Correlative Species Distribution Models Research Articles

Is there always space at the top? Ensemble modeling reveals climate-driven high-altitude squeeze for the vulnerable snow trout Schizothorax richardsonii in Himalaya

Mountain systems throughout the globe are conspicuously sensitive to on-going climate alterations. This condition is much more detrimental in the Himalaya, where the rate of warming and thus the glacier meltdown is much higher than elsewhere. The Himalayan coldwater species are concerningly most vulnerable to these changes because of their limited thermal range. Whilst climate studies strongly prognosticate the altered distribution of plants and mammals in this region, the impact on coldwater fishes still remains unknown. We used snow trout (Schizothorax richardsonii), a Himalayan coldwater specialist as a model organism to predict the current suitability and climate-driven potential range shift in an ensemble-based modeling framework. We collated occurrence data from a long term, extensive field-based sampling with additional records derived from an in-depth literature survey. A comprehensive input data set including topographic, hydrogeomorphic and climatic variables were used to build correlative species distribution models for the current and future time-periods (2050 and 2070) under three-carbon emission scenarios (Representative Concentration Pathways) by integrating five General Circulation Models. Furthermore, we predicted the elevation-specific range shifts of the snow trout in response to climate change. We predict that a wide-ranging mid-elevation river network is currently suitable for the snow trout in Himalaya, however a significant part of its current distributional range would be lost over time. Our results highlight that snow trout would expand their range upwards into the high-altitude streams with a concurrent predominant range contraction in most of their lagging edges, ultimately creating a high-altitude squeeze. The net habitat loss under three RCP scenarios (RCPs 2.6, 4.5 and 8.5) was estimated to range from 7.41% to 16.29% for the year 2050 which would further increase in the year 2070 ranging from 9.46% to 26.56%. These results provide a strategic information on prioritizing climate-adaptive actions to target the currently suitable habitats and future refugia identified. Our modeling framework also provides a foremost basis to conserve not only the snow trout but also to evaluate the climate impact on several other coldwater species, which are equally vulnerable and ecologically important in the Himalaya.

Predictability of Species Distributions Deteriorates Under Novel Environmental Conditions in the California Current System

Spatial distributions of marine fauna are determined by complex interactions between environmental conditions and animal behaviors. As climate change leads to warmer, more acidic, and less oxygenated oceans, species are shifting away from their historical distribution ranges, and these trends are expected to continue into the future. Correlative Species Distribution Models (SDMs) can be used to project future habitat extent for marine species, with many different statistical methods available. However, it is vital to assess how different statistical methods behave under novel environmental conditions before using these models for management advice, and to consider whether future projections based on these techniques are biologically reasonable. In this study, we built SDMs for adults and larvae of two ecologically important pelagic fishes in the California Current System: Pacific sardine (Sardinops sagax) and northern anchovy (Engraulis mordax). We used five different SDM methods, ranging from simple (thermal niche model) to complex (artificial neural networks). Our results show that some SDMs trained on data collected between 2003 and 2013 lost substantial predictive skill when applied to observations from more recent years, when ocean temperatures associated with a marine heatwave were outside the range of historical measurements. This decrease in skill was particularly apparent for adult sardine, which showed non-stationary relationships between catch locations and sea surface temperature through time. While sardine adults and larvae shifted their distributions markedly during the marine heatwave, anchovy largely maintained their historical spatiotemporal distributions. Our results suggest that correlative relationships between species and their environment can become unreliable during anomalous conditions. Understanding the underlying physiology of marine species is therefore essential for the construction of SDMs that are robust to rapidly changing environments. Developing distribution models that offer skillful predictions into the future for species such as sardine and anchovy, which are migratory and include separate sub-stocks, may be particularly challenging.

