Species Distribution Models Studies Research Articles

Cross‐validation matters in species distribution models: a case study with goatfish species

In an era of ongoing biodiversity, it is critical to map biodiversity patterns in space and time for better‐informing conservation and management. Species distribution models (SDMs) are widely applied in various types of such biodiversity assessments. Cross‐validation represents a prevalent approach to assess the discrimination capacity of a target SDM algorithm and determine its optimal parameters. Several alternative cross‐validation methods exist; however, the influence of choosing a specific cross‐validation method on SDM performance and predictions remains unresolved. Here, we tested the performance of random versus spatial cross‐validation methods for SDM using goatfishes (Actinopteri: Syngnathiformes: Mullidae) as a case study, which are recognized as indicator species for coastal waters. Our results showed that the random versus spatial cross‐validation methods resulted in different optimal model parameterizations in 57 out of 60 modeled species. Significant difference existed in predictive performance between the random and spatial cross‐validation methods, and the two cross‐validation methods yielded different projected present‐day spatial distribution and future projection patterns of goatfishes under climate change exposure. Despite the disparity in species distributions, both approaches consistently suggested the Indo‐Australian Archipelago as the hotspot of goatfish species richness and also as the most vulnerable area to climate change. Our findings highlight that the choice of cross‐validation method is an overlooked source of uncertainty in SDM studies. Meanwhile, the consistency in richness predictions highlights the usefulness of SDMs in marine conservation. These findings emphasize that we should pay special attention to the selection of cross‐validation methods in SDM studies.

Incorporating physiological knowledge into correlative species distribution models minimizes bias introduced by the choice of calibration area

Correlative species distribution models (SDMs) are important tools to estimate species’ geographic distribution across space and time, but their reliability heavily relies on the availability and quality of occurrence data. Estimations can be biased when occurrences do not fully represent the environmental requirement of a species. We tested to what extent species’ physiological knowledge might influence SDM estimations. Focusing on the Japanese sea cucumber Apostichopus japonicus within the coastal ocean of East Asia, we compiled a comprehensive dataset of occurrence records. We then explored the importance of incorporating physiological knowledge into SDMs by calibrating two types of correlative SDMs: a naïve model that solely depends on environmental correlates, and a physiologically informed model that further incorporates physiological information as priors. We further tested the models’ sensitivity to calibration area choices by fitting them with different buffered areas around known presences. Compared with naïve models, the physiologically informed models successfully captured the negative influence of high temperature on A. japonicus and were less sensitive to the choice of calibration area. The naïve models resulted in more optimistic prediction of the changes of potential distributions under climate change (i.e., larger range expansion and less contraction) than the physiologically informed models. Our findings highlight benefits from incorporating physiological information into correlative SDMs, namely mitigating the uncertainties associated with the choice of calibration area. Given these promising features, we encourage future SDM studies to consider species physiological information where available.

Travelling away from home? Joining global change and recovery scenarios to anticipate the marine distribution of diadromous fish

Species Distribution Models (SDM) are useful tools providing results that can be extrapolated to anticipate species range shifts, under climate change scenarios. SDM studies integrating spatial constraints are significantly lacking in the marine environment, leading to optimistic predictions. This is particularly true for anadromous species in which marine distributions can be driven by their affinity to their natal rivers. Acipenser sturio is a critically endangered anadromous fish for which two stocked populations are currently maintained in the Gironde-Garonne-Dordogne (France) and Elbe (Germany) river systems. Benefiting from bycatch reports of A. sturio, we applied a SDM process that explicitly considers distance to home when evaluating habitat suitability. More precisely, we included the variable ‘distance to mouth of the natal river system’ into SDM inputs to test and characterize its influence on the marine distribution of A. sturio. We used this model to obtain the marine distribution under current climatic conditions with the two source populations and under population recovery scenarios (functional populations hypothesized to exist in ten currently unoccupied river systems). We projected the model under future conditions with two climatic scenarios (RCP 4.5 and 8.5) and three time slices over the 2023–2099 period. Constrained-ranges of both existing and hypothetical populations are projected to expand in the future. We observed an overall increase of habitat suitability, with new suitable sectors localized further from natal river mouths. By informing on the suitable marine surface that each hypothetical population holds and adds to the existing ones, our approach aims at informing about the feasibility of species recovery and marine habitats protection strategies. Our findings highlight the need for including dispersal information in marine SDM. The application of our dispersal-constrained approach may be considered for other less-well-known species for which dispersal point sources are identifiable, such as other diadromous species in different study areas.

