Abundance, habitat use and food consumption of seabirds in the high-Arctic fjord ecosystem

To monitor the rapid changes occurring in Arctic ecosystems and predict their direction, basic information about the current number and structure of the main components of these systems is necessary. Using boat-based surveys, we studied the numbers and distribution of seabirds foraging in Hornsund (SW Spitsbergen) during three summer seasons. The average number of seabirds foraging concurrently in the whole fjord was estimated at 28,000. Little Auks Alle alle were the most numerous, followed by Northern Fulmars Fulmarus glacialis, Brünnich’s Guillemots Uria lomvia and Black-legged Kittiwakes Rissa tridactyla. The pelagic zone was exploited by some 75% of the birds. Their density was the highest (> 400 ind. km−2) in the tidewater glacier bays, where kittiwakes were predominant, and the lowest in the coastal glacier bays. The seabirds in Hornsund daily consumed c. 12.7 tons of food, i.e. c. 0.2% of the summer mesozooplankton and fish standing stocks available in the fjord. This food consisted primarily of copepods, amphipods and molluscs (c. 70%), whereas fish made up < 15%. More than 50% of this biomass was ingested by pursuit divers, while surface feeders took c. 29% and benthophages c. 13%. About three-quarters of the food biomass was taken from the pelagic zone. This paper describes, for the first time in quantitative terms, the structure and function of a seabird community foraging in an Arctic fjord. It also provides a baseline for future studies on climate-induced changes in the importance of seabirds in the Arctic food web.


Introduction
Climate-induced changes in environmental conditions are occurring very rapidly and are most pronounced in the Arctic. They are manifested by increasing air and seawater temperatures, resulting in rapid shrinking of the sea ice range and in tidewater glacier retreat (Serreze et al. 2007). The dramatically diminishing area of ice cover in both the sea and the coastal zone will have detrimental consequences for ice-associated algae, invertebrates, fish, and marine birds and mammals (Wassmann et al. 2011;Moody et al. 2012;Sydeman et al. 2012;Post et al. 2013;Barber et al. 2015;Amélineau et al. 2016;CAFF 2017;Pratte et al. 2019).
Rapid glacier recession and melting rate are substantially changing the topography of fjords, as well as their hydrology and productivity, including fish and invertebrates, which constitute an important food resource for seabirds (Dahl et al. 2003;Kumar et al. 2018;Błaszczyk et al. 2019). Changing environmental conditions in such waters modify the food resources and foraging habitats used by seabirds, which could affect the magnitudes of their populations and species composition (Kitaysky and Golubova 2000;Willis et al. 2006;Dalpadado et al. 2012;Sydeman et al. 2012;Grémillet et al. 2015;Kortsch et al. 2015;Descamps et al. 2017;Ramirez et al. 2017;Mallory et al. 2018;Renaud et al. 2018;Vihtakari et al. 2018;Hovinen et al. 2019). Nonetheless, the consequences of this process may differ. On the one hand, it may result in a population decline of some vulnerable species, such as the Ivory Gull Pagophila eburnea (Gilg et al. 2016) or Kittlitz's Murrelet Brachyramphus brevirostris, which has been found to be linked to the retreat of tidal glaciers (Kuletz et al. 2003). On the other hand, rapid glacier melting periodically creates attractive foraging spots for seabirds, and glacier recession unveils new habitats accessible 1 3 to birds, thus enhancing species diversity (CAFF 2017;Stempniewicz et al. 2017;Urbanski et al. 2017).
