How integrated are behavioral and endocrine stress response traits? A repeated measures approach to testing the stress-coping style model

It is widely expected that physiological and behavioral stress responses will be integrated within divergent stress-coping styles (SCS) and that these may represent opposite ends of a continuously varying reactive–proactive axis. If such a model is valid, then stress response traits should be repeatable and physiological and behavioral responses should also change in an integrated manner along a major axis of among-individual variation. While there is some evidence of association between endocrine and behavioral stress response traits, few studies incorporate repeated observations of both. To test this model, we use a multivariate, repeated measures approach in a captive-bred population of Xiphophorus birchmanni. We quantify among-individual variation in behavioral stress response to an open field trial (OFT) with simulated predator attack (SPA) and measure waterborne steroid hormone levels (cortisol, 11-ketotestosterone) before and after exposure. Under the mild stress stimulus (OFT), (multivariate) behavioral variation among individuals was consistent with a strong axis of personality (shy–bold) or coping style (reactive–proactive) variation. However, behavioral responses to a moderate stressor (SPA) were less repeatable, and robust statistical support for repeatable endocrine state over the full sampling period was limited to 11-ketotestosterone. Although post hoc analysis suggested cortisol expression was repeatable over short time periods, qualitative relationships between behavior and glucocorticoid levels were counter to our a priori expectations. Thus, while our results clearly show among-individual differences in behavioral and endocrine traits associated with stress response, the correlation structure between these is not consistent with a simple proactive–reactive axis of integrated stress-coping style. Additionally, the low repeatability of cortisol suggests caution is warranted if single observations (or indeed repeat measures over short sampling periods) of glucocorticoid traits are used in ecological or evolutionary studies focussed at the individual level.