Lowland forest loss and climate-only species distribution models exaggerate a forest-dependent species' vulnerability to climate change

Climate-only species distribution models are commonly used to evaluate a species' vulnerability to future climate change. However, these models may overestimate a species' vulnerability when occurrence data comes from only a subset of the species' historical range due to anthropogenic range contractions. This study investigated the impacts of using climate-only species distribution models with anthropogenically truncated occurrence data on future predictions, using the little-studied leopard cat (Prionailurus bengalensis) of mainland Tropical Asia as a case study. Like most other species in mainland Tropical Asia, the leopard cat's current distribution represents a non-random subset of its historical range due to the disproportionate loss of lowland forests in the region. We found that climate-only species distribution models trained on current occurrence data significantly overestimated the leopard cat's vulnerability to future climate change, predicting considerable range contractions by 2081–2100 due to rising summer temperatures. A similar range contraction was observed using a virtual species whose occurrence was solely predicated on extant forest cover and not climate. This suggested the leopard cat's apparent vulnerability to future climate change was driven by a spurious relationship with summer temperatures, as modern occurrence data primarily came from refugial forest cover in the region's cooler uplands rather than the warmer, human-dominated lowlands. This bias was not apparent, however, from our current predictions, which had high predictive accuracy and were spatially consistent with current knowledge of the species' distribution; an observation highlighting the inadequacy of predictive accuracy alone as a metric for selecting models in studies where transferability to novel environments is critical. We next considered two potential strategies for detecting and mitigating the effects of anthropogenically truncated occurrence data: 1) the use of hybrid models that incorporate both climatic and non-climatic variables, and 2) the use of climate-only models with historical datasets. Despite using different data, both proposed strategies produced spatially similar predictions of present and future leopard cat occurrence with little-to-no range contractions predicted. We thus recommend all species distribution modeling studies interested in assessing a species' vulnerability to climate change compare future predictions derived from climate-only and hybrid models as well as compare predictions from current and historical datasets, as differences in predictions can highlight the presence of potential biases and other limitations. Finally, we discuss our findings as they apply to both our focal species and to the overall field of species distribution modeling.

Amplifying the Power of eDNA by Making it FAIR

Environmental DNA (eDNA) is a fast-growing biomonitoring approach to detect species and map their distributions, with the number of eDNA publications exponentially increasing in the past decade. While millions of DNA sequences are often generated and assigned to taxa in each publication, these records are stored in numerous locations (e.g., supplementary materials at journals’ servers, open data publishing platforms such as Dryad) and in various formats, which makes it difficult to find, access, re-use and integrate datasets. Making eDNA data FAIR (findable, accessible, interoperable, re-usable) has vast potential to improve how the biological environment is measured and how change is detected and understood. For instance, it would allow biomonitoring and species distribution modelling studies across extended space and time scales, which is logistically difficult or impossible for individual projects. It would also shed light on “dark” (unassigned) DNA sequences by facilitating their storage and re-analyses with updated ever-growing DNA reference databases. Several challenges are associated with making eDNA FAIR, including how to standardise data formats and bioinformatics workflows, and simplifying the process of post-publication data archiving so that it is acceptable for eDNA practitioners to adopt. Over the next three years, we plan to work closely with biodiversity data platforms such as the Global Biodiversity Information Facility (GBIF) and Atlas of Living Austrlia (ALA), eDNA science journals, and eDNA practitioners, to solve these challenges and enable eDNA to achieve its revolutionary potential as a unified source of information that supports environmental management.

Does adding community science observations to museum records improve distribution modeling of a rare endemic plant?

AbstractUnderstanding the ranges of rare and endangered species is central to conserving biodiversity in the Anthropocene. Species distribution models (SDMs) have become a common and powerful tool for analyzing species–environment relationships across geographic space. Although evaluating the distribution of rare species is integral to their conservation, this can be difficult when limited distribution data are available. Community science platforms, such as iNaturalist, have emerged as alternative sources for species occurrence data. Although these observations are often thought to be of lower quality than those of natural history collections, they may have potential for improving SDMs for species with few occurrence records from collections. Here, we investigate the utility of iNaturalist data for developing SDMs for a rare high‐elevation plant, Telesonix jamesii. Because methods for modeling rare species are limited in the literature, five different modeling techniques were considered, including profile methods, statistical models, and machine learning algorithms. The inclusion of iNaturalist data doubled the number of usable records for T. jamesii. We found that a random forest (RF) model using ensemble training data performed the highest of any model (area under curve = 0.98). We then compared the performance of RF models that use only natural history training data and those that use a combination of natural history (herbarium specimens) and iNaturalist training data. All models heavily relied on climate data (mean temperature of driest quarter, and precipitation of the warmest quarter), indicating that this species is under threat as climate continues to change. Validation datasets affected model fits as well. Models using only herbarium data performed slightly poorer when evaluated with cross‐validation than when validated externally with iNaturalist data. This study can serve as a model for future SDM studies of species with similar data limitations.

