Tidewater glacier fronts are an important foraging ground for an Arctic marine predator

In nature, food resources are distributed heterogeneously, and understanding where and how animals utilize these resources has long been a fundamental question in ecology¹. Foraging is an ecologically critical behavior, as it is a major determinant of species’ energetic efficiency, reproductive success, and survival rate². Since the foraging ecology of animals is shaped by external factors such as prey availability and environmental conditions, recent environmental changes have had cascading effects on the foraging ecology of top predators^3,4,5. However, unlike terrestrial animals, whose behaviors can be directly observed in the natural environment, the interactions between marine mammals, the environment, and prey species are poorly understood, making it difficult to predict the impact of climate change on marine mammals.

During the open-water season in the Arctic, brownish murky spots that spread in areas where glaciers meet the sea are commonly observed^6,7. At the front of tidewater glaciers, the subglacial discharge of low-density buoyant freshwater creates upwelling water flows along the calving front. This strongly upwelling glacial meltwater, typically released into the water at a depth of several hundred meters, delivers nutrients, plankton, and sometimes fish from the bottom of the sea to the surface^8,9,10,11. This process creates biomass-rich water zones at glacier fronts and provides easily available prey for predators, making these areas forage hotspots during the summer months¹². Indeed, large gathering events of marine predators—including seabirds, seals, whales, and in some subpopulation, polar bears—have been observed at glacier fronts in various parts of the Arctic^{8,13,14,15,16,17}, including our study area, Inglefield Bredning^18,19. Biologging studies have further supported this hypothesis, showing the frequent diving behavior of predators near glaciers, which is likely associated with foraging^{8,16,20,21,22,23}.

However, it is important to emphasize the considerable lack of direct evidence to confirm actual foraging activities or identify their diet in these areas. Although biologging studies allow swimming behavior observations, determining whether it constitutes foraging or whether foraging has been successful is difficult²⁴. Furthermore, although acoustic monitoring can be used to observe the feeding behavior of cetaceans that produce foraging sounds, this technique is not applicable to pinnipeds, as they do not echolocate²⁵. Therefore, two important questions remain: how important are glacier fronts as foraging grounds for seals, and what are they eating there? The aim of our study was to fill these knowledge gaps through a spatial analysis of ringed seal (Pusa hispida) stomach contents.

Ringed seals are ice-associated seals and the most abundant pinnipeds in the Arctic²⁶. They play a crucial role in Arctic marine ecosystems as predators of various fish and invertebrates and as the main prey for polar bears (Ursus martimus)²⁷. Moreover, ringed seals are the most frequently hunted marine mammals in the Inuit communities of northern Greenland for various purposes—such as food, sled dog feed, and fur—positioning them as a central species for Greenland cultural identity, food security, and socioeconomic systems. Ringed seals generally give birth in late March and April²⁸, and undergo molting from May to early July, during which they engage in minimal feeding²⁹. Both the breeding and molting seasons are energetically costly, resulting in marked body mass loss during spring; therefore, to replenish their fat deposits, ringed seals feed most intensively during summer³⁰.

Inglefield Bredning, situated in northwest Greenland (Fig. 1), is home to the northernmost Inuit communities. This fjord serves as the primary hunting ground for the local Inuit, where hunting takes place throughout the year. In most places, except the Arctic, the stomachs of marine mammals are collected from stranded individuals, making it impossible to determine their specific location and date of death. In this study, we collaborated with the Inuit communities around Inglefield Bredning to collect ringed seal stomachs with a detailed site of capture during 2022–2023.

The digestive speed of ringed seals is quite high, and their stomachs become empty within ~4 h of prey ingestion³¹. This rapid digestion rate has been regarded as a drawback because only a snapshot of feeding can be considered³². This study attempted to take advantage of this, making it possible to compare and clarify habitat use and diet within a short timeframe. Here, we performed stomach content analysis of ringed seals to investigate their diet and its relationship with glacier fronts around Inglefield Bredning. We also conducted a hydroacoustic survey to document the prey field available for seals in the region. Our study provides direct evidence that tidewater glaciers serve as vital foraging grounds for seals, and highlights the importance of these areas for the Arctic marine ecosystem.

