The diet of high-arctic seabirds in coastal and ice-covered, pelagic areas near the Svalbard archipelago F. MEHLUM and G. W. GABRIELSEN Mehlum, F. & Gabrielsen, G. W. 1993: The diet of high-arctic seabirds in coastal and ice-covered, pelagic areas near the Svalbard archipelago. Polar Research 12(1), 1-20. Food samples from six high-arctic seabird species were collected during spring and summer seasons between 1982 and 1990 in the Svalbard region. The material came from coastal localities on the island of Spitsbergen and the marginal ice zone in eastern Svalbard waters. Polar cod Boreogadus saida was the most frequently occurring prey in the ice-covered areas. Analysis of otoliths showed that most polar cod were one- or two- year olds. These year classes are known to associate with sea ice. Other ice-associated (sympagic) organisms, such as gammarid amphipods, were not found to be of high importance as prey for seabirds in the study area. However, the sea ice occurring in the area was mainly one year old. Such ice contains a less developed sympagic fauna than multi-year ice. The pelagic amphipod Parathemisto libellula, which is not sympagic but occurs in the water column, was also found to be an important prey in the marginal ice zone, especially for the Briinnich’s guillemot Uria lomuia. The smallest of the seabird species studied, the little auk Alle alle, differed from the other five species in its diet, preying mainly upon smaller items such as copepods and young stages of amphipods, euphausiids and decapods. The diet of the various seabird species was in general more diverse in the coastal areas than in the marginal ice zone. F. Mehlum and G. W. Gabrieken., Norwegian Polar Institute, P.O. Box 5072 Majorstua, N-0301 Oslo, Norway. Introduction The waters around the Svalbard archipelago sup- port large populations of several species of sea- birds (Mehlum 1989; Mehlum & Bakken in press). These populations consume a variety of marine organisms. There is little knowledge avail- able about the impact of the seabirds on the marine ecosystem in this part of the Arctic and the functional role of seabirds in the ecosystem as predators on marine organisms. The diets of seabirds in ice-covered, pelagic and coastal areas around the Svalbard archipelago and the northern Barents Sea have not previously been studied in detail. A general review of the available infor- mation up to 1959 is given by Lavenskiold (1964). The most comprehensive published studies of sea- bird diets in coastal parts of the Svalbard area are those by Hartley & Fisher (1936) and Lydersen et al. (1989). Some preliminary results of parts of our own studies have been published elsewhere (Mehlum & Gjertz 1984; Gjertz et al. 1985; Mehlum 1990). Several authors have recently emphasised the importance of ice-edges and the marginal ice zone in polar seas as foraging areas for seabirds (see Hunt 1991 for review). These habitats are abun- dant in the waters near Svalbard and may be crucial for the subsistence of the seabirds in this area. In the ice-related habitats seabirds may exploit the pelagic zooplankton and fish species as well as the fauna associated with the under-ice surface (sympagic fauna), dominated by gamma- rid amphipods and young polar cod Boreogadus saida (Bradstreet 1980). This paper describes the diet of seabirds in the Svalbard area during spring and summer seasons between 1982 and 1990. We present analyses of foods found in the stomachs of six species of seabirds: northern fulmar Fulmarus glacialis, black-legged kittiwake Rissa tridacfyla, ivory gull Pagophila eburnea, Briinnich’s guillemot Uria lomuia, black guillemot Cepphus grylle, and little auk Alle alle. All seabird species except the ivory gull were collected both in coastal areas with no sea ice and in the marginal ice zone in the northern Barents Sea. The ivory gull was collected only in the marginal ice zone. Three of these species are surface feeders and three pursuit diving sub- surface feeders. In particular, we wanted t o inves- tigate the importance of fauna associated with the sea ice subsurface to seabirds foraging in the marginal ice zone, and to compare the diets of Present address: Norwegian Institute for Nature Research, c/o Troms0 Museum, N-9ooo Troms), Norway. 2 F. Mehlum & G . W . Gabrielsen B J ~ ~ R N ~ J Y A 0 birds foraging in coastal versus ice-covered, pel- agic areas. 0 A u g . 1 9 8 2 0 J u l y / A u g . 1 9 8 4 + Apr./May 1 9 8 5 4 May/June 1 9 8 6 rn Febr./March 1 9 8 7 V A u g . 1 9 8 7 Physical and biological setting Seabirds are present in the Svalbard region all year, both in water areas that are open and partly ice-covered, but the number of individuals in the area are highest in the summer months. Fulmars and several species of alcids and larids breed on steep cliffs along the coasts. The largest colonies are found in the northwestern, southwestern and southeastern parts of the archipelago, including the two islands, Hopen and Bj~rnBya (Bear Island) (Fig. 1). The Svalbard archipelago is located north of the shallow and productive Barents Sea. These islands, except for B j ~ m ~ y a (74”30‘N), are located between 76 to 81 degrees north latitude (Fig. 1). They experience a high-arctic climate, with maximum mean summer temperatures reaching 45°C. Despite the extreme northern location, the area is one of the most easily access- ible for seabirds of the High Arctic. This is pri- marily due to the presence of open water during 0 0 0 0 Kongs- fjorden - Q DET 0 m NORTHERN B A R E N T S SEA 4 rn HOPEN- 5 0 0 4 m o o 6 0 0 20 00 26 00 S o 0 0 5 6 0 0 4000 Fig. 1. Sketch map of the study area and locations of sampling, stations. Diet of high-arctic seabirds 3 species are shown in Table 1. Most seabirds were collected at sea. All specimens shot were fully grown individuals if not otherwise indicated. Actively feeding birds were preferred, but most of the specimens were collected while flying. Within 30min after being shot, the birds were placed whole in a deep-freeze, or the stomachs and esophagus tracts were removed and either stored in a deep-freeze or preserved in 70% ethanol. Food samples from Kittiwakes at breeding col- onies in the Kongsfjorden area were collected by catching adults and chicks on the nests and inducing the birds to regurgitate their food into plastic bags. Similarly, chick-feeding little auks in the Hornsund and Isfjorden areas were captured upon returning to their nests and their sublingual pouches were emptied and the contents placed in plastic bags. These samples were placed in a deep- freeze (-20°C). A third method of collecting data, used in 1989 at Kovalskifjellet in Storfjor- den, was direct observation of food brought by parents of young Brunnich's guillemots to the breeding ledges. The primary prey brought to the young were fish, and the different taxa could be determined in most instances (96%). a large part of the year, especially in the western parts of the archipelago. Throughout the year, a branch of the North Atlantic Current transports enormous masses of relatively warm water north- wards into the Barents Sea and along the western coasts of Svalbard (Loeng 1989). These water masses mix with the cold polar water, and in the frontal areas a rich marine production occurs (Loeng 1989). The resultant large biomass of pelagic invertebrates and fish provides the food basis for the seabird populations inhabiting the area. The extent of the ice cover in these waters varies from year to year and at different times of the year (Vinje 1985). Marginal ice zones are very productive areas due to high levels of inorganic nutrients and little vertical mixing of near surface water (Sakshaug & Skjoldal 1989). The phyto- plankton bloom is grazed by copepods and other zooplankters, which in turn provide food for larger organisms such as fish and seabirds. During the spring and summer seasons, the productive areas gradually recede northwards with the mar- ginal ice zone. Materials and methods Data collection The sampling periods, geographical areas and the numbers of samples obtained from each seabird Data analyses For identification of prey, the stomach contents were first washed into a 0.5 mm sieve. All items were then sorted to the lowest possible taxon using available keys and, when possible, reference Table 1. Sampling localities, month/year, and sample sizes of six species of seabirds. Only birds containing identifiable prey taxa are included. Locality/Month/year Briinnich's Black Fulmar Kittiwake Ivory gull guillemot guillemot Little auk n = 5 0 n = 1 2 3 n = 1 9 n = 276 n = 73 n = 178 Marginal ice-zone Svalbard E, Aug. 19x2 Svalbard E, July/Aug. 1984 Svalbard E, Apr./May 1985 Svalbard E, May/June 1986 Svalbard E. Febr./March 1987 Svalbard E. Aug. 1987 Coastal areas Kongsfjorden, July 1982/83 Kongsfjorden, June 1984 Kongsfjorden, June/July 1985 Kongsfjorden, July/Aug. 1987 Hornsund, March/Apr. 1985 Hornsund. July/Aug. 1987 Kovalskifjellet, Storfj., Aug. 1989 Isfjorden, July 1990 14 8 0 0 0 0 0 0 0 0 28 0 0 0 13 18 14 12 0 0 14 5 0 22 25 0 0 0 6 3 10 0 0 0 0 0 0 0 0 0 0 0 1 3 9 28 29 9 0 5 7 0 4 0 161 0 7 2 2 10 0 0 0 0 23 0 29 0 0 0 21 0 7 8 0 0 0 0 14 0 15 99 0 14 4 F. Mehlurn & G . W . Gabrielsen specimens. Identification of prey was primarily based on fish otoliths, squid beaks, polychaete jaws, crustacean exoskeletons and intact speci- mens. All otoliths were counted and (if intact) lengths measured to the nearest 0.1 mm using a micro- scope with a measuring ocular. Two otoliths dif- fering less than 0.2 mm in length were considered to be from the same fish. The number of fish ingested was estimated as half the number of otoliths within each 0.2 mm length category. In the samples’ from Hornsund in spring 1985, the number of fish ingested was based only on the total number of otoliths divided by two. The lengths of polar cod specimens were estimated according to Frost & Lowry (1981) where fish length (cm) = 2 . 