Survival probability estimates for immature green turtles Chelonia mydas in the Bahamas

Green turtles Chelonia mydas in the Atlantic undergo a complex series of developmental migrations, moving long distances among foraging grounds. The study of the ecology and demography of these foraging populations is hampered by a lack of estimates of survival probabilities for immature stages. In addition, for all species of sea turtles, estimates of survival probabilities have repeatedly been identified as the greatest requirement for improving models of population dynamics and plans for the management and conservation of these endangered and threatened species. We use data from long-term capture-mark-recapture studies of 2 sampling populations of immature green turtles in the Bahamas—Union Creek, Great Inagua from 1978 through 2001, and Conception Creek, Conception Island from 1989 through 2001—to generate estimates of annual survival probabilities. We employ a joint analysis of live-recapture and dead-recovery data (Burnham model) to generate estimates of true survival and permanent emigration probabilities for the Union Creek sampling population. Because fewer dead-recovery data were available for the Conception Creek sampling population, we used a Cormack-Jolly-Seber modeling approach to estimate an apparent annual survival probability of 0.680, which confounds mortality and permanent emigration. Our best estimate of true annual survival probability for Union Creek green turtles, protected from humaninduced mortality, is 0.891. High survival probabilities in immature stages are necessary for species such as sea turtles with long lifespans and late sexual maturity to maintain stable populations. After green turtles emigrated from Union Creek, annual survival probability declined to 0.761 as the turtles were no longer protected from human-induced mortality, demonstrating the negative effect of this mortality on the future of green turtle populations in the Caribbean. This study reports the first application of the joint analysis of live-recapture and dead-recovery data to sea turtle populations and demonstrates the advantages of this modeling approach.