Evaluating 318 continental‐scale species distribution models over a 60‐year prediction horizon: what factors influence the reliability of predictions?

Abstract

Aim Species distribution models (SDMs) are currently the most widely used tools in ecology for evaluating the suitability of environments for biodiversity in the face of future environmental change. In this study we seek to provide an assessment of the predictive performance of SDMs over time. How well do SDMs predict for future time periods and what factors influence predictive performance? Innovation We used a historical spatially explicit database of 1.8 million occurrence records for 318 tetrapod species from across continental Australia over the period 1950–2013. We fitted distribution models for each species to data from four multi‐decadal time slices and used these to predict the species distributions up to 60 years after the data collection period for the fitted models. We evaluated predictions against observed data from the relevant time period. Predictions were made assuming either complete knowledge of changes in climatic and environmental conditions or assuming the environment and climate remained unchanged between the fitting and evaluation time periods. We used generalized linear mixed models to model variation in the predictive performance of SDMs over time in relation to a variety of factors, including the length of time between fitting and evaluation, species traits, taxonomic group and attributes of the dataset used to fit models. Main conclusions We found that most models provided useful predictions even when the period between model fitting and evaluation was 60 years (area under the receiver operator characteristic curve > 0.7 in 80% of the species evaluated). Variation in predictive performance over time was strongly related to the species range breadth (models for species with broad geographical ranges tended to perform worse than models for locally restricted species) and to the environmental coverage of occupancy data. Conversely, taxonomic group, habitat preferences and body size were not highly influential in describing the variation in predictive performance over time.

Modelling the spatial distribution of cetaceans in New Zealand waters

“…Some species with widely distributed recorded locations had poorer model fits than species with restricted ranges, perhaps reflecting the cosmopolitan distribution of the former (e.g., moderate explained deviance and AUC scores for killer whale and bottlenose dolphin) and the more aggregated nature of others for the latter (e.g., high explained deviance and AUC scores for Māui dolphin, Hector's dolphin). Evidence from previous studies have indicated that species with limited geographic ranges and/or environmental tolerances are generally better modelled than those with greater ranges (Morán‐Ordóñez, Lahoz‐Monfort, Elith, & Wintle, ; Stephenson et al, ; Thomson et al, ) because widespread species are less likely to have sharp easily identifiable environmental thresholds that clearly delineate their environmental niche (Morán‐Ordóñez et al, ). For species with limited ranges, the best model fits were commonly located closer to shore where sampling effort was highest.…”

Modelling species responses to extreme weather provides new insights into constraints on range and likely climate change impacts for Australian mammals

“…Predictive performance was assessed in terms of discrimination ability measured using the area under the receiver-operator characteristic curve (AUC; Hanley and McNeil 1982) adapted for use with presence-background samples (Phillips et al 2006). This metric is suited to presence-background data, since calibration cannot be assessed and applying thresholds to predictions loses information (Guillera-Arroita et al 2015, Morán-Ordóñez et al 2016. We calculated AUC using the ten-fold cross-validation provided in Maxent.…”

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

“…SDMs are a widely used tool for predicting species distributions in response to changing climate (Engler et al, 2011;Hof, Araújo, Jetz, & Rahbek, 2011;Hof et al, 2018; Morán-Ordóñez, Lahoz-Monfort, Elith, Wintle, & Guisan, 2017;Peterson et al, 2002;Thuiller, Guéguen, Renaud, Karger, & Zimmermann, 2019;Zurell, Graham, Gallien, Thuiller, & Zimmermann, 2018). SDMs are based on statistical correlations between species occurrences and environmental predictor variables (Elith & Leathwick, 2009;Guisan & Thuiller, 2005;Guisan & Zimmermann, 2000), which can then be transferred into future time periods to predict future species distributions under climate change.…”