Arctic fjords are among the most climate-sensitive regions (Serreze et al. 2007;Kumar et al. 2018;Błaszczyk et al. 2013). The response of a fjord ecosystem including marine birds and mammals to a rapidly changing environment is a subject that has been attracting increased attention (Hop et al. 2002;Willis et al. 2006;Stempniewicz et al. 2007Stempniewicz et al. , 2017Arimitsu et al. 2012;Lydersen et al. 2014;Urbanski et al. 2017;Hovinen et al. 2019;Duarte et al. 2019). Baseline information on the abundance and distribution of seabirds is essential if we wish to identify population changes, implement efficient conservation management programmes, and assess their importance in the Arctic ecosystem's food webs (Karnovsky and Hunt 2002;Diemer et al. 2011;Renaud et al. 2018). However, current knowledge of population sizes of seabirds using fjords as foraging grounds is insufficient for most Arctic areas (CAFF 2017;Descamps et al. 2017;Vihtakari et al. 2018). When constructing models of food web/ trophic relationships in a fjord ecosystem, one usually uses numbers of local breeding birds. Such an approach, however, does not take into account the fact that some seabirds forage exclusively in the fjord, while others exploit feeding grounds in both the fjord and the open sea in proportions depending on current food availability and weather conditions Urbanski et al. 2017). Our present research has been based exclusively on surveys of seabirds actually foraging in the fjord.
To address the above shortcomings, systematic seabird transect surveys were carried out in Hornsund, SW Spitsbergen, an area of high biological production and species diversity (Węsławski et al. 2006). Hornsund, too, is experiencing the effects of strong global warming, which are manifested in the fastest retreat of tidal glaciers in the whole of Svalbard (Błaszczyk et al. 2013). The objectives of this study were: (1) to estimate the abundances of seabirds foraging in Hornsund, including their principal foraging guilds and habitats; (2) to identify their community structure, distribution and foraging habitat preferences in the fjord; (3) to estimate the biomasses of different food types taken daily by seabirds from the fjord during the chick feeding period.

Study area
Our study area was Hornsund (SW Spitsbergen), a rapidly deglaciating high-Arctic fjord. The largest seabird colonies in the area are at Ariekammen, Dotten, Krykkjestupet, Luciapynten, Skoddefjellet and Sofiekammen. Hornsund hosts one of the largest worldwide concentrations of breeding Little Auks Alle alle (ca 6,00,000 breeding pairs; Keslinka et al. 2019). There are also several big colonies of Black-legged Kittiwakes Rissa tridactyla (size category: 15-20,000 breeding pairs) and Brünnich's Guillemots Uria lomvia (5.5-10,000 pairs). However, due to inaccessibility and hidden nesting sites, the breeding population size of many common seabirds such as the Northern Fulmar Fulmarus glacialis, the Black Guillemot Cepphus grylle and the Atlantic Puffin Fratercula arctica is poorly understood. Common Eiders Somateria mollissima, Arctic Terns Sterna paradisea and Glaucous Gulls Larus hyperboreus are also common though less numerous breeders in the fjord area, with populations consisting of several dozen to several hundred breeding pairs (Norwegian Polar Institute, unpublished database).
The study area comprised both the pelagic and coastal parts of the fjord. The latter consisted of non-glaciated (NGCL) and iced coastline, including glaciers in different stages of retreat. Sea-terminating tidewater glaciers (STTG), calving intensively and receding rapidly, provide substantial amounts of drifting ice. They usually have deep glacier bays and long frontal zones. Glaciers advanced in their retreat, the fronts of which are largely on the coastline (coastlineterminating glaciers, CLTG), have shallow glacier bays and exhibit little if any calving activity (Fig. 1).

Field methods
We conducted systematic boat-based surveys of the coastline and pelagic sectors ( Fig. 1) between 12 July and 1 August 2014-2016, i.e. during the chick-rearing period of the seabird species we studied. Two persons, one noting the observations and the other steering the boat (a rubber Bombard dinghy, travelling at about 15 km h −1 ), conducted the surveys between 11:00 and 18:30 (UTC + 2); the assumption was that seabirds forage more or less evenly throughout the polar day (Wojczulanis- Jakubas et al. 2020). The height of the observer was 1.5 m above the water, which ensured good observation conditions for the identification and counting of birds along the entire width of the transect. Surveys did not take place on windy days with strong wave action. We used line-transect methods (Tasker et al. 1984) to record the species and numbers of all foraging birds observed except geese and waders. To be consistent with data collected during less favourable weather conditions, we truncated all our observations to a distance of 150 m on either side of the boat. The observations during the coastal surveys encompassed a distance of 150 m between the boat and the coastline, and another 150 m on the opposite side of the boat, the bird densities being calculated assuming perfect detection within this 300 m strip.