1 9 8 ~ + 1.588 (x is otolith length in mm). The sizes of polar cod found in different bird species in spring and summer were compared to the classification of age classes given by L a m e & Gulliksen (1989). They found that the mean fork length of one-year-old specimens was 81.6 mm (SD = 9.3) in the period 25 M a y 4 June, and 102.0mm (SD = 12.6) in the period 31 July- 7 August. Two-year-old polar cod had cor- respondingly mean lengths of 108.7 mm (SD = 9.1) and 125.3 mm (SD = 21.6). L0nne & Gul- liksen found no older age-classes associated with sea ice. Student t-test was used for determining differences in sizes of polar cod taken by different seabird species. A count was made of the crustaceans in each stomach or estimated from subsamples when present in large numbers. The length of whole crustaceans was measured to the nearest mm. When only fragments were encountered, the length of the posterior part was measured to the nearest 0.1 mm by means of microscope and measuring ocular (Bradstreet 1980). Estimates of total length based on the measurements of posterior parts were made in accordance with the equations given by Bradstreet (1980). For the summer samples of little auks, wet weights of crustaceans were used. Dry weights of amphipods were estimated from total lengths according to the equations given by Bradstreet (1980). Wet weights were then estimated from the dry weights. For the water content of Parathemisto libellula we used the value 82.4% given by Ikeda & Skjoldal (1989), and for unidentified amphipods we used an estimated value of 80%. The different age groups of the euphausiid Thysanoessa inerrnis were determined according to Dalpadado (1989), who classified first year individuals at 13-18 mm and second year individuals at 20-25 mm. In the little auk samples from Isfjorden in 1990, the total length (LT, in mm) of copepods was measured from the tip of the head to the end of the furca (accuracy 0.01 mm). Conversion from length to dry weight (pg) of Calanus was based on the relationships given by Franz & Die1 (1984): A water content of 73.5%, which was given for Calanus finmarchicus adult females (Ikeda & Skjoldall989), was used for estimating wet weight from dry weight of C. finrnarchicus/glacialis. The number of polychaetes ingested was deter- mined by counting the number of polychaete jaws present in the stomach sample. Similarly, squid were identified by their beaks. When both upper and lower halves of beaks were present in the stomach sample, the number of squid ingested was determined by counting the lower beaks of squid present. In order to reduce the number of prey categories, the food items recorded were grouped into the following taxa: nereids (Polychaeta, Nereidae), gastropods, cephalopods, copepods, gammarids (Gammaridea), hyperiids (Hyperii- dea), Amphipoda indet., euphausiids, decapods, polar cod, other fishes, Pisces indet., and others. Remains of crustaceans which could not be identified to a lower taxonomic level than Amphipoda were classified under “Amphipoda indet.”. Similarly polar cod is treated separately from other fishes because of its importance in seabird diets. “Other fishes” includes all ident- ifiable fishes except for polar cod. “Pisces indet.” comprises fish remains not identified, which may also include polar cod remains. We mainly used the frequency of occurrence and the numerical abundance of different prey taxa in the analysis of the diet of the different seabird species (Duffy & Jackson 1986). Fre- quency of occurrence was determined for each bird species as the percentage of sampled birds in which the prey type occurred (empty stomachs were not included). Numerical abundance was defined as the percentage in numbers of a prey type compared to the total number of identified prey items. Percentage by weight was calculated by dividing the weight of a particular prey type by the total weight of all prey combined. The latter measure was applied for the little auk samples, only. Dry weight = 10(1.167179+0.36W73LT). Diet of high-arctic seabirds 5 In comparing similarity in diets we used Mori- sita’s index, “C”, of overlap (Horn 1966; Dia- mond 1983) on our frequency of occurrence data. This index ranges from 0 (no overlap) t o 1 (com- plete overlap). This index is usually applied t o numerical abundance data. We chose rather t o use frequency of occurrence because in many of our samples good values of numerical abundance of some taxa could not be obtained due t o decay of the food items in the stomachs of the birds. We believe that this use of the index is suitable for determining which species are most similar in diet. Cluster analyses using the single linkage method (Pielou 1984) were made, and tree dia- grams were constructed, based upon indices of dissimilarity, “D” = 1 - C. Difjiculties in data interpretation In the analysis of seabird stomach samples, single measures of diet carry bias and have the potential to distort the true diets of the birds. For example, often only fragments of prey items are found, mainly the least digestible parts such as fish oto- lithis, squid beaks, and polychaete jaws. Thus, although the frequency of occurrence measure- ment is easily obtainable, it will over-represent the most persistent prey species found in the stomach samples as well as prey which are only present in low numbers. If the sample contains considerable differences in prey sizes, the numeri- cal abundance measure of diet tends to over- estimate the importance of small prey. Adequate measurement of a seabird’s diet would be the mass or the nutritional value of freshly ingested matter of the different prey taxa. However, such information is usually not available because of the different states of digestion of the prey items. For some prey species it is possible, on the basis of allometrical relationships, to estimate the total length or mass of specimens from the lengths of different body parts. These problems have been discussed in detail by Duffy & Jackson (1986). Such diet measurement problems do not apply to seabird species which carry food to their young in their beaks or gular pouches. In our study, only little auks feeding their young give suitable data on prey composition by mass. This study is based on material collected over several years. In the Results section we have combined data from the various years, and our interpretation of the diets of the different seabird species may be biased by inter-annual variation in food availability. We have combined the data because of low sample sizes in individual sampling periods. The samples collected in ice-covered areas were collected over a wider geographical area in the marginal sea ice zone in eastern parts of Svalbard rather than at a fixed geographical locality. In addition, spring samples were obtained in areas farther south than the summer samples because of seasonal differences in ice- cover. Results Diet composition of seabird species in coastal areas versus the marginal ice zone Fulmar Fulmarus glacialis. - The diet of 50 birds included nereids, cephalopods, crustaceans and fish (Fig. 2). Nereids (Nereis irrorata), cepha- lopods (Gonatus fabricii) and polar cod occurred most frequently in the pelagic samples, while cephalopods, decapods and fish other than polar cod dominated the coastal samples. The pelagic amphipod Parathemisto libellula occurred only in samples from the marginal ice zone, 22.7% of those examined. Nereids occurred in 45.5% of fulmar samples from the marginal ice zone, and none were found in samples from coastal areas. Cephalopods occurred in 31.8% and 75.0% of the birds from ice-covered and coastal areas, respectively. Kittiwake Rissa tridactyla. - Polar cod was the dominant food taxon in the marginal ice zone (70.2% of the birds), but lower (28.8%) in coastal areas (Table 2). In the total material, polar cod was found in 48.0% of the birds. The pelagic amphipod Parathemisto, occurred in 30.3% of the bird specimens from coastal areas, but was not found in birds from the marginal ice zone. Other taxa present in the coastal samples were cepha- lopods, copepods and euphausiids. In the coastal spring samples (Hornsund in MarchIApril 1985), polar cod occurred in 44.0% of the birds along with other fish including lantern fish Benthosema sp., Atlantic cod Gadus morhua, and snailfish Liparidae (Fig. 3). Other prey of Hornsund birds were squid ( Gonatus fabricii) and polychaetes (Nereis irrorata), with smaller num- bers of amphipods and decapods. During chick-rearing, euphausiids and amphi- pods were more frequent in the birds sampled in 6 F. Mehlum & G . W . Gabrielsen Table 2. Frequency of occurrence of prey taxa (% of birds investigated) in kittiwakes Rksa fndactylu, Briinnich’s guillemots Uria lomuia, black guillemots Cepphlls grylle, and little auks Alle alle, in birds collected in the marginal ice zone (MIZ) versus coastal areas (C). Chick diet of Briinnich’s guillemots from Kovalskifjellet is shown separately (K). Sample sizes in brackets. Kittiwake Br. guillemot B1. guillemot Little auk Prey taxon Nereids Gastropods Cephalopods Copepods Amphipoda indet. Gammarids Hyperiids Euphausiids Decapods Polar cod Other fishes Pisces indet. Other 12.3 16.7 1.1 11.5 0 0 53.8 0 1.8 1.5 0 3.8 0 0 17.3 0 0 6.1 0 11.5 0.5 0 0 0 0 3.0 4.5 0 0 4.8 3.8 66.7 1.8 4.6 22.5 15.4 0 0 21.1 8.3 0 1.5 4.5 38.5 0 9.5 30.8 16.7 0 30.3 64.0 3.8 0 0 3.8 30.6 0 15.2 7.9 38.5 0 0 3.8 0 7.0 6.1 14.6 7.7 0 0 32.7 0 70.2 28.8 28.1 34.6 83.9 71.4 9.6 5.6 22.8 18.2 1 . 