Physiological predictors of reproductive performance in the European Starling (Sturnus vulgaris): II. Multivariate Analysis

Abstract

Physiological variation is generally thought or supposed to underlie variation in fitness related traits in wild animals, including reproductive effort, reproductive success, and survival. However, physiological markers of individual quality have proven elusive. In this paper and its companion, we use data on 14 physiological parameters measured in 152 observations of 93 individual European starlings (Sturnus vulgaris) over 2 years in an attempt to understand how physiological variation relates to variation in current breeding productivity, future fecundity, and survival. The companion paper, focusing on univariate analysis, showed that individual physiological parameters have little relationship with these performance measures. Here, we used more sophisticated statistical approaches in an attempt to extract a multivariate signal from the biomarkers - physiological dysregulation as calculated via statistical distance, and a number of principal components analysis approaches. Broadly speaking, there was a surprising lack of association between physiology and performance: while some physiological summary measures were associated with some performance measures, the associations were not particularly strong or robust given the large number of statistical tests conducted. This implies either that there are relatively few links between physiology and performance, or, more likely, that the complexity of these relationships exceeds our ability to measure and model it, even using state-of-the-art statistical approaches. This is likely particularly true because our population was quite heterogeneous; we nonetheless urge caution regarding the over-interpretation of isolated significant findings in the literature.