Based on the fjord's bathymetry, we took a line 300 m from the shore to be the boundary of the coastal zone. This zone is shallow enough (30-40 m) for benthophages to dive and feed on the bottom. Thus, all the surveyed areas farther from the shore and deeper were assigned to the pelagic zone. We included all observations of foraging birds within the surveyed area. Benthophages (ducks, Black Guillemots) and pursuit divers (Little Auks, Brünnich's Guillemots, Atlantic Puffins) observed alternately on the water surface and diving were considered to be foraging. In the case of surface feeders (Black-legged Kittiwakes, Northern Fulmars, Arctic Terns), feeding birds were considered to be those that were picking food off the water surface, or circling over a specific place and diving for food that appeared on the surface. Birds flying over the survey area were not taken into account. All the observers have had many years of experience in conducting marine bird surveys in Svalbard.
The numbers and locations of the surveys were deemed representative of each habitat. The general principle was adopted that the degree of sampling should correspond to the degree of habitat diversity and temporal variability: relatively more surveys were therefore carried out in the more diverse coastal habitats, whereas the very extensive and homogeneous pelagic habitat was sampled relatively less often. As a result, the total area of the relatively small glacier bays (STTG and CLTG) were surveyed often enough that their total surveyed surface area (sum of areas surveyed each time) exceeded their actual area (Table S1). To estimate the total numbers of seabirds foraging in the particular habitats and in the whole fjord, we calculated the mean density of birds for each habitat (the sum of birds observed during all surveys divided by the accumulated area of the habitats) and then multiplied it by the total area of the given habitat in the fjord. The distribution of habitats and the total area surveyed are shown in Fig. 1 and listed in Table S1. The size of the surveyed areas in particular habitats was large enough to eliminate variation in local bird density within the habitats (foraging hot spots). Based on foraging strategies, we allocated the seabirds to three primary foraging guilds, i.e. surface feeders (SF), pursuit divers (PD) and benthic feeders (BF), and into two groups with regard to habitat foraging preferences, i.e. coastal feeders (CF) and pelagic feeders (PF) (Ainley 1977;Simberloff and Dayan 1991).
The seabird surveys in Hornsund were carried out during three summer seasons in 2014-16. To calculate the total seabird food consumption in the fjord, we used survey data from the 2016 summer season (n = 15 surveys), as this was the most representative of the habitat coverage. Based on long-term databases from the Polish Polar Station and the Institute of Oceanology (Polish Academy of Sciences), this season also appeared to be typical as regards both meteorological and hydrographical conditions and bird number fluctuations (Promińska et al. 2018;Stempniewicz et al. 2021). Nevertheless, well-sampled data from the coastal habitat surveys performed in all three seasons were compared to demonstrate the interannual variability of seabird numbers and food consumption.

Calculating seabird food consumption
We estimated the food consumption by seabirds on the basis of published values of Daily Energy Expenditure (DEE) or Field Metabolic Rate (FMR) for particular species (adults and chicks), either measured directly (e.g. Gabrielsen et al. 1991;Uttley et al. 1994) or obtained from the equation relating DEE and seabird body mass (r 2 = 0.91): log(DEE) = 1.1 482 + 0.6521 × (log body mass) or the relationship between body mass and field metabolic rate: FMR = 16.69 m 0.651 (Mehlum and Gabrielsen 1995;Nagy et al. 1999;Crocker et al. 2002;Ellis and Gabrielsen 2002). Next, we calculated the overall daily energy requirements of the "family unit" (ADFFI; two parent birds and their chicks) using our own and literature average values of mean brood size (MBS) in the particular species from the study area (Mehlum and Gabrielsen 1995;Barrett et al. 2002) (Table S2).