1 19.2 11.5 14.3 15.4 0 24.6 30.3 14.6 19.2 4.0 28.6 21.2 13.9 1.8 1.5 0 0 0 0 5.8 19.5 Fulmar Nereids G a s t r o p o d s C e p h a l o p o d s C o p e p o d s Gammarids Boreo a d u s Other ?ishes P i s c e s indel. Other 0 20 4 0 60 80 7 100 % I I pelagic. summer (n- 22) Fulmar Nereids G a s t r o p o d s C e o h a l o o o d s ‘Copepods Gammarids Olher 0 20 4 0 60 80 100% Fig. 2. Frequency of occurrence and numerical abundance (hatched) of prey taxa found in fulmars Fulmarus glacialis. ~ 0 0.7 0 88.7 4.2 24.6 68.3 23.9 63.4 0 0 18.3 3.5 coastal areas than in spring (Fig. 3). In the July/ August 1987 samples from the Kongsfjorden area, the pelagic amphipod Parathemisto libellula occurred most frequently (46.3%), while euphau- siids (mainly Thysanoessa inermis), polar cod and polychaetes (Nereis sp.) were found in 24.4%, 19.5% and 19.5% of the samples, respectively. In the marginal ice zone (Fig. 3), birds were sampled in the spring (April 1985 and May/June 1986) and summer (August 1982 and July/August 1984). Fish dominated the diet during both periods with polar cod, the species most fre- quently encountered during spring (frequency of occurrence 42.3%), supplemented by coalfish Pollachius oirens, haddock Melanogrammus aeglefinus, lantern fish Benthosema sp., capelin Mallotus villosus, snailfish (Liparis sp., Care- proctus sp.), and blennies (Lumpenidae, Zoar- cidae). The polychaete Nereis sp. was found in 23.1% of the birds. In July/August 1984 polar cod was found in 93.5% of the birds; other fish species occurred in 22.6% of the birds. Ivory Gull Pagophila eburnea. - A total of 19 specimens of ivory gull were collected during three different periods (Fig. 4), all in pelagic areas within the marginal ice zone. Fish was the dominant prey found, and polar cod occurred in 52.6% of the birds examined. Other fish species Diet of high-arctic seabirds 7 Kittiwake coastal. spring (n- 25) Kittiwake coastal. summer (n- 41) Nereids Gastropods Cephalopods Copepods Gammarids Boreo adus Other fishes Pisces indet. All Nereids Gastropods Cephalopods Copepods Gammarids Hyperiids Boreo adus Other fishes Pisces indet. Other Other 0 20 40 60 80 100 % 0 20 40 60 80 100 % Nereids Gastropods Cephalopods Copepods Gammarids Hyperiids Boreo adus Other fishes Pisces indet. Other E=i=L= 0 20 40 J 60 80 100 % pelaglc. summer (n= 31) Nereids Gastropods Cephalopods Copepods Gammands Hyperiids Boreo adus Othergshes Pisces indet. Other 0 20 40 60 80 100 % Fig. 3. Frequency of Occurrence and numerical abundance (hatched) of prey taxa found in kittiwakes R b a tridactyla. (blenny Lycodes sp., Atlantic cod, and coalfish) occurred in 26.3% of the stomach samples, while mammal fat (probably seal) was found in 21.1% of birds. Six Gammarus wilkitzkii, a sympagic amphipod, were found in one bird. Echinoderms occurred in two birds, but was suspected of being obtained by scavenging from seal excrement. Ivory Gull Nereids Gastropods Cephalopods Copepods Gammarids H periids Amphi od% indet. gphausiids Decapods Boreo adus Olhereishes Pisces indet. Other 0 20 40 60 80 100 % Fig. 4. Frequency of Occurrence and numerical abundance (hatched) of prey taxa found in ivory gulls Pagophila eburnea. Briinnich's Guillemot Uria lomvia. - The pelagic amphipod Parathemisto occurred in the highest frequency (64.0%) among samples from the mar- ginal ice zone, while gammarid amphipods and euphasiids had the highest occurrence (38.5%) in coastal samples (Table 2). Gammarid amphipods and euphausiids were encountered in lower occur- rence in the marginal ice zone (4.5% and 7.9%). Polar cod was found in 28.1% and 34.6% of the birds in ice-covered and coastal areas, respect- ively. Amphipods and euphausiids were encountered most frequently in the early spring samples from Hornsund (April/May 1985), occurring in 71.4% and 64.3% of the 14 birds, respectively. However, euphausiids (mainly Thysanoessa inermis) dom- inated in numbers and accounted for 68.5% of all identifiable prey items (Fig. 5). The pelagic amphipod Parathemisto was not encountered, and the dominant amphipods in the samples were Gammarellus homari and Anonyx nugax, both benthic living species. Polar cod was also a com- mon prey, occurring in 42.9% of the birds. In the 12 summer samples from coastal areas from 8 F. Mehlum & G . W . Gabrielsen I Brunnich’s Guillemoc I coastal. spring ( n - 14) Nereids Gastropods Cephalopods Copepods Gammarids Hyperiids Boreo adus Other ?ishas Pisces indei. Other 0 20 40 60 80 100 Briinnichs Guillemot Nereids Gastropods Cephalopods Copepods Gammarids Hypenids Amphi oda indet. gohausiids n % Nereids Gastropods CeDhalooods Conebods1 I Pisces indet. Other -_ J l Brunnich’s Guillemot pelagic, spring (n= 76) Nereids GastroDods 0 20 40 60 80 100% pelagic. summer (n= 13) Cephalopods Copepods Gammarids Boreogadus Other fishes Pisces indet. Other 0 20 40 60 80 100 % Fig. 5. Frequency of Occurrence and numerical abundance (hatched) of prey taxa found in Brunnich’s guillemots Uria lomuia. Kongsfjorden (June/July 1984-85), benthic amphipods occurred most frequently (33.3%), while polar cod and other fish species (mainly blennies and capelin) occurred in 25.0% of the birds. The samples from the marginal ice zone during the spring period (Fig, 5) comprise a total of 76 birds from Feb./March 1987, April 1985 and late May/early June 1986. The most frequently occur- ring taxon was the amphipod Parathemisto, occur- ring in 63.2% of the samples, while polar cod occurred in 27.6% of the birds. The euphausiid Thysanoessa inermis and the amphipod Para- themisto libellula accounted for 51.3% and 36.6% of the 1486 identifiable prey items. In all but one bird the euphausiids were confined to the samples from Feb./March 1987. Decapods were encoun- teredin 15.8% of the birds. In the April/May 1985 sample, however, the decapod Pandalus borealis was the most frequent prey: 45.6% of the stomachs. The samples collected in ice-covered waters during the summer seasons (Fig. 5) comprise 13 birds, of which 7 were first-year birds still accompanied by one of their parents. Para- themisto was the dominant prey in the sample occumng in 69.2% of the birds and accounting for 200 of the 213 identified items (93.9%). Polar cod was the second most important and occurred in 30.8% of the birds. Only one prey specimen of gammarid amphipods was found, the sympagic species Gammarus wilkitzkii. No difference was found between the diet of adult and first-year- birds. Parathemisto occurred in four out of seven first-year birds (57%). Briinnich’s guillemot chick diet was studied at Kovalskifjellet in 1989 (Table 2). In the 161 food deliveries observed the chicks were almost exclu- sively fed fish, of which 83.9% were polar cod, 9.9% snakeblennies (Lumpenidae), 1.5% scul- pins (Cottidae), and 4.3% unidentified fishes; one chick meal delivered was a squid. Black Guillemot Cepphus grylle. - The diet of Diet of high-arctic seabirds 9 and 9.5% of the birds, respectively. Polychaetes, gastropods, Parathemisto, euphausiids, mysiids and decapods were only encountered in the coastal samples. black guillemots was generally more diverse in coastal areas than in the marginal ice zone (Table 2). Benthic dwelling prey were frequent in the coastal samples, whereas polar cod was dominant in pelagic, ice-covered areas. In inshore waters at Hornsund (March-April 1985) and Kongsfjorden (July-Aug. 1985), the diversity of prey was high with a high proportion of benthic invertebrates in addition t o polar cod and benthic sculpins (Fig. 6). The polychaete Nereis irrorata had the highest frequency of occur- rence (53.8%), followed by benthic amphipods (48.3%) and decapods (32.7%). In Hornsund birds polychaetes were dominant: frequency of occurrence = 96.6%, and numbers = 87.3%. Amphipods were also numerous, mainly Gam- marellus homari, a species usually found in shal- low water in the Hornsund region (Opalinski & Weslawski 1989). In offshore, ice-covered areas (Fig. 6), polar cod was the dominant prey, occurring in 71.4% of the birds, with other fishes (coalfish and Lycodes) and sympagic amphipods