Stomach contents of ringed seals

A total of 42 ringed seal stomachs were collected during the summers of 2022 (n = 11) and 2023 (n = 31) (Supplementary Table S1). Twelve out of the 42 stomachs (29%) were empty, and no significant difference in the proportion of empty stomachs was found between the sexes (Wilcoxon rank sum test, W = 207.5, p = 0.68) or among age groups (Kruskal–Wallis rank sum test, chi-squared  = 0.30244, df = 2, p = 0.86). Fifteen prey taxa were found in the 30 stomachs with visible prey remains (Table 1). Polar cod (Boreogadus saida) was the most dominant species in total biomass (B_i = 83.5%) found in half of all seals (Table 1). Zooplankton were the second most dominant in total biomass (B_i = 6.3%), found in 35.7% of seals (Table 1).

Distance to the nearest glacier vs diet

The median distance to the nearest glacier front from the site of capture (DNG) across all collected samples was 4 km (0.5 < DNG < 73.6 km; Fig. 2A, B), and no significant difference was found between the sexes (Wilcoxon rank sum test, W = 194, p = 0.471) and among age groups (Kruskal–Wallis rank sum test, chi-squared = 0.41795, df = 2, p = 0.811). The median DNG was significantly shorter for seals with history of feeding: 2.4 and 7.8 km for seals with prey found in their stomach and with empty stomachs, respectively (Wilcoxon rank sum test, W = 98, p < 0.05; Fig. 2B). Based on the median DNG (4 km), when all samples were divided into two groups—seals hunted closer to the glaciers and those caught farther away—seals hunted closer to the glaciers had significantly heavier stomachs (Wilcoxon rank sum test, W = 269, p < 0.05; Fig. 2C, D). The DNG had a significant effect on stomach content weight and consumed prey diversity (p < 0.01, respectively), with stomach content weight decreasing and prey diversity increasing as distance to the glacier increased (Supplementary Table S2 and Fig. 2E).

Based on thirty samples after excluding empty stomachs, seals with traces of feeding on polar cod were hunted significantly closer to the glacier than those without (Wilcoxon rank sum test, W = 1, p < 0.001), with a median DNG of 1.6 and 20.7 km, respectively (Fig. 3A, B). The significantly negative correlation between DNG and polar cod consumption is also observed (p < 0.001, R² =  0.52; Fig. 3C). In contrast to the pattern for polar cod, seals with traces of feeding on zooplankton were caught relatively farther from the glacier than those without (Fig. 3D, E), with only marginal statistical support (Wilcoxon rank sum test, W = 156, p = 0.07). A median DNG was 18.8 and 2.0 km with and without zooplankton, respectively (Fig. 3D, E), and positive correlation between DNG and zooplankton consumption was observed (p < 0.005, R² = 0.28; Fig. 3F).

Distribution of potential prey in Inglefield Bredning

Acoustic registrations on all nine transects surveyed in Inglefield Bredning (Fig. 4A) were generally characterized by a diffuse but dense layer of scatterers in the upper 50 m, occasionally extending to a depth of 100 m (Supplementary Fig. S1). The relative frequency response in the upper 50 m was ~1:1.5, between 38 and 200 kHz, indicating the predominance of zooplankton^33,34. This was corroborated by ground truthing, where a range of invertebrate zooplankton groups—Limacina helicina., Thysanoessa sp., Mysis sp., and Themisto sp.—and small fish larvae were observed. Zooplanktons were heterogeneously distributed across the fjord (Supplementary Fig. S1), and no consistent relationship was observed between the DNG and its abundance (Supplementary Fig. S2 and Supplementary Table S3).

Additionally, several dense patches with stronger reflections at 38 kHz (up to 1:0.7) were observed (Fig. 4B), representing loose aggregations of small swim-bladdered fish, such as polar cod, the only pelagic fish that aggregate in large schools in this region and the high Arctic³⁵. Only three distinct dense backscatter aggregations with a high ratio of 38:200 were observed (Fig. 4A, B). These aggregations were ~3 m in height and 10–30 m in length, observed near the surface at 6–13 m depth. Two were occurred 7.4 km from the nearest glacier front and another at 7.6 km, all along the transect that achieved our closest approach to the glacier fronts (Fig. 4A).

