How well do species distribution models predict occurrences in exotic ranges?. Issue 6 (10th March 2022)

• Record Type: Journal Article
• Authors: Nguyen, Dat; Leung, Brian
• Other Names: Schrodt Franziska handlingEditor.
• Is Part Of: Global ecology & biogeography. Volume 31:Issue 6(2022)
• Page Start: 1051
• Page End: 1065
• Publication Date: 2022-03-10
• Subjects: GBIF -- global -- invasive species -- presence‐only data -- species distribution model -- transferability; Ecology -- Periodicals; Biogeography -- Periodicals; Biodiversity -- Periodicals; Macroevolution -- Periodicals; 577
• DOI: 10.1111/geb.13482
• Languages: English
• ISSNs: 1466-822X

Abstract

Abstract: Aim: Species distribution models (SDMs) are widely used predictive tools to forecast potential biological invasions. However, the reliability of SDMs extrapolated to exotic ranges remains understudied, with most analyses restricted to few species and equivocal results. We examine the spatial transferability of SDMs for 647 non‐indigenous species extrapolated across 1, 867 invaded ranges, and identify what factors may help differentiate predictive success from failure. Location: Global. Time period: Current. Major taxa studied: Six hundred and forty‐seven terrestrial species; eight taxonomic classes. Methods: We performed a large‐scale assessment of the transferability of SDMs using two modelling approaches: generalized additive models (GAMs) and MaxEnt. We fitted SDMs on the native ranges of species and extrapolated them to exotic ranges. We examined the influence of general factors and factors related to biological invasions on spatial transferability. Results: Despite both modelling approaches performing well in the range of the species used for fitting, we observed moderate to low spatial transferability on average (mean area under the receiver operating characteristic curve [AUC] ~ .7) when extrapolating to their invaded ranges. Transferability differed between taxonomic classes and invaded continents and was positively influenced by the performance of the model and environmental generalism in the native range, and the year of first record. Models performed worse with greater environmental coverage in the exotic range, gross domestic product and number of occurrences in the native range, geographic distance between ranges and when extrapolating to islands. Main conclusions: After controlling for sampling bias, half of SDMs were only weakly predictive, which should affect how SDM‐based forecasts are interpreted. Performance differed based on characteristics of the data, species, and ranges, and can suggest when SDMs may be reliable and when we should be most cautious. These considerations touch directly upon the potential use of SDMs for management of biological invasions. We discuss possible mechanisms of these findings.