Competition and demography rather than dispersal limitation slow down upward shifts of trees’ upper elevation limits in the Alps

Abstract Species range limits are expected to be dramatically altered under future climate change and many species are predicted to shift their distribution upslope to track their suitable conditions (i.e. based on their niche). However, there might be large discrepancies between the speed of the upward shift of the climatic niche and the actual migration velocity of the species, especially in long‐lived organisms such as trees. In fact, most studies did not find any significant upward shift of the distributional limits of temperate forest trees over the last decades. It therefore beckons the questions why trees are moving upslope much slower than their bioclimatic envelope and what are the implications for ecosystem functioning. Here, we compared the simulations of the upslope displacement of the bioclimatic envelope of 16 tree species inhabiting temperate mountain forests under ongoing and future climate change obtained by correlative species distribution models (SDMs) to those from a dynamic forest model accounting for dispersal, competition and demography. We then partitioned the discrepancy in upslope migration velocity between the SDMs and the dynamic forest model into different components by manipulating dispersal limitation, interspecific competition and demography. Tree species in the dynamic forest model migrated only slowly upslope in contrast to the SDMs. Most of the difference in migration velocity can directly be attributed to tree's demography (long life cycle), followed by effects of competition and only a marginal contribution of dispersal limitation. Additionally, lower elevation species (‘non‐treeline’) shifted slower upslope than high elevation species (‘treeline’) indicating a strong effect of interspecific competition at their leading edge. Synthesis. Forests have a high inertia to climate change because of their longevity and ability to acclimatize to high climatic fluctuations. Lower elevation tree species (deciduous) only slowly establish in stands at higher elevation where coniferous species dominate and likely profit from facilitation by disturbance events. Therefore, forest ecosystems seem to persist, even if climate becomes unfavourable, until they approach a tipping point at which an extreme event (e.g. drought, storm or insect attack) leads to a large dieback and resource change enabling new suitable species to spread and establish.

Clearing up the Crystal Ball: Understanding Uncertainty in Future Climate Suitability Projections for Amphibians

Amphibian populations are threatened globally, and one of the hypotheses for these declines is climate change. Species distribution models are frequently used to predict changes in suitable habitat as a result of changing climates; however, these projections can be heavily influenced by choice of modeling approach. To evaluate global predictions for amphibians, we conducted a literature review of studies that utilize correlative species distribution models to project changes in climate suitability under various climate change scenarios. We paid particular attention to the use of model selection in choosing among candidate species distribution models (SDMs) so as to control for overparameterization of SDMs. In addition, we conducted a case study with three species of Slender Salamanders (Batrachoseps) to further investigate the impact of differences in modeling decisions on projected amphibian climate suitability. We found 83 studies (including the present case study) in which projections of future climate suitability were made for amphibian species. Of those studies, 36 included estimates of percent change in climate suitability and thus were included in our meta-analysis. These studies included projections for over 1000 species or species complexes, with the majority being Anurans (86%), and encompassed five continents with the most representation in South America, Europe, and North America. Across these studies, average projected change in climate suitability ranged from －70% to 167%, and these projected changes varied with dispersal assumptions representative concentration pathway (RCP) used, the projection year, and taxonomic order. Only three of the 36 studies reported the use of Akaike information criterion (AIC)-based model selection to choose a best-fit SDM. However, our case study demonstrated that predicted change in climate suitability varied whether the best-fit or default SDM was used for projections. Further, this result varied among species (ΔAIC = 0), suggesting that the impact of overparameterization differs across species. Our results illustrate that there is a pressing need to project climate suitability across more species and more geographic regions. In addition, we may need to revisit projections for previously investigated species to evaluate additional climate scenarios and whether overparameterization may have influenced projections. Our ability to accurately model future changes in climate suitability will be essential for successful conservation and management plans for amphibians.

Can seedlings' physiological information improve vegetation distribution predictions at local scales?

