Food consumption estimates of Barents Sea harp seals

The consumption of various prey species, required by the Barents Sea harp seal (Phoca groenlandica) stock in order to cover their energy demands, has been estimated by combining data on the energy density of prey species and on seasonal variations in the energy expenditure and body condition of the seals. Data on diet composition and body condition were collected in the period 1990-1996 by sampling harp seals during different seasons, in various areas of the Barents Sea. All diet composition data were based on reconstructed prey biomass, and adjustments were made for differences in digestibility of crustaceans and fish. The number of seals representing different age and sex groups were calculated for the entire population, and the monthly food requirements were estimated. In 1998, the total Barents Sea harp seal stock was estimated to comprise 2.22 million seals based on a mean production of 30 1,000 pups. After adjustments for a pup mortality of 30% its total annual food consumption was estimated to be in the range of 3.35-5 .05 million tonnes (depending on choice of input parameters). Assuming that there are seasonal changes in basal metabolic rate associated with changes in body mass, and that the field metabolic rate of the seals corresponded to two times their predicted basal metabolic rate, the annual food consumption of the Barents Sea harp seal stock was estimated. If capelin (Mallotus villosus) was assumed to be abundant, the annual total consumption was estimated to be 3.35 million tonnes, of which 1,223 ,800 tonnes were crustaceans, 807,800 tonnes were capelin, 605,300 tonnes were polar cod (Boreogadus saida), 212,400 tonnes were herring (Clupea harengus), 100,500 tonnes were cod (Gadus morhua) and 404,200 tonnes were "other fish". A very low capelin stock in the Barents Sea (as it was in the period 1993-1996) led to switches in seal diet composition, with increased consumption of polar cod (from ca. 16%-18 % to ca. 23%-25 % of total consumption), other gadoids (dominated by cod, but also including haddock (Melanogrammus aeglefinus) and saithe (Pollachius virens», herring, and "other fish" . Using the same set of assumptions as in the previous estimate, the total consumption would have been 3.47 million tonnes, divided between various prey species as follows (in tonnes): polar cod 876,000, codfish (cod, saithe and haddock) 359,700, "other fish" 618,800, herring 392,500, and crustaceans 1,204,200. Overall , the largest quantities of food were estimated to be consumed in the period June-September. In 1999, the total Barents Sea harp seal stock size was estimated to be 2.18 (95% CI, l.79 to 2.58) million animals, which would give an annual food consumption in the range of 2.69 3.96 million tonnes (based on upper and lower 95% confidence limits and adjusted for a pup mortality rate of 0.3) if capelin is assumed to be abundant. Nilssen, K.T. , Pedersen, O.-P. , Folkow, L.P. and Haug, T. 2000. Food consumption estimates of Barents Sea harp seals. NAMMCD Sci. Publ. 2:9-27. ------------------------------------------------------9 NAMMCO SCientific Publications, Volume 2 Fig. I: Moulfing hmp seals occur in great abundance in the loose pack ice of the southern Barents Sea in March, April and May. Photo:KjeJl Arlie Fagerheim Introduction Increased attention is being directed towards the study of multispecies interactions for the management of marine resources, This has encouraged studies of the feeding ecology of important top predators, including marine mammals. The harp seal (Phoca groenlandica) is the most abundant seal species in the Barents Sea (Fig. 1). Consequently, studies of its role as a top predator is considered important within a management context (Anonymous 1991). The harp seal is now included in a multi species model (MULTSPEC) which may provide the basis for future management of marine resources in the area (Bogs tad et al. 1997). Bioenergetic models have been used in previous attempts to assess the possible impact of Barents Sea harp seals on fish stocks (Markussen and 0ritsland 1991; Nord0Y et al. 1995a). The energy requirements of individual seals have been estimated and subsequently extrapolated to the entire seal stock. However, information on the feeding habits of Barents Sea harp seal was limited when these analyses were performed. Studies of harp seal feeding habits were carried out in 1986-1988 (Haug et al. 1991; Lydersen et al. 1991 ; Nilssen et al. 1992), and more comprehensive studies have continued in the period 1990-1996 to assess the feeding habits of Barents Sea harp seals throughout the year (Haug and Nilssen 1995; Nilssen 1995; Nilssen et al. 1995a,b, 1998a,b; Lindstr0m et al. 1998) . Seasonal variations in the distribution (Haug et al.1994) and body condition of the seals (Nilssen et al. 1997) have also been evaluated (Fig. 2). The aim of the present study was to app ly the new knowledge concerning seasonal changes in feeding habits , distributions