occurring in only 14.3% Little Auk Alle alle. - Little auk foods were mainly small invertebrates, with copepods dom- inant by occurrence and numbers. The diet in coastal waters was more diverse than in the mar- ginal ice zone (Table 2) comprising gastropods, copepods, amphipods, euphausiids, decapods, chaetognaths and fish larvae. Gastropods, euphausiids and decapods were not present in the pelagic samples. The only coastal area spring sample is from Hornsund in March/April 1985. Copepods were found in 66.7% of the birds. However, amphipods other than Parathemisto and euphausiids were also frequent in the diet, and the latter accounted for 92.4% of the number of prey items recorded except for copepods. Copepods were present in 91.3% of all coastal samples in summer and they accounted for 85.3% in the wet weight of the prey (Fig. 7). The pelagic Black Guillemot coastal. spring (n- 29) Black Guillemo! coastal, summer (n= 23) Nereids Gastropods Cephalopods Copepods Gammands Hyperiids 0 20 4 0 60 80 100 n Nereids Gastropods Cephalopods Copepods Gammarids Nereids Gastropods Cephalopods Copepods Gammands H periids AmohiDoc& indet. % 0 20 40 60 80 1 0 0 % I Black Guillemo: pelagic, spring (n= 12) Nereids fl Black Guillemot pelagic, summer (n= 9) Hyperiids Boreo adus Otherqishes Pisces indet. Other 0 20 40 60 80 100 Ai % Gastropods Cephalopods Copepods Gammarids Hyperiids Boreo adus Other %shes Pisces indet. Other 0 20 4 0 60 d 80 100 % Fig. 6. Frequency of occurrence and numerical abundance (hatched) of prey taxa found in black guillemots Cepphus grylle. 10 F. Mehlum & G . W . Gabrielsen A Nereids Gastropods Nereids 1 Gastropods I Cephalopods Copepods Gammarids I Hyperiids Boreo adus Other?ishes [-J Boreo adus Pisces indet. Othertshes 0 20 40 60 80 100% Other Other / I J Pisces indet. 0 20 40 60 80 100 % Nereids Gastropods Cephalopods Copepods Gammarids Hyperiids Boreo adus Otherfshes Pisces indet. Other Little Auk p c - + J 0 20 40 60 80 100% Nereids 1) Gastropods Cephalopods Copepods Gammarids HvDeriids Amphi oda indet. ADhausiids I r // becapods Bore0 adus Other Bishes Pisces indet. Other 0 20 40 60 80 100 % Fig. 7. Frequency of occurrence and numerical abundance (hatched) of prey taxa found in little auks Alle alle. For the coastal summer samples wet weight i s presented instead of numerical abundance. amphipods Parathemisto spp. and decapod larvae were the second and third most frequent prey items encountered, and were found in 75.6% and 70.9% of the birds, respectively. Copepods were found in all birds collected during summer in Hornsund (1987) and Isfjorden (1990), while they occurred only in 21.4% of the birds in Kongsfjor- den (1985). In the Kongsfjorden sample Para- themisto was the dominant prey, occurring in 57.1% of the birds, and followed by gammarids with 28.6%. The July-Aug. Hornsund birds con- tained 86.1% wet weight copepods, while larval stages of decapods (mainly Eupagurus) comprised the second most important prey taxon with 70.1% of the macroplankton wet weight (with copepods excluded). Parathemisto spp. constituted 25.3% of the macroplankton wet weight and were the most important taxon relative to frequency in numbers; they accounted for 52.5% of the macro- planktonic prey items. Two species of Para-them- isto were encountered. P. abyssorum occurred more frequently than P. libellula (40.0% and 12.1% of macroplanktonic prey items, respect- ively) and contributed more to the total weight of macroplankton consumed by the birds (15.4% vs. 9.9%). Euphausiids were of minor importance in the Hornsund sample and accounted for only 1.0% of the macroplankton weight. Copepods accounted for 70.9% of the total wet weight of the prey in the Isfjorden sample, while Parathemisto spp. were second (18.5%), followed by Thy- sanoessa sp. (6.4%) and decapod larvae (3.9%). Parathemisto abyssorurn was the dominant Para- themisto species (198 of 206 individuals, or 96%). The summer 1990 material from Isfjorden was analysed in order to describe the composition of different Calanus stages and species, and to determine their weights. Of 4903 whole specimens of copepods, C. jinmarchicuslglacialisis dominated with 96.0% copepodite stage V (mean dry weight per individual copepod 408 pg), 1.9% females stage VI (mean dry weight 539pg), and 1.8% copepodite IV (mean dry weight 187pg). Only five specimens of C . hyperboreus were found, three of copepodite IV (mean dry weight 685 pg) and two copepodite V (mean dry weight 2267 pg). Diet of high-arctic seabirds 11 “ 1 3 2 l 2 6 m X x x x x x x x X xxx x x x x x x x The estimated dry weights of C . finmarchicus/ glacialis from the pelagic samples of summer 1982 averaged 608 pg (range: 192-928 pg, n = 20). In the spring samples from the marginal ice zone, copepods and chaetognaths were most fre- quently encountered and occurred in 60.0% and 46.7% of the birds, respectively (Fig. 7, chaeto- gnaths are listed as “others”). The pelagic amphi- pod Parathemisto libellula, sympagic amphipods, and one-year old individuals of polar cod were found in much lower percentages of the birds. Late summer samples from pelagic, ice-covered areas (Fig. 7) contained only copepods, Par- athemisto libellula and sympagic, gammarid amphipods (mainly Apherusa glacialis and Gam- marus wilkitzkii). Estimates of dry weights of the different prey taxa showed that gammarid amphipods, which were found in 28.6% of the birds, accounted for as much as 78.9% of the total weight of the prey. v 1 - 4 ” Comparison of diet between seabird species and habitats A matrix of the frequency of occurrence of dif- ferent prey taxa in the six seabird species inves- tigated are shown in Table 3 where they are separated by habitat: ice-covered versus coastal areas. Polar cold was important for all seabird species (except the little auk) in the pelagic, partly ice-covered areas, but less important in the coastal samples where all bird species had a more varied diet. A second important prey taxon in the marginal ice zone was hyperiid amphipods. Parathemisto libellula was the most common hyperiid in Briinnich’s guillemots, and also frequent in ful- mars and little auks in the marginal ice zone, and in kittiwakes from coastal areas. On the other hand, P. abyssorum was dominant in little auks from coastal areas. Nereids were frequent in ful- mars and kittiwakes in offshore areas, and in kittiwakes, Briinnich’s guillemots and black guillemots in coastal waters. Cephalopods occurred in three seabird species, but the fre- quency of occurrence was > l o % in only two species. They were encountered in very high fre- quency in fulmars in coastal areas, and less fre- quent in Briinnich’s guillemots and kittiwakes. In the marginal ice zone they were found only in fulmars, but at a lower level than along the coast. Certain prey taxa had a frequency of occurrence of larger than 10% in only a single seabird species: copepods and chaetognaths in little auks, gastro- pods in coastal black guillemots, and mammal fat in ivory gulls. Table 3. Summary table of the frequency of Occurrence of different prey taxa in seabirds sampled in the marginal ice zone and :oastal areas. Frequency of Occurrence in bird stomachs x = 1040% XX = 40-70% XXX > 70% MARGINAL ICE ZONE Fulmar Kittiwake Ivory gull Briinnich’s guillemot Black guillemot Little auk _____ COASTAL AREAS Fulmar Kittiwake Briinnich’s guillemot Black guillemot Little auk 12 F. Mehlum & G . W . Gabrielsen Table 4. Morisita’s indices for dietary overlap between seabirds sampled in the marginal ice zone and coastal areas in the Svalbard region. The indices are based on frequencies of occurrence. MARGINAL ICE ZONE Species FUGLA RITRI PAEBU URLOM CEGRY Fuhnar 0 Kittiwake 0.62 0 Ivory gull 0.46 0.90 0 Briinnich’s guillemot 0.54 0.41 0.37 0 Black guillemot 0.53 0.97 0.88 0.41 0 Little auk 0.22 0.12 0.11 0.47 0.20 COASTAL AREAS Species FUGLA RITRI URLOM CEGRY Fuhnar 0 Kittiwake 0.38 0 Briinnich’s guillemot 0.38 0.68 0 Black guillemot 0.28 0.53 0.61 0 Little auk 0.17 0.34 0.26 0.33 ~ (FUGLA = Fulmar, RITRI = Kittiwake, PAEBU = Ivory gull, URLOM = Briinnich’s guillemot, CEGRY = Black guillemot, ALALL = Little auk). The Morisita’s indices of dietary overlap between the six seabird species studied in the marginal ice zone and five species in coastal regions are shown in Table 4. These matrices are calculated from the frequencies of occurrence of each of the 14 prey categories listed in Table 3. The corresponding dendrograms of dissimilarity (Fig. 8) show that in the marginal ice zone there is a high similarity in diet between black guille- mots and kittiwakes, both of which exploit ice- associated prey. Ivory gulls also showed a diet similar to these two species, while the most distant species in diet similarity was the little auk, which mainly fed upon copepods. The dendrograms indicate a lower similarity in diet between the species in coastal areas than in offshore areas. In coastal birds the highest similarity in diet was between Briinnich’s guillemots and kittiwakes; both fulmars and little auks had very low dietary overlap indices in coastal regions. The highest dietary overlap between marginal ice zone and coastal habitats by applying