Repeatability of glucocorticoid hormones in vertebrates: a meta-analysis

We often expect that investigations of the patterns, causes, and consequences of among-individual variation in a trait of interest will reveal how selective pressures or ecological conditions influence that trait. However, many endocrine traits, such as concentrations of glucocorticoid (GC) hormones, exhibit adaptive plasticity and, therefore, do not necessarily respond to these pressures as predicted by among-individual phenotypic correlations. To improve our interpretations of among-individual variation in GC concentrations, we need more information about the repeatability of these traits within individuals. Many studies have already estimated the repeatability of baseline, stress-induced, and integrated GC measures, which provides an opportunity to use meta-analytic techniques to investigate (1) whether GC titers are generally repeatable across taxa, and (2) which biological or methodological factors may impact these estimates. From an intensive search of the literature, we collected 91 GC repeatability estimates from 47 studies. Overall, we found evidence that GC levels are repeatable, with mean repeatability estimates across studies ranging from 0.230 for baseline levels to 0.386 for stress-induced levels. We also noted several factors that predicted the magnitude of these estimates, including taxon, sampling season, and lab technique. Amphibians had significantly higher repeatability in baseline and stress-induced GCs than birds, mammals, reptiles, or bony fish. The repeatability of stress-induced GCs was higher when measured within, rather than across, life history stages. Finally, estimates of repeatability in stress-induced and integrated GC measures tended to be lower when GC concentrations were quantified using commercial kit assays rather than in-house assays. The extent to which among-individual variation in GCs may explain variation in organismal performance or fitness (and thereby inform our understanding of the ecological and evolutionary processes driving that variation) depends on whether measures of GC titers accurately reflect how individuals differ overall. Our findings suggest that while GC titers can reflect some degree of consistent differences among individuals, they frequently may not. We discuss how our findings contribute to interpretations of variation in GCs, and suggest routes for the design and analysis of future research.