Marine mammals, the apex predators in marine ecosystems, can be used as indicators of changes in marine environments and ecosystems³⁶. In the context of rapid glacial decline due to accelerating climate change, assessing the role of glaciers in marine ecosystems is essential for predicting their future impact. Since 1862, it has been hypothesized that glacier fronts are important foraging grounds for marine predators, but remained to be explained with evidence¹³. Here, collaboration with Inuit communities where traditional seal hunting is practiced allowed us to collect a comprehensive dataset, and provided an unique opportunity to address these vital ecological questions. By combining stomach content analysis and capture records, we provided direct evidence that glacier fronts, which are gaining increased attention owing to accelerating climate change, serve as vital foraging grounds for seals. Our study suggested that seals foraged intensively at the glacier fronts, and the prey they consumed differed between the glacier front and other areas.

The most commonly eaten prey was polar cod (Table 1). While the result was consistent with those from other high-Arctic regions, including northern Greenland³⁷, southern Greenland^37,38, Svalbard, and the Canadian High Arctic^39,40, our study revealed hotspots where polar cod were actually eaten (Fig. 3A). Seals fed on polar cod intensively at glacier fronts, the only place in the fjord where we found aggregations of polar cod (Fig. 4).

Polar cod, the most ubiquitous fish in the High Arctic, can be found in a variety of habitats, including deep waters⁴¹, shallow coastal waters^42,43, and sea ice^44,45. Glacier fronts have been suggested as potential habitats during summer^46,47; however, due to the extreme difficulty and danger of conducting observations near the glacier front with the risk of calving, these areas have been the least studied habitats for polar cod. From this perspective, stomach contents from ringed seals that easily swim very close to glaciers provide a rare opportunity to investigate the importance of glacier fronts as a habitat for polar cod. Our results showed that these areas are of high importance, which calls for future studies on the functional role of these areas in the life cycle of polar cod. The distribution of polar cod presented here is a snapshot, as it reflects data obtained over a limited survey period. Therefore, we cannot rule out the possibility of their presence in unsurveyed areas, and the temporal dynamics of their distribution remain unrevealed. To investigate the distribution of polar cod in greater detail, surveys expanded in both temporal and spatial scope are required. This is particularly important in the context of the ongoing borealization of the Arctic fish communities due to climate change, which may intensify competition for prey and habitats among fish in the region and affect the future distribution of polar cod^48,49.

While polar cod was consumed predominantly closer to the glaciers, this tendency was not observed for zooplankton (Fig. 3). Rather, individuals without zooplankton were caught relatively closer to the glaciers (Fig. 3E, F). Furthermore, consumed prey diversity increased with greater distance to the glacier (Supplementary Table S2), which is attributed to the fact that seals fed almost exclusively on polar cod near the glacier.

These differences in feeding areas may reflect the seal’s feeding preference for polar cod near glaciers, driven by the energy content, foraging costs, and distribution of each prey species^50,51. Data from satellite transmitters mounted on ringed seals in the area indicate that seals spend most of their time in water shallower than 50 m⁵², where their dominant zooplankton prey—Themisto spp., Mysis sp., and Thysanoessa sp.—are distributed^53,54. Since both polar cod and zooplankton were detected at depths shallower than 50 m in our acoustic survey (Fig. 4 and Supplementary Fig. S1), foraging costs to dive to target prey may not differ between zooplankton and polar cod. Moreover, although the energy content per gram of polar cod and zooplankton is similar (Supplementary Table S4), the size difference between these prey types makes it more efficient to eat one polar cod than to eat numerous zooplankton (e.g., one polar cod provides an equivalent energy of >30 Themisto sp. or 1000 copepods⁵⁵). Therefore, near the glacier fronts, which may have a relatively abundant polar cod biomass, it would be suboptimal to spend time and energy on foraging zooplankton. In contrast, in areas where the polar cod biomass is not abundant, seals may feed on zooplankton or other available prey species instead. Hence, it is reasonable that seals consumed polar cod considerably closer to the glacier fronts, whereas zooplankton were relatively less consumed in the area. Our study supports the hypothesis that marine mammals target aggregations of fish for high energy potential with low foraging costs, as has been suggested in many studies^42,52,56,57. Our study also suggests that the results of diet composition vary depending on the sampling location, which should be considered in future dietary research. For example, samples collected only near the glacier fronts are likely to show a dietary bias towards polar cod.

