Environmental stress reduces shark residency to coral reefs

Coral reefs have undergone significant declines in coral cover globally over the past two decades due to disease, cyclones, and bleaching. Such bleaching events can increase mortality, reduce coral cover and structural complexity, and alter biodiversity, community composition, and ecosystem function. Many shark species, including grey reef sharks, rely on coral reefs for feeding, breeding, and social refugia, implying that climate-driven habitat change could strongly affect predator behavior and population vulnerability. Reef sharks often show routine habitat use and residency—largely uninterrupted occupancy of a limited area over time—which is tied to habitat quality, trophic interactions, and population persistence. However, the environmental drivers of residency, particularly environmental stress, remain poorly understood, despite implications for population dynamics, predator–prey landscapes, nutrient transfer, dispersal, and conservation. Grey reef sharks (Carcharhinus amblyrhynchos), an Endangered Indo-Pacific reef-associated species, are site-attached central place foragers with predictable movements from core residency areas, making them an ideal model for acoustic telemetry-based investigation of residency. This study investigates how environmental stress on coral reefs influences reef shark residency across the Chagos Archipelago using a composite satellite-derived environmental stress exposure index that integrates cloud cover, current, depth, salinity, multiple sea surface temperature (SST) metrics, and wind. The authors hypothesize that increased environmental stress reduces reef shark residency as a behavioral adjustment to locate more suitable habitat or resources (e.g., prey, physical or thermal refugia).

Prior work documents global declines in coral cover and the ecological consequences of bleaching on reef structure and function. Reef sharks provide key ecological roles on reefs but face rising extinction risk; yet links among habitat quality, environmental variability, and movement ecology are understudied. Residency behavior is known to mediate ecological processes and depend on habitat quality, but evidence on environmental drivers for reef sharks is limited and mixed. Some studies found SST and related temperature metrics influence shark presence/residency in other species, while others reported limited effects of environmental variables on reef shark activity space. Composite indices (e.g., RESET) capture multiple stressors and their interactions, which vary regionally (e.g., SST, current, wind in Chagos; depth and SST in the Red Sea; DHW, SST, current in the Gilbert Islands). This contextualizes the need to assess multi-stressor effects on reef shark residency over long timescales.

