Sexually selected signaling in birds: a case for bayesian change-point analysis of behavioral routines

Responses of organisms to environments or to conspecifics may abruptly change once the organism has changed its state. For example, the expression of sexually selected signals often depends on the pairing status of the sender. A likely change in signaling routines at the point of pair formation should thus be taken into account when investigating a sexually selected trait like birdsong. However, female breeding behavior is elusive in many species, and the date of pair formation may be unknown. We developed a change-point model in a Bayesian context to analyze the seasonally changing singing activity in male Nightingales (Luscinia megarhynchos) for which the pairing status was assumed to be unknown. We used our change-point singing activity model to estimate the pairing status of the males as well as the dates of pair formation (i.e., the unobserved switches of states). We obtained results on pairing success and date of pair formation that were consistent with our data from mist netting and with the results of earlier studies. We also found that the peak in nocturnal and dawn singing activity was after the period of female arrival and was later in the season in unpaired males than in paired males. On the basis of our analyses, we argue that change-point models are powerful analytical tools for many fields of research and can be used whenever animals abruptly switch behavioral routines.