Physiological information has been successfully included in marine Species Distribution Models (SDM) before, but few have considered a previous development stage that could have affected the present-day distribution of the species at local scales. The aim of this study is to analyze the inclusion of physiological information of seedling survival on a correlative SDM based on adult present-day presences. The species were the invasive shrub Baccharis halimifolia and the native saltmarsh Juncus maritimus. For each species, five SDM were established using different approaches: using only experimentally derived physiological data, a correlative model with environmental predictors, additive combinations of presence/absence maps derived from the previous models, and a correlative model with the physiological data as a predictor variable. For B. halimifolia, the inclusion of the physiological data as a predictor variable yielded better results than with the other approaches; with J maritimus, this inclusion achieved an accuracy as high as the model with only environmental variables as predictors. The additive combinations generated less accurate models but offered possible advantages in future specialized studies. The results for B. halimifolia could extrapolate to other invasive species that rely on spreading high amounts of individuals and are more vulnerable in their early stages than in their growing and adult phases. Thus, this approach can improve the capacity for mapping invasive species’ distributions at local scales, and the conservation efforts to control biological invasions in estuaries and coastal ecosystems.

Future range dynamics of the red alga Capreolia implexa in native and invaded regions: contrasting predictions from species distribution models versus physiological knowledge

Correlative species distribution modelling is a widely used method to predict potential species ranges but can suffer from limitations in integrating species’ fundamental niches. Therefore, they might underestimate suitable ranges, but including physiological information can improve accuracy of predictions and help identify mechanisms of e.g. range limitation. However, approaches using both, results from correlative as well as physiological investigations are rare, especially in research on seaweeds. Here, we provide results from both approaches to predict the suitable habitat range of Capreolia implexa (Rhodophyta) in its native range (Australia and New Zealand) and invaded range (Chile) under present and future climate scenarios (year 2100, rcp 2.6 and rcp 8.5). We used the Maxent modelling technique and physiological knowledge from a temperature tolerance experiment (2–20 °C) for thermal niche estimation. Results from both approaches suggest larger suitable habitat ranges under present day conditions for both regions than currently occupied. Abiotic range limitation in the native range led to underestimation of the suitable temperature range by Maxent (here lower temperature limit = 8.3 °C). Predictions based on the laboratory temperature tolerance experiment suggest additional suitable habitat in colder regions (here lower temperature limit = 6.6 ± 0.4 °C). Under future climate conditions, both native and invaded ranges should shift southward, which will lead to an overall loss of suitable habitat in the native range. Like that, rcp 8.5 conditions should reduce the native range to 50% of the present-day extent. We demonstrate the limitation of correlative SDM modelling for species that live on continental margins and that physiological experiments can help to identify species’ niches beyond correlative analyses, providing valuable information for range projections. Furthermore, we provide valuable insights relevant for both invasion management and conservation.Graphic abstract

Combining correlative and mechanistic niche models with human activity data to elucidate the invasive potential of a sub‐Antarctic insect

AbstractAimCorrelative species distribution models (SDMs) are subject to substantial spatio‐temporal limitations when historical occurrence records of data‐poor species provide incomplete and outdated information for niche modelling. Complementary mechanistic modelling techniques can, therefore, offer a valuable contribution to underpin more physiologically informed predictions of biological invasions, the risk of which is often exacerbated by climate change. In this study we integrate physiological and human pressure data to address the uncertainties and limitations of correlative SDMs and to better understand, predict and manage biological invasions.LocationWestern archipelagos of the Southern Ocean and martime Antarctica.TaxonEretmoptera murphyi (Chironomidae), invertebrates.MethodsMahalanobis Distances were used for correlative SDM construction for a species with few records. A mechanistic SDM was built around different fitness components (larval survival and life stage progression) as a function of temperature. SDM predictions were combined with human activity levels in Antarctica to generate a site vulnerability index to the assess colonization risk of E. murphyi. Future scenarios of ecophysiological suitability were built around the warming trends in the region.ResultsBoth SDMs converge to predict high environmental suitability in the species' native and introduced ranges. However, the mechanistic model indicates a slightly larger invasive potential based on larval performance at different temperatures. Human activity levels across the Antarctic Peninsula play a key role in discerning site vulnerabilities. Niche suitability in Antarctica grows considerably under long‐term climate scenarios, leading to a substantially higher invasive threat to the Antarctic ecosystems. In turn changing conditions result in growing physiological mismatches with the environment in the native range in South Georgia.Main conclusionsLong‐term studies of invasion potential under climate benefit from integrating correlative predictions with physiological experiments, as the invasion potential varies depending on the area and the timescale examined. This study also highlights a conservation paradox whereby the accidental introduction of an insect represents a threat to the Antarctic ecoystems that contrasts with its endangered status at the native range.