Our findings were obtained by leveraging a traditional limitation of stomach content analysis— the fact that it captures only a snapshot of feeding. A remaining limitation that must be considered is that, while our approach could reveal the horizontal patterns in feeding ecology, it does not capture vertical structure. This gap could be addressed by integrating biologging as a mutually complementary approach, yielding a more robust depiction of feeding ecology that links horizontal feeding space with three-dimensional habitat use.

Global warming causes Arctic glaciers to rapidly melt^58,59,60,61, resulting in increasing amounts of freshwater into the ocean^62,63. The glaciers at our study site have also experienced increased melting during recent decades^64,65, and several glaciers have reduced in size to the extent that they no longer reach the sea. Unlike tidewater glaciers, where low-density, buoyant meltwater is released from the glacier base and drives upwelling plumes along the calving front, meltwater from land-terminating glaciers does not produce upwelling and instead forms a stratified surface freshwater layer⁹. Consequently, glaciers have markedly different effects on marine productivity depending on whether they extend into the sea or retreat onto land⁹. Indeed, seals prefer marine- to land-terminating glaciers, even if they are adjacent to each other²³. If tidewater glaciers further retreat beyond the coastline, important foraging grounds for marine mammals disappear and may consequently alter their habitat use. Here, we focused on ringed seals, as they are the most abundant pinnipeds in the Arctic and one of the most important species for local communities and marine ecosystems. To further investigate the importance of glacier fronts for Arctic marine top predators, it is necessary to study how other marine mammals, such as bearded seals (Erignathus barbatus), harp seals (Pagophilus groenlandicus), and narwhals (Monodon monoceros), utilize these areas. The role of glacier fronts under climate change may vary between species depending on their patterns of use and intensity⁴⁶.

Our spatial analysis of stomach contents contributes to our understanding of diet, habitat use, foraging strategies, and prey distribution, and providing direct evidence of the importance of tidewater glacier fronts as foraging grounds for a marine predator and their diet in this spot. Our results suggest that glacier retreat induced by recent warming may reduce important foraging habitat for seals and may lead to changes in their diet, behavior, and distribution. It could also influence polar bear distribution, given that ringed seals are their primary prey^17,27. As a next step toward a deeper understanding of the effects of glacier retreats on marine ecosystems, it is necessary to expand research both geographically and temporally. Given that Arctic glaciers are changing rapidly on annual timescales, continued long-term monitoring is essential to capture and quantify their spatiotemporal dynamics. Moreover, in view of regional variation in seal diets^37,40,66, broader geographic sampling and cross-regional comparisons will be important. Although no age or sex differences in DNG were found in this study, the sample size may not be sufficient to draw firm conclusions regarding such differences. Further investigation will be needed to clarify the effects of glacier retreat across age and sex classes.

In conclusion, we would like to emphasize the importance of collaborative research with local people, which made our study possible. Such collaboration facilitates the collection of data that would be difficult for scientists alone to obtain—both quantitatively and qualitatively, and provides opportunities to learn more about the surrounding natural environment through their knowledge and experience⁶⁷. A key feature of our study is the collection of stomach samples with verified capture locations, which was uniquely enabled by the cooperation of Inuit hunters.