To calculate the overall food consumption, we used published information on the dietary composition of each species, with preference given to data from Hornsund or nearby fjords (Stempniewicz 1981;Lydersen et al. 1989;Stempniewicz and Węsławski 1992;Mehlum and Gabrielsen 1993). Most seabirds are opportunistic within the food type as to available prey items. However, the number of prey species available in the high Arctic, and therefore the choice of food type for foraging seabirds, is very limited. Nevertheless, taking into account possible local and temporal variations in the bird's diet, we divided this into just seven rather broad food categories (Table S3).
The calculation of the food biomass taken by birds included calorific values of food, digestive efficiency and water content. The mean energy density and moisture content were calculated for each of the seven food categories. Based on the literature, the following mean calorific values of seabird food were assumed: small fish-6.5 kJ g −1 , crustaceans-4 kJ g −1 and squid-3.5 kJ g −1 .
Food moisture contents were assumed to be 71.1% for fish and 77.3% for invertebrates (Mehlum and Gabrielsen 1995;Furness and Bryant 1996;Hunt and Furness 1996;Crocker et al. 2002). The avian assimilation efficiency, i.e. 76% for auks, 79% for gulls and terns, and 87% for Northern Fulmars, was taken from Bairlein (1999). Using these values, we obtained biomass equivalents of particular food types and then calculated the daily consumption by the most numerous seabird species. The daily food intake was calculated using the following equation (Nagy 1999): Having estimated the mean numbers of foraging birds (assuming them to be breeding birds feeding nestlings), we obtained their total daily food consumption in the entire Hornsund fjord, taking account of the species in question, the different guilds and the various habitats.

Statistics
The pattern of multivariate seabird community structure was illustrated using Principal Coordinate Analysis (PCOA). In addition, one-way PERMANOVAs were used to test for differences in the seabird community structure Daily food intake (wet g) = Daily energy expenditure e (kJ) Energy in food kJ g −1 × (1 − Moisture) × Assimilation efficiency among habitats. The PERMDISP test was used to test the homogeneity of multivariate sample dispersions between the habitats. All the statistical analyses were performed using PRIMER 6 and PERMANOVA (Clarke and Warwick 2001;Anderson et al. 2008).

Seabird numbers, community structure and habitat use
We estimated the average total number of seabirds foraging concurrently in the whole Hornsund area at c. 28,000 with an overall density of c. 90 ind. km −2 . Little Auks (c. 13,500; density c. 43 ind. km −2 ) followed by Northern Fulmars (c. 4500; c. 14 ind. km −2 ), Brünnich's Guillemots (c. 3700; c. 12 ind. km −2 ) and Black-legged Kittiwakes (c. 3300; c. 10 ind. km −2 ) were the most numerous seabird species observed foraging in the fjord and constituted c. 90% of the total seabird community. Pursuit divers, above all Little Auks (c. 48%) and then Brünnich's Guillemots (c. 13%), were predominant in the Hornsund seabird community, making up a total of c. 64%. The proportion of surface-feeders, represented mainly by Black-legged Kittiwakes and Northern Fulmars, was estimated at 28%, while benthic-feeders (almost exclusively Black Guillemots) made up c. 8% (Table 1; Fig. 2c). The largest habitat, i.e. the pelagic zone, was exploited by the great majority of seabirds (c. 21,000 individuals; c. 75%) with an overall density of c. 86 ind. km −2 . Coastal habitats, which provided foraging areas for c. 7000 birds (c. 25%) at an overall density of c. 105 in ind. km −2 , included tidal glacier bays (c. 3700), non-glaciated shores (c. 3100) and the least exploited coastline glacier bays (c. 300 individuals) ( Table 1). Table 1 Total estimated number (N) and density (N km −2 ) of common seabirds including foraging guilds (benthic feeders-BF, pursuit divers-PD, surface feeders-SF) observed in Hornsund, including the pelagic zone (P) and coastal habitats (sea-terminating tidal glacier bays-STTG, coastal-terminating glacier bays-CLTG, non-glaciated coastline-NGCL) (mean ± SE; n = 15 surveys) * Total values also contain less numerous species, which are not included in the  1 3 We found significant effects of habitat on both the species composition and guild structure of the seabird community (    seabirds from different for a ging guilds. The pelagic zone (P) was