Assessing the exposure of forest habitat types to projected climate change-Implications for Bavarian protected areas.

AimDue to their longevity and structure, forest ecosystems are particularly affected by climate change with consequences for their biodiversity, functioning, and services to mankind. In the European Union (EU), natural and seminatural forests are protected by the Habitats Directive and the Natura 2000 network. This study aimed to assess the exposure of three legally defined forest habitat types to climate change, namely (a) Tilio‐Acerion forests of slopes, screes, and ravines (9180*), (b) bog woodlands (91D0*), and (c) alluvial forests with Alnus glutinosa and Fraxinus excelsior (91E0*). We analyzed possible changes in their Bavarian distribution, including their potential future coverage by Natura 2000 sites. We hypothesized that protected areas (PAs) with larger elevational ranges will remain suitable for the forests as they allow for altitudinal distribution shifts.MethodsTo estimate changes in range size and coverage by PAs, we combined correlative species distribution models (SDMs) with spatial analyses. Ensembles of SDM‐algorithms were applied to two climate change scenarios (RCP4.5 and RCP8.5) of the HadGEM2‐ES model for the period 2061–2080.ResultsOur results revealed that bog woodlands experience the highest range losses (>2/3) and lowest PA coverage (max. 15% of sites with suitable conditions). Tilio‐Acerion forests exhibit opposing trends depending on the scenario, while alluvial forests are less exposed to climatic changes. As expected, the impacts of climate change are more pronounced under the “business as usual” scenario (RCP8.5). Additionally, PAs in flat landscapes are more likely to lose environmental suitability for currently established forest habitat types.Main conclusionsBased on these findings, we advocate the expansion of the Natura 2000 network particularly in consideration of elevational gradients, connectivity, and projected climatic suitability. Nonclimatic stressors on forest ecosystems, especially bog woodlands, should be decreased and climate change mitigation efforts enhanced. We recommend transferring the approach to other habitat types and regions.

Correlative climatic niche models predict real and virtual species distributions equally well.

Climate is one of the main factors driving species distributions and global biodiversity patterns. Obtaining accurate predictions of species' range shifts in response to ongoing climate change has thus become a key issue in ecology and conservation. Correlative species distribution models (cSDMs) have become a prominent tool to this aim in the last decade and have demonstrated good predictive abilities with current conditions, irrespective of the studied taxon. However, cSDMs rely on statistical association between species' presence and environmental conditions and have rarely been challenged on their actual capacity to reflect causal relationships between species and climate. In this study, we question whether cSDMs can accurately identify if climate and species distributions are causally linked, a prerequisite for accurate prediction of range shift in relation to climate change. We compared the performance of cSDMs in predicting the distributions of 132 European terrestrial species, chosen randomly within five taxonomic groups (three vertebrate groups and two plant groups), and of 1,320 virtual species whose distribution is causally fully independent from climate. We found that (1) for real species, the performance of cSDMs varied principally with range size, rather than with taxonomic groups and (2) cSDMs did not predict the distributions of real species with a greater accuracy than the virtual ones. Our results unambiguously show that the high predictive power of cSDMs can be driven by spatial autocorrelation in climatic and distributional data and does not necessarily reflect causal relationships between climate and species distributions. Thus, high predictive performance of cSDMs does not ensure that they accurately depict the role of climate in shaping species distributions. Our findings therefore call for strong caution when using cSDMs to provide predictions on future range shifts in response to climate change.

