The Importance of Including Spatial Autocorrelation When Modelling Species Richness in Archipelagos: A Bayesian Approach

Abstract

One of the aims of island biogeography theory is to explain the number of species in an archipelago. Traditionally, the variables used to explain the species richness on an island are its area and distance to the mainland. However, increasing evidence suggests that accounting for other variables is essential for better estimates. In particular, the distance between islands should play a role in determining species richness. This work uses a Bayesian framework using Gaussian processes to assess whether distance to neighbouring islands (spatial autocorrelation) can better explain arthropod species richness patterns in the Azores Archipelago and in the Canary Islands. This method is flexible and allows the inclusion of other variables, such as maximum altitude above sea level (elevation). The results show that accounting for spatial autocorrelation provides the best results for both archipelagos, but overall, spatial autocorrelation seems to be more important in the Canary archipelago. Similarly, elevation plays a more important role in determining species richness in the Canary Islands. We recommend that spatial autocorrelation should always be considered when modelling an archipelago’s species richness.