Modelling ecological niches and species distributions can be useful when predicting the potential spread of invasive species and species’ response to changing climates. Numerous methods for modelling are used, but the choice of modelling approach depends on many factors, and no single method is likely to excel in all cases.
Relying on the near-global distribution of Olea europea, researchers used primarily GBIF-mediated occurrence data to compare performance of seven modelling approaches across a broad and varied landscape. First, the authors calibrated the models using global climate and occurrence data, and then narrowed in on a smaller study area in Asia to test the transferability of models into future climate scenarios.
The overall results of the comparative study showed very different predictions in future climates depending on modelling approach. While six out of seven tested models yielded predictions significantly better than random, the authors concluded that models based on Maxent and SVM (Support Vector Machines) were best at distiguishing between suitable and unsuitable distributional potential.