Incorporating intraspecific variation into species distribution models improves distribution predictions, but cannot predict species traits for a wide‐spread plant species

The most common approach to predicting how species ranges and ecological functions will shift with climate change is to construct correlative species distribution models (SDMs). These models use a species’ climatic distribution to determine currently suitable areas for the species and project its potential distribution under future climate scenarios. A core, rarely tested, assumption of SDMs is that all populations will respond equivalently to climate. Few studies have examined this assumption, and those that have rarely dissect the reasons for intraspecific differences. Focusing on the arctic‐alpine cushion plant Silene acaulis, we compared predictive accuracy from SDMs constructed using the species’ full global distribution with composite predictions from separate SDMs constructed using subpopulations defined either by genetic or habitat differences. This is one of the first studies to compare multiple ways of constructing intraspecific‐level SDMs with a species‐level SDM. We also examine the contested relationship between relative probability of occurrence and species performance or ecological function, testing if SDM output can predict individual performance (plant size) and biotic interactions (facilitation). We found that both genetic‐ and habitat‐informed SDMs are considerably more accurate than a species‐level SDM, and that the genetic model substantially differs from and outperforms the habitat model. While SDMs have been used to infer population performance and possibly even biotic interactions, in our system these relationships were extremely weak. Our results indicate that individual subpopulations may respond differently to climate, although we discuss and explore several alternative explanations for the superior performance of intraspecific‐level SDMs. We emphasize the need to carefully examine how to best define intraspecific‐level SDMs as well as how potential genetic, environmental, or sampling variation within species ranges can critically affect SDM predictions. We urge caution in inferring population performance or biotic interactions from SDM predictions, as these often‐assumed relationships are not supported in our study.

Genes on the edge: A framework to detect genetic diversity imperiled by climate change.

Ongoing global warming is disrupting several ecological and evolutionary processes, spanning different levels of biological organization. Species are expected to shift their ranges as a response to climate change, with relevant implications to peripheral populations at the trailing and leading edges. Several studies have analyzed the exposure of species to climate change but few have explored exposure at the intraspecific level. We introduce a framework to forecast exposure to climate change at the intraspecific level. We build on existing methods by combining correlative species distribution models, a model of species range dynamics, and a model of phylogeographic interpolation. We demonstrate the framework by applying it to 20 Iberian amphibian and reptile species. Our aims were to: (a) identify which species and intraspecific lineages will be most exposed to future climate change; (b) test if nucleotide diversity at the edges of species ranges are significantly higher or lower than on the overall range; and (c) analyze if areas of higher species gain, loss, and turnover coincide with those predicted for lineages richness and nucleotide diversity. We found that about 80% of the studied species are predicted to contract their range. Within each species, some lineages were predicted to contract their range, while others were predicted to maintain or expand it. Therefore, estimating the impacts of climate change at the species level only can underestimate losses at the intraspecific level. Some species had significant high amount of nucleotide at the trailing or leading edge, or both, but we did not find a consistent pattern across species. Spatial patterns of species richness, gain, loss, and turnover were fairly concurrent with lineages richness and nucleotide diversity. Our results support the need for increased attention to intraspecific diversity regarding monitoring and conservation strategies under climate change.