exploited mainly by pursuit divers, the most common of which were Little Auks. The non-glaciated shore was used primarily by benthic feeding Black Guillemots and Little Auks. Black Guillemots prevailed in the bays formed by retreating glaciers (CLTG), while surface feeding Black-legged Kittiwakes were absolutely predominant in the tidewater glacier bays (STTG). The numbers and densities of foraging seabirds were the highest in the STTG bays, exceeding 3660 birds and 400 ind. km −2 , above all Black-legged Kittiwakes, lower in the nonglaciated coastal habitats (over 3000 and c. 60 ind. km −2) , and the lowest in CLTG (c. 260 birds and 44 ind. km −2 , respectively) ( Table 1). The results of the PERMDISP analysis revealed a statistically significant variation in multivariate dispersion across the habitats in terms of species composition (F = 10.739, df1 = 3, df2 = 83, p(perm) = 0.001) and guild structure (F = 5.05, df1 = 3, df2 = 56, p(perm) = 0.009). Non-glaciated seashore habitats (NGCL) and retreating coastal glacier bays (CLTG) showed greater variability in the community structure than tidal glacier bays (STTG) and the pelagic zone (P) ( Table S4; Figs. 2b, 3b).

Interannual variability
During the three study seasons, significant variations in the numbers of birds from different species and guilds feeding in the coastal habitats of the fjord were observed. The most abundant year was 2015, when over 20,000 simultaneously foraging birds were recorded, whereas c. 5390 and 7000 individuals were found in 2014 and 2016, respectively.
Black-legged Kittiwakes and Little Auks were the most variable, with numbers varying from 5 to 38 times between the years. Thus, the guilds of pursuit divers and surface feeders were quantitatively the most variable. The number of benthophages in the three seasons remained the most stable. In terms of the numbers of foraging seabirds, the tidal glacier bays (STTG) were the most variable of the habitat types, whereas the coastal glacier bays (CLTG) were the most stable (Table S5).

Food consumption
Based on the mean number of seabirds foraging in Hornsund, their species and guild composition, as well as our own and literature data on their diets and food requirements, we estimated the total amounts of different food types taken daily by seabirds in Hornsund during the chick feeding period. Seabirds foraging in the fjord took c. 12.7 tons of food daily. The most important consumers were Little Auks (35.4%), followed by Brünnich's Guillemots (20.1%), Northern Fulmars (19.1%) and Black-legged Kittiwakes (9.3%). Over half of the food biomass (58%) was ingested by pursuit divers. Surface feeders took c. 29% and benthic feeders c. 13%. About 75% of daily food was taken from the pelagic zone and c. 25% from coastal habitats, including non-glaciated habitats (c. 13%) and tidal glacier bays (c. 11%). Seabirds took less than 1% of their daily food from the coast-terminating glacier bays (Table 3).

Discussion
Huge numbers of seabirds (c. 28 thousand individuals) foraged concurrently in Hornsund, which was a result of the unprecedented overall density of foraging seabirds (c. 90 ind. km −2 ). The density in the pelagic zone (c. 86 ind. km −2 ) was only slightly lower than that in coastal areas (c. 105 ind. km −2 ). These values considerably exceeded those obtained in other Arctic fjords. Diemer et al. (2011), using the same survey method and transect width (300 m) in Cumberland Sound fjords (Baffin Island, Nunavut, Canada), recorded 5.5 ind. km −2 , while Schoen et al. (2013) found 6.65 ind. km −2 in Yakutat Bay, Alaska. The distribution of foraging birds in the Hornsund fjord was patchy; only in the case of tidal glacier bays was it very distinct and constant. Along the non-glaciated coast (NGCL) and in the pelagic zone (P), bird flocks were also observed, but during successive surveys they appeared at different locations. This was probably related to the moving aggregations of available prey on which the birds were feeding. As a consequence of their high numbers, the seabirds foraging in Hornsund ingested as much as c. 12.7 t. of food daily, which constitutes c. 0.2% (daily) or c. 6% (monthly) of the summer mesozooplankton and fish standing stock available in the fjord ). In the North Water Polynya ecosystem, where Little Auks forage in huge numbers (15-30 M breeding pairs), they take only 0.3-0.6% of the primary pelagic production. However, locally and temporarily, this figure could be as high as 5-14% (Karnovsky and Hunt 2002).
