Thermal vulnerability of sea turtle foraging grounds around the globe

The study addresses how climate change may impact key foraging habitats of highly mobile marine megafauna, specifically sea turtles, and evaluates the thermal stability of these habitats under future ocean warming. With marine biodiversity facing a crisis—over 1500 marine species are listed by the IUCN as threatened and many extinctions already recorded—yet fewer than 8% of marine taxa assessed, projections suggest that by 2100 more than 90% of assessed marine species will face increased extinction risk and habitat contraction. For wide-ranging species like sea turtles, conservation assessments have largely focused on nesting sites, providing fragmented insights into overall population status. Foraging conditions critically determine reproductive output because accumulation of fat at foraging grounds fuels breeding migrations and nesting, potentially influencing population trends across regions. As foraging productivity is tied to thermal conditions, climate change could degrade habitat quality and carrying capacity, impacting body condition and reproduction. This work compiles a global database of satellite-tracked adult sea turtles to delineate foraging grounds for all seven extant species and evaluates the thermal novelty these habitats will face by 2100.

Prior work has widely used satellite telemetry to track sea turtles, but global syntheses of foraging habitats and integrated threat assessments remain limited. Sea turtle assessments often focus on nesting beaches, missing offshore and neritic foraging dynamics. Thermal conditions influence prey availability and turtle body condition; altered climate may reduce foraging habitat quality and productivity. Existing frameworks such as Regional Management Units (RMUs) help structure conservation, yet an international action plan that integrates foraging habitat information across multinational ranges has been delayed. Studies have documented temperature-driven movements (e.g., hawksbills shifting to deeper, cooler waters during warm periods; olive ridleys moving northward during El Niño; loggerheads shifting to cooler foraging areas) and species-specific dietary sensitivities (e.g., green turtles’ reliance on seagrass; hawksbills’ spongivory), suggesting differential vulnerability to warming. Calls for improved ocean spatial planning and expansion of effective MPAs for marine megafauna underscore the need for comprehensive, climate-smart conservation that includes offshore habitats.

Data sources and selection: The authors conducted a literature search (Google Scholar; terms: “sea turtle” or “marine turtle”, “satellite telemetry”, “foraging”) spanning 1982–2020, including peer-reviewed and gray literature. Publications were retained if they tracked adult turtles between foraging habitats and breeding sites. Only complete tracks (arriving at foraging locations) were included, resulting in 1035 individuals and 4817 foraging location points across seven species and 54 of 58 sea turtle RMUs. For pelagic foragers, foraging sites were inferred from clustering of transmissions and directional changes. Georeferencing and digitization: Maps from sources were georeferenced within a GIS by aligning identifiable control points to a georeferenced base, adjusting position/rotation/scale, and assigning coordinates. From each track, at least one foraging point was identified; multiple points were extracted for complex tracks with multiple stops or pronounced trajectory shifts. Delineation of foraging hotspots: Points were clustered by species and RMU. Kernel density estimation (ArcGIS v10.6.1, Spatial Analyst) with region-specific bandwidths (variable-bandwidth approach) produced utilization distributions; the 50% isopleth defined high-use foraging hotspots. Analyses used a cylindrical equal-area projection. Proximity to coastline: The Euclidean distance from each hotspot centroid to the coastline was calculated using NOAA’s GSHHS shoreline dataset. Protection coverage and high seas: Hotspot polygons were overlaid with the World Database on Protected Areas (WDPA) to quantify MPA coverage. Hotspot centroids were overlaid with EEZ boundaries (marineregions.org) to identify those in high seas. Thermal novelty analysis: Daily sea surface temperature (SST) data were obtained from CMIP6 GFDL-CM4 at 0.25° resolution. Baseline period: 2000–2014; future: 2085–2100 under SSP585. Monthly minima and maxima SST time series were computed for each hotspot cell for baseline and future. Thermal novelty (TNo) was quantified using the Hellinger distance (HD) between baseline and future monthly distributions, averaged across min and max SST per grid cell, then averaged across cells within each hotspot. HD ranges 0–1, with ≥0.5 indicating moderate novelty and ≥0.8 indicating near-alien conditions. The percentage of total foraging area experiencing novel conditions was estimated. Statistics: Distance-to-coast differences among species were tested with Kruskal–Wallis; pairwise comparisons used Dunn’s method. The relationship between foraging area size and distance to coast used Spearman correlation on log-transformed area; linear regression examined area vs. latitude. Analyses were conducted in R v4.3.2; mapping in ArcGIS v10.6.1. Statistical significance was set at p<0.05.

