Chicken or egg? Outcomes of experimental manipulations of maternally transmitted hormones depend on administration method - a meta-analysis.

Steroid hormones are important mediators of prenatal maternal effects in animals. Despite a growing number of studies involving experimental manipulation of these hormones, little is known about the impact of methodological differences among experiments on the final results expressed as offspring traits. Using a meta-analytical approach and a representative sample of experimental studies performed on birds, we tested the effect of two types of direct hormonal manipulations: manipulation of females (either by implantation of hormone pellets or injection of hormonal solutions) and manipulation of eggs by injection. In both types of manipulation we looked at the effects of two groups of hormones: corticosterone and androgens in the form of testosterone and androstenedione. We found that the average effect on offspring traits differed between the manipulation types, with a well-supported positive effect of egg manipulation and lack of a significant effect of maternal manipulation. The observed average positive effect for egg manipulation was driven mainly by androgen manipulations, while corticosterone manipulations exerted no overall effect, regardless of manipulation type. Detailed analyses revealed effects of varying size and direction depending on the specific offspring traits; e.g., egg manipulation positively affected physiology and behaviour (androgens), and negatively affected future reproduction (corticosterone). Effect size was negatively related to the dose of androgen injected into the eggs, but unrelated to timing of manipulation, offspring developmental stage at the time of measuring their traits, solvent type, the site of egg injection and maternal hormone delivery method. Despite the generally acknowledged importance of maternal hormones for offspring development in birds, the overall effect of their experimental elevation is rather weak, significantly heterogeneous and dependent on the hormone and type of manipulation. We conclude by providing general recommendations as to how hormonal manipulations should be performed in order to standardize their impact and the results achieved. We also emphasize the need for research on free-living birds with a focus on fitness-related and other long-term effects of maternal hormones.