Turtle bycatch in the pelagic longline fishery off southern Africa

Capture by pelagic longline fisheries has been identified as a key threat to turtle populations. This study is the first assessment of turtle bycatch in the South African pelagic longline fishery for tunas Thunnus spp. and swordfish Xiphias gladius. A total of 181 turtles was caught on observed sets between 1998 and 2005, at a rate of 0.04 per 1 000 hooks (0–15.5 per 1 000 hooks, SD = 1.28). Loggerhead turtles Caretta caretta comprised 60.0% of the total turtle capture and were caught at rate of 0.02 per 1 000 hooks. The second most commonly caught species was the leatherback turtle Dermochelys coriacea (33.8%), which were caught at rate of 0.01 per 1 000 hooks. Five hawksbill turtles Eretmochelys imbricata were caught at a rate of 0.001 per 1 000 hooks and three green turtles Chelonia mydas at a rate of 0.001 per 1 000 hooks. Catches were clustered, with 70% of turtles caught on 1% of sets. Apart from one set on the Agulhas Bank, on the southern coast of South Africa, all sets that caught three or more turtles were on the Walvis Ridge and on the shelf edge north of the Orange River (25°–31° S and 0°–15° E). Most of the variance in turtle bycatch was accounted for by 'vessel'. Five vessels (of a total of 50) caught 65% of turtles, at a rate of 0.4 per 1 000 hooks. The target species (swordfish or tunas) was the second most important explanatory variable; 89.5% of turtles were caught by swordfish-directed vessels at a rate of 0.15 per 1 000 hooks. Season was the third most important explanatory variable, with more turtles caught between January and June (0.13 per 1 000 per hooks) than in the remainder of the year (0.03 per 1 000 hooks), although leatherback turtles tended to be caught throughout the year. Extrapolations based on stratification by 5° grid cell, by season and by target species estimated that a total of 190 turtles was caught per year (approximately 100 loggerheads and 50 leatherbacks). Using three different techniques, the extrapolations varied between 190 and 560 turtles per year. However, if the proposed increase in fishing effort to 50 rights-holders is effected, turtle bycatch is likely to increase to about 770 turtles per year. Leatherback turtles caught by the South African pelagic longline fisheries are likely to be from the local nesting population. That population has been protected at its nesting beaches but has not recovered as expected. The overlap of turtle tracks and fishing effort suggests that the longline fishery could be partially responsible for the slow recovery.