Fieldwork

We collaborated with Inuit hunters from the communities of Siorapaluk, Qaanaaq, and Qeqertat in northwest Greenland to sample the stomach contents of ringed seals between August and September of 2022 (n = 11) and 2023 (n = 31) (Fig. 1). During the open-water season, daily hunting activities are conducted by boats throughout the fjord. Seal stomach contents were sampled by hunters or biologists, and the site of capture for each seal was recorded by the hunters. Sex was recorded based on the reproductive organs. The esophagus and duodenum were closed, and whole stomachs were carefully removed from the carcasses. We examined stomachs within 24 h after the animals were caught without freezing, as this may destroy prey remains and make it difficult to identify the species. We also collected canine teeth on the lower jaws for age estimation. Each jaw was boiled for approximately 10 min, after which the teeth were extracted. The teeth were thoroughly washed and stored at room temperature until further processing. Our study was approved by the Ethics Committee of Kyoto University, Japan.

Stomach content analysis

The collected stomachs were cut open, and their contents weighed before washing with a series of 4.75, 1.7, and 0.5 mm mesh sieves. Fleshy pieces and hard parts remaining on the sieve, including cephalopod beaks, otoliths of fish, skeletons of invertebrates, and crustaceans, were collected and rinsed carefully with water. The collected items were sorted, counted, and identified to the lowest possible taxonomic level using identification reference ^68,69,70 and a collection of otoliths from fish caught in a survey around Greenland provided by the Greenland Institute of Natural Resources, Nuuk, Greenland. During species identification, all examinations were performed under a stereomicroscope (Micronet, YS03C) using a micrometer.

The collected otolith lengths with minimal or no erosion were measured using ImageJ software (National Institute of Health) along their longest axis to the nearest 0.01 mm^71,72. From the measured otolith length, the original wet mass of fish were calculated by using regressions from previous studies (Supplementary Table S4). For other species with no regression available, we used either the mean mass from Finley et al.⁷³ or the mean mass of all specimens of that species measured in this study (Supplementary Table S4). Fish numbers and total biomass were determined by dividing counted otolith numbers by two, because teleost fish have two sagittal otoliths. Based on the number of each prey species found and the mass of each specimen measured, the (1) frequency of occurrence (FO_i %) and (2) relative frequency of occurrence by biomass (B_i %) were calculated for each prey item (i) to examine the dietary composition and importance of each species as follows:

where S_i and S_t represent the number of seals that consumed prey type i and the total number of seals, respectively, and

where b_i and b_t represent the total biomass of prey type i consumed by seals and the total biomass of all estimated prey, respectively.

The diversity of prey types eaten by each individual was calculated using Simpson’s diversity index based on the number of prey, following Chambellant et al.⁶⁶ as follows:

where n_i represents the number of individuals of prey type i, and n_t represents the total number of prey found from the stomachs. This index yields a value of 0 when the individual feeds on only a single type of prey, and increases with both the number of prey types and the evenness with which they are consumed⁷⁴. Therefore, larger values suggest that the individual preys on a greater variety of prey.

Age estimate analysis

The age of each seal was estimated by counting the number of growth layers in the cementum of the canine teeth. Before the analyses, the teeth were decalcified with 5% HNO₃ solution for at least 48 h, then kept in water for 24 h after decalcification. Decalcified teeth were cut into 14-μm slices with a freeze-microtome (Leica CM 1520 or Yamato Kouki FX-801) at −20 °C. The sliced teeth were carefully arranged on a glass slide and stained with toluidine blue/NaHCO₃ solution. The number of layers in the cementum was counted to estimate the age of each seal, following Stewart et al.⁷⁵. Ringed seals were then grouped into three age classes as follows: pup; 1-year individuals born in the spring; juvenile; 1–5 years old; adult; sexually mature individuals 6 years and older⁷⁶. The ages of four individuals were recorded as unknown because we could not collect teeth from them (Supplementary Table S1).

Spatial and statistical analysis

All geographic information layers were created using the open geographical tool QGIS (version 3.16.7). The capture sites were mapped, and the shortest distance from each to the nearest glacier was calculated (Landsat 9 September 7, 2023).

All statistical analyses were performed using R version 4.3.2. A generalized linear model based on gamma distribution (log link) was performed on stomach contents weights and consumed prey diversity to examine the potential effects of age, sex, date and the distance to the nearest glacier from site of capture (DNG). Since DNG was the only variable with a significant effect (Supplementary Table S2), subsequent analyses focused on this factor.