Climate extreme variables generated using monthly time‐series data improve predicted distributions of plant species

Extreme weather can have significant impacts on plant species demography; however, most studies have focused on responses to a single or small number of extreme events. Long‐term patterns in climate extremes, and how they have shaped contemporary distributions, have rarely been considered or tested. BIOCLIM variables that are commonly used in correlative species distribution modelling studies cannot be used to quantify climate extremes, as they are generated using long‐term averages and therefore do not describe year‐to‐year, temporal variability. We evaluated the response of 37 plant species to base climate (long‐term means, equivalent to BIOCLIM variables), variability (standard deviations) and extremes of varying return intervals (defined using quantiles) based on historical observations. These variables were generated using fine‐grain (approx. 250 m), time‐series temperature and precipitation data for the hottest, coldest and driest months over 39 years. Extremes provided significant additive improvements in model performance compared to base climate alone and were more consistent than variability across all species. Models that included extremes frequently showed notably different mapped predictions relative to those using base climate alone, despite often small differences in statistical performance as measured as a summary across sites. These differences in spatial patterns were most pronounced at the predicted range margins, and reflect the influence of coastal proximity, continentality, topography and orographic barriers on climate extremes. Species occupying hotter and drier locations that are exposed to severe maximum temperature extremes were associated with better predictive performance when modelled using extremes. Understanding how plant species have historically responded to climate extremes may provide valuable insights into our understanding of contemporary distributions and help to make more accurate predictions under a changing climate.

Validating Species Distribution Models With Standardized Surveys for Ixodid Ticks in Mainland Florida.

Tick-borne pathogens are of growing concern. The U.S. Centers for Disease Control and Prevention (CDC) developed guidelines standardizing surveys of tick vectors to better monitor the changes in their occurrences. Unbiased surveillance data, from standardized surveys, are presumed critical to generate valid species distribution models (SDMs). We tested previously generated SDMs from standardized protocols for three medically important ticks [Amblyomma americanum (Linnaeus, Ixodida, Ixodidae), Ixodes scapularis (Say, Ixodida, Ixodidae), and Dermacentor variabilis (Say, Ixodida, Ixodidae)]. These previous models ruled out a quarter to half of the state as having these species, with consensus occurrence in about a quarter of the state. New surveys performed throughout 2019 on 250 transects at 43 sites indicated the rule-out functions were 100% accurate for I. scapularis and D. variabilis and 91.9% for A. americanum. As SDM concordance increased, the proportion of transects yielding ticks increased. Independent surveys of SDMs provide external validation—an aspect missing from many SDM studies.

Toward reliable habitat suitability and accessibility models in an era of multiple environmental stressors.

Global biodiversity declines, largely driven by climate and land‐use changes, urge the development of transparent guidelines for effective conservation strategies. Species distribution modeling (SDM) is a widely used approach for predicting potential shifts in species distributions, which can in turn support ecological conservation where environmental change is expected to impact population and community dynamics. Improvements in SDM accuracy through incorporating intra‐ and interspecific processes have boosted the SDM field forward, but simultaneously urge harmonizing the vast array of SDM approaches into an overarching, widely adoptable, and scientifically justified SDM framework. In this review, we first discuss how climate warming and land‐use change interact to govern population dynamics and species’ distributions, depending on species’ dispersal and evolutionary abilities. We particularly emphasize that both land‐use and climate change can reduce the accessibility to suitable habitat for many species, rendering the ability of species to colonize new habitat and to exchange genetic variation a crucial yet poorly implemented component of SDM. We then unite existing methodological SDM practices that aim to increase model accuracy through accounting for multiple global change stressors, dispersal, or evolution, while shifting our focus to model feasibility. We finally propose a roadmap harmonizing model accuracy and feasibility, applicable to both common and rare species, particularly those with poor dispersal abilities. This roadmap (a) paves the way for an overarching SDM framework allowing comparison and synthesis of different SDM studies and (b) could advance SDM to a level that allows systematic integration of SDM outcomes into effective conservation plans.

