Hawksbill turtles (Eretmochelys imbricata) migrate between nesting beaches and feeding habitats that are often associated with tropical reefs, but it is uncertain which nesting colonies supply which feeding habitats. To address this gap in hawksbill biology, we compile previously published and new mitochondrial DNA (mtDNA) haplotype data for 10 nesting colonies (N = 347) in the western Atlantic and compare these profiles to four feeding populations and four previously published feeding samples (N = 626). Nesting colonies differ significantly in mtDNA haplotype frequencies (Phi(ST) = 0.588, P < 0.001), corroborating earlier conclusions of nesting site fidelity and setting the stage for mixed-stock analysis. Feeding aggregations show lower but significant structure (Phi(ST) = 0.089, P < 0.001), indicating that foraging populations are not homogenous across the Caribbean Sea. Bayesian mixed-stock estimates of the origins of juveniles in foraging areas show a highly significant, but shallow, correlation with nesting population size (r = 0.378, P = 0.004), supporting the premise that larger rookeries contribute more juveniles to feeding areas. A significant correlation between the estimated contribution and geographical distance from nesting areas (r = -0.394, P = 0.003) demonstrates the influence of proximity on recruitment to feeding areas. The influence of oceanic currents is illustrated by pelagic stage juveniles stranded in Texas, which are assigned primarily (93%) to the upstream rookery in Yucatan. One juvenile had a haplotype previously identified only in the eastern Atlantic, invoking rare trans-oceanic migrations. The mixed-stock analysis demonstrates that harvests in feeding habitats will impact nesting colonies throughout the region, with the greatest detriment to nearby nesting populations.

Analyses of mitochondrial control region polymorphisms have supported the presence of several demographically independent green turtle (Chelonia mydas) rookeries in the Greater Caribbean region. However, extensive sharing of common haplotypes based on 490-bp control region sequences confounds assessment of the scale of natal homing and population structure among regional rookeries. We screened the majority of the mitochondrial genomes of 20 green turtles carrying the common haplotype CM-A5 and representing the rookeries of Buck Island, St. Croix, United States Virgin Islands (USVI); Aves Island, Venezuela; Galibi, Suriname; and Tortuguero, Costa Rica. Five single-nucleotide polymorphisms (SNPs) were identified that subdivided CM-A5 among regions. Mitogenomic pairwise φ(ST) values of eastern Caribbean rookery comparisons were markedly lower than the respective pairwise F(ST) values. This discrepancy results from the presence of haplotypes representing two divergent lineages in each rookery, highlighting the importance of choosing the appropriate test statistic for addressing the study question. Haplotype frequency differentiation supports demographic independence of Aves Island and Suriname, emphasizing the need to recognize the smaller Aves rookery as a distinct management unit. Aves Island and Buck Island rookeries shared mitogenomic haplotypes; however, frequency divergence suggests that the Buck Island rookery is sufficiently demographically isolated to warrant management unit status for the USVI rookeries. Given that haplotype sharing among rookeries is common in marine turtles with cosmopolitan distributions, mitogenomic sequencing may enhance inferences of population structure and phylogeography, as well as improve the resolution of mixed stock analyses aimed at estimating natal origins of foraging turtles.

Coral reef community structure has remained remarkably stable over a 10 yr period within a small protected marine area despite repeated hurricane impacts. Local community dynamics have been highly variable, however. Sites that were destroyed by disease in the 1970s are showing little or no recovery, while sites less than a kilometre away that were devastated by Hurricane Hugo in 1989 are recovering well. Strong coral recruitment has occurred in shallow, exposed areas that showed the greatest hurricane impacts, and these areas are now more species rich than in 1988, although coral cover has not reached pre-hurricane levels. Coral colony survivorship has been high throughout most of the study area. Partial mortality rates were elevated for several years following Hurricane Hugo, but significant whole coral-head mortality only occurred during periods with hurricane impacts and only at the most exposed sites. Overall, the coral community has proved resilient to closely repeated major hurricane impacts. From a single case study we cannot attribute this resilience to the relatively low level of human impacts, but grazing fish populations have apparently remained high enough to keep macroalgae in check despite the mass mortality of the herbivore Diadema antillarum in the 1980s.