Climate‐driven shifts in the distribution of koala‐browse species from the Last Interglacial to the near future

The koala's Phascolarctos cinereus distribution is currently restricted to eastern and south‐eastern Australia. However, fossil records dating from 70 ± 4 ka (ka = 103 yr) from south‐western Australia and the Nullarbor Plain are evidence of subpopulation extinctions in the southwest at least after the Last Interglacial (~128–116 ka). We hypothesize that koala sub‐population extinctions resulted from the eastward retraction of the koala's main browse species in response to unsuitable climatic conditions. We further posit a general reduction in the distribution of main koala‐browse trees in the near future in response climate change. We modelled 60 koala‐browse species and constructed a set of correlative species distribution models for five time periods: Last Interglacial (~128–116 ka), Last Glacial Maximum (~23–19 ka), Mid‐Holocene (~7–5 ka), present (interpolations of observed data, representative of 1960–1990), and 2070. We based our projections on five hindcasts and one forecast of climatic variables extracted from WorldClim based on two general circulation models (considering the most pessimistic scenario of high greenhouse‐gas emissions) and topsoil clay fraction. We used 17 dates of koala fossil specimens identified as reliable from 70 (± 4) to 535 (± 49) ka, with the last appearance of koalas at 70 ka in the southwest. The main simulated koala‐browse species were at their greatest modelled extent of suitability during the Last Glacial Maximum, with the greatest loss of koala habitat occurring between the Mid‐Holocene and the present. We predict a similar habitat loss between the present and 2070. The spatial patterns of habitat change support our hypothesis that koala extinctions in the southwest, Nullarbor Plain and central South Australia resulted from the eastward retraction of the dominant koala‐browse species in response to long‐term climate changes. Future climate patterns will likely increase the extinction risk of koalas in their remaining eastern ranges.

Testing species assemblage predictions from stacked and joint species distribution models

AbstractAimPredicting the spatial distribution of species assemblages remains an important challenge in biogeography. Recently, it has been proposed to extend correlative species distribution models (SDMs) by taking into account (a) covariance between species occurrences in so‐called joint species distribution models (JSDMs) and (b) ecological assembly rules within the SESAM (spatially explicit species assemblage modelling) framework. Yet, little guidance exists on how these approaches could be combined. We, thus, aim to compare the accuracy of assemblage predictions derived from stacked and from joint SDMs.LocationSwitzerland.TaxonBirds, tree species.MethodsBased on two monitoring schemes (national forest inventory and Swiss breeding bird atlas), we built SDMs and JSDMs for tree species (at 100 m resolution) and forest birds (at 1 km resolution). We tested accuracy of species assemblage and richness predictions on holdout data using different stacking procedures and ecological assembly rules.ResultsDespite minor differences, results were consistent between birds and tree species. Cross‐validated species‐level model performance was generally higher in SDMs than JSDMs. Differences in species richness and assemblage predictions were larger between stacking procedures and ecological assembly rules than between stacked SDMs and JSDMs. On average, predictions were slightly better for stacked SDMs compared to JSDMs, probabilistic stacks outperformed binary stacks, and ecological assembly rules yielded best predictions.Main conclusionsWhen predicting the composition of species assemblages, the choice of stacking procedure and ecological assembly rule seems more decisive than differences in underlying model type (SDM vs. JSDM). JSDMs do not seem to improve community predictions compared to SDMs or improve predictions for rare species. Still, JSDMs may provide additional insights into community assembly and may help deriving hypotheses about prevailing biotic interactions in the system. We provide simple rules of thumb for choosing appropriate modelling pathways. Future studies should test these preliminary guidelines for other taxa and biogeographical realms as well as for other JSDM algorithms.

Beyond the Model: Expert Knowledge Improves Predictions of Species’ Fates under Climate Change

