic conservation approaches require corridors suitable for multiple species. Modelling habitat linkages for wildlife is based on several modelling steps (each involving multiple choices), and in the case of multi-species corridors, an approach to optimize single species corridors to few or a single functional corridor for multiple species. To model robust corridors for multiple species and simultaneously evaluate the impact of methodological choices, we develop a multi-method approach to delineate corridors that effectively capture movement of multiple wildlife species, while limiting the area required. Using wildlife presence data collected along ground-based line transects between Lake Manyara and Tarangire National Parks, Tanzania, we assessed species-habitat association in both ensemble and stacked species distribution frameworks and used these to estimate linearly and non-linearly scaled landscape resistances for seven ungulate species. We evaluated habitat suitability and least-cost and circuit theory-based connectivity models for each species individually and generated a multi-species corridor. Our results revealed that species-habitat relationships and subsequent corridors differed across species, but the pattern of predicted landscape connectivity across the study area was similar for all seven species regardless of method (circuit theory or least-cost) and scaling of the habitat suitability-based cost surface (linear or non-linear). Stacked species distribution models were highly correlated with the seven species for all model outputs (r = 0.79 to 0.97), while having the greatest overlap with the individual species least-cost corridors (linear model: 61.6%; non-linear model: 60.2%). Zebra was the best single-species proxy for landscape connectivity. Overall, we show that multi-species corridors based on stacked species distribution models achieve relatively low cumulative costs for savanna ungulates as compared to their respective single-species corridors. Given the challenges and costs involved in acquiring data and parameterizing corridor models for multiple species, zebra may act as a suitable proxy species for ungulate corridor conservation in this system.

Predicting species distributions and community composition using satellite remote sensing predictors

Biodiversity is rapidly changing due to changes in the climate and human related activities; thus, the accurate predictions of species composition and diversity are critical to developing conservation actions and management strategies. In this paper, using satellite remote sensing products as covariates, we constructed stacked species distribution models (S-SDMs) under a Bayesian framework to build next-generation biodiversity models. Model performance of these models was assessed using oak assemblages distributed across the continental United States obtained from the National Ecological Observatory Network (NEON). This study represents an attempt to evaluate the integrated predictions of biodiversity models—including assemblage diversity and composition—obtained by stacking next-generation SDMs. We found that applying constraints to assemblage predictions, such as using the probability ranking rule, does not improve biodiversity prediction models. Furthermore, we found that independent of the stacking procedure (bS-SDM versus pS-SDM versus cS-SDM), these kinds of next-generation biodiversity models do not accurately recover the observed species composition at the plot level or ecological-community scales (NEON plots are 400 m2). However, these models do return reasonable predictions at macroecological scales, i.e., moderately to highly correct assignments of species identities at the scale of NEON sites (mean area ~ 27 km2). Our results provide insights for advancing the accuracy of prediction of assemblage diversity and composition at different spatial scales globally. An important task for future studies is to evaluate the reliability of combining S-SDMs with direct detection of species using image spectroscopy to build a new generation of biodiversity models that accurately predict and monitor ecological assemblages through time and space.

Diversity patterns of palms in Mexico using species distribution models

ABSTRACT The family of palms (Arecaceae) comprises around 2,400 species distributed throughout the world, from which nearly 100 species have been reported to occur in Mexico. Given their importance and the lack of information about their distribution patterns in the country, we applied stacked species distribution models to estimate the current distribution patterns of palms in Mexico. Only 47 species had enough presence records for their modeling. About 50% of the models showed that Annual Precipitation had the greatest contribution to the potential distribution. From the species analyzed, 63% are distributed in the southeast of the country with Roystonea regia as the species with the greatest extent (367,550 km2) and Coccothrinax readii occupying the smaller potential distribution (9,850 km2). It was possible to identify regions of the country with high species richness and where the establishment of new natural protected areas would help to the conservation of palm trees in Mexico. The southeast of Mexico represents the highest richness (>10 species) with about 130,000 km2, and the central slope of the Mexican Pacific, a fragmented landscape with a medium potential distribution (>5 species). Our results represent an important step to guide the establishment of conservation areas for the family Arecaceae in Mexico.

