Predicting the spatio-temporal pattern of range expansion under lack of equilibrium with climate

Abstract

Early detection of invasive species in regions under threat is key for biodiversity conservation. Here we conduct a retrospective study to assess whether correlative Species Distribution Models (SDMs) can predict the spatio-temporal range of expansion in an invasive species, the Asian hornet. Given that modelling invasive species distributions defies one of the main assumptions of SDMs, the equilibrium with climate, we also introduce a quantitative assessment of climatic disequilibrium in the invaded area based on hypervolume comparison between the native and invaded areas (both unoccupied and occupied territories). We evaluate the ability of three different modelling approaches (presence-only, presence-background and presence-absence) calibrated with data until 2013 to predict the species distribution observed the following years (2015, 2017, 2019, and 2021). Our results show that presence-only models based on the BIOCLIM algorithm can effectively predict the spatio-temporal pattern of invasion when predictions are rescaled based on percentiles (i.e. ranked suitability) and the model is calibrated with data from both native and invaded areas. These models predicted higher suitability values for sites that were invaded earlier in time and, importantly, they did not predict low suitability values for sites that were eventually occupied years later. Thus, they can be very useful for decision-makers and managers, as they provide a probabilistic prediction of both (i) the temporal dimension of species range expansion, and (ii) the potential distribution range in the long term. Such information would allow prioritizing monitoring efforts in the short term without losing sight of the potential risks in the long term