and variations in body condition of Barents Sea harp seals, to estimation of the consumption of different prey species by the seal stock. Fig. 2: The condition of hmp seals is monitored by measuring blubber thickness. Photo: Tore Hallg . , Different assumptions concerning the energy requirements for activity and basal metabolic rate were used in order to determine the sensitivity of the consumption estimates to variations in metabolic parameters. MATERIAL AND METHODS Sampling of seals The sampling of seals accounted for the seasonal migration pattern of Barents Sea harp seals (Fig. 3) (Chapskii 1961 ; Benjaminsen 1979; Haug et al. 1994; Nord0Y et al. MS 1995b). Harp seals, taken as by-catch in gill net fisheries were collected in coastal areas of northern Norway in March and April. Harp seals were sampled in the southern Barents Sea (the East Ice) and White Sea from February to early May, and along the pack-ice belt in the northern parts of the Barents Sea between June and October (Table 1). Details on methods of capture, condition measurements, age determination, and the laboratory methods used for analyses of stomach and intestinal contents are given in Haug et al. (1991) and Nilssen et al. (l995a,b, 1997). Model inputs and assumptions Modelling strategy The energy of the food is converted to heat or mechanical work through oxidative processes, or is deposited in tissues through growth, stored in blubber deposits and, in pregnant females, transferred to the growing foetus and thereafter to the 10 ----------------------------------------------------Minke whales, hmp and hooded seals: Major predators ill the North Atlantic ecosystem growing pup during lactation. Energy is also transferred to heat during conversion of food to blubber, and when stored blubber is being mobilized. The sum of these energy requirements, the metabolizable energy, is, however, less than the gross energy intake, because energy is also lost in the urine and faeces (Lavigne et ai. 1982). The total consumption of the various prey items required by the harp seals to meet their energy demands is estimated by combining information on diet composition (Nilssen et ai. 1995a,b, 1998a,b; Lindstr0m et ai. 1998), with information on the energy density of various prey species (Pechenik et al. 1973; Martensson et al. 1996). Diet composition data were calculated as reconstructed prey biomass, and adjustments were made for differences in digestibility between crustaceans and fish (Martensson et al. 1994). The number of seals belonging to different age and sex groups were calculated, and monthly food requirements modelled based on dietary data, the energy densities of prey, and data on energy requirements. Stock size, age composition and length groups The total stock size of Barents Sea harp seals has been estimated to comprise 2.22 million seals, based on a mean production of 301,000 pups (ICES 1999). This estimate was used in the present study. We furthermore assumed that the number of pups (O-year old animals) would be 210, 700 (adjusted for a pup mortality rate of 0.3), and that all pups belonged to the 10 1-11 0 cm length group. The age-composition of 1,189 male and 508 female harp seals (aged 1+ years) taken by Norwegian sealers in the East Ice moulting lairs in April-May 1995 showed a very low representation of one and two years old animals (Nilssen et al. 1998a). The year classes born during the late 1980s were also very scarce, in particular the 1987 and 1988 year classes (Kjellqwist et ai. 1995). We assumed that seals 6 years old (approximately 150 cm standard body length) and younger were immature. Based on the observed 1995 age-composition (Nilssen et ai. 1998a) and available length at age data, the immatures (excluding the pups) were calculated to comprise 20% (384,400 animals) of the stock. These were assumed to be distributed with 96,lOO animals in each of the four body length groups 111-120, ... ,141-150 cm. We assumed that the number of pregnant females were equal to the 301,000 born pups and that they were equally distributed with 75,250 animals in the four body length groups 151160, ... ,181-190. Non-pregnant females were treated as males, and together they were calculated to comprise 1,236,600 animals, which yields 309,150 animals in each of the four body length groups longer than 150 cm. Seasonal variations in body and blubber masses Previous observations suggest that both blubber thickness and condition of adult harp seals vary on a seasonal basis (Sivertsen 1941; Sergeant 1973, 1991). Recent analyses of harp seal body condition in the Barents and White Seas confirm this seasonality (Nilssen et al. 1997). These new body condition studies revealed that harp seals generally are in a poor condition in spring and early summer (May-June). Condition improves during the course of the summer, and the seals are in good condition in August (Nilssen et ai. MS 1998b) and during the autumn. The energy Fig. 3: Annlla/migratioll pattern oj the Barellts Sea harp seal


