SEARS, Michael W.; BAKKEN, George S.; ANGILLETTA, Michael J.; FITZGERALD, Lee A.; University of Nevada, Reno; Indiana State University; Indiana State University; Texas A&M University: Using artificial neural networks to model the operative temperatures of small animals in a spatially-explicit context

Ecologists have long known that thermal resources have profound impacts on an individual’s physiological performance, behavioral options, and, ultimately, fitness. Unfortunately, the tools available to study the spatio-temporal distribution of operative temperatures in an ecological setting have provided only a very coarse portrayal of the thermal environment, and thus many questions regarding the quality of habitat from a thermal perspective have been largely glossed over. Here, we illustrate a novel method for mapping the operative temperatures of small animals at a scale relevant for the study of the behavior and physiology of individuals. We use standard micrometeorological data (solar radiation, wind speed, air temperature) and topographical data (slope, aspect, elevation, vegetative cover) to predict the temperatures of hollow copper models of small lizards using artificial neural networks. We demonstrate the improved performance of artificial neural networks compared to other statistical modeling techniques for the prediction of operative temperatures. Furthermore, we illustrate how artificial neural network modeling techniques can be combined with aerial photography (or other geographical data) to produce centimeter to meter scale resolution thermal maps over areas covering square kilometers. Such techniques will allow detailed investigation of how the spatio-temporal distribution of operative temperatures influences many aspects of the ecology of individuals including thermoregulation, physiological performance, habitat selection, habitat quality, movement patterns, and species interactions.