Species distribution models (SDMs) are the primary tools used to model and predict changes to species’ ranges, and are often used to provide a quantitative baseline for conservation measures. However, most SDM methods and frameworks have been primarily designed for use with species with relatively large amounts of occurrence data and covering broad continental ranges.Here, we undertake a systematic review of the literature (224 published studies) to assess the appropriate use of SDMs in island biogeography, specifically focusing on marine islands. We divide species into different insular distribution categories (i.e., chorotypes: single island/archipelago endemics, non-endemic natives, and non-natives) in order to provide chorotype-specific SDM recommendations. We highlight how to navigate three fundamental considerations related to the application of SDMs in island environments. 1) Response variables, specifically the issue of small sample sizes for many island species. 2) Predictor variables, including (i) the selection of relevant environmental predictors at appropriate spatial grains, and (ii) addressing the truncation of environmental extent across the entire species range, especially for non-endemic species. 3) Model building, particularly, in the context of limited occurrence data for many island species, how to (i) approach the uncertainty related to the choice of modelling method, and (ii) avoid overfitting.We also examine the data sources used in island SDM studies, finding that there are strong geographical biases in study location. Alongside this, we evaluate the potential of the GBIF database – a comprehensive global database of species occurrences – in island SDM research. We find that GBIF has been potentially underutilised in island SDM studies so far, and represents a useful resource for filling island distribution data gaps for several taxa going forward.Based on the insights we obtained from our systematic review, we propose a set of SDM recommendations tailored to insular species and environments.
Full Text Finder