Morista’s index was for little auks (0.72), followed by kit- tiwakes (0.70), and fulmars (0.57). The lowest overlap value was for the black guillemot (0.29), probably reflecting its benthic versus under-ice foraging habits. The value for Briinnich’s guille- mot was intermediate (0.43). Sizes of polar cod, Parathemisto spp. and Thysanoessa inermis consumed by seabirds A comparison of prey sizes for different seabird species and seasons allows the predominant prey cohorts and differences between predator species to be determined. Length distributions of polar cod based on otolith lengths (Fig. 9) and L ~ n n e & Gulliksen’s (1989) classification indicate that in spring Briinnich’s guillemots and little auks feed mainly on one-year-old cod. The mean lengths of polar cod in the Briinnich’s guillemots collected during spring varied from 75 to 96 mm between samples, whereas those from the single sample set of little auks averaged 81 mm. The polar cod taken by kittiwakes and black guillemots com- prised individuals of both one-year-old and older. Mean lengths in spring varied from 78 to 115 mm for black guillemots, and from 83 to 131mm between samples in kittiwakes. In Hornsund (April/May 1985) kittiwakes caught polar cod significantly larger than both Briinnich’s and black Diet of high-arctic seabirds 13 0.6 Coastal areas (Single linkage method! ’ 0.5 o.6 t 0.2 0.1 0.0 T 5 m Marainal ice-zone Fig. 8. Cluster diagram of dissimilarity in diet between five scabird species in coastal areas and six species in the marginal ice zone. The dissimilarity is measured a s one minus Morisita’s index, using the frequency of occurrence data. guillemots (t = 11.3, p < 0.001; and t = 6.1, p < 0.001, respectively). In summer, mean lengths of polar cod in kitti- wakes varied from 106 to 134mm (three data sets), compared to from 97 to 136mm in black guillemots (two data sets). The majority of the polar cod eaten by kittiwakes and black guillemots in summer 1984 were probably one-year-old fishes (mean lengths: 97 mm and 106 mm, respectively). In the other samples, however, the proportion of older polar cod was higher (mean lengths: 118 to 136 mm). Length distributions of Parathemisto libellula in samples from different seabird species and seasons are shown in Fig. 10. In spring and sum- mer Briinnich’s guillemots preyed upon relatively large amphipods, with mean lengths from 25.0 to 31.5mm. Kittiwakes, little auks and fulmars seemed to have taken smaller-sized amphipods during the summer, though the small sample sizes for kittiwakes and fulmars makes things uncer- tain. In the little auk samples collected from Isfjorden in July 1990, the mean length of P. libellula was 11.5 mm, while the mean length of the much more frequent prey P. abyssorum was 6.1 mm. The only two measured P. libellula in kittiwakes from the spring samples were, like most specimens found in the Briinnich’s guille- mot, from the larger cohort. The majority of the Thysanoessa inermk taken by Briinnich’s guillemots and little auks in spring and kittiwakes in summer were two-year-old indi- viduals based on Dalpadado’s (1989) classification of age groups of this euphausiid (Fig. 10). In the coastal summer sample of little auks from Isfjorden 1990, the mean lengths was 6.1mm (range: 5.0-15.0 mm), which indicates that these birds preyed upon the 1990 year’s production of Thysanoessa inermis. Discussion In this study, which is based on 719 food samples from six seabird species in the Svalbard region, we have demonstrated the diversity of prey taken by seabirds in the region and also documented the dominance of certain key prey species, such as polar cod and Parathemisto, which are con- sumed by most seabirds. We found the diet of the smallest species studied, the little auk, least similar to the diet of the other species. It fed mainly on copepods. .’ Fishes The polar cod is a principal prey species for sea- birds in pelagic, ice-covered waters in the eastern Svalbard region (Table 3), as has been described for seabirds in the Canadian High Arctic (Brad- street & Cross 1982). The year classes I and I1 of this species associate with ice and may be found in crevices in the ice or between sandwiched ice- flows (L@nne & Gulliksen 1989). Older polar cod are confined to the open water masses. Thus, the year classes of young polar cod may be treated as part of the sympagic fauna. The distribution of young polar cod in the ice-covered areas around Svalbard is not known in detail. However, a spawning area of the western component of the Barents Sea population of polar cod has been K it ti w nk e L en gt hs o f P ol ar C od 50 10 0 I5 0 20 0 (m m ) 1 : : : : : : : : : : : : : : 1 (2 9) M ar ch lA pr . 19 85 (1 1) - A pr ./M ay I 98 5 (1 1 ) -=t= - M ay /J un e 19 86 (4 ) t Ju ly 1 98 2/ 83 (2 6) A ug . 19 82 (2 9) - July/Aug. 19 84 (1 1 ) & A ug . 19 87 B ri in ni ch 's C ui ll em ot 50 1 0 0 I5 0 20 0 (m m ) t : : : : : : : : : : : : : : 1 Fe br ./M ar ch 1 98 7 M ar cW A pr . I9 85 A pr .lM ay 1 98 5 M ay /J un e 19 86 (1 56 ) = (1 43 ) (1 7) F R - (2 2) - (2 ) -- Ju ly 1 98 5 B la ck G ui ll em ot 5 0 10 0 15 0 20 0 (m m ) M ar cM A pr . 19 85 A pr ./M ay 1 98 5 M ay /J un e 19 86 Ju Iy /A ug . 19 82 (1 3) - (6 ) - (1 1 ) (7 ) - - (2 4) - A uy . 19 84 L it tl e A uk 50 10 0 15 0 20 0 (m m ) I : : : : : : : : : : : : : : { - M ay /J un e 19 86 (5 ) Fi g. 9. E st im at ed l en gt hs ( m ea n & st an da rd d ev ia tio n, a nd ra ng e) o f po la r co d B or eo ga du s sa id 0 in d if fe re nt s ea bi rd sp ec ie s. T he e st im at es a re b as ed u po n ot ol ith l en gt hs ( L m ne & G ul lik se n 19 89 ). Sa m pl e si ze s ar e gi ve n in b ra ck et s. P nr rt he m is tn li he llu ln 5 15 2 5 35 ( m m ) I : : : : : : : : : : : : : : : : { + Fe br /M ar ch 1 98 7 M ay /J un e I9 86 a . - A ug . 19 84 3 s A ug 19 82 s - - - 4+ Ju ly 1 99 0 A u g . 19 82 T h ys n n oe ss n in er m is 5 IS 25 3 5 (m m ) 1 : : : : : : : : : : : : : : : : 1 (9 ) - Ju ly 1 98 2 Fi g. 1 0. E st im at ed l en gt hs ( m ea n 2 s ta nd ar d de vi at io n, a nd ra ng e) of t he a m ph ip od P ar ar he m is ro li be llu h an d th e eu ph au - si id T hy su no es sa in er m is i n di ff er en t se ab ir d sp ec ie s. S am pl e si ze s a re g iv en in b ra ck et s. M ar ch /A pr . 1 98 5 M ar ch lA pr . I9 85 Ju ly 1 99 0 Diet of high-arctic seabirds 15 P. libellula, presumably the young of the year, becomes available t o seabirds. P. libellula is not regarded as part of the sympagic fauna, but it may be found in the water column just below sea ice (L0nne & Gulliksen 1991a). Wing (1976) studied the ecology of P. libellula and P . pacijica in south- eastern Alaskan waters and found that both species migrate vertically between day and night most of the year. He suggested that P. libellula made an upward movement of 15&200m each night with a significant part of the population between the surface and 50m. Similarly, Gul- liksen (1984) reported that the biomass of this amphipod under land-fast ice in eastern Svalbard varied between day and night, with the highest biomass around midnight. This vertical migration may influence the availability of the amphipod as prey for seabirds, making them available in highest numbers for surface or near-surface feeders at night. The smaller hyperiid P. abyssorurn was found only in little auks taken along the western coast of Spitsbergen (Hornsund and Isfjorden). This amphipod was dominant in both areas during summer, though the specimens taken were much smaller than P. libellula. Zelichman & Golovkin (1972) also found a high frequency of P . abys- sorurn in the diet of little auks at Novaja Zemlja. The P. libellula found in the summer samples of little auks were of the small cohort, probably the young of the year, similar in size to those found in little auks in Arctic Canada (Bradstreet 1982). Weslawski & Kwasniewski (1983) suggested that P. abyssorurn was an indicator of Atlantic Water in the Svalbard area, because in their study, the species was found only in Atlantic Water in the western parts and not in the Arctic Water in the eastern parts of the archipelago. However, the species is also widespread in the Arctic Ocean as well as in deep water in the sub-Arctic (Ekman 1953). In Lancaster Sound, Northwest Terri- tories, P. abyssorurn occurred in deeper water than P . libellula and were not thought to be important as prey for other seabirds with the possible exception of the deep-feeding Briinnich’s guillemot (Bradstreet 1976). Krill, mainly Thysanoessa inerrnis, were also important prey species in Briinnich’s guillemot , but only in the spring. This is probably because these arcto-boreal euphausiids are primarily found in Atlantic Water masses, which around Svalbard are normally not ice-covered during summer (Dalpadado & Skjoldall991). Birds were located east of Svalbard, in the region north of Hopen and east of Edge0ya (Loeng 1989). It is likely that the young polar cod from this locality will spread into large areas. Capelin was not encountered frequently in the diet of any of the seabird samples, but was found in kittiwakes and Briinnich’s guillemots in low numbers. This pelagic fish is an important