Modeling the Natural Distributions of Stonefly Species (Plecoptera) In Midwest Streams

Understanding the natural distributions of aquatic species is critical for evaluating the biological conditions of streams and the impacts of climate changes. We compiled 5628 records of stonefly species in Midwest streams/rivers based on the collections of Illinois Natural History Survey and other institutions, and established a dataset for a range of GIS-based environmental variables, including watershed geology, climate, topography, hydrology, soil, natural land covers, and glacial history. Both datasets are organized at 12-digit hydrological units (HUC 12). Using IL data we developed two sets of Maxent models for 22 stonefly species using standard options and a recently-developed AICc-based selection procedure (ENMTools), respectively. Species presence-absence data from 29 units that were historically well sampled were used to evaluate model performances. Models based on AICc often over-predicted the occurrences and richness. Models calibrated based on default options under- or over-predicted, depending on the threshold used and sites. The AICc-procedure may not improve the utility of Maxent. Estimates of species richness in a HUC12 cell and occurrences of individual species derived from standard Maxent models and appropriately selected thresholds can be used as a robust index of each of these two measures, respectively. Based on this pilot study, we modeled the occurrences of >140 stonefly species in five Midwest states.