Explanative power of variables used in species distribution modelling: an issue of general model transferability or niche shift in the invasive Greenhouse frog (Eleutherodactylus planirostris)

Abstract

Listen

Translate article

The use of species distribution models (SDMs) to predict potential distributions of species is steadily increasing. A necessary assumption when projecting models throughout space or time is that climatic niches are conservative, but recent findings of niche shifts during biological invasion of particular plant and animal species have indicated that this assumption is not categorically valid. One reason for observed shifts may relate to variable selection for modelling. In this study, we assess differences in climatic niches in the native and invasive ranges of the Greenhouse frog (Eleutherodactylus planirostris). We analyze which variables are more 'conserved' in comparison to more 'relaxed' variables (i.e. subject to niche shift) and how they influence transferability of SDMs developed with Maxent on the basis of ten bioclimatic layers best describing the climatic requirements of the target species. We focus on degrees of niche similarity and conservatism using Schoener's index and Hellinger distance. Significance of results are tested with null models. Results indicate that the degrees of niche similarity and conservatism vary greatly among the predictive variables. Some shifts can be attributed to active habitat selection, whereas others apparently reflect variation in the availability of climate conditions or biotic interactions between the frogs' native and invasive ranges. Patterns suggesting active habitat selection also vary among variables. Our findings evoke considerable implications on the transferability of SDMs over space and time, which is strongly affected by the choice and number of predictors. The incorporation of 'relaxed' predictors not or only indirectly correlated with biologically meaningful predictors may lead to erroneous predictions when projecting SDMs. We recommend thorough assessments of invasive species' ecology for the identification biologically meaningful predictors facilitating transferability.