prey species for seabirds breeding at Bj0rnaya (V. Bakken, pers. comm.) and Hopen (76”30’N) (R. Barrett, pers. comm.). The Barents Sea capelin spawns along the coasts of North Norway and the Kola Peninsula, while the young capelin mature in the central or northern Barents Sea. Capelin is known to follow the retreating ice-edge north- ward during the summer, preying upon the grow- ing populations of copepods (Loeng 1989). Capelin is normally not abundant in Svalbard waters except for the southeastern parts after the summer disappearance of sea ice. However, capelin may occasionally be found locally along the western coasts of Spitsbergen. The seabirds sampled for this study were collected outside regions where capelin is normally abundant at that time of year. Another factor that might con- tribute to the lack of capelin in the diet of the seabirds sampled is that the capelin stock in the Barents Sea was dramatically depleted in the 1980s and the population crashed in 1985-86 (Hamre 1988). Pelagic crustaceans Our results indicate that Parathemisto libellula is also a key prey species for several species of seabirds in the Svalbard and northern Barents Sea areas, a finding similar to that for the Canadian Arctic (Bradstreet & Cross 1982). This amphipod also comprises a significant proportion of the diets of ringed seals Pusa hkpida (Gjertz & Lydersen 1986) and harp seals Phoca groen- landica in Svalbard and the northern Barents Sea (Lydersen et al. 1991). According t o our data, large sized cohorts, probably adults of P . libellula, form a staple food of Briinnich’s guillemots in the marginal ice zone during spring. This prey must be very abundant in the marginal ice zone east of Svalbard. Large groups of Briinnich’s guillemots, numbering thousands of birds, have been observed foraging in open leads in this area (per- sonal obs.). Of the birds collected from these leads for dietary analysis, most were gorged with P. libellula. During summer a smaller cohort of 16 F. Mehlum & G . W . Gabrielsen sampled only north of the frontal area between Atlantic and Arctic water masses during the sum- mer. Calanoid copepods were important as prey only for the little auk, although they also occurred infrequently in kittiwakes, Briinnich’s and black guillemots. Three Cafanus species are common in Svalbard waters: C . finmarchicus, confined to Atlantic Water masses, and C . glacialis and the larger C . hyperboreus in Arctic Water masses. One might expect the larger C. hyperboreus to be selected by little auks if equally accessible and abundant, because of the higher energy content per prey individual taken by the bird. In north- western Baffin Bay, little auks selected the largest species of copepods (C. hyperboreus) and the largest life stage of both C . hyperboreus and C . glacialis (Bradstreet 1982). However, in eastern Lancaster Sound, copepods were absent and the diet was dominated by amphipods (Bradstreet 1976). In our July samples from Isfjorden, C. glacialis (copepodite stage V ) totally dominated in numbers, and only five specimens of C. hyper- boreus were found. However, nothing is known about the composition of plankton available t o little auks at the time of collection. Similarly, in ice-filled waters east of Svalbard in August 1982, the size distribution suggests that the C . finmarchicus/glacialis taken belonged t o the larg- est stages (copepodite V and adults). Decapod larvae were also frequent as prey for little auks in the coastal samples. The pre- dominating species encountered was Eupagurus pubescens, which is abundant as zoea larvae in coastal Spitsbergen from April t o September (Weslawski 1987). Polychaetes, squid and other soft-bodies prey Nereids (the only species identified was Nereis irrorata) were found in seabirds sampled both in coastal and ice-filled areas. Only the chitinous jaws were present, which have low digestibility and thus may remain in the digestive tract for long time. In the coastal samples, nereids are probably taken by black guillemots while feeding on benthic dwelling prey. Nereids were also re- corded as prey for surface feeding seabirds such as kittiwakes and fulmars. These polychaetes are known to “swarm” in surface waters during repro- duction and may thus be available t o surface feeding seabirds (Pettibone 1954). The only remains of cephalopods found in the birds were their beaks. All beaks identified were from the squid Gonafus fabricii, a widely dis- tributed species. Cephalopods were the dominant prey in coastal samples of fulmars, but only occurred at low levels in pelagic samples. Squid were also found in kittiwakes and Briinnich’s guillemots, but only in coastal samples and in low numbers. The distribution and ecology of G . fabricii in Svalbard is unknown, and its apparent difference in abundance in our samples from sea- birds from coastal areas and the marginal ice zone cannot be explained. However, the presence of beaks in the stomach of fulmars agree with pre- vious studies in Arctic regions (Duffy & Sergant 1950; Belopolskii 1957; de Korte 1972; Bradstreet 1976; Lydersen et al. 1989). G. fabricii is found throughout the water column, with mainly small individuals occurring near the surface (Nesis 1965). The mean length of G . fabricii eaten by fulmars in Lancaster Sound was 143 mm (60 g), which corresponds t o an estimated age of 12 months (Bradstreet 1976). No soft parts of nereids and squid were found in the samples. Thus, the importance of these taxa as prey for seabirds might be less than indicated by the frequency of occurrence and numerical abundance measures. Other soft-bodied prey that disappear rapidly from the digestive tract of the seabirds would be difficult to identify if not exam- ined immediately after ingestion. We did not encounter organisms such as pteropods, cteno- phores and scyphozoan jellyfish in our samples, but we suspect that they are taken by seabirds in the Svalbard region equally as much as by seabirds in other regions (Harrison 1984). We have observed fulmars feeding on jellyfish in western Spitsbergen coastal waters (F. Mehlum, pers. obs.). The importance of ice-associated prey The marginal ice zone in the waters around Sval- bard comprises important habitats for several sea- bird species (Mehlum 1989). Certain species, such as the ivory gull, Ross’ gull Rhodostethia rosea and black guillemot, are mainly associated with the sea ice in pelagic areas, while others, such as the little auk, Briinnich’s guillemot, kittiwake, and fulmar, are distributed both in the marginal ice zone and in ice-free regions (Mehlum 1989). During the summer season a large proportion of Diet of high-arctic seabirds 17 the birds feeding in these areas are presumably non-breeding individuals (Mehlum 1989). In the marginal ice zone seabirds forage deep in the water column as well as on sympagic organ- isms associated with sea ice. The sympagic fauna available as prey for seabirds have been classified into two types (Melnikov & Kulikov 1980): (1) autochthonous ice-fauna associated with multi- year ice and including gammarid amphipods such as Gammarus wilkitzkii, Apherusa glacialis and Onisimus sp.; and (2) allochthonous ice-fauna, found only temporarily in the ice, which in the Barents Sea includes polar cod and is associated both with first-year ice and multi-year ice. In Svalbard, sympagic amphipods were not the main food source for any of the six seabird species studied. These amphipods were most frequent in the little auk and the black guillemot, but they occurred only in 16.7 and 9.5% of the birds exam- ined, respectively. In little auks, the percentage by weight of the different prey taxa in the summer samples showed that the sympagic amphipods A. glacialis and G . wilkitzkii accounted for as much as 78.9% of the total weight. However, this high abundance by weight was biased by the fact that all the 297 specimens were found in 6 of 21 birds (all 6 from the summer 1982 sample). It therefore seems that sympagic amphipods (the autoch- thonous ice-fauna) contribute little to the diet of seabirds foraging in ice-covered waters in the eastern Svalbard region. According to L m n e and Gulliksen (1991a), the abundance of the sympagic fauna depends on the age and the structure of the subsurface of the ice. They found high densities of sympagic amphipods under multi-year ice, while there were few or none under first-year ice. The proportion of multi- year ice in the Barents Sea varies from year to year but is generally small in comparison with the waters north of Svalbard and in the Fram Strait in the northern part of the Greenland Sea. Based on Vinje’s (1985) estimates, the maximum out- flow of multi-year ice from the Polar Basin to the Barents Sea is 30,000 km2, or about 15% of the marginal ice zone at its maximum extent. The relatively low proportion of multi-year ice in the Barents Sea compared to the waters north of Svalbard suggests a higher abundance of sym- pagic organisms north of Svalbard than in our sampling areas in the northern Barents Sea. This agrees with the results of Lenne & Gulliksen (1991a, b). In multi-year ice north of Svalbard, L ~ n n e & Gabrielsen (1992) found that