Breaking the fast: first report of dives and ingestion events in molting southern elephant seals

Molting is a widespread, energetically demanding process across taxa that involves partial or total renewal of integuments and depends on seasonal, genetic, and environmental conditions. In pinnipeds, especially phocids, molting requires elevated skin temperatures and increased peripheral blood flow, which raise metabolic heat loss, especially in cold water. Southern elephant seals (SES) undergo a catastrophic molt, traditionally assumed to be accompanied by fasting and prolonged haul-out on land to minimize thermoregulatory costs. However, SES are exposed to variable weather while ashore, and thermoregulatory costs can be substantial. The study asks whether molting female SES truly fast and remain on land, whether they enter the water during molt, and what environmental factors influence this behavior. The authors hypothesize that females would mostly remain on land fasting early in molt, that warm/low-wind conditions might induce at-sea behavior, and that only females in good body condition would spend time at sea due to energetic trade-offs.

Prior work indicates molting elevates metabolic rate in phocids, with heat loss in water greatly exceeding that in air. Elephant seals have been considered to fast and remain ashore during molt based on high thermoregulatory costs and observed mass loss. Environmental conditions (temperature, wind, solar radiation) modulate heat flux and behavior in molting seals, with social thermoregulation and habitat choice (grass beds, pools) mitigating heat loss. Stomach temperature pills (STP) and time–depth recorders (TDRs) have successfully identified ingestion (feeding/drinking) in seabirds and pinnipeds, but had not been applied to molting SES. Previous studies suggest some molting pinnipeds may drink seawater and possibly feed opportunistically; classification of ingestion types can be performed using metrics of temperature drop and recovery. Overall, the literature supports high energetic constraints during molt, the potential role of environmental drivers on behavior, and limited direct evidence for ingestion during molt in SES.

Study period and sites: Austral summer molts (December–February) from 2014 to 2022 at Kerguelen Archipelago (Surmance 2014–2019; Estacade 2020 and 2022; Port-aux-Français 2021). Ethics approvals obtained from the French Polar Institute. Subjects and instrumentation: Adult female SES were captured during molt and anesthetized (tiletamine–zolazepam, 10 mg/kg). Animals were equipped with Wildlife Computers devices: TDRs (TDR10-LX-309D; 76×55×32 mm; 125 g) and stomach temperature pills (STP; TDR-STP-207; 63×21 mm; 3 g). Argos locations (classes 3,2,1,0) were obtained from transmissions. Data processing – diving/surface swimming: Poor Argos locations precluded fine track use; dive profiles derived from depth, light, and wet/dry sensors. Seasonal decomposition of light levels (statsmodels seasonal_decompose) removed diel effects. At-sea behavior required a submerged phase, a maximum depth, and a surface phase. Surface swimming was defined as maximum depth 1–5 m with ≤20 min between submerged phases. Dives had maximum depth >5 m, with distinct descent–bottom–ascent phases. A dive cycle comprised ≥2 successive dives separated by <20 min at surface (break around 17 min). Metrics: durations, depths, counts, and frequencies of dives and surface swimming events/cycles. Stomach temperature analysis: Ingestion events were identified by sharp declines in stomach temperature below 36.5 °C (T2), from a prior stable baseline (T1), followed by logarithmic recovery to baseline sustained ≥10 min. Close events were merged; 12 anomalous events (non-logarithmic or sensor error) were removed. STP robustness required ≥3 h constant temperature; outliers filtered by online average and IQR bounds. STP drift over time was assessed by downsampling (1 per 15 min) and Kendall’s tau; mean drift was negligible (≈±0.1 °C per 10 days), so no correction applied. Ingestions were assigned to behavioral context (dive, surface swimming, on land, or unknown) using synchronized wet/dry and light sensors with the diving classification. Ingestion type classification used published thresholds: index of recovery rate (I = t0.5/(T1–T2), threshold 250 s·°C−1), area above the curve (integral >3000 s·°C), recovery time trec >35 min, and ΔT >7.4 °C to infer likely prey vs water ingestion. Hyperthermia detection on land: Mean on-land stomach temperature was computed; hyperthermia defined as values >97.5% quantile sustained >1 h. Weather covariates: Daily temperature, wind speed, and sunshine duration (min/day; used as a proxy for solar radiation due to lack of W/m² data). Statistical analyses: R 4.3.1. A centered–scaled PCA summarized swimming behavior: PC1 (59.0% variance) loaded on number of dives, number of surface swims, and mean dive duration (proxy for diving score); PC2 (23.8%) represented surface swimming duration (proxy for surface swimming score). Linear mixed models (LMM) assessed effects of molt stage at capture, BMI, and monitoring duration on PC1 (diving) and PC2 (surface swimming); individual ID as random effect. Generalized linear mixed models (negative binomial) tested effects of molt stage, BMI, and monitoring duration on ingestion counts; ID as random effect. GLMs (quasi-binomial or binomial) evaluated environmental effects on proportion of diving females and probability of surface swimming. LMMs assessed predictors (swimming score, ingestion counts, body mass/length, monitoring stage, monitoring duration) of body mass loss (kg) as a proxy for condition, with ID as random effect. Significance assessed by Type I ANOVA; results reported as mean ± SD unless otherwise noted.