• Global delineation: 133 foraging hotspots for seven sea turtle species were identified between 50°N and 40°S from 4817 foraging locations (n=1035 individuals).
• Proximity to coast: 79/133 hotspots lie within 100 km of shore. Distances varied significantly among species (Kruskal–Wallis H=55.133, p<0.01). Green turtle hotspots were significantly closer to coast than leatherback and olive ridley (Dunn’s pairwise p<0.01). Leatherbacks (mid-Atlantic) and loggerheads (NW Pacific) had the most distant hotspots.
• Area–distance relationships: Foraging area size increased with distance from coast (p<0.01). Only 14/133 hotspots were in the high seas, yet these represented 32.6% of total foraging area, including 47.6% of leatherback and 36.7% of loggerhead foraging surfaces.
• Protection coverage: Only ~2% of total foraging hotspot surface falls within existing MPAs; 57% of total surface is essentially unprotected (<5% MPA overlap). Flatback hotspots had comparatively higher protection (36% of surface with >50% MPA coverage); all other species had minimal overlap (0–7%). One-third of hotspot coverage is in high seas, where conservation is challenging.
• Thermal novelty by 2100: 68.6% of total foraging habitat area is projected to experience novel SST conditions under SSP585 by 2085–2100. TNo highest near the equator, decreasing toward the poles; no clear longitude trend.
• Species-specific thermal novelty (median [range] TNo and extent): • Hawksbill: 0.66 [0.31–0.98]; 83.1% with TNo≥0.5; alien (TNo>0.8) in 4.8% (Pacific and NE South America). • Olive ridley: 0.63 [0.38–0.92]; 98.2% with TNo≥0.5; alien in 42.2% (notably NE South America coasts). • Green: 0.54 [0.33–0.86]; 68.9% with TNo≥0.5; alien in 16% (Pacific and Indian Oceans). • Leatherback: 0.53 [0.17–0.90]; 69.2% with TNo≥0.5; alien in 11.1% (Pacific and Indian Oceans). • Flatback: 0.52 [0.42–0.70]; 71.2% with TNo≥0.5; no areas with TNo≥0.8. • Loggerhead: 0.45 [0.29–0.87]; 28.2% with TNo≥0.5. • Kemp’s ridley: 0.27; 0% with TNo≥0.5.
• Spatial constraints: Some subtropical hotspots (e.g., northern Mediterranean loggerhead sites) may be unable to shift poleward due to land barriers, making even moderate SST changes impactful.

The study provides the first global delineation of adult sea turtle foraging hotspots and demonstrates that these key habitats are largely unprotected and thermally vulnerable under end-century warming. The findings support conservation planning goals, including expansion of MPAs toward 30% ocean coverage by 2030, by identifying priority foraging areas—including remote, high-seas sites—that are currently neglected. Novel or alien thermal conditions are projected for most hotspots, especially in tropical regions where many hawksbill and olive ridley foraging sites occur, indicating elevated climate risk. While turtles exhibit physiological and behavioral mechanisms (e.g., deeper diving, nocturnal foraging, migration) that can buffer thermal stress, site fidelity and geographic constraints can limit adaptive redistribution. Dietary specialization further modulates sensitivity: green turtles (seagrass-dependent) and hawksbills (spongivorous) may be more vulnerable to climate-driven changes in prey/habitats than omnivores (flatback, loggerhead, Kemp’s ridley, olive ridley). The pronounced gap in MPA coverage—particularly offshore—highlights the need for conservation approaches that transcend coastal bias, include high-seas governance solutions, and integrate dynamic climate risks to ensure resilience of critical foraging habitats. Incorporating these mapped hotspots into climate-smart spatial planning could enhance protection efficacy for sea turtles globally.

This work compiles a comprehensive global dataset of sea turtle adult foraging grounds and quantifies their exposure to future thermal novelty, revealing that most hotspots are unprotected and many will face novel or alien SSTs by 2100. One-third of hotspot area lies in the high seas, emphasizing the necessity for international, coordinated conservation frameworks and MPA expansion that include offshore regions. The study provides actionable spatial products to guide climate-smart, systematic conservation planning that integrates both terrestrial (nesting) and marine (foraging) habitats and accounts for ocean three-dimensionality. Future research should incorporate additional climatic variables and multiple climate models to bracket uncertainty, improve positional accuracy by using original tracking data, and complement telemetry with alternative data sources (aerial surveys, strandings) to capture missing foraging areas and refine prioritization.

• Data coverage: Not all nesting sites or populations have been tracked; individuals from the same nesting site may use different foraging areas, so some hotspots are likely missing.
• Inclusion criteria: Some known foraging areas were excluded because available information did not meet methodological criteria.
• Georeferencing precision: Foraging locations were digitized from published maps with varying accuracy; spatial precision depends on source map quality and georeferencing.
• Climate inputs: Thermal novelty assessments used only SST (monthly min/max) from a single CMIP6 model (GFDL-CM4) at 0.25° under SSP585; other variables/models and scenario uncertainty were not explored.
• Behavioral and ecological complexity: Potential adaptive behaviors, prey dynamics, and trophic interactions were not explicitly modeled beyond SST-driven novelty.