Introduction
Increased attention is being directed towards the study of multispecies interactions for the management of marine resources, This has encouraged studies of the feeding ecology of important top predators, including marine mammals.The harp seal (Phoca groenlandica) is the most abundant seal species in the Barents Sea (Fig. 1).Consequently, studies of its role as a top predator is considered important within a management context (Anonymous 1991).The harp seal is now included in a multi species model (MULTSPEC) which may provide the basis for future management of marine resources in the area (Bogs tad et al. 1997).Bioenergetic models have been used in previous attempts to assess the possible impact of Barents Sea harp seals on fish stocks (Markussen and 0ritsland 1991;Nord0Y et al. 1995a).The energy requirements of individual seals have been estimated and subsequently extrapolated to the entire seal stock.However, information on the feeding habits of Barents Sea harp seal was limited when these analyses were performed.
Studies of harp seal feeding habits were carried out in 1986-1988 (Haug et al. 1991;Lydersen et al. 1991 ;Nilssen et al. 1992), and more comprehensive studies have continued in the period 1990-1996 to assess the feeding habits of Barents Sea harp seals throughout the year (Haug and Nilssen 1995;Nilssen 1995;Nilssen et al. 1995aNilssen et al. ,b, 1998a,b;,b;Lindstr0m et al. 1998) .Seasonal variations in the distribution (Haug et al.1994) and body condition of the seals (Nilssen et al. 1997) have also been evaluated (Fig. 2).
The aim of the present study was to app ly the new knowledge concerning seasonal changes in feeding habits , distributions and variations in body condition of Barents Sea harp seals, to estimation of the consumption of different prey species by the sea l stock.
Fig. 2: Th e condition of hmp seals is monitored by measuring blubber thickness.
Photo: Tore Hallg ., Different assumptions concerning the energy requirements for activity and basal metabolic rate were used in order to determine the sensitivity of the consumption estimates to variations in metabolic parameters.

Sampling of seals
The samp ling of seals accounted for the seasonal migration pattern of Barents Sea harp seals (Fig. 3) (Chapskii 1961 ; Benjaminsen 1979;  Haug et al. 1994; Nord0Y et al.MS 1995b).Harp seals, taken as by-catch in gill net fisheries were collected in coastal areas of northern Norway in March and April.Harp seals were sampled in the southern Barents Sea (the East Ice) and White Sea from February to early May, and along the pack-ice belt in the northern parts of the Barents Sea between June and October (Table 1).Details on methods of capture, condition measurements, age determination, and the laboratory methods used for analyses of stomach and intestinal contents are given in Haug et al. (1991) and Nilssen et al. (l995a,b, 1997).

Modelling strategy
The energy of the food is converted to heat or mechanical work through oxidative processes, or is deposited in tissues through growth, stored in blubber deposits and, in pregnant females, transferred to the growing foetus and thereafter to the 10 -----------------------------------------------------growing pup during lactation.Energy is also transferred to heat during conversion of food to blubber, and when stored blubber is being mobilized.The sum of these energy requirements, the metabolizable energy, is, however, less than the gross energy intake, because energy is also lost in the urine and faeces (Lavigne et ai. 1982).
The total consumption of the various prey items required by the harp seals to meet their energy demands is estimated by combin ing information on diet composition (Nilssen et ai.1995a,b,  1998a,b; Lindstr0m et ai.1998), with information on the energy density of various prey species (Pechenik et al. 1973;Martensson et al. 1996).Diet composition data were calculated as reconstructed prey biomass, and adjustments were made for differences in digestibility between crustaceans and fish (Martensson et al. 1994).The number of seals belonging to different age and sex groups were calculated, and monthly food requirements modelled based on dietary data, the energy densities of prey, and data on energy requirements.

