Mark-recapture of the endangered franciscana dolphin (Pontoporia blainvillei) killed in gillnet fisheries to estimate past bycatch from time series of stranded carcasses in southern Brazil

Incidental fishery mortality estimates of franciscana based on stranding data are biased downwards, as only a fraction of the total bycatch ends up ashore. We estimated the probability of a franciscana incidentally killed by the coastal gillnet fisheries in southern Brazil to wash ashore and used this as a correction factor to back-calculate fishing related mortality from a dataset of carcasses collected between 1979 and 1998. From November 2005 to January 2009, 145 franciscanas incidentally killed in nets were tagged and returned to the sea. Only 11 of the tagged animals were found during beach surveys. Generalized Linear Models were used to model the probability of a tagged franciscana reaching the shore as a function of the covariates wave period, wind direction and intensity, distance from coast and the target species of the fishery. The target species had a significant effect on the stranding probability. The stranding probability of a tagged franciscana was higher in the fishery targeting white croaker (Micropogonias furnieri) (median = 0.105; 95% CI = 0.05–0.18) rather than weakfish (Cynoscion guatucupa) (0.013; 0.0003–0.069). As the stranding probability estimate for weakfish was imprecise (wide credible interval) we decided to hind cast the number of franciscanas incidentally killed for white croaker season only. The corrected estimate of franciscana mortality was approximately 10 times higher than previous estimates based solely on stranding data. Finally, this novel mark-recapture approach provides a useful correction factor to reduce the bias in incidental mortality estimates derived from stranding data.