The need to proactively manage landscapes and species to aid their adaptation to climate change is widely acknowledged. Current approaches to prioritizing investment in species conservation generally rely on correlative models, which predict the likely fate of species under different climate change scenarios. Yet, while model statistics can be improved by refining modeling techniques, gaps remain in understanding the relationship between model performance and ecological reality. To investigate this, we compared standard correlative species distribution models to highly accurate, fine-scale, distribution models. We critically assessed the ecological realism of each species' model, using expert knowledge of the geography and habitat in the study area and the biology of the study species. Using interactive software and an iterative vetting with experts, we identified seven general principles that explain why the distribution modeling under- or overestimated habitat suitability, under both current and predicted future climates. Importantly, we found that, while temperature estimates can be dramatically improved through better climate downscaling, many models still inaccurately reflected moisture availability. Furthermore, the correlative models did not account for biotic factors, such as disease or competitor species, and were unable to account for the likely presence of micro refugia. Under-performing current models resulted in widely divergent future projections of species' distributions. Expert vetting identified regions that were likely to contain micro refugia, even where the fine-scale future projections of species distributions predicted population losses. Based on the results, we identify four priority conservation actions required for more effective climate change adaptation responses. This approach to improving the ecological realism of correlative models to understand climate change impacts on species can be applied broadly to improve the evidence base underpinning management responses.

Multi-Scenario Species Distribution Modeling.

Correlative species distribution models (SDMs) are increasingly being used to predict suitable insect habitats. There is also much criticism of prediction discrepancies among different SDMs for the same species and the lack of effective communication about SDM prediction uncertainty. In this paper, we undertook a factorial study to investigate the effects of various modeling components (species-training-datasets, predictor variables, dimension-reduction methods, and model types) on the accuracy of SDM predictions, with the aim of identifying sources of discrepancy and uncertainty. We found that model type was the major factor causing variation in species-distribution predictions among the various modeling components tested. We also found that different combinations of modeling components could significantly increase or decrease the performance of a model. This result indicated the importance of keeping modeling components constant for comparing a given SDM result. With all modeling components, constant, machine-learning models seem to outperform other model types. We also found that, on average, the Hierarchical Non-Linear Principal Components Analysis dimension-reduction method improved model performance more than other methods tested. We also found that the widely used confusion-matrix-based model-performance indices such as the area under the receiving operating characteristic curve (AUC), sensitivity, and Kappa do not necessarily help select the best model from a set of models if variation in performance is not large. To conclude, model result discrepancies do not necessarily suggest lack of robustness in correlative modeling as they can also occur due to inappropriate selection of modeling components. In addition, more research on model performance evaluation is required for developing robust and sensitive model evaluation methods. Undertaking multi-scenario species-distribution modeling, where possible, is likely to mitigate errors arising from inappropriate modeling components selection, and provide end users with better information on the resulting model prediction uncertainty.

Species distribution modelling: Bombina bombina (Linnaeus, 1761) and its main invasive threat Perccottus glenii (Dybowski, 1877) in Latvia under global climate change

Amphibians have been greatly affected by invasive species through direct predation and competition. The invasion in Eurasia of Perccottus glenii (the Chinese sleeper) triggered by climate change may have detrimental consequences for the Fire-bellied toad (Bombina bombina) because the invader can feed upon B. bombina eggs, toads and juveniles. We used correlative species distribution models (SDMs) to infer competitive interaction between the invader and native species and found a significant level of correlation (r=0.636) between both SDMs, meaning there is a considerable overlap of areas of similar habitat suitability related to the species and a potential for competition. For strategical management purposes we mapped areas in Latvia where there are lesser chances for negative interactions between the both species, and where unwanted contact and/or competition between the invader and the toad can be avoided. DOI: http://dx.doi.org/10.5755/j01.erem.74.4.21093

Beyond the model: expert knowledge improves predictions of species' fates under climate change.