The meteorological and hydrological conditions, glacier activity and available food resources in the fjord change significantly from year to year (Kwasniewski et al. 2012;Błaszczyk et al. 2013;Promińska et al. 2018). This results in a different abundance of feeding grounds in different years, and hence, the numbers of seabirds foraging in the fjord . During the three study seasons, we noted significant differences in the numbers of seabirds foraging in the coastal habitats of Hornsund. This applied to individual species as well as foraging guilds and habitats. In the most abundant year (2015), an average of over 20,000 birds were foraging simultaneously in the coastal habitats; in the other two years, their numbers were much lower: c. 5390 in 2014 and c. 7000 in 2016 (Table S5). This significant interannual variability in the number of foraging seabirds must have translated into the amounts of food consumed. Thus, the amounts of food consumed daily by seabirds in coastal habitats varied between 2.4 t (2014) and 8.9 t (2015). If the numbers of seabirds foraging in the pelagic zone in the three years varied proportionally with those observed in the coastal habitats, the total food consumption in the whole Hornsund fjord could have ranged from 9.8 t in 2014 to as much as 36.4 t in 2015.
The greatest variability was displayed by the numbers of surface-feeding Black-legged Kittiwakes and pursuitdiving Little Auks, two species which forage alternately in the fjord and on the shelf outside the fjord. The number of benthophages in the three seasons remained the most stable. Tidal glacier bays (STTG) turned out to be the most variable habitat in terms of the numbers of feeding birds, while coastal glacier bays (CLTG) appeared to be the most stable (Table S4).
Food is taken mainly from the pelagic zone (9.6 t) because of its great extent, while the tidewater glacier bays are the most important of the coastal habitats. In the longer term, however, the gradual disappearance of glacier bays will likely reduce the communities of seabirds associated with this habitat. The rapidly progressing deglaciation of the fjord's shores will favour the growth of ice-free coastal habitats (NGCL) and hence, the numbers of mainly benthophagic birds that feed there. Nowadays, the community they form, while not very numerous, is the most diverse . Almost 60% of the prey biomass was taken by pursuit divers, with Little Auks alone taking 35.4%. Little Auks, however, avoid glacier bays because of the turbid water occurring there, which makes it difficult or impossible for them to visually search for tiny copepods (Stempniewicz et al. 2013). Water transparency (Secchi depth) and total suspended matter (TSM) near the tidal glacier fronts in Hornsund ranged between 0.6-1.0 m and 220-439 mg d,… m −3 (Dragańska-Deja et al. 2020).
Compared to seabirds, the abundance of marine mammals, and therefore, the biomass of food consumed by them in Hornsund are negligible. During the three study years, we recorded four species of marine mammals, represented in total by 20 ringed seals Pusa hispida, nine bearded seals Erignathus barbatus, 30 belugas Delphinapterus leucas and three polar bears Ursus maritimus. Their distribution was clearly heterogeneous, with the highest concentrations observed in the tidal glacier bays , previously found in the case of ringed seals (Hamilton et al. 2016).
The species composition and proportions of seabirds foraging in Hornsund partly reflected the size of local breeding populations as well as their habitat preferences. The majority of seabirds feed exclusively in the fjord area (e.g. Black Guillemots, Common Eiders, Arctic Terns) . However, some of them, including the most numerous Little Auks as well as Brünnich's Guillemots and Black-legged Kittiwakes, exploit alternative foraging grounds situated outside the fjord (Jakubas et al. , 2014. Variable proportions of their local populations feed within the fjord, depending largely on the current weather (mainly the direction and strength of the wind and the state of the sea) and the available food resources for a given species ). Just c. 3% of the 6,00,000 breeding pairs of Little Auks foraged in the fjord during the chick rearing period in the typical 2016 season. On the other hand, no less than 10% of the local Black-legged Kittiwake population usually feeds in the fjord. However, based on observations of c. 10,000 Black-legged Kittiwakes feeding simultaneously on 20 July 2015 in the forefield of the Storbreen glacier, the bulk of the breeding population (c. 15,000 breeding pairs) can feed in Hornsund, making use of exceptionally active subglacial discharges and the calmer fjord waters during bad weather in the open sea. They temporarily exploit the highly attractive tidewater glacier bays as important alternative feeding grounds (Urbanski et al. 2017).
