Green turtle somatic growth model: Evidence for density dependence

The green turtle, Chelonia mydas, is a circumglobal species and a primary h erbivore in marine ecosystems. Overexploitation as a food resource for huma n populations has resulted in drastic declines or extinction of green turtl e populations in the Greater Caribbean, Attempts to manage the remaining po pulations on a sustainable basis are hampered by insufficient knowledge of demographic parameters, In particular, compensatory responses resulting fro m density-dependent effects have not been evaluated for any sea turtle popu lation and thus have not been explicitly included in any population models. Growth rates of immature green turtles were measured during an 18-yr study in Union Creek, a wildlife reserve in the southern Bahamas, We have evaluat ed the growth data for both straight carapace length (SCL) and body mass wi th nonparametric regression models that had one response variable (absolute growth rate) and five potential covariates: sex, site, year, mean size, an d recapture interval. The SCL model of size-specific growth rates was a goo d fit to the data and accounted for 59% of the variance. The body-mass mode l was not a good fit to the data, accounting for only 26% of the variance, In the SCL model, sex, site, year, and mean size all had significant effect s, whereas recapture interval did not. We used results of the SCL model to evaluate a density-dependent effect on somatic growth rates. Over the 18 yr of our study, relative population dens ity underwent a sixfold increase followed by a threefold decrease in Union Creek as a result of natural immigration and emigration. Three lines of evi dence support a density-dependent effect. First, there is a significant inv erse correlation between population density and mean annual growth rate, Se cond, the condition index (mass/(SCL)(3)) of green turtles in Union Creek i s positively correlated with mean annual growth rates and was negatively co rrelated with population density, indicating that the green turtles were nu trient limited during periods of low growth and high population densities. Third, the population in Union Creek fluctuated around carrying capacity du ring our study and thus was at levels likely to experience density-dependen t effects that could be measured. We estimate the carrying capacity of pastures of the seagrass Thalassia tes tudinum, the major diet plant of the green turtle, as a range from 122 to 4 439 kg green turtles/ha or 16-586 million 50-kg green turtles in the Caribb ean. Because green turtle populations are probably regulated by food limita tion under natural conditions, carrying capacity can serve as a baseline to estimate changes in green turtle populations in the Caribbean since pre-Co lumbian times and to set a goal for recovery for these depleted populations . Finally, we compare the growth functions for green turtle populations in th e Atlantic and Pacific oceans. Not only does the form of the size-specific growth functions differ between the two regions (monotonic declining in the Atlantic and nonmonotonic in the Pacific), but also small juvenile green t urtles in the Atlantic have substantially higher growth rates than those in the Pacific. Research is needed to evaluate the causes of these difference s, but our results indicate that demographic parameters between ocean basin s should only be extrapolated with great caution.