Modeling habitat distribution from organism occurrences and environmental data: case study using anemonefishes and their sea anemone hosts

Abstract

We demonstrate the KGSMapper (Kansas Geological Survey Mapper), a straightforward, web-based biogeographic tool that uses environmental conditions of places where members of a taxon are known to occur to find other places containing suitable habitat for them. Using occurrence data for anemonefishes or their host sea anemones, and data for environmental parameters, we generated maps of suitable habitat for the organisms. The fact that the fishes are obligate symbionts of the anemones allowed us to validate the KGSMapper output: we were able to compare the inferred occurrence of the organism to that of the actual occurrence of its symbiont. Characterizing suitable habitat for these organisms in the Indo-West Pacific, the region where they naturally occur, can be used to guide conservation efforts, field work, etc.; defining suitable habitat for them in the Atlantic and eastern Pacific is relevant to identifying areas vulnerable to biological invasions. We advocate distinguishing between these 2 sorts of model output, terming the former maps of realized habitat and the latter maps of potential habitat. Creation of a niche model requires adding biotic data to the environmental data used for habitat maps: we included data on fish occurrences to infer anemone distribution and vice versa. Altering the selection of environmental variables allowed us to investigate which variables may exert the most influence on organism distribution. Adding variables does not necessarily improve precision of the model output. KGSMapper output distinguishes areas that fall within 1 standard deviation (SD) of the mean environmental variable values for places where members of the taxon occur, within 2 SD, and within the entire range of values; eliminating outliers or data known to be imprecise or inaccurate improved output precision mainly in the 2 SD range and beyond. Thus, KGSMapper is robust in the face of questionable data, offering the user a way to recognize and clean such data. It also functions well with sparse datasets. These features make it useful for biogeographic meta-analyses with the diverse, distributed datasets that are typical for marine organisms lacking direct commercial value.