Stock size, age composition and length groups
The total stock size of Barents Sea harp seals has been estimated to comprise 2.22 million seals, based on a mean production of 301,000 pups (ICES 1999).This estimate was used in the present study.We furthermore assumed that the number of pups (O-year old animals) would be 210, 700 (adjusted for a pup mortality rate of 0.3), and that all pups belonged to the 10 1-11 0 cm length group.
The age-composition of 1,189 male and 508 female harp seals (aged 1+ years) taken by Norwegian sealers in the East Ice moulting lairs in April-May 1995 showed a very low representation of one and two years old animals (Nilssen et al. 1998a).The year classes born during the late 1980s were also very scarce, in particular the 1987 and 1988 year classes (Kjellqwist et ai. 1995).We assumed that seals 6 years old (approximately 150 cm standard body length) and younger were immature.Based on the observed 1995 age-composition (Nil ssen et ai. 1998a) and available length at age data, the immatures (excluding the pups) were calculated to comprise 20% (384,400 animals) of the stock.These were assumed to be distributed with 96,lOO animals in each of the four body length groups 111-120, ... ,141-150 cm.
We assumed that the number of pregnant females were equal to the 301,000 born pups and that they were equally distributed with 75 ,250 animals in the four body length groups 151-160, ... , 181-190.Non-pregnant females were treated as males, and together they were calculated to comprise 1,236,600 animals, which yields 309,150 animals in each of the four body length groups longer than 150 cm.

Seasonal variations in body and blubber masses
Previous observations suggest that both blubber thickness and condition of adult harp seals vary on a seasonal basis (Sivertsen 1941;Sergeant 1973Sergeant , 1991)).Recent analyses of harp seal body condition in the Barents and White Seas confirm this seasonality (Nilssen et al. 1997).These new body condition studies revealed that harp seals generally are in a poor condition in spring and early summer (May-June).Condition improves during the course of the summer, and the seals are in good condition in August (Nilssen et ai. MS 1998b) and during the autumn.The energy

___________________________________________________ 11
Table J: Infonnati on about harp seals from collections along the Norwegian coast and in the Barents andWhite Seas during 1990-1996.The data on diets include relative composition of the diet based on stomach and intestine contents.The condition data include total body mass, standa rd length, girth, dorsal bubber thickness and blubber mass.If total body mass of the animals is known, and assuming that the equation is valid for all seasons, the blubber mass can be estimated for all length groups.Mean blubber masses were calculated for all length groups lO 1-11 0, ... ,181-190 em, and for all months for which data were available.Linear interpolations between months were carried out in periods where data were lacking (Table 2).The energy density of blubber (wet weight) was taken as 35 kJ/g (Nord0Y and Blix 1985).

Year
A similar procedure was used for calculation of total body masses for seals in all length groups (Table 3).

Growth
The resting metabolic rate of juvenile harp seals appears to be two times the BMR predicted according to Kleiber (1975) (e.g., Irving and Hart 1957;Folkow and Blix 1989).We assumed that the excess energy metabolism reflects the energy required for growth, and we have consequently added a growth factor corresponding to 3.4 .MO .75 (W) to the predicted BMR for all length groups less than 151 cm (immature animals) to account for such growth.This seems to be valid, as judged from the results obtained in studies of free living ringed seals (Phoca hispida) (Lydersen and Hammill 1993), and in cap- tive harbour sea l (Phoca vitulina) pups (Markussen et al. 1990).We assumed that no extra energy was required for growth in adult seals (animals longer than 150 cm).

Foetal growth
We assumed that foetal growth resulted in a foetal body mass of 10 kg in early March.
Growth was assumed to be exponential, and occurring mainly in January and February.The energy density of the foetus was assumed to be similar to minke whale foetuses, and constant at 3.8 kJ/g (Nord0y et al. 1995c).Pregnant females were assumed to use 130 MJ to meet the requirements of foetal growth in each of the months of January and February.This value includes both the heat of gestation and the energy content of the foetus.

Lactation
Energy losses incurred by females during lactation were not included specifically in the model.This is because the energy required for this purpose is mainly drained from blubber deposits (Lydersen and Kovacs 1996), so these energy requirements will largely be reflected in the change in blubber mass of females during lactation.

Urine and faeces
The metabolizable energy represents approximately 92% of the digestible energy, the remaining 8% of which is assumed to be lost in the urine (Lavigne et al. 1982).In addition, energy is lost in the faeces.This energy loss corresponds to about 8% of the energy content of fish and about 18% of the energy content of crustaceans (Nord0y et al. 1993 ;Mtutensson et al. 1994).Thus, only between 74% and 84% of the energy content of a meal is available for body growth and energy expenditure.This was taken into account in the model.