• Movement and diving: Of 39 monitored females, the majority went to sea (abstract: 79% swam). Fifteen females (39%) performed dives, totaling 660 dives across 77 dive cycles; 226 surface swimming cycles and 375 surface swimming events were detected. Mean dive duration 7.7 ± 3.6 min; mean surface duration 2.2 ± 2.5 min; dive cycles averaged 1.4 ± 1.7 h. Table 1 reported maximum dive depth around 12.5 m (range 1–25 m). At-sea time averaged 49.2 ± 23.9% of monitored time, with at-sea time ranging roughly from about 90.0 to 92.8% of a day during at-sea days; 57% of dives and 66% of surface swimming occurred during daytime.
• Stomach temperature and ingestion: Mean stomach temperature at sea was higher than on land (37.1 ± 0.3 °C vs 36.5 ± 0.2 °C; Student’s t-test t = –7.3, df = 49.6, p < 0.0001). On land, 61.5% of animals (n = 24) exhibited hyperthermia events (1–3 per individual), with mean stomach temperature 38.1 ± 0.3 °C; 33 hyperthermia events lasted 110 ± 39 min; the first subsequent at-sea event occurred within 2.45 ± 2.10 h on average. A total of 87 ingestion events were recorded from 80% of individuals (n = 31), averaging 2.8 ± 2.1 per individual; 70% occurred during daytime. Of these, 12% occurred during dives (n = 10), 49% during surface swimming (n = 42), and 39% in unknown contexts (likely on land/tide zone). Based on published classification metrics, 25–53% of ingestion events were consistent with potential prey ingestion; the remainder were likely water ingestion.
• Environmental drivers: Probability of surface swimming increased with higher air temperature and decreased with higher wind speed (GLMs: temperature positive, p < 0.001; wind negative, p = 0.03). Proportion of diving females decreased with wind speed (p < 0.001) while temperature had no significant effect on diving proportion (p = 0.64). Sunshine duration did not significantly initiate at-sea movements but correlated with number of ingestion events.
• Physiology and condition: Initial physiological parameters (BMI, body mass) did not significantly influence swimming score (PC1/PC2) or ingestion counts. Body mass loss did not differ between females that remained on land and those that went to sea. Among females that went to sea, higher diving score (PC1) and greater number of ingestion events were associated with lower body mass loss (Table 6: number of ingestions p = 0.03; additional significant predictors included monitoring duration p = 0.001).

Findings challenge the long-held assumption that molting SES fast and remain strictly on land. Many females engaged in at-sea behavior (surface swimming and shallow dives) during molt, with ingestion events indicating frequent drinking and occasional prey capture. Environmental conditions, particularly warmer temperatures and lower wind speeds, increased the likelihood of surface swimming, consistent with the hypothesis that seals use water immersion to dissipate heat during thermally stressful periods. Sunshine duration correlated with ingestion events, further suggesting drinking to manage hydration under heat stress. Body condition (BMI, mass) and molt stage at capture did not influence at-sea behavior or ingestion, indicating environmental rather than physiological drivers. Contrary to expectations that cold-water immersion would impose prohibitive thermoregulatory costs, body mass loss did not increase in at-sea females; among those going to sea, more diving and ingestion correlated with reduced mass loss, implying net energetic benefits. Shallow, short dives and proximity to the Kerguelen Plateau during summer suggest access to shallow prey (e.g., krill), which could supplement energy stores when molt is extended. Overall, results indicate that thermal stress management and opportunistic ingestion (water and potentially prey) are important behavioral strategies during molt, with implications under ongoing climate warming.

This study provides the first direct evidence that molting female southern elephant seals commonly enter the water, drink, and sometimes ingest prey, contrary to the prevailing paradigm of strict fasting and haul-out during molt. At-sea behavior is modulated by environmental conditions (warmer, less windy days) and does not increase body mass loss; in fact, increased diving and ingestion are associated with reduced mass loss among at-sea females. These findings warrant reconsideration of the fasting paradigm and its presumed fitness consequences during molt. Future research should: (1) identify ingested items and quantify ingestion rates using complementary sensors (e.g., video, sonar, eco-sensors), (2) directly measure energetic costs and thermoregulation during molt in air vs water, (3) assess how time spent at sea influences molt duration and subsequent foraging success, and (4) evaluate how climate-driven changes in thermal environment alter molting strategies and fitness outcomes.

• Location and behavior classification were constrained by poor-quality Argos locations, leaving 39% of ingestion events unassigned to precise context (at sea vs on land).
• Solar radiation data (W/m²) were unavailable; sunshine duration was used as a proxy, limiting precision in thermal analyses.
• Molt stage was difficult to determine consistently via visual checks, precluding analysis of how at-sea time affects molt duration.
• STP data quality issues: 12 ingestion events removed due to anomalous curves; eight seals ejected the pill before recapture; average retention was relatively short (7.2 ± 2.9 days).
• Several reported parameters exhibit textual inconsistencies (e.g., proportions and depth metrics), reflecting data/reporting limitations.
• The study focused on adult females at specific Kerguelen sites; generalizability to other populations, sexes, or regions may be limited.