How well do SDMs predict occurrences in exotic ranges?
Native distributions are unreliable for predicting invaded ranges
In our last research summary, Liu et al. (2020) argued that native ranges are generally good predictors of the invaded range. This week, Nguyen & Leung (2022) use a different approach and reveal more methodological challenges. PDF
Nguyen, D. and B. Leung. 2022. How well do species distribution models predict occurrences in exotic ranges? Global Ecology and Biogeography 31(6): 1051-1065.
Summary
Researchers and managers often use species distribution models (SDMs) to predict the potential ranges of invasive species - including locations of possible future invasions under different climate change scenarios. In these situations, researchers would like to be able to use occurrences in the native range to inform risk in the introduced range. However, there is ongoing debate as to whether this is a viable approach because it assumes that the conditions that influence distributions in the native range are the same as the ones that influence distributions in the introduced range (i.e. ‘niche conservatism’). Nguyen and Leung created species distribution models based on the native ranges of 647 species and used these models to predict the locations of the nonnative ranges of these species. Then, they compared the predicted nonnative ranges to actual occurrences to determine the accuracy of the models. Nguyen and Leung found that the models they created with native range data adequately reflected the climatic conditions of each species’ native range, but were unable to accurately predict introduced ranges. Although Liu et al. (2020) argued that the climate ‘niche’ was likely similar in the native and introduced ranges, Nguyen and Leung (2022)’s analysis suggest that we should still be cautious when interpreting invasion risk from models based on the species’ native range.
Take home points
Generally, species distribution models based on native range data poorly predict introduced ranges.
Model reliability might be slightly higher for plants, and for species with a lot of distribution data in the introduced range (for example, those invading North America).
Management implications
Managers should use caution when interpreting maps of invasion risk based on the species’ native range.
Numerous modeling challenges likely contribute to poor prediction including different environmental conditions in the native vs. introduced ranges, lack of distribution data in either native or introduced ranges, and insufficient time for a species to spread in the introduced range (‘lack of equilibrium’).
Keywords
Range Expansion; Species distribution models (SDMs); Global distributions; Niche conservatism; Big Data