Identifying marine invasion hotspots using stacked species distribution models

Early detection and management of aquatic invasive species requires identification of those areas most at risk of invasion (i.e., hotspots). Here we identify present-day and future hotspots of invasion risk for marine invertebrates and algae in nearshore habitats of the northwest Atlantic and northeast Pacific using more than 12 years of monitoring data in conjunction with other occurrence data and stacked species distribution models. The stacked species distribution models predicted the general patterns of observed invasive species richness in both study areas (Atlantic: r2 = 0.52, Pacific: r2 = 0.42). In the northwest Atlantic, we identified hotspots through much of Massachusetts, New Hampshire and southern Maine, and in several bays in southwestern New Brunswick and Nova Scotia. In the northeast Pacific, much of the southern Salish Sea was identified as a hotspot, as were a few areas along the outer coast of Washington and Oregon. Projecting our species distribution modelling results to 2075 (climate scenario RCP 8.5), we found that existing hotspots are likely to expand slightly in the Atlantic, while in the Pacific existing hotspots are predicted to shift or expand, new hotspots are likely to appear, and areas with few invasive species attaining moderate invasive species richness. Our results suggest that climate change will have larger effects on the distributions of our focal invasive species on the Pacific coast compared to the Atlantic. Resultant hotspot maps provide an integrated perspective and guidance to managers tasked with prioritizing locations for monitoring and implementing policy related to marine invasive species, with projected hotspots making planning for future changes in invasion risk possible.

Using stacked SDMs with accuracy and rarity weighting to optimize surveys for rare plant species

Effective conservation of rare species requires reasonable knowledge of population locations. However, surveys for rare species can be time-intensive and therefore expensive. We test a methodology using stacked species distribution models (S-SDMs) to efficiently discover the greatest number of new rare species’ occurrences possible. We used S-SDMs for 22 rare plant species in southern Ontario, Canada to predict the best survey locations among individual 1-ha cells. For each cell, we weighted distribution model outputs by accuracy and species rarity to create an efficiency value. We used these efficiency values as an index to determine the locations of our field surveys. We conducted field surveys in multi-species cells, “MSC” (areas with high predicted efficiency for multiple species) and single species cells, “SSC” (areas with high probability for only one species) to determine the relative efficiency of a multi-species survey approach. MSC were more than twice as likely as SSC to have at least one rare plant species discovered. Efficiency ranks were also useful in directing surveyors toward incidental discoveries of other rare species that were not modeled. Our technique of using S-SDMs can help direct surveys to more efficiently find rare species occurrences.

Is prediction of species richness from stacked species distribution models biased by habitat saturation?

Abstract 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.

Do hotspots fall within protected areas? A Geographic Approach to Planning analysis of regional freshwater biodiversity

Abstract Identification of biodiversity hot spots is necessary for developing conservation strategies. The south‐eastern U.S.A. harbours high aquatic biodiversity; however, this diversity is threatened as a result of anthropogenic activities. We report patterns of fish and mussel biodiversity, locate hotspots of diversity and endemism, and identify gaps in protection of these hotspots. Maxent was used to generate species distribution models for native fish and mussel species in river basins throughout east Texas. Stacked species distribution models were used to assess spatial patterns of biodiversity and examine the concordance between fish and mussel hotspots. A Geographic Approach to Planning analysis was used to identify areas of high conservation concern. Predicted geographical variation in species richness of mussels corresponds to patterns shown in fish; however, there was discordance among endemicity hotspots between the two taxa. Because overall endemicity was low in both groups, species richness is likely to be the more informative measure of biodiversity at these regional scales. While hotspots represent a small percentage of the study area, almost all richness and endemicity hotspots (&gt;95%) in east Texas occur outside of protected areas. The results of this study suggest that protected areas designated based on terrestrial habitat may fail to protect aquatic biodiversity. Studies such as this provide a baseline for development of conservation and management strategies for protection of regional freshwater biodiversity, while also suggesting that designating conservation lands should include consideration of both terrestrial and aquatic biodiversity.