Evaluating the capacity of species distribution modeling to predict the geographic distribution of the mangrove community in Mexico.

Mangroves are highly productive ecosystems that provide important environmental services, but have been impacted massively in recent years by human activities. Studies of mangroves have focused on their ecology and function at local or landscape scales, but little has been done to understand their broader distributional patterns or the environmental factors that determine those distributions. Species distribution models (SDMs), have been used to estimate potential distributions of hundreds of species, yet no SDM studies to date have assessed mangrove community distributions in Mexico (the country with the fourth largest extent of this ecosystem). We used maximum entropy approaches to model environmental suitability for mangrove species distributions in the country, and to identify the environmental factors most important in determining those distributions. We also evaluated whether this modeling approach is adequate to estimate mangrove distribution as a community across Mexico. Best models were selected based on statistical significance (AUC ratio), predictive performance (omission error of 5%), and model complexity (Akaike criterion); after this evaluation, only one model per species met the three evaluation criteria. Environmental variable sets that included distance to coast yielded significantly better models; variables with strongest contributions included elevation, temperature of the coldest month, and organic carbon content of soil. Based on our results, we conclude that SDMs can be used to map mangrove communities in Mexico, but that results can be improved at local scales with inclusion of local variables (salinity, hydroperiod and microtopography), field validations, and remote sensing data.

Testing whether ensemble modelling is advantageous for maximising predictive performance of species distribution models

Predictive performance is important to many applications of species distribution models (SDMs). The SDM ‘ensemble’ approach, which combines predictions across different modelling methods, is believed to improve predictive performance, and is used in many recent SDM studies. Here, we aim to compare the predictive performance of ensemble species distribution models to that of individual models, using a large presence–absence dataset of eucalypt tree species. To test model performance, we divided our dataset into calibration and evaluation folds using two spatial blocking strategies (checkerboard‐pattern and latitudinal slicing). We calibrated and cross‐validated all models within the calibration folds, using both repeated random division of data (a common approach) and spatial blocking. Ensembles were built using the software package ‘biomod2’, with standard (‘untuned’) settings. Boosted regression tree (BRT) models were also fitted to the same data, tuned according to published procedures. We then used evaluation folds to compare ensembles against both their component untuned individual models, and against the BRTs. We used area under the receiver‐operating characteristic curve (AUC) and log‐likelihood for assessing model performance. In all our tests, ensemble models performed well, but not consistently better than their component untuned individual models or tuned BRTs across all tests. Moreover, choosing untuned individual models with best cross‐validation performance also yielded good external performance, with blocked cross‐validation proving better suited for this choice, in this study, than repeated random cross‐validation. The latitudinal slice test was only possible for four species; this showed some individual models, and particularly the tuned one, performing better than ensembles. This study shows no particular benefit to using ensembles over individual tuned models. It also suggests that further robust testing of performance is required for situations where models are used to predict to distant places or environments.

