Using GIS to create accurate species lists based on Kenyan park attributes and range maps
The intersection of species range maps and study sites are often used as evidence for species presence in ecological studies. While this method is readily accessible through the freely available IUCN range maps for all known species, it tends to vastly overestimate the true alpha diversity of sites. In fact, this method produces a more accurate picture of the regional species pool (‘metacommunity’) or the time-averaged community over decades or even centuries. If, however, we are interested in the true local scale of a site or its faunal community over a more highly resolved time interval, it is inaccurate to assume a positive presence when a species’ range intersects with a site. In this study, we use comprehensive, specimen-verified mammal species lists for 14 Kenyan national parks to test the probability that a mammal species is present in a park given the spatial relationship between the site and its published range.
We show that species are just as likely to be present in a nearby site that it has no intersection with than in a site that is entirely within its range. In addition, intersection with a range provides only about 30% probability that a species will be in a site. We present a set of models based on spatial juxtaposition, amount of overlap, and topographic ruggedness differential that allows us to replicate specimen-verified mammal species lists with at least 70% accuracy. Our model provides critical information about how current range map methodologies may generate bias in presence-absence data and how species are distributed within their IUCN-recognized ranges. Comprehensive and reliable species lists for national parks are extremely time-consuming to compile from specimen records and publications. Assuming that at least some well-researched, specimen-based species lists exist as controls, our method can be easily replicated and the models calibrated with other taxa to greatly improve compliation accurate, comprehensive presence-absence data for community ecology studies.