Green and hawksbill turtle abundance and population dynamics at foraging grounds in Bonaire, Caribbean Netherlands

Green turtles Chelonia mydas and hawksbill turtles Eretmochelys imbricata are neg- atively impacted by natural and anthropogenic disturbances. Unknown numbers of turtles are killed annually in the coastal waters of Bonaire and Klein Bonaire, Caribbean Netherlands. We used N-mixture models, conventional distance sampling and the multiple Lincoln-Petersen method to estimate abundance from transect-count and net-capture surveys. Maximum likelihood and Bayes- ian generalised linear models were used to assess trends in annual abundance in 2003−2018, and a Bayesian state-space logistic model was developed to generate the posterior distributions of pop- ulation parameters and make abundance predictions for 2019−2030. Mean ± SE annual abun- dance was 555 ± 149 green turtles (2.5th and 97.5th percentiles = 337, 943) and 70 ± 13 hawksbill turtles (49, 101), and there were no trends in western Bonaire and Klein Bonaire in 2003−2018. Mean annual abundance was 348 ± 135 green turtles (171, 731) and there was a positive trend inside Lac Bay, southeast Bonaire, 2003−2018. Green turtles have higher population growth rate and carrying capacity, and therefore can sustain higher human-induced mortality than hawksbill turtles. However, under low mortality rates (<0.100), both species can fluctuate stably between the lower and upper limits of the carrying capacity. The methodology implemented can be adapted to estimate sea turtle abundance, monitor and model their population dynamics, and assess the neg- ative impact of human-induced mortality in other Caribbean islands.

KEY WORDS: Chelonia mydas · Eretmochelys imbricata · Abundance · N-mixture model · Distance sampling · Multiple Lincoln-Petersen method · Generalised linear model · Bayesian state-space logistic model