A comparison of macroecological and stacked species distribution models to predict future global terrestrial vertebrate richness

AbstractAimPredicting future changes in species richness in response to climate change is one of the key challenges in biogeography and conservation ecology. Stacked species distribution models (S‐SDMs) are a commonly used tool to predict current and future species richness. Macroecological models (MEMs), regression models with species richness as response variable, are a less computationally intensive alternative to S‐SDMs. Here, we aim to compare the results of two model types (S‐SDMS and MEMs), for the first time for more than 14,000 species across multiple taxa globally, and to trace the uncertainty in future predictions back to the input data and modelling approach used.LocationGlobal land, excluding Antarctica.TaxonAmphibians, birds and mammals.MethodsWe fitted S‐SDMs and MEMs using a consistent set of bioclimatic variables and model algorithms and conducted species richness predictions under current and future conditions. For the latter, we used four general circulation models (GCMs) under two representative concentration pathways (RCP2.6 and RCP6.0). Predicted species richness was compared between S‐SDMs and MEMs and for current conditions also to extent‐of‐occurrence (EOO) species richness patterns. For future predictions, we quantified the variance in predicted species richness patterns explained by the choice of model type, model algorithm and GCM using hierarchical cluster analysis and variance partitioning.ResultsUnder current conditions, species richness predictions from MEMs and S‐SDMs were strongly correlated with EOO‐based species richness. However, both model types over‐predicted areas with low and under‐predicted areas with high species richness. Outputs from MEMs and S‐SDMs were also highly correlated among each other under current and future conditions. The variance between future predictions was mostly explained by model type.Main conclusionsBoth model types were able to reproduce EOO‐based patterns in global terrestrial vertebrate richness, but produce less collinear predictions of future species richness. Model type by far contributes to most of the variation in the different future species richness predictions, indicating that the two model types should not be used interchangeably. Nevertheless, both model types have their justification, as MEMs can also include species with a restricted range, whereas S‐SDMs are useful for looking at potential species‐specific responses.

Using a species-centered approach to predict bird community responses to habitat fragmentation

The relative importance of habitat fragmentation versus loss on species richness has been much debated. However, recent findings that fragmentation effects are relatively weak may be an artifact of using human-classified vegetation rather than adopting a species-eye view to measure landscape structure. We present the first example of a species-centered approach for examining fragmentation effects on ecological communities. We tested hypotheses relating to the relative influence of habitat amount, configuration, and focal patch size on southwest Oregon bird communities. We used boosted regression trees based on unclassified Landsat TM to create ‘stacked’ species distribution models (S-SDMs) for a large pool of avian species and nested subset of habitat specialists. We tested the relative importance of S-SDM-derived habitat amount, patch number, mean patch size, and focal patch size in explaining species richness. We compared this approach to metrics based on generic land-cover classifications. Species-centered models had greater statistical support than land-cover models. In species-centered models, species richness increased as a function of focal patch size and decreased with patch number, supporting the hypothesis of negative effects of fragmentation per se. Land-cover based models indicated inconsistent support for habitat amount but a positive effect of fragmentation. The species-centered approach identified habitat configuration relationships obscured by land-cover based approaches. While positive land-cover based fragmentation effects were consistent with recent synthesis work, the species-centered approach consistently revealed strong negative effects of fragmentation matching traditional theoretical expectations. S-SDMs may offer promise for generalizing ecological theory to real species distributions.

Disentangling the processes driving tree community assembly in a tropical biodiversity hotspot (New Caledonia)

AbstractAimUnderstanding the drivers of community diversity and composition is a major question in ecology, which is of particular significance for the conservation of narrowly distributed taxa facing habitat alteration. In this study, we examined three ecological processes (environmental and biotic filtering, and dispersal limitation) hypothesized to drive community assembly in the island biodiversity hotspot of New Caledonia, and whether the relative importance of these processes changes along environmental gradients.MethodsFirst, we assessed environmental filtering by stacking binary presence/absence species distribution models generated for 678 tree species from c. 40,000 occurrences. Second, we acknowledged the influence of dispersal limitation by applying a buffer around each occurrence. Third, we modelled a local hierarchy of performances by ranking species in decreasing order of their predicted habitat suitability, up to the predicted local species richness. Predictions of stacked species distribution models (S‐SDMs) accounting for different combinations of the filtering processes were compared with the composition of 12 forest plots.ResultsThe three filters appear to be important driving forces. Accounting for environmental filtering provided fairly accurate assemblage predictions but species richness was overestimated. Considering dispersal limitation increased assemblage specificity (the proportion of correctly predicted absences) while considering differences in local performances increased assemblage sensitivity (the proportion of correctly predicted presences). The predictive ability of all S‐SDMs decreased with elevation, which suggests that unpredictable stochastic processes influenced biodiversity dynamics in more productive mountain habitats.Main conclusionsNew Caledonian tree communities appear to be similar to those of other tropical regions in their structuring processes (important role of dispersal and biotic filtering, increasing role of neutral processes with productivity). Our findings stress the potential of using S‐SDMs to understand and predict current and future biodiversity patterns.