Feeding habits of harp seals
Our knowledge concerning the feeding habits of harp seals in winter (November-January) and in summer (July-August) is still limited.The decline in condition in spring and early summer (Nilssen et al. 1997) suggests that energy intake decreases during this period.The observation that harp seals improve in condition from June to August/September (Nilssen et al. 1997;Nilssen et al. MS 1998b) indicates that the summer and autumn are periods of intense feeding.The slow decrease in body mass during October-February (Table 3), suggests that feeding is reduced in this period.Such seasonal changes in food intake have been documented in captive harp seals (Lager et al. 1994) and grey seals (Nord0Y and Blix 1988), maintained under simulated natural light-darkness cycles.
Tables 4 and 5 summarize data on harp seal diets in terms of bulk biomass, based on studies carried out in the Barents Sea and in coastal areas of northern Norway during the period 1990-1996 (Nilssen et al. 1995a,b;1998a;Lindstmm et al. 1998).The Barents Sea stock of capelin (Mallotus villosus) collapsed in 199211993 (Hamre 1994;Gj0sreter 1995).For this reason, two versions of diet composition were used as inputs in the model.1998) when the capelin stock was at a low level.

Enelgy densities of prey
The energy densities of prey species were taken from Martensson et al. (1996).In months with several observations of energy densities for a given prey species, mean values were used.For "other crustaceans" the energy density for krill was used.For saithe (Pollachius virens) and haddock (Melanogrammus aeglefinus) , for which data were lacking, the energy density.for cod (Gadus morhua) was used, and for "other fish" the energy density was assumed to be 5 kJ/g.Energy density values for polar cod (Boreogadus saida) were based on regressions between fat content and energy density (Pechenik et al. 1973).The energy densities for polar cod va ried between 5.5 and 7.8 kJ/g, with the highest values in the period July-December (1. S. Chri stiansen, Norwegian Institute of Fisheries and Aquaculture, Troms0, Norway, pers.comm.).

RESULTS
Estimates of total food consumption The total food consumption of the Barents Sea harp seal stock (assumed to compri se 2.1 3 million seals including adjustments for a pup mortality of30%) was estimated to be in the range of 3.35-4.84 million tonnes (depending on choice of input parameters) with capelin abundant in the ecosystem in the Barents Sea.With a very low capelin stock in the Barents Sea, the estimated food consumption increased slightly, to val ues ranging between 3.47-5.05 million tonnes (Tables 6 and 7).
When calculati ons ofBMR were based on annual average body masses for each length group, consumption estimates increased by approx imately 1 % to 2.5%, compared to a situation where calculations of BMR were based on monthly variations in body masses for each length group throughout the year.
When the value for field metabolic rate was increased from 2 times BMR to 3 times, the consumption estimates increased by between 40.6% and 41.9%.When the consequences of monthly variations in body mass for the BMR estimates were taken into account, the total food consumption was found to be approximately 3.35 million tonnes with FMR = 2 .BMR, compared to an estimated consumption of approximately 4.72 million tonnes with FMR = 3 .BMR.Running the model with both a BMR and FMR that varied throughout the year resulted in consumption estimates 16.7% to 17.4% hi gher than doing so on the assumption of a variable BMR and a fixed FMR of 2 .BMR (Tabl e 6).
The model predicted that the largest quantities of food were consumed in the period June-September.Use of variable versus constant BMR input values resulted in minor differences but the estimated consumption in June was higher based on annual average M for each length group compared with energy requirements based on mean M for each length group in each month (Fig. 4 and 5).
In 1999, the total Barents Sea harp sea l stock size was estimated to be 2. 18 (95% CI, 1.79 to 2.58) milli on animals (lCES 1999).Assuming that the f ield metabolic rate of the sea ls corresponds to two times their estimated basa l metabolic rate, the annual food consumption of the Barents Sea harp sea l stock in 1999 was estimated to be in the range of 2.69 -3.96 milli on tonnes (based on upper and lower 95% confidence limits and adjusted for a pup mortality of 30%) if capelin is ass umed to be abundant (Table 8).

Consumption estimates of various prey species
For each of the two diets used in the various model runs, only minor variations occurred in the relative contribution of diffe rent prey species to the total harp seal consumption estimates.When cape lin was abundant in the ecosystem, the contribution of crustaceans was ca.35% to 37% of the total consumption estimates.Capelin contributed ca.23% to 27%, polar cod ca.17% to 18%, herring (Clupea harengus) ca.6%, cod ca.3% and "other f ish" ca.11 % to 12%, to the total consumption estimates (Table 6) ..
When taking monthly vari ations in body mass into account in predicting BMR, and assuming a FMR of 2 .BMR, the estimated consumption by harp seals of crustaceans was ca.1,223 ,800 tonnes, capelin ca.807,800 tonnes, polar cod ca.605 ,300 to nnes, herring ca. 2 12,400 tonnes, cod ca.100,5 00 tonnes and "other f ish" ca.404,200 tonnes (Fig. 6).It should be noted that "other fish" also includes gadoids such as cod, haddock and sa ithe, which were impossible to identify to species because of the extent of the digestion of the otoliths recovered in the stomach contents of the seals (N ilssen et af.1995a,b).