Study area and data collection: The Chagos Archipelago (central Indian Ocean) lies within the British Indian Ocean Territory Marine Protected Area. Following strong El Niño events (2015–2016), reefs experienced widespread bleaching. An acoustic tagging program of grey reef sharks was conducted from 2013 to 2021 across five atolls (Benares, Blenheim, Peros Banhos, Salomon, Victory Bank) using an array of 54 receivers (52 used after filtering). Receivers were spaced >500 m apart (mean nearest-neighbor ~2.15 km) to avoid detection overlap. Range testing was not feasible; similar arrays report 300–500 m detection ranges. Tagging and animal handling: 122 grey reef sharks (81 female, 41 male; total length 70–159 cm; mean 117.9 cm) were implanted intraperitoneally with Innovasea V16 69 kHz coded transmitters (nominal delays 30–90 s or 60–180 s; battery life ~10 years). Capture used handlines/barbless hooks; fish were measured and handled with seawater irrigation and minimal handling time (<5 min). Tag delay differences were standardized following Jacoby et al. Receivers were serviced annually (except 2017). Data preparation and residency index: From 2013-03-01 to 2020-11-30, detections from known IDs only were retained. Filters removed: single-detection animals; detections with gaps <30 s (second detection removed); implausible transitions exceeding 6.9 m s−1 (10× minimum sustainable speed). The first 24 h post-tagging were removed to avoid capture effects. Residency was calculated per shark per month at each receiver as days detected (minimum 2 days) divided by days the receiver was active that month, enabling spatial and temporal comparisons. Environmental stress exposure (SE) index: Environmental stress was quantified using the Reef Environmental Stress Exposure Toolbox (RESET), integrating nine variables from satellite remote sensing and Google Earth Engine: cloud cover, current, depth, salinity, wind, and four SST-based metrics (SST, Degree Heating Weeks, SST anomaly, SST variability). Products were resampled by bilinear interpolation to 500 m to match receiver detection range. Variables were combined using ecological/health thresholds and fuzzy logic to yield a composite index scaled 0–1 (low–high stress). RESET values ≥0.3 indicate considerable stress at Chagos. The index monitors relative temporal changes in stress and does not directly quantify reef health. Statistical analyses: Temporal variation in environmental stress was modeled via a GAMM with environmental SE as response, numerical day (from 01/01/2014) as predictor, and atoll as random effect (mgcv). Spatial differences in stress were tested with beta regression using atoll and atoll×year interactions (betareg), followed by post hoc emmeans comparisons. Residency drivers: A GLMM (glmmTMB; binomial, logit link) modeled residency proportion (days detected/days in month) with weights = days per month. Fixed effects: combined environmental SE (scaled), season (wet: Oct–Mar; dry: Apr–Sep), year (factor), sex, total length (scaled). Random effects: individual ID and receiver ID (station). Collinearity was checked (VIF ≤1.05). Model selection used MuMIn dredge with nesting to exclude overly complex nested models. Models with ΔAICc <2 comprised the confidence set; if best model weight <0.9, model averaging was applied and relative importance computed by summed Akaike weights. Marginal and conditional R² were calculated (MuMIn). Diagnostics checked heteroscedasticity, autocorrelation, and binomial fit. A secondary analysis removed El Niño periods (2014-07-01–2016-06-30; 2018-07-01–2019-06-30) to test robustness. Lagged effects and space use: Cross-correlation functions (ccf) between monthly mean SE and residency examined lagged responses (t months). Kernel Utilization Densities (KUD) at 50%, 75%, 95% were computed (adehabitatHR) and plotted through time. Absence durations were quantified as median days between detections per individual per month and compared between El Niño and non-El Niño periods using the Brunner–Munzel test. Population-level residency distributions were summarized in sixth-month bins across the study.

Data summary: 714,810 detections from 122 grey reef sharks across 52 receivers were analyzed. Residency indices ranged 0.03–1.00 (mean 0.34 ± 0.33 SD). Environmental SE index at receivers ranged 0.03–0.60 (mean 0.22 ± 0.09). Model selection and fixed effects: Two candidate GLMMs (ΔAICc <2) indicated environmental SE, season, sex, and year as important predictors; total length had zero relative importance. Model averaging showed:

• Environmental SE (scaled): significant negative effect on residency (estimate ≈ −0.12, z = −10.48, p < 0.001), indicating decreased residency with increased stress. The trend persisted after excluding El Niño periods (estimate = −0.07, z = −4.34, p < 0.001).
• Season: wet season reduced residency relative to dry (estimate = −0.40, z = −21.73, p < 0.001).
• Year effects (vs. 2013 baseline): significant differences in most years except 2014 and 2018 (2015: 0.62, p < 0.001; 2016: 0.60, p < 0.001; 2017: 0.69, p < 0.001; 2019: 0.27, p = 0.01; 2020: 0.26, p = 0.02; 2014: 0.11, p = 0.26; 2018: 0.06, p = 0.59).
• Sex: not significant (male estimate = 0.50, z = 1.87, p = 0.06). Random effects and fit: Variance (SD) for ID = 1.90 (1.38) and station = 1.58 (1.26) on logit scale. R²m = 0.02; R²c = 0.52, indicating high variability attributable to random effects. Conditional random effect modes showed 56% (29/52) of receivers with residency significantly different from the intercept, with some receivers exhibiting increased residency relative to the global mean. Lagged effects: Cross-correlations revealed significant negative correlations between environmental SE and residency at lags t = 0 to t = +16 months (correlation coefficients −0.16 to −0.37), indicating reduced residency can persist up to 16 months after stress peaks. Space use and absence: During elevated stress, KUD areas at 50%, 75%, and 95% expanded almost immediately, suggesting more diffuse space use. Sharks were absent from forereefs for longer during high-stress periods (Brunner–Munzel P*(1235.9) = −2.8336, p = 0.0047). Reported probability that sharks remained away longer during stress was 0.4661. Temporal and spatial stress patterns: GAMM indicated significant temporal variability in environmental SE (edf 8.9, F = 209.0, p < 0.001; adjusted R² = 0.28; deviance explained 27.6%), with lows in Mar 2013, Jan 2017, Sep 2020 and peaks in May 2015 and May 2016 aligning with El Niño. The proportion of low-residency sharks increased during high-stress periods. Beta regression revealed significant atoll differences by year; e.g., Blenheim experienced less stress than Salomon and Peros Banhos (2013–2016), and Victory Bank less than Salomon and Peros Banhos in multiple years. Receivers at Benares Shoal had higher environmental SE than Victory Bank in 2016.

