Joint estimation of survival and breeding probability in female dolphins and calves with uncertainty in state assignment

Abstract While the population growth rate in long‐lived species is highly sensitive to adult survival, reproduction can also significantly drive population dynamics. Reproductive parameters can be challenging to estimate as breeders and nonbreeders may vary in resighting probability and reproductive status may be difficult to assess. We extended capture–recapture (CR) models previously fitted for data on other long‐lived marine mammals to estimate demographic parameters while accounting for detection heterogeneity between individuals and state uncertainty regarding reproductive status. We applied this model to data on 106 adult female bottlenose dolphins observed over 13 years. The detection probability differed depending on breeding status. Concerning state uncertainty, offspring were not always sighted with their mother, and older calves were easier to detect than young‐of‐the‐year (YOY), respectively, 0.79 (95% CI 0.59–0.90) and 0.58 (95% CI 0.46–0.68). This possibly led to inaccurate reproductive status assignment of females. Adult female survival probability was high (0.97 CI 95% 0.96–0.98) and did not differ according to breeding status. Young‐of‐the‐year and 1‐year‐old calves had a significantly higher survival rate than 2‐year‐old (respectively, 0.66 CI 95% 0.50–0.78 and 0.45 CI 95% 0.29–0.61). This reduced survival is probably related to weaning, a period during which young are exposed to more risks since they lose protection and feeding from the mother. The probability of having a new YOY was high for breeding females that had raised a calf to the age of 3 or lost a 2‐year‐old calf (0.71, CI 95% 0.45–0.88). Yet, this probability was much lower for nonbreeding females and breeding females that had lost a YOY or a 1‐year‐old calf (0.33, 95% CI 0.26–0.42). The multievent CR framework we used is highly flexible and could be easily modified for other study questions or taxa (marine or terrestrial) aimed at modeling reproductive parameters.

Abstract licence: CC BY

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.