sympagic amphipods were more important as prey for little auks and Briinnich’s guillemots than in first-year ice in the northern Barents Sea. In the coastal areas and fjords in western Spits- bergen there is usually no sea ice during the summer months, and thus the sympagic fauna is not a potential food source for seabirds foraging in these areas. Even though the sympagic fauna of amphipods may not be very important as prey for most sea- birds in ice-covered waters near Svalbard, they may be indirectly important as part of the prey of polar cod, at least in multi-year ice ( L m n e & Gulliksen 1989). However, the diet of polar cod was dominated by calanoid copepods and Para- themisro in their study. Bradstreet & Cross (1982) reported similarly that the diet of polar cod caught at the landfast ice-edge at offshore locations in Arctic Canada was dominated by copepods and Parathemisto, with only a small proportion of sympagic organisms present. The occurrence of sympagic amphipods in the diet of little auks during late summer agrees with Bradstreet (1982) for northwestern Baffin Bay. Bradstreet (1982) also described a seasonal shift in prey taken by little auks, where calanoid cope- pods dominated the May samples and Para- themisto, Apherusa glacialis and polar cod dominated in the August samples. It is apparent from our data and from Lydersen et al. (1989) that a similar seasonal shift in diet occurs in the Svalbard region from summer to autumn. Cope- pods, which dominated the diet in our spring and summer samples, were absent in little auks collected by Lydersen et al. (1989) in September- October in the Hornsund region of Svalbard; Parathemisto and polar cod and striped snailfish Liparis liparis dominated instead. The seasonal change in diet may be due to differences in avail- ability of prey. Diet similarities among species When comparing the diets of the different seabird species, we found the greatest similarity in the diets of black guillemots and kittiwakes foraging in ice-filled waters. Both feed on ice-associated prey, primarily polar cod. In contrast, their prey are quite different in coastal areas, where the black guillemot takes mainly suprabenthic prey. At Hornsund, Lydersen e t al. (1989) found the highest similarity in the autumn diets of the little auk and Briinnich’s guillemot, both ‘which feed 18 F. Mehlum & G . W, Gabrielsen mainly on polar cod and Parathemisto. The three pursuit-diving species studied (i. e. the three acids) differ considerably in size and feeding ecology. The Briinnich’s guillemot feeds on medium and large sized pelagic prey, whereas the little auk takes small pelagic prey, and the black guillemot suprabentic prey (mainly fish and crustaceans) in inshore regions and ice-associated prey in the marginal ice zone. In all the three surface feeding seabird species, fish, especially polar cod, are an important part of the diet. In the ice-covered areas these seabird species probably take advan- tage of the high abundance of polar cod associated with sea ice. Kittiwakes in Hornsund (spring 1985) consumed significantly larger polar cod than both Briinnich’s and black guillemots sampled during the same period. We have no good explanation for this difference, but the kittiwakes might have foraged in separate areas with a higher proportion of cod aged two-year-old or more. Foraging areas of birds collected in coastal regions During the breeding season, seabirds are con- strained in their foraging range by distance to the breeding site. During the chick-rearing period, the young have to be fed at certain rates in order to maintain normal growth before fledging. We have no detailed information of the feeding locations used most frequently by seabirds in Sval- bard and their relationship to where we collected birds in coastal areas. In southeast Svalbard, we have recorded Briinnich’s guillemots flying north- ward with fish in their beaks as far as 95km southeast of the nearest breeding colony (F. Mehlum unpubl.). We have also observed large numbers of seabirds feeding in coastal areas in Kongsfjorden and Hornsund, indicating that at least some prey are taken close to the colonies. The black guillemot, unlike other seabirds studied, feeds only in coastal areas close to the breeding site, usually less than 1 km distant (Cairns 1987). In contrast to other Atlantic alcids, their foraging range is constrained by the high workload needed to feed their two-chick clutches which they rear all the way to fledging (Mehlum et al. 1993). Some arctic seabird species are attracted t o special features along the coast where food is abundant. These include glacier faces and river outlets where prey organisms, especially crus- taceans, are concentrated and easily obtainable. The existence of such feeding areas in front of glaciers in Svalbard was observed at Nor- denskioldbreen in Billefjorden and Mittag-Lef- flerbreen in Wijdefjorden by Hartley & Fisher (1936). In their study, the two dominant seabird species foraging in the fjords along the glacier fronts were fulmar and kittiwake, which mainly preyed upon the euphausiid Thysanoessa inermis. Similar observations have been made elsewhere in Svalbard. Norderhaug et al. (1977) described the phenomenon at Smeerenburgbreen and Mon- acobreen in northwest Spitsbergen, while Mehlum (1984) reported concentrations of sea- birds along the front of the large glacier Austfonna on Nordaustlandet, as well as along Kongsvegen and the outlets of several small rivers in Kongsfjorden. The literature also contains similar reports from Greenland (Hartley & Dun- bar 1938) and the Canadian Arctic (McLaren & Renaud 1982; Renaud & McLaren 1982). The exact nature of the mechanism responsible for the concentration of prey organisms in such areas is not known. Mehlum (1984) suggests that the aggregation of organisms is caused by passive upwelling in connection with fresh water outlets beneath the glacier or at river outlets. The outflow of fresh water generates a reaction current of salt water (below the fresh water) towards the glacier face or river outlet, and the organisms are trans- ported upwards to the freshwater layer. When they experience lower salinity, they may be more or less immobilized and thus easy prey for the birds. In 1983, Mehlum (1984) observed large numbers of kittiwakes at river outlets in Kongsfjorden feeding upon the amphipod Par- athemisto libellula. Numbers of amphipods were found dead, apparently dying by the low salinity. Aarset & Aunaas (1987) have shown that this species is relatively intolerant t o low salinity water. The flow of fresh water from arctic glaciers and rivers varies through the summer and is cor- related to air temperature and precipitation (Repp 1979). The concentration of prey organ- isms most likely occurs only in periods with high fresh water outflow and may thus be an unreliable food source for the birds due t o possible variations between years. For example, the concentrations found outside Nordenskioldbreen in 1933 (Hart- ley & Fisher 1936) were not found in the summers of 1921,1926 and 1937 (F. C. Stott, pers. comm.). Similarly, seabird foraging activity at Austfonna was lower in 1982 than in 1981 (Mehlum 1984). Kittiwake concentrations at river outlets in Kongsfjorden were found in 1983 but not in 1984 Diet of high-arctic seabirds 19 (Mehlum 1984). We suggest that prey con- centrations at river outlets and glacier faces are utilised opportunistically by seabirds in years when these phenomena occur. The high pro- portion of Parathemisto and Thysanoessa found in our coastal samples of the surface feeding kittiwakes may be explained by such opportunistic foraging. Concluding remarks We would add that caution should be taken in the interpretation of our results dealing with the relative importance of different prey types in the diet of the six seabird species studied. In Table 2 we have for each seabird species pooled the data for all individuals sampled in coastal versus off- shore, ice-covered regions (see Materials and Methods). Thus, differences in diet composition among samples sets in each of the two habitat types were not accounted for. Furthermore, sample sizes of the single data sets were different, which might bias the averaged frequency of occur- rence presented in Table 2. These problems were partly eliminated in Figs. 2-7, by splitting the data into spring and summer samples. Because of this potential bias we did not perform statistical tests for comparing the pooled data sets. Because of low sample sizes, we also were careful in our interpretation of differences in occurrence or numerical abundance between single data sets. Even though we were not fully able to dem- onstrate the variability in the seabirds’ diet from one year to the other and despite the shortcomings of pooling sample sets, our extensive total material should give a reasonable general picture of the main prey types taken by the six seabirds species in the study areas and indicate differences among years and sampling localities. An interesting follow up study would be to investigate the importance of capelin to seabirds foraging in the marginal ice zone in the northern Barents Sea in years when the stock of this key fish species in the Barents Sea ecosystem is abundant (Sakshaug & Skjoldal 1989). Acknowledgements. -We thank H. Dahlen, I . Gjertz, C. Lyd- ersen, V. Ree, and E. Soglo for collecting seabirds, and I . Gjertz, 0. Kaasa, J. Koszteyn, 0. J. LBnne, and J. M. Wes- lawski for help with the analysis of stomach samples. We also want to thank M. Konarzewski and J. Taylor for providing the material of little auks from Hornsund 1987. R. Barrett, I. Gjertz, 0. J. LBnne, G. L. Hunt, and two anonymousreviewers made valuable comments to the manuscript. The study was partly supported by the Norwegian Research Program for Mar- ine Arctic Ecology (Pro Mare) of the Norwegian Research Council for Science and the Humanities. References Aarset A. V. & Aunaas, T. 1987: Osmotic responses to hypos- motic stress in the amphipods Gummarus wilkitzkii, Onisimus glacialis and Parathemisto libellula from Arctic Waters. Polar Biol. 7 , 18%193. Belopolskii. L. 0. 