Findings support the hypothesis that increased environmental stress on coral reefs reduces grey reef shark residency, leading to broader and more diffuse space use and longer absences from the forereef. The detected lag of up to 16 months suggests that behavioral consequences of stress persist well beyond peak events, potentially reflecting delayed reef recovery and cascading ecological changes. Such reductions in residency may alter trophic interactions, predator–prey dynamics, and cross-ecosystem nutrient subsidies that reef-associated predators bring from offshore, potentially influencing reef resilience during stress. Conservation implications include possible increased vulnerability to illegal, unregulated, and unreported (IUU) or commercial fisheries if sharks spend more time offshore or outside MPA cores; alternatively, reduced aggregation could lower encounter rates with fishers. Spatial heterogeneity in residency responses (some sites showing increased residency) points to local factors such as reef resilience, seabird-derived nutrient subsidies (notably higher around rat-free islands), and hydrodynamic shelter from wave exposure that may buffer stress effects and support higher residency. While acoustic detectability can be influenced by environmental conditions (e.g., wind), the long, variable time series likely minimizes bias. The mechanisms linking environmental drivers (notably SST metrics) to behavior remain to be disentangled, and results add early evidence that multi-stressor environmental change can impact movement and residency in grey reef sharks.

By integrating 8 years of acoustic telemetry for 122 grey reef sharks with a composite satellite-derived environmental stress index, the study demonstrates that environmental stress, season, and interannual variability significantly predict reef shark residency. Increased environmental stress reduces residency, expands space use, and extends absences from reefs, with effects persisting up to 16 months after stress peaks. Spatial variability suggests some locations are more resilient, offering potential refugia. As climate change intensifies stress and bleaching frequency, these behavioral shifts could affect reef ecosystem functioning (e.g., nutrient subsidies) and shark conservation outcomes through altered exposure to fisheries. Future research should disentangle direct versus indirect effects of specific environmental variables (especially SST metrics), incorporate additional stressors (e.g., turbidity, chlorophyll-a, pH, UV) as products improve, leverage longer-term bleaching maps (e.g., Allen Coral Atlas) for causal inference, assess energetic and demographic consequences of reduced residency, and quantify interactions with IUU/commercial fishing under changing movement patterns.

• Acoustic detectability can vary with environmental conditions (e.g., wind), and no formal range testing was conducted at the site; however, the long time series and variable conditions likely mitigate bias.
• The environmental SE index is a composite proxy of stress exposure and does not directly quantify coral health; it omits some relevant stressors (turbidity, chlorophyll-a, pH, UV) due to data limitations in shallow waters.
• Some variables in the index may influence both habitat quality and shark movement directly, complicating separation of direct vs. indirect effects.
• The study did not identify the specific environmental factors driving reduced residency; mechanisms underlying immediate and lagged responses remain unresolved.
• Receiver coverage focuses on shallow forereefs; increased time offshore or deeper may fall outside detection ranges, potentially affecting observed residency.
• One annual servicing expedition (2017) did not occur, and analyses focus on five atolls within the archipelago, which may limit generalizability.