The most visible effect of climate warming in the Arctic is the shrinkage of the extent / area of sea ice (Serreze et al. 2007). In the case of Arctic fjords, this is taking the form of rapid glacier retreat. In Hornsund, this process is the fastest in the whole of Svalbard. In the period 1899-2010, the fjord lost c. 172 km 2 of ice cover. The rate of glacier retreat has increased significantly in recent years to c. 3 km 2 (area) and c. 70 m (linear) per year (Błaszczyk et al. 2013). Since 2000, Hornsund's largest glacier, Hornbreen, has retreated at an average rate of 106 m a −1 (Grabiec et al. 2018). So far, the intense melting of glaciers favours the maintenance of attractive feeding grounds for seabirds. However, rapid glacier recession in the Arctic fjords will ultimately result in the disappearance of tidal glacier bays, where subglacial discharges create foraging hotspots with the highest observed foraging seabird densities. As a consequence, we can expect a decrease in surface feeding Black-legged Kittiwakes foraging in the fjord and increasing proportions of pelagic seabirds exploiting planktonic food resources in the fjord (Lydersen et al. 2014;Stempniewicz et al. 2017;Urbanski et al. 2017).
Available food resources, and consequently, the numbers and community structure of seabirds foraging in the fjord and total food consumption can change year to year for reasons such as the irregular advection of oceanic water, the fluctuating input of freshwater of glacial and riverine origin, and the weather conditions (Stempniewicz et al. , 2021Promińska et al. 2018). Increasing proportions of warm water masses reaching Svalbard fjords by advection will probably transform the proportions of Arctic prey communities into Atlantic ones and will influence the numbers and specific structure of foraging seabirds (Vihtakari et al. 2018;Hop et al. 2019).
Currently, planktonic invertebrates, such as copepods, amphipods and cephalopods, constitute the most important prey of seabirds in Hornsund. The lesser importance of fish in the diet compared to lower latitudes and the predominance of planktonic invertebrates, especially crustaceans, in the trophic web is a specific feature of Arctic marine ecosystems (Hobson and Welch 1992;Wassmann et al. 2011;Kortsch et al. 2015;Renaud et al. 2018). According to future climateinduced scenarios, the proportions of fish in the seabird diet are expected to increase and, consequently, the importance of plankton-feeding seabirds to decrease (Stempniewicz et al. 2007). The results of this paper provide a baseline for future studies on climate-induced changes in the importance of seabirds in the Arctic food web.
Acknowledgements We wish to thank our colleagues Lech Iliszko, Mateusz Barcikowski and Mariusz Cymerski for their contributions to the surveys and invaluable help during the fieldwork. This work was funded by a grant from the Polish National Centre for Research and Development-PNRF (GLAERE, POL-NOR/199377/91/2014) and carried out by permission of the Governor of Svalbard. The work of MG was supported by a grand no. 2017/01/X/NZ8/00379 National Science Centre, Poland and by the Ministry of Science and Higher Education Outstanding Young Scientist Scholarship. We would like to thank Dr Kyle Hamish Eliot and two anonymous reviewers for their thorough evaluation of the manuscript and very constructive comments, greatly improving the final version of our work.
Author contributions LS developed the initial conception of the research and wrote the preliminary draft; LS, MiG, and DK carried out the seabird surveys in the field; MiG prepared the data base; MaG and JMW compiled data on marine invertebrates and fish; MaG and DK performed statistical analyses. All the authors contributed to the interpretation of results and to editing the manuscript.

Conflict of interest
The authors do not have any conflict of interest to declare.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.