Lizard Thermal Physiology Drives Abundance Peaks along Climate Gradients but Only Weakly Predicts Distributional Limits.

AbstractLaboratory measurements of physiological traits have long been used to infer the thermal limits and preferences of species in the field. However, it remains unclear how well individual physiological traits scale up to explain broad distribution patterns of species, such as their climatic limits, the breadth of temperatures they occur in, and the conditions at which population abundances are highest. We address these gaps by combining laboratory-measured thermal traits (critical thermal minimum [CT_min], critical thermal maximum [CT_max], and thermal preference [T_pref]) with occurrence and abundance data from 21 species of Anolis lizards collected from extensive mark-resight surveys of communities across the Caribbean islands of Puerto Rico and Hispaniola. Our findings suggest that thermal limits do map to distribution boundaries, such that CT_max and CT_min are significant predictors of maximum and minimum environmental temperatures at which species occur in nature, albeit with substantial error. Curiously though, physiological niche breadth (${\mathrm{CT}}_{\max }-{\mathrm{CT}}_{\min }$) does not positively correlate with climatic niche breadth. This means that species able to tolerate a wide range of temperatures do not always occur across a broad range of climates, limiting our ability to make clear-cut statements about what constitutes a thermal generalist or specialist. The climatological temperature where population abundance is maximized is the geographic feature best predicted by physiology, yet counterintuitively T_pref performs worse than critical thermal limits at predicting where this abundance peak occurs. Together, our findings suggest that individual physiological responses to temperature do not always translate to distribution patterns in predictable ways, suggesting a substantial role for other factors, such as competition, predation, nonthermal habitat characteristics, and behavioral buffering, in setting range-wide distribution patterns.