Thermal Responses Differ across Levels of Biological Organization

Temperature is one of the most important environmental factors driving the genome-to-phenome relationship. Metabolic rates and related biological processes are predicted to increase with temperature due to the biophysical laws of chemical reactions. However, selection can also act on these processes across scales of biological organization, from individual enzymes to whole organisms. Although some studies have examined thermal responses across multiple scales, there is no general consensus on how these responses vary depending on the level of organization, or whether rates actually follow predicted theoretical patterns such as Arrhenius-like exponential responses or thermal performance curves (TPCs) that show peak responses. Here, we performed a meta-analysis on studies of ectotherms where biological rates were measured across the same set of temperatures, but at multiple levels of biological organization: enzyme activities, mitochondrial respiration, and/or whole-animal metabolic rates. Our final dataset consisted of 235 pairwise comparisons between levels of organization from 13 publications. Thermal responses differed drastically across levels of biological organization, sometimes showing completely opposite patterns. We developed a new effect size metric, "organizational disagreement" (OD) to quantify the difference in responses among levels of biological organization. Overall, rates at higher levels of biological organization (e.g., whole animal metabolic rates) increased more quickly with temperature than rates at lower levels, contrary to our predictions. Responses may differ across levels due to differing consequences of biochemical laws with increasing organization or due to selection for different responses. However, taxa and tissues examined generally did not affect OD. Theoretical TPCs, where rates increase to a peak value and then drop, were only rarely observed (12%), possibly because a broad range of test temperatures was rarely investigated. Exponential increases following Arrhenius predictions were more common (29%). This result suggests a classic assumption about thermal responses in biological rates is rarely observed in empirical datasets, although our results should be interpreted cautiously due to the lack of complete thermal profiles. We advocate for authors to explicitly address OD in their interpretations and to measure thermal responses across a wider, more incremental range of temperatures. These results further emphasize the complexity of connecting the genome to the phenome when environmental plasticity is incorporated: the impact of the environment on the phenotype can depend on the scale of organization considered.