Species distribution models can be highly sensitive to algorithm configuration

In pursuit of a more robust provenance in the field of species distribution modelling, an extensive literature search was undertaken to find the typical default values, and the range of values, for configuration settings of a large number of the most commonly used statistical algorithms available for constructing species distribution models (SDM), as implemented in the R script packages (such as Dismo and Biomod2) or other species distribution modelling programs like MaxEnt. We found that documentation of SDM algorithm configuration option settings in the SDM literature is, overall, very uncommon, and the justifications for these settings were minimal, when present. Such settings were often the R default values, or were the result of trial and error. This is potentially concerning since: (i) it detracts from the robustness of the provenance for such SDM studies; (ii) a lack of documentation of configuration option settings in a paper prevents the replication of an experiment, which contravenes one of the main tenets of the scientific method; (iii) inappropriate or uninformed configuration option settings are particularly concerning if they represent a poorly understood ecological variable or process, and if the algorithm is sensitive to such settings, this could result in erroneous and/or unrealistic SDMs. Therefore, this study sets out to comprehensively test the sensitivity of eight widely used SDM algorithms to variation in configuration options settings: MaxEnt, Artificial Neural Network (ANN), Generalized Linear Model (GLM), Generalized Additive Model (GAM), Multivariate Adaptive Regression Splines (MARS), Flexible Discriminant Analysis (FDA), Surface Range Envelope (SRE) and Classification tree analysis (CTA). A process of expert elicitation was used to derive a range of appropriate values with which to test the sensitivity of our algorithms. We chose to use species occurrence records for two species - Koala (Phascolartos cinereus) and Thorny Devil (Moloch horridus) - in order to investigate how algorithm sensitivity depends on the species being modelled. Results were assessed by comparing the modelled distribution of the control SDM (default settings) to the modelled distribution from each sensitivity test SDM (i.e. non-default configuration settings). This was done using the visual and statistical measures of predictive performance available in the Biodiversity and Climate Change Virtual Laboratory (BCCVL), including the area under the (receiver operating characteristic) curve. The aim of our study was to be able to draw conclusions as to how the sensitivity of SDM algorithms to their configuration option settings may detract from the reliability of SDM results, given the often unjustified and unscrutinized use of the default settings, and generally infrequent and largely perfunctory attendance to this issue in most of the published SDM literature. Our results indicate that all of the algorithms tested showed sensitivity to alternative (non-default) values for some of their configuration settings and that often this sensitivity is species-dependent. Therefore we can conclude that the choice of configuration settings in these widely used SDM algorithms can have a large impact on the resulting projected distribution. This has important ramifications for decision-making and policy outcomes wherever SDMs are used to inform species and biodiversity management plans and policy settings. This study demonstrates that assigning suitable values for these settings is a very important consideration and as such should always be published along with the model. Documenting all configuration settings is necessary to increase the scientific robustness, transparency and reproducibility of species distribution modelling studies.

Using Unsupervised Artificial Neural Networks to Detect Sibling Species: A case in Myxomycetes

Species distribution modelling (SDM) --i.e. the prediction of species potential geographic distributions based on correlations between known presence records and the environmental conditions at occurrence localities-- is one of the most freqently cited developments in recent years in the realm of biodiversity studies (Web of Science 2019). The reasons for the explosion in SDM studies reside in: a) the pressure to know species distributions (under both present and future climate change scenarios) with precision to satisfy scientific as well as societal objectives; b) the impossibility of knowing every occurrence of all but the most conspicuous and/or special interest species; c) the availability of primary occurrence and environmental data in unprecedented amounts, thanks to initiatives like the Global Biodiversity Information Facility (GBIF); and d) the development of algorithms and software, along with the computing power, which allow inference of species distribution models from the available data. The standard methods to produce such models are based on environmental feature vectors, and some well-established algorithms such as distance-based machine learning, regression or a combination of these (Tsoar et al. 2007). In this presentation, we explore deep learning techniques (LeCun et al. 2015), particularly how those developed in recent years could contribute to the study of species distribution (see also Botella et al. 2018; Deneu et al. 2018). In this contribution we aim to identify sibling species on the basis of their ecological preferences. In this exercise, we prepare an image-based environmental representation space using an unsupervised classification approach, instead of a set of environmental feature vectors. For our case study, we chose a well-known cosmopolitan myxomycete (i.e., mycetozoan) species: Hemitrichia serpula (Scop.) Rostaf., whose ecological preferences --involving several biomes-- may suggest that it could comprise several sibling species.

Improving Models of Species Ecological Niches: A Remote Sensing Overview

Effective conservation capable of mitigating global biodiversity declines require thorough knowledge on species distributions and their drivers. A species ecological niche determines its geographic distribution, and species distribution models (SDMs) can be used to predict them. For various reasons, e.g., the lack of spatial data on relevant environmental factors, SDMs fail to characterize important ecological relationships. We argue that SDMs do not yet include relevant environmental information, which can be measured with remote sensing (RS). RS may benefit SDMs because it provides information on e.g., ecosystem function, health and structure, complete spatial assessment, and reasonable temporal repeat for the processes that determine geographical distributions. However, RS data is still seldom included in such studies with the exception of climate data. Here we provide a guide for researchers aiming to improve their SDM studies, describing how they might include RS data in their specific study. We propose how to improve models of species ecological niches, by including measures of habitat quality (e.g., productivity), nutritional values, and seasonal or life-cycle events. To date, several studies have shown that using ecologically-relevant environmental predictors derived from RS improve model performance and transferability, and better approximate a species ecological niche. These data, however, are not a panacea for SDMs, as there are cases in which RS predictors are not appropriate, too costly, or exhibit low predictive power. The integration of multiple environmental predictors derived from RS in SDMs can thus improve our knowledge on processes driving biodiversity change and improve our capacity for biodiversity conservation.