Predicting Optimal Sites for Ecosystem Restoration Using Stacked-Species Distribution Modeling

Habitat restoration is an important tool for managing degraded ecosystems, yet the success of restoration projects depends in part on adequately identifying preferred sites for restoration. Species distribution modeling using a machine learning approach provides novel tools for mapping areas of interest for restoration projects. Here we use stacked-species distribution models (s-SDMs) to identify candidate locations for installment of manmade reefs, a useful management tool for restoring structural habitat complexity and the associated biota in marine ecosystems. We created species distribution models for 21 species of commercial, recreational, ecological, or conservation importance within the Southern California Bight based on observations from long-term reef surveys combined with high resolution (200 x 200m) geospatial environmental data layers. We then combined the individual species models to create a stacked-species habitat suitability map, identifying over 800 km2 of potential area for reef restoration within the Bight. When considering only the 21 focal species, s-SDM scores were positively associated with observed bootstrap species richness not only on natural reefs (linear model: slope = 0.27, 95% CI = 0.17 - 0.36, w = 1), but also this result was supported by two independent test datasets. The predicted richness from this linear model was associated with observed species richness when considering only the focal species on manmade reefs (linear model: slope = 0.52, 95% CI = 0.13 - 0.92, w = 1) and also when considering 204 other non-focal species on both natural and manmade reefs in southern California (slope = 3.65, 95% CI = 2.93 - 4.37, w = 1). Finally, our results demonstrate that the existing manmade reefs included in our study on average are located in regions with habitat suitability that is not only less suitable than natural reefs (t-value = -5.4; p < 0.05) only, but also only slightly significantly better than random (p < 0.05), demonstrating a need for more biologically informed placement of manmade reefs. The stacked-species distribution model provides insight for marine restoration projects in southern California specifically, but more generally this method can also be widely applied to other types of habitat restoration including both marine and terrestrial.

Using macroecological constraints on spatial biodiversity predictions under climate change: the modelling method matters

The prevailing method for estimating the potential impact of future climate change on biological communities is to stack binary predictions from species distribution models (binary stacked species distribution models, bS-SDM). However, it has been argued that bS-SDM may overestimate species richness and, hence, community composition. Alternative approaches, such as SESAM (‘Spatially Explicit Species Assemblage Modelling’), explicitly incorporate limits to species richness, preventing overestimation. We compared richness and taxonomic composition estimates as predicted by SESAM and bS-SDM for Mediterranean bird communities both in the present day and as projected in the future under simulated climate change scenarios. We trained single-species distribution models (S-SDM) and direct macroecological richness models (MEM) for 81 bird species, using climate, topographic, land-use and human-pressure indicators as predictors. Then, we compared and evaluated the models’ predictions. Species richness as predicted by bS-SDM was more accurate than under SESAM for present-day communities. Taxonomic composition was well predicted under both methods. However, we detected significant differences in future projections. Under bS-SDM, increased suitable area for a number of species leads to important changes in community composition and predicts higher levels of diversity in the future. In stark contrast, SESAM predicts lower species richness in the future and strong homogenization of bird communities across space. This study shows how the choice of the modelling approach drives substantially different expectations about future community composition under climate change. We therefore recommend contrasting predictions generated under different modelling approaches to gain better understanding of possible future scenario of biodiversity change. In addition, trasferability tests (i.e. hindcasting to past or predicting from the past to the present) should be used to effectively compare S-SDM and SESAM predictive abilities.