Abstract Metabolic scaling fundamentally sets the pace of life in almost all organisms. Research on the metabolic scaling exponent has been largely limited to theoretical and controlled experimental studies, yet causal understanding of how ecological factors shape metabolic scaling in real‐world conditions is lacking. In this paper, our aim is to develop and test an inferential model for the metabolic scaling exponent of free‐living animals. Our inferential model is based on a constraint‐based ecological theory and uses the empirical individual size distribution (ISD) as data input since it contains ecological information on the resource partitioning among individuals within a population. We apply the parameterized Maximum Entropy Theory of Ecology (pMETE) to predict the ISDs and estimate the intraspecific metabolic scaling exponent using individual size data of wild brown trout populations in more than 2000 stream samples in France. Using structural causal modelling, we assess the causal effect of water temperature on the metabolic scaling exponent of brown trout, taking into account other biological and environmental variables. Our findings reveal that the exponent estimated from pMETE averages 0.83, exceeding theoretical expectations (e.g. 0.67 or 0.75), yet aligning with some previous metabolic scaling studies on trout species. Controlling the confounding biases of season and human footprint does not alter the negative effect of temperature on the exponent, but renders it insignificant and reduces its effect size by 54%–62%. This suggests that the true causal impact of temperature on metabolic scaling can be overestimated when season and human footprint are neglected. Our study provides new avenues to infer the metabolic scaling of free‐living animals and to assess the causal effects of environmental factors on the metabolic scaling exponent at large spatial and temporal scales, broadening metabolic scaling research in the context of global change. Our modelling framework can be applied to other wildlife populations where individual size data are available.
Xu et al. (Tue,) studied this question.