1957: Ecology of sea colony birds of the Barents Sea. [Transl. from Russian, 1961. Israel Prog. Sci. Tansl., Jerusalem.] Bradstreet, M. S. W. 1976: Summer feeding ecology of seabirds in Eastern Lancaster Sound, 1976. LGL Limited, Environ- mental research associates. Toronto, Canada. 187 pp. Bradstreet, M. S. W. 1980: Thick-billed m u m s and black guillemots in the Barrow Strait area, N.W.T., during spring: diets and food availability along ice edges. Can. 1. Z o o f . 58, 2120-2140. Bradstreet, M. S. W. 1982. Pelagicfeeding ecology of Dovekies, Alle alle. in Lancaster Sound and Western Baffin Bay. Arctic 35,126140. Bradstreet, M. S. W. & Cross, W. E. 1982: Trophic relation- ships at High Arctic ice edges. Arctic 35, 1-12. Cairns, D. K. 1987: The ecology and energetics of chick pro- visioning by black guillemots. Condor 89, 627-635. Dalpadado, P. 1989 Some observations on moulting, growth and maturation of krill (Thysanoessa inermis) from the Bar- ents Sea. 1. Plankton Res. I ! , 133-139. Dalpadado, P. & Skjoldal, H. R. 1991: Distribution and life history of krill from the Barents Sea. Polar Res. 10,443460. de Korte, J. 1972: Birds, observed and collected by “De Ned- erlanse Spitsbergen Expeditie” in West and East Spitsbergen, 1967 and 1968’69; first part. Beauforria 19, 113-150. Diamond, A. W. 1983: Feeding overlap in some tropical and temperate seabird communities. Studies in Aoian Biology 8, 24-46. Duffy, D. C. & Jackson, S. 1986: Diet studies of seabirds: a review of methods. Colon. Waterbirds 9, 1-17. Duffy, E. & Sergeant, D. E. 1950 Field notes on the birds of Bear Island. Ibis 92, 554563. Ekman, S. 1953: Zoogeography of the sea. Sidgwick and Jackson. London. 417 pp. Franz, H. G. & Diel, S. 1984: Secondary production of Calanus finmarchicus (Copepoda: Calanoidea) in a transitional system of the Fladen Ground Area (Northern North Sea) during the spring of 1983. XIX EMBS, Plymouth, Devon, K. 16-21 September 1984. Frost, K. J. & Lowry, L. F. 1981: Trophic importance of some marine gadids in northern Alaska and their body-otolith size relationships. Fish. Bull. U.S. 79, 187-192. Gjertz, I . & Lydersen, C. 1986. The ringed seal (Phoca hispida) spring diet in northwestern Spitsbergen, Svalbard. Polar Res. 4 , 53-56. Gjertz, I . , Mehlum F. & Gahrielsen G. W. 1985: Food sample analysis of seabirds collected during the ‘Lance’-cruise in ice- filled waters in Eastern Svalbard 1984. Norsk. Polarinst. Rapp. 23. Gulliksen, B. 1984: Under-ice fauna from Svalbard waters. Sarsia 69, 17-23. Hamre, J. 1988: Some aspects of the interrelation between the 20 F. Mehlum & G . W . Gabrielsen herring in the Norwegian Sea and the stocks of capelin and cod in the Barents Sea. lnternational Council for the Explo- ration of the Sea. C.M. 1988/H:42, 1-15. (Mimeo). Harrison, N. M. 1984: Predation on jellyfish and their associates by seabirds. Limnol. Oceanogr. 29, 13351337. Hartley, C. H. & Dunbar, M. J . 1938: On the hydrographic mechanisms of the so-called brown zones associated with tidal glaciers. J . Mar. Res. I , 305-311. Hartley. C. H. and Fisher, J . 1936: The marine foods of birds in an inland fjord region in west Spitsbergen. Part 2. Birds. J . Animal. Ecol. 5 , 370-389. Horn, H. S. 1966: Measurement of “overlap” comparative ecological studies. A m . Nut. 100, 41W24. Hunt, G . L. Jr. 1991: Marine birds and ice-influenced environ- ments of polar oceans. 1. Mar. Syst. 2 . 233-240. Ikeda, T. & Skjoldal, H. R. 1989: Metabolism and elemental composition of zoo-plankton from the Barents Sea during early Arctic summer. Mar. Biol. 100, 173-183. Loeng, H. 1989: Ecological features of the Barents Sea. Pp. 327-365 i n Rey, L. & Alexander, V. (eds.): Proc. 6th Conf. of the Comitt! Arctique Int; 13-15 May 1985. E. J . Brill. Leiden. Lydersen, C.. Angatyr, L. A,. Wiig, 0. & (britsland, T. 1991: Feeding habits of Northeast Atlantic harp seals (Phoca groen- landica) along the summer ice edge of the Barents Sea. Can. J. Fkh. Ayuat. Sci. 48, 2180-2183. Lydersen, C., Gjertz, 1. & Weslawski, J . M. 1989: Stomach contents of autumn-feeding marine vertebrates from Hornsund, Svalbard. Polar Rec. 25, 107-1 14. Lbnne, 0. J . & Gabrielsen, G. W. 1992: Summer diet of seabirds feeding i n sea-ice covered waters near Svalbard. Polar Biol. 12,68%92. Lsnne. 0. J . & Gulliksen, B. 1989: Size, age and diet of polar cod, Boreogadus saida (Lepechin 1773). in ice covered waters. Polar Biol. 9 , 187-191. Lsnne, 0 . J . & Gulliksen, B. 1991a: On the distribution of sympagic macro-fauna in the seasonally ice covered Barents Sea. Polar Biol. 1 1 . 4 5 7 4 9 . Lsnne, 0 . J. & GuUiksen, B. 1991b: Sympagic macro-fauna from multiyear sea-ice near Svalbard. Polar Biol. 1 1 , 471- 477. Lbvenskiold, H. L. 1964: Avifauna svalbardensis. Norsk Polarinst. Skrifter 129. 460 pp. McLaren, P. & Renaud, W. E. 1982: Seabird concentrations in late summer along the coasts of Devon and Ellesmere Islands, N.W.T. Arctic 35, 112-117. Mehlum, F. 1984: Konsentrasjoner av sjbfugl langs kanten av isbreer og utenfor breelver pi3 Svalbard. Fauna 37, 156-160. Mehlum, F. 1989: Summer distribution of seabirds in northern Greenland and Barents Seas. Norsk Polarinst. Skrifter 191, Mehlum F. 1990: Seabird distribution in the northern Barents Sea marginal ice-zone during late summer. Polar Res. 8, 61- 65. Mehlum F. & Bakken, V. (in press): Seabirds in Svalbard: status, recent changes and management. In Nettleship, D. 1-56. N., Burger, J . & Gochfeld, M. (eds.): Seabirds on Islands: Threats, Case Sfudies & Action Plans. International Council for Bird Preservation Techn. Publ. Cambridge, UK. Mehlum F., Gabrielsen, G. W. & Nagy, K. A. 1993: Energy expenditure by black guillemots (Cepphus g y l l e ) during chick-rearing. Colon. Waterbirds 16 (in press). Mehlum F. & Gjertz 1. 1984: Feeding ecology of seabirds in the Svalbard area - a preliminary report. Norsk. Polarinst. Rapp. 16. Melnikov, I. A. & Kulikov, A. S. 1980: The cryopelagic fauna of the central Arctic Basin. Biol Tsental‘nogo Arkticheskogo Basseina. Pp. 97-111. Nauka, Moscow. (Canadian translation of Fisheries and Aquatic Sciences 4910). Nesis, K. S. 1965: Distribution and feeding of young squids Gonatus fabricii (Licht.) in the Labrador Sea and the Norweg- ian Sea. Oceanology 5 , 102-108. Norderhaug. M., Brun. E. & Mbllen, G. U. 1977: Bar- entshavets sjbfuglressurser. Norsk Polarinst. Medd. 104, 1- 119. Opalinski, K. W. & Weslawski, J. M. 1989: Ecology, metabolic rate and metabolic adaptations in Spitsbergen amphipods. Pol. Arch. Hydrobiol. 36. 333-350. Pettibone, M. H. 1954: Marine polychaete worms from Point Barrow, Alaska, with additional records from the north Atlantic and north Pacific. Proc. U.S. NationalMuseum 103, 203-356. Pielou, E. C. 1984: The Interpretation of Ecological Data. A Primer on Classifcation and Ordination. Wiley & Sons, New York, 263 pp. Renaud, W. E. & McLaren, P. 1982: lvory Gull (Pagophila eburnea) distribution in late summer and autumn i n Eastern Lancaster Sound and Western Baffin Bay. Arctic 35, 141- 148. Repp. K. 1979: Breerosjon, glacio-hydrologi og materialutslipp i et hbyarktisk miljb, Brbggerbreen, Vest-Spitsbergen. Upubl. Ph.D. Thesis, Univ. Oslo (in Norwegian). Sakshaug, E. & Skjoldal, H. R. 1989: Life at the Ice Edge. Ambio 18, 60-67. Vinje, T. 1985: Drift, composition, morphology and distribution of the sea ice fields in the Barents Sea. Norsk. Polarinst. Skrifter 179C. 26 pp. Weslawski, J . M. 1987: Distribution of Decapoda (Crustacea) in South Spitsbergen coastal waters with remarks on their ecology and breeding biology. Polish Polar Res. 8. 121-134. Weslawski, J. M. & Kwasniewski 1983: Application of bio- logical indicators to determine the reach and origin of sea currents within the region of Spitsbergen. Pol. Arch. Hyd- robiol. 30, 189-197. Wing, B. L. 1976: Ecology of Parathemisto libellula and P. pacifca (Amphipoda: Hyperiidea) in Alaskan Coastal Waters. Northwest Fisheries Center, Auke Bay Fisheries Laboratory, Processed Report March 1976. NOAA. Zelickman, E. A. & Golovkin, A. N. 1972: Composition, structure and productivity of neritic plankton communities near the bird colonies of the northern shores of Novaya Zemlya. Marine Biology 17, 265-274.