Dynamic occupancy modeling of temperate marine fish in area-based closures.

Species distribution models (SDMs) are commonly used to model the spatial structure of species in the marine environment, however, most fail to account for detectability of the target species. This can result in underestimates of occupancy, where nondetection is conflated with absence. The site occupancy model (SOM) overcomes this failure by treating occupancy as a latent variable of the model and incorporates a detection submodel to account for variability in detection rates. These have rarely been applied in the context of marine fish and never for the multiseason dynamic occupancy model (DOM). In this study, a DOM is developed for a designated species of concern, cusk (Brosme brosme), over a four‐season period. Making novel use of a high‐resolution 3‐dimensional hydrodynamic model, detectability of cusk is considered as a function of current speed and algae cover. Algal cover on the seabed is measured from video surveys to divide the study area into two distinct regions: those with canopy forming species of algae and those without (henceforth bottom types). Modeled estimates of the proportion of sites occupied in each season are 0.88, 0.45, 0.74, and 0.83. These are significantly greater than the proportion of occupied sites measured from underwater video observations which are 0.57, 0.28, 0.43, and 0.57. Individual fish are detected more frequently with increasing current speed in areas lacking canopy and less frequently with increasing current speed in areas with canopy. The results indicate that, where possible, SDM studies for all marine species should take account of detectability to avoid underestimating the proportion of sites occupied at a given study area. Sampling closed areas or areas of conservation often requires the use of nonphysical, low impact sampling methods like camera surveys. These methods inherently result in detection probabilities less than one, an issue compounded by time‐varying features of the environment that are rarely accounted for marine studies. This work highlights the use of modeled hydrodynamics as a tool to correct some of this imbalance.

Evaluating collinearity effects on species distribution models: An approach based on virtual species simulation.

The increasing use of species distribution modeling (SDM) has raised new concerns regarding the inaccuracies, misunderstanding, and misuses of this important tool. One of those possible pitfalls − collinearity among environmental predictors − is assumed as an important source of model uncertainty, although it has not been subjected to a detailed evaluation in recent SDM studies. It is expected that collinearity will increase uncertainty in model parameters and decrease statistical power. Here we use a virtual species approach to compare models built using subsets of PCA-derived variables with models based on the original highly correlated climate variables. Moreover, we evaluated whether modelling algorithms and species data characteristics generate models with varying sensitivity to collinearity. As expected, collinearity among predictors decreases the efficiency and increases the uncertainty of species distribution models. Nevertheless, the intensity of the effect varied according to the algorithm properties: more complex procedures behaved better than simple envelope models. This may support the claim that complex models such as Maxent take advantage of existing collinearity in finding the best set of parameters. The interaction of the different factors with species characteristics (centroid and tolerance in environmental space) highlighted the importance of the so-called “idiosyncrasy in species responses” to model efficiency, but differences in prevalence may represent a better explanation. However, even models with low accuracy to predict suitability of individual cells may provide meaningful information on the estimation of range-size, a key species-trait for macroecological studies. We concluded that the use of PCA-derived variables is advised both to control the negative effects of collinearity and as a more objective solution for the problem of variable selection in studies dealing with large number of species with heterogeneous responses to environmental variables.

Does scale matter? A systematic review of incorporating biological realism when predicting changes in species distributions.

BackgroundThere is ample evidence that biotic factors, such as biotic interactions and dispersal capacity, can affect species distributions and influence species’ responses to climate change. However, little is known about how these factors affect predictions from species distribution models (SDMs) with respect to spatial grain and extent of the models.ObjectivesUnderstanding how spatial scale influences the effects of biological processes in SDMs is important because SDMs are one of the primary tools used by conservation biologists to assess biodiversity impacts of climate change.Data sources and study eligibility criteriaWe systematically reviewed SDM studies published from 2003–2015 using ISI Web of Science searches to: (1) determine the current state and key knowledge gaps of SDMs that incorporate biotic interactions and dispersal; and (2) understand how choice of spatial scale may alter the influence of biological processes on SDM predictions.Synthesis methods and limitationsWe used linear mixed effects models to examine how predictions from SDMs changed in response to the effects of spatial scale, dispersal, and biotic interactions.ResultsThere were important biases in studies including an emphasis on terrestrial ecosystems in northern latitudes and little representation of aquatic ecosystems. Our results suggest that neither spatial extent nor grain influence projected climate-induced changes in species ranges when SDMs include dispersal or biotic interactions.ConclusionsWe identified several knowledge gaps and suggest that SDM studies forecasting the effects of climate change should: 1) address broader ranges of taxa and locations; and 1) report the grain size, extent, and results with and without biological complexity. The spatial scale of analysis in SDMs did not affect estimates of projected range shifts with dispersal and biotic interactions. However, the lack of reporting on results with and without biological complexity precluded many studies from our analysis.

