Resource information
Models to predict species’ ranges have chiefly been limited to abiotic variables. However, the full ecological niche depends on a myriad of factors, both biotic and abiotic, that often correspond to completely different data types. We applied a methodology based on data mining techniques to construct ecological niche models composed of biotic as well as abiotic variables using three quite different sets of variables: climatic layers, maps of land cover and point collections of Mexican mammals. We show how potential ecological interactions can be inferred from geographic data using co-occurrences as proxies, and generate corresponding distribution models. We consider two case studies: an insect genus (Lutzomyia sp.) and a mammal species (Lynx rufus). We show that for both examples model predictability is higher using biotic versus abiotic variables, but even higher when both variable types are integrated together. Also, by identifying those variables that are most relevant in describing the suitable (niche) and unsuitable (anti-niche) areas we can establish an ecological profile for any geographic location and quantify the relative influence of each location and its impact on species. In conclusion, we show that including both abiotic and biotic factors not only leads to a fuller more comprehensive understanding of the niche, but also leads to more accurate prediction models.
