Physical calculations of resistance to water loss improve predictions of species range models

Species ranges are constrained by the physiological tolerances of organisms to climatic conditions. By incorporating physiological constraints, species distribution models can identify how biotic and abiotic factors constrain a species' geographic range. Rates of water loss influence species distributions, but characterizing water loss for an individual requires complex calculations. Skin resistance to water loss (r(i)) is considered to be the most informative metric of water loss rates because it controls for experimental biases. However, calculating ri requires biophysical equations to solve for the resistance of the air that surrounds an organism, termed the boundary layer resistance (r(b)). Here, we compared theoretical and empirical methods for measuring skin resistance to water loss of a Plethodon salamander collected from nature. For the empirical methods, we measured r(b) of agar replicas at five body sizes, two temperatures, three vapor pressure deficits, and six flow rates using a flow-through system. We also calculated r(b) using biophysical equations under the same experimental conditions. We then determined the ecological implications of incorporating skin and boundary layer resistance into a species range model that estimated potential activity time and energy balance throughout the geographic range of the study species. We found that empirical methods for calculating r(b) resulted in negative values of r(i), whereas biophysical calculations produced meaningful values of r(i). The species range model determined that ignoring realistic boundary layer and skin resistances reduced average estimates of energy balance by as much as 64% and potential activity time by 88% throughout the spatial extent of the model. We conclude that the use of agar replicas is an inadequate technique to characterize skin resistance to water loss, and incorporating boundary layer and skin resistances to water loss improves estimates of activity and energetics for mechanistic species distribution models. More importantly, our study suggests that incorporating the physical processes underlying rates of water loss could improve estimates of habitat suitability for many animals.