In search of relevant predictors for marine species distribution modelling using the MarineSPEED benchmark dataset

AbstractAimIdeally, datasets for species distribution modelling (SDM) contain evenly sampled records covering the entire distribution of the species, confirmed absences and auxiliary ecophysiological data allowing informed decisions on relevant predictors. Unfortunately, these criteria are rarely met for marine organisms for which distributions are too often only scantly characterized and absences generally not recorded. Here, we investigate predictor relevance as a function of modelling algorithms and settings for a global dataset of marine species.LocationGlobal marine.MethodsWe selected well‐studied and identifiable species from all major marine taxonomic groups. Distribution records were compiled from public sources (e.g., OBIS, GBIF, Reef Life Survey) and linked to environmental data from Bio‐ORACLE and MARSPEC. Using this dataset, predictor relevance was analysed under different variations of modelling algorithms, numbers of predictor variables, cross‐validation strategies, sampling bias mitigation methods, evaluation methods and ranking methods. SDMs for all combinations of predictors from eight correlation groups were fitted and ranked, from which the top five predictors were selected as the most relevant.ResultsWe collected two million distribution records from 514 species across 18 phyla. Mean sea surface temperature and calcite are, respectively, the most relevant and irrelevant predictors. A less clear pattern was derived from the other predictors. The biggest differences in predictor relevance were induced by varying the number of predictors, the modelling algorithm and the sample selection bias correction. The distribution data and associated environmental data are made available through the R package marinespeed and at http://marinespeed.org.Main conclusionsWhile temperature is a relevant predictor of global marine species distributions, considerable variation in predictor relevance is linked to the SDM set‐up. We promote the usage of a standardized benchmark dataset (MarineSPEED) for methodological SDM studies.

Incorporating plant fossil data into species distribution models is not straightforward: Pitfalls and possible solutions

The increasing development of species distribution models (SDMs) using palaeodata has created new prospects to address questions of evolution, ecology and biogeography from wider perspectives. Palaeobotanical data provide information on the past distribution of taxa at a given time and place and its incorporation on modelling has contributed to advancing the SDM field. This has allowed, for example, to calibrate models under past climate conditions or to validate projected models calibrated on current species distributions. However, these data also bear certain shortcomings when used in SDMs that may hinder the resulting ecological outcomes and eventually lead to misleading conclusions. Palaeodata may not be equivalent to present data, but instead frequently exhibit limitations and biases regarding species representation, taxonomy and chronological control, and their inclusion in SDMs should be carefully assessed. The limitations of palaeobotanical data applied to SDM studies are infrequently discussed and often neglected in the modelling literature; thus, we argue for the more careful selection and control of these data. We encourage authors to use palaeobotanical data in their SDMs studies and for doing so, we propose some recommendations to improve the robustness, reliability and significance of palaeo-SDM analyses.

Topographic variables improve climatic models of forage species abundance in the northeastern United States

AbstractQuestionSpecies distribution modelling has most commonly been applied to presence‐only data and to woody species. Can similar methods be used to create detailed predicted abundance maps for forage species? These predictions would be of great value for agricultural management and land‐use planning.LocationNortheastern USA.MethodsWe used field data from 31 grazed farms to model abundances for six forage species with three statistical methods: GLM, GAM and Random Forest models. A hierarchical ecological framework encompassing climatic, edaphic and topographic variables related to the plant species requirements for water, light and temperature was used to guide variable selection.ResultsAlthough many species distribution modelling studies have used only climatic variables, the inclusion of topography greatly improved explanatory power. Edaphic variables contributed little more beyond the information already provided by climate and topography. Random Forest models had higher overall predictive capability, and were used to produce the final potential abundance maps for the six forage species.ConclusionsClimate‐only predictions may be suitable for state or regional planning, but topographic variables must be included in species distribution models used to support decision‐making at the farm and field scales.