Considering this wide range of inter-female variability in yolk T levels, we can assume that several genes were targeted by the selection. One set of genes could encode the enzymes of the steroidogenic pathway, since a recent study showed that aromatase mRNA expression in ovarian follicles of female house sparrows is negatively correlated with yolk T concentrations in both the largest F1 follicles in the ovary and in eggs laid. In spite of the different duration of selection experiments, mean egg T concentrations were comparable among the STI, HSR and HET lines suggesting that there is a common factor that limits the selection for high egg T content. Such constraints have been discussed at the level of either parental or offspring generations or both. In the mother, the constraints of high egg T deposition could arise from an inability to independently regulate yolk and plasma T concentrations and the resulting costs of elevated plasma T on maternal physiology and behaviour. Our previous study demonstrated that selection for increased yolk T AMN107 levels did not lead to a parallel increase in circulating T concentrations, suggesting independent regulation of both pools. However, the costs of up-regulated steroidogenesis in females can be mediated not only through high plasma T, but also through its conversion into other active metabolites, namely estradiol. In offspring, the constraints of high egg T can relate to its pleiotropic effects on diverse morphological, physiological and behavioural characteristics, resulting in potential trade-offs among these offspring traits. Moreover, we found lower egg mass in lines selected for decreased fearfulness and increased social motivation as compared to their oppositely selected lines and simultaneously these line differences in egg mass were inverted to line differences in yolk T levels. In precocial birds, this inverse pattern between egg mass and yolk T levels was not reported yet and underlying mechanisms need to be further explored. In the second part of our study, we examined the hypothesis that natural selection favours mutually adjusted deposition of different egg components, i.e. yolk androgens and yolk antibodies, to adaptively modulate trans-generational maternal effects. We found significant differences in yolk IgY levels between oppositely-selected lines, but these inter-line differences in yolk IgY were not consistent with the pattern of inter-line differences in yolk T deposition across the three selection experiments. Lower yolk IgY levels were detected in the STI, LSR and HET lines as compared to their oppositely-selected LTI, HSR and LET lines. Until now, only a few studies have investigated mutually related variations in yolk androgens and immunoglobulins in free living birds, and they showed inconsistent results. An inverse pattern of yolk T and IgY levels was found within the laying sequence in the black-headed gull, but there was no correlation between these egg components at the level of the laying sequence or clutches. In the black-legged kittiwake, food supplemented females produced eggs with lower A4 and IgY levels in their replacement clutches as compared to non-supplemented females, resulting in a positive correlation between yolk A4 and IgY concentrations.