Fig 1.
For most sites, observed abundance decreases with size class.
We observe that juveniles are most common, followed by subadults, and adults. New recruits are rare. Notable exceptions, are the dominance of green turtle adults on Rota (37% of surveyed turtles), Rose Atoll (36%), and Oahu (49%), and adult hawksbills at Palmyra Atoll (54%). Though low-level nesting can be widespread, for both species, only Rose Atoll and Pearl & Hermes (underlined labels) are documented nesting sites (> 30 nesters per season) in these regions. For green turtles, remote Wake Atoll had the lowest proportion of adults (5%), Hawaii Island had the largest proportion of new recruits (9%). Ta’u had the most hawksbill new recruits (7%). Smoothed histograms show the proportion of turtles in each size class, calculated from diver estimates of straight carapace lengths. Histogram and label colors saturate as size class increases. Labels give the percentage of each class, with the most abundant size classes in bold. Only sites with > 20 turtles of a given species observed over all survey years were analyzed.
Fig 2.
Densities and distributions of sea turtles in the U.S. Pacific Islands.
The greatest sea turtle density is in the (A) Pacific Remote Island Areas (268 turtles per 1000 tow segments), followed by (B) the Mariana Archipelago (127), (C) American Samoa (93), and (D) the Hawaiian Archipelago (27). Jarvis Island had an astonishing density of 844 turtles; more than twice the density of any other site. Negative binomial models were consistently the highest-ranked across all sites, best describing the zero-inflated, heavy-tailed data. Histograms are shaded according to the modeled site-level density. Where the above panels model the densest turtle populations in each region, (E) maps the densities for all sites. The protected and remote Northwestern Hawaiian Islands have a surprisingly low sea turtle population densities. Squares show sites shaded by modeled density per 1000 tow segments (10 m x ~220 m).
Fig 3.
Predicted number of individual turtles detected in surveys varied by species and region.
Density models predicted > 14 times as many green turtles (576) as hawksbills (41) predicted across all regions: the Pacific Remote Island Areas (PRIA), American Samoa (AMSM), the Mariana Archipelago (MARI), and the Hawaiian Archipelago (HIIS). PRIA had the greatest number of green turtles (219), followed by MARI (193), HIIS (82), and AMSM (82). While densities were far greater for PRIA than MARI, number of green turtles were similar. For hawksbills, AMSM had the greatest number (15), followed by MARI (13), PRIA (11) and HIIS (2). Higher abundance of detected green turtles in the PRIA and hawksbills in AMSM occurred despite the relatively few number of sites surveyed for each of these regions. Correspondingly, HIIS showed low numbers despite more sites, substantially larger available habitat area, and a greater total survey distance than all other regions (HIIS: 772, MARI: 364, PRIA: 286, AMSM: 257 km). We multiplied densities per site (Fig 2) by average number of tow segments per annual survey to calculate predicted individuals per site, then summed all site values for each region for aggregate predicted individuals. Green turtles are shown in green, and hawksbills in orange.
Fig 4.
From 2002–2015, regional green turtle densities were stable or increased.
Box plots (mean, 1 sd, min-max) show fitted density estimates for sites in each region, by year. All regions show growth from 2002–2008, declines from 2008–2012, and growth thereafter. From these times series, stochastic exponential growth models calculate rates of change. We list the fitted parameters. Population growth rate was inversely correlated with population size, which though not unexpected, may indicate density dependence. Black line is a LOESS model with standard error (gray band).
Fig 5.
SST and productivity are the highest-ranked drivers of sea turtle densities.
We used RF models to assess how SST, ecosystem productivity, available habitat, and human impacts affected sea turtle densities. (A) Scatter plots of single factors reflect expected relationships. Densities show a thermal envelope peaking at 27.5°C, increase with ecosystem productivity, decrease with an index of human impact, and surprisingly show little effect from habitat area. Filled circles are site densities, colored lines are LOESS fits (gray bands are standard error). Results from the RF model formalize these relationships in (B) ICE plots and (C) two variable PDPs. Shaded polygons represent the predictor space, while filled circles show the observed density values. The modeled habitat-density effect now appears positive, but noisy. Partial ŷ values assess the average change in the predicted value from each model factor, thin gray lines represent each site, and colored lines are the mean relationship. For some sites, (C) the negative influence of human impacts to turtle density appears mitigated by optimal SST. (D) Model variables are ranked according to the percent increase in MSE of removing each variable. The average was taken across each iteration of a LOOCV. Historical overexploitation, conservation efforts, and spatial population structure likely contribute to the unexplained error, and quantifying these factors may improve model performance.