Decomposition of the maximum entropy niche function - A step beyond modelling species distribution

Some algorithms used to model species' distributions are often only considered as black boxes; coefficients of the underlying niche functions are often not interpreted ecologically. Here, we focus on the maximum entropy approach that is commonly and successfully applied in order to model species' distribution. By means of an eigenanalysis, we decompose the niche function into independent factors that can be interpreted separately. In addition, we derive parameters that can be used to characterize the species' niche, especially considering the steepness of the modelled niche function and the sensitivity of the considered species against changes in certain environmental conditions. On the example of three well-studied Taiga forest tree species we illustrate the capability and scope of our approach. Given the easy availability of environmental data and species occurrences, the presented approach seems to be a feasible way to gain deeper insights into the factors that are related to species' distributions.