Comment on an analysis of endotherm thermal tolerances: systematic errors in data compilation undermine its credibility

Understanding the vulnerability of animal populations and communities to rapid climate warming is a critically important endeavour and has recently been informed by a variety of meta-analyses (e.g. [1]). One current area of interest is the synthesis of thermoregulatory data from mammals and birds to examine geographical variation in thermal limits, by comparing thermoregulatory parameters such as thermoneutral zone (TNZ) breadth and variation in the upper and lower critical temperatures that bound the TNZ [2–5]. Here, we report wide-ranging errors in the dataset that is the foundation of Khaliq et al .'s [2] analysis.Khaliq et al . [2] presented a large meta-analysis that extracted data on thermoregulation in birds (161 species) and mammals (297 species) from the physiological literature, with the goal of examining the relationships between physiological capacities and geographical variation in climate. From these individual studies, they compiled data on the lower (LCT) and upper (UCT) critical temperatures, which represent the upper and lower boundaries of the TNZ, the range of environmental temperatures over which resting metabolic rate is constant in endothermic homeotherms.Our discovery that many unsuitable or non-existent data had been included in the dataset compiled by Khaliq et al . [2] was initially triggered by our surprise at the large number of mammal UCT data ( N = 297). At the time, we were conducting a comprehensive review of mammalian thermoregulation in the heat and had found far fewer studies (approx. 100) that contained data permitting the determination of a UCT. We then examined the methods section in the electronic supplementary material, figure S1 and accompanying text, which indicated that Khaliq et al . had indeed used the appropriate definitions of LCT and UCT. Given this disparity, we reviewed the source data from …