The need to proactively manage landscapes and species to aid their adaptation to climate change is widely acknowledged. Current approaches to prioritizing investment in species conservation generally rely on correlative models, which predict the likely fate of species under different climate change scenarios. Yet, while model statistics can be improved by refining modeling techniques, gaps remain in understanding the relationship between model performance and ecological reality. To investigate this, we compared standard correlative species distribution models to highly accurate, fine-scale, distribution models. We critically assessed the ecological realism of each species' model, using expert knowledge of the geography and habitat in the study area and the biology of the study species. Using interactive software and an iterative vetting with experts, we identified seven general principles that explain why the distribution modeling under- or overestimated habitat suitability, under both current and predicted future climates. Importantly, we found that, while temperature estimates can be dramatically improved through better climate downscaling, many models still inaccurately reflected moisture availability. Furthermore, the correlative models did not account for biotic factors, such as disease or competitor species, and were unable to account for the likely presence of micro refugia. Under-performing current models resulted in widely divergent future projections of species' distributions. Expert vetting identified regions that were likely to contain micro refugia, even where the fine-scale future projections of species distributions predicted population losses. Based on the results, we identify four priority conservation actions required for more effective climate change adaptation responses. This approach to improving the ecological realism of correlative models to understand climate change impacts on species can be applied broadly to improve the evidence base underpinning management responses.

A comparison of joint species distribution models for presence–absence data

Abstract Joint species distribution models (JSDMs) account for biotic interactions and missing environmental predictors in correlative species distribution models. Several different JSDMs have been proposed in the literature, but the use of different or conflicting nomenclature and statistical notation potentially obscures similarities and differences among them. Furthermore, new JSDM implementations have been illustrated with different case studies, preventing direct comparisons of computational and statistical performance. We aim to resolve these outstanding issues by (a) highlighting similarities among seven presence–absence JSDMs using a clearly defined, singular notation; and (b) evaluating the computational and statistical performance of each JSDM using six datasets that vary widely in numbers of sites, species, and environmental covariates considered. Our singular notation shows that many of the JSDMs are very similar, and in turn parameter estimates of different JSDMs are moderate to strongly, positively correlated. In contrast, the different JSDMs clearly differ in computational efficiency and memory limitations. Our framework will allow ecologists to make educated decisions about the JSDM that best suits their objective, and enable wider uptake of JSDM methods among the ecological community.

Integrated modelling of Atlantic mackerel distribution patterns and movements: A template for dynamic impact assessments

Modelling is important for impact assessments of anthropogenic pressures on wildlife. Models are particularly useful when dealing with complex dynamic systems (as pelagic ecosystems) where data are limited and if various ‘what if’ scenarios should be tested. The aim of this study was to produce and implement an integrated modelling approach, linking high resolution hydrodynamic models (HDM) of the marine environment with correlative species distribution models (SDM) and agent-based models (ABM), for describing the spatio-temporal distribution and movements of Atlantic mackerel (Scomber scombrus) in the Norwegian Sea. The SDM was fitted with scientific mackerel trawl data as response variables (collected in July and August 2006–2014) and temperature (from the HDM), water depth and time period as predictors of spatial distributions. The SDM was able to produce dynamic predictions of a similar order of magnitude as observed catch per unit effort (CPUE) as well as realistic large-scale distribution patterns, when tested on independent data (not included in the modelling). The ABM was calibrated, with normalized SDM predictions (habitat suitability as a proxy for food availability) and hydrodynamics as input and simulated on a single year (2013) for the period May–October, when the migratory mackerel is present in the study area. A pattern-oriented modelling (POM) approach was used to verify if the model reproduced multiple observed real-world patterns. The ABM produced similar patterns as observed regarding migration timing, growth and large scale geographic distribution. Fine scaled information on mackerel movement and behaviour is limited, which is also reflected in the results. More data and knowledge are therefore required to improve the patterns emerging from fine scaled processes. The potential of the model for assessing an impact of a single seismic survey (mimicking a real survey) was finally evaluated. The exercise allowed estimating the number of affected fish (within 50 km from the sound source) and potential changes in local migrations, with the specific assumed minimum sound pressure thresholds (resulting in a fleeing reaction by the mackerel) set to 165 dB re 1 μPa. The model framework was shown to be useful by allowing simulations of impact scenarios in a realistic and dynamic environment. The model can be further updated when data on fine scale movements of mackerel and most importantly when improved data on response behaviour to impacts of sound become available.

How much should one sample to accurately predict the distribution of species assemblages? A virtual community approach

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.