First, we evaluated the normality of the data distribution of stomach contents weight and DNG (km) using the Shapiro–Wilk test. As the results indicated that the data did not follow a normal distribution (W = 0.697, p < 0.001), non-parametric tests were applied for further analysis. A Wilcoxon rank sum test was performed to investigate whether DNG (km) differed between male and female. A Kruskal–Wallis rank sum test was performed to explore the difference among age classes (pups, juveniles, and adults). As the preliminary analysis revealed no sex and age differences in DNG (km), all individuals were grouped together in further analyses.

To examine the importance of glacier fronts as feeding grounds, we compared the DNG between empty stomachs (n = 12) and stomachs with prey (n = 30) (Wilcoxon rank sum test). We also examined whether stomach content weight differed between two groups of seals, divided all samples at the median DNG (4 km): those caught closer to the glacier and those caught farther away (Wilcoxon rank sum test). To examine where and what seals were feeding on, we excluded empty stomachs and compared the DNG between the presence/absence of two major prey in the stomach, polar cod and zooplankton (Wilcoxon rank sum test). To investigate the association between the DNG and the consumption of polar cod and zooplankton, we fitted a linear regression between the DNG and the biomass (g) of polar cod and zooplankton found in the stomach.

Acoustic survey

We conducted exploratory survey between 1 and 12 September 2023 at Inglefield Bredning. Acoustic data were collected from a depth of 5 to 700 m using a Simrad EK80Portable (Simrad ES38-18/200-18C, produced by Kongsberg Maritime) with a dual-frequency transducer mounted on a pole at the side of a 7-m dinghy at an approximate depth of 1 m. The transducer emitted 38 and 200 kHz continuous wave pulses every second, with a pulse duration of 1.024 ms and a beam opening angle of 18°. The 38 kHz signal was received as a split beam with three sectors, whereas the 200 kHz signal was received as a single beam.

For practical reasons, data were collected only during the day under good weather and sea conditions. The survey was designed in consideration of local concerns that active acoustic measurements might disturb marine mammals. To minimize the duration of the operation while ensuring representative coverage of the various parts of the fjord, the survey was segmented into a series of 9 transects with a speed varying between 2.5 and 5.0 kts. The specific locations chosen in consultation with local hunters, given that many areas were difficult to approach due to numerous icebergs. The acoustic data were scrutinized using MAREC LSSS v.2.16.0, in which the fish and zooplankton layers were detected using the relative frequency response between the two frequencies. Swim-bladdered fish have a higher acoustic backscatter from the 38 kHz signal than the 200 kHz signal, whereas the opposite is true for zooplankton⁷⁷. Therefore, we used a threshold ratio of 1:1 (38:200 kHz) between the two to assign the acoustic layers to one of the two groups. Calibration of the echo sounder was performed off-site in Øresund, Denmark at a depth of ~6 m using a 38.1 mm tungsten carbide sphere according to standard methods⁷⁸.

Visual ground truthing of the acoustic registrations was performed 1–2 times per day in areas with high densities of acoustic back-scatterers. The equipment for ground truthing included cameras mounted on a remotely operated vehicle (Trident from Sofar Technology) down to a depth of 100 m and a mini-camera (Waterwolf underwater fishing camera) on a fishing line that could be deployed down to a depth of 150 m.

Scattering intensity was extracted from the acoustic data and used as an indicator of potential prey abundance. To investigate the relationship between prey distribution and distance to glacier fronts, a generalized additive model assuming a normal distribution was applied, with scattering intensity as the response variable and the distance to the nearest glacier from the observation site as the explanatory variable.

Ethics

All samples were obtained from ringed seals harvested during legally permitted Inuit subsistence hunting in Greenland. Sampling was conducted by Inuit hunters as part of their regular subsistence activities, and samples were subsequently provided to the researchers. The research was overseen by the Animal Research Ethics Committee of the Wildlife Research Center, Kyoto University, Japan. Relevant permissions and official confirmations regarding sample collection and handling were obtained from the Government of Greenland (Nanoq–ID nr.: 23210863) and the Ministry of Environment of Denmark (MIM ID nr.: 44261).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.