To breed or not: a novel approach to estimate breeding propensity and potential trade-offs in an Arctic-nesting species

摘要

Breeding propensity, i.e., the probability that a mature female attempts to breed in a given year, is a critical demographic parameter in long-lived species. Life-history theory predicts that this trait should be affected by reproductive trade-offs so that the probability of future reproduction should depend on the current reproductive investment. However, breeding propensity is one of the most difficult parameters to estimate because nonbreeders are often absent from the breeding area, thereby requiring the inclusion of unobservable states in the analysis. We developed a new methodological approach by integrating a robust design sampling scheme within the multi-event capture-recapture framework. Our new model accounted for uncertainty in state assignation while allowing for departure of individuals between secondary sampling occasions. We applied this model to a long-term data set of female Greater Snow Geese (Chen caerulescens atlantica) to estimate breeding propensity and to investigate potential reproductive costs. We combined resightings during the nesting stage and recapture at the end of the breeding season to estimate breeding propensity and nesting success, and added recoveries to improve survival probability estimates. We found that both breeding propensity and nesting success depended upon breeding status in the previous year, though not survival. Successful breeders had a lower breeding propensity than failed breeders in the following year, but a higher nesting success. Individuals absent from the breeding colony had a low breeding propensity, but a high nesting success the following year. Our results suggest a cost of reproduction on breeding propensity in the next year, but once females decide to breed, nesting success is likely driven by individual quality. An added benefit of our model is that, unlike previous models with unobservable states, all parameters were identifiable when survival and breeding probabilities were fully state dependent. Our new multi-event framework is a flexible tool that can be applied to a large range of species to estimate breeding propensity and to investigate reproductive trade-offs.