Plumage manipulation alters the integration of social behavior, physiology, internal microbiome, and fitness.
Signals often covary with physiological and behavioral traits to form an axis of integrated phenotypic variation associated with reproductive performance. This pattern of phenotypic integration could result from intrinsic between-individual differences that are causally related to signal production, physiology, and behavior. Alternatively, signal expression itself might generate dynamic feedback between physiology, behavior, and the experienced social environment, resulting in an integrated phenotype. Here, we manipulated the plumage of female tree swallows (Tachycineta bicolor) to decouple the expression of a social signal from any pre-existing behavioral or physiological differences. We collected a time series of physiological samples, monitored social interactions with a sensor network, and tracked reproductive performance. Relative to sham controls, dulled females experienced an altered social environment; overall, these females were visited more by conspecific females and less by males. Dulled females subsequently changed their own behavior by initiating fewer interactions and increasing nestling provisioning. These differences resulted in an altered internal microbiome and glucose levels and, ultimately, dulled females produced more offspring. Moreover, dulled females produced larger clutches than control females in the year after the manipulation. Thus, signal variation alone--independent from any pre-existing differences--had a sustained causal affect on a suite of integrated traits. This finding suggests that dynamic feedback may play an important role in coordinating an integrated signaling phenotype. Our results have implications for understanding how variation in signal expression arises and is maintained and the extent to which the information encoded in signals is contingent upon their use in a social environment.