DISCUSSION
Barents Sea harp seals are known to display opportunistic feeding patterns in that different species are consumed in different areas and at different times of the year (Haug et al.1991 ;Lydersen et al. 1991;Nilssen 1995;Nilssen et al. 1992Nilssen et al. , 1995aNilssen et al. ,b, 1998a,b;,b;Lindstr0m et al. 1998).
However, the consumption estimates in this study, based on results from diet studies (Nilssen et al. 1995a(Nilssen et al. ,b, 1998a;;Lindstr0m et al. 1998), energy densities in the various prey species (Pechenik et al. 1973;Martensson et al. 1996), and energy requirements (e.g., Lager et al. 1994) revealed that the bulk of the harp seal diet comprised relatively few species, in particular capelin, polar cod, herring, cod, krill and Parathemisto libellula.
Crustaceans, mainly krill and Parathemisto libellula, seem to be important prey of harp seals during summer and autumn (July-October).The estimates of consumption of crustaceans showed little relative variation between the two diets applied in the model (Tables 6 and 7).The consumed biomass of crustaceans was estimated to be between ca.1,223,800 and ca.1,717,100 tonnes, or ca.32% to 37% of the total consumption.This is comparable to the estimate made by Nord0Y et al. (1995a), who found that crustaceans contributed about 35% of the biomass eaten by the Barents Sea harp seals.However, it should be kept in mind that the knowledge about the harp seal diet in June and July is poor, and that the contribution of krill to the harp seal diet in July was based only on the assumption that their diet in July was equivalent to that observed in August 1996 (Nilssen et al. MS 1998b).
Among the commercial harvested fish species in the Barents Sea, capelin, when abundant, seems to be one of the most important prey species of harp seals.Capelin contributed ca.24% to 27% to the total harp seal consumption.The consumed capelin biomass was estimated to vary between ca.807,800 -1,293,300 tonnes, when the species was available in the ecosystem.This is substantial when compared to the total fisheries (1,094,000 tonnes) and estimated total stock biomass (3.9 million tOlmes of 2+) of Barents Sea capelin in 1992 (Anonymous 1993).
The capelin stock collapsed during the winter of 199211993 (Hamre 1994;Gj0sa:ter 1995;Gj0sa:ter et al. 1998).The total Barents Sea capel in stock size was estimated to be only ca.195,000 tonnes in 1995, and with very poor production in the period 1993-1995 (ICES 1996).
The annual harp seal consumption estimates of capelin, after this, decreased significantly to approximately 22,900 -41,800 tonnes.
Krill of the genus Thysanoessa are known to be an important part of the diet of many commercially and ecologically important fish species in the Barents Sea, including the capelin (Gj0sa:ter 1998).Coinciding with a strong reduction in the older age-groups of the capelin stock in the late 1980s (Gj0sa:ter et al. 1998), there was a subsequent increase in abundance and biomass of Thysanoessa, and a possible predator -prey interrelationship between capelin and krill in the Barents Sea was suggested (Dalpadado and Skjoldal 1996).This may imply that a reduction of the capelin stock is more or less automatically followed by an increase in krill abundance, and may also to some extent explain the apparent large consumption of krill by harp seals in the area in the mid-1990s.
The lack of capelin as potential harp seal prey in the period 1993 -1996, seems also to have been compensated for by increased consumption of polar cod.The annual consumption of polar cod was estimated to increase from a level of ca.605,300 -805,200 tonnes in 1990-1992 to ca. 876,000 -1,2 12,000 tonnes in 1993-1996, depending on input parameters.This is very high considering the stock size evaluations (ca.600,000-1,000,000 tonnes) of polar cod in the eastern Barents Sea in 1992-1993 (Sunnana and Christiansen 1997).The polar cod is known to be an important prey for many of the top predators in Arctic marine ecosystems (Ponomarenko 1968;Bradstreet and Cross 1982).In the Barents Sea, the polar cod spawn in early winter along the west and south coast of Novaja Zemlja (Ponomarenko 1968).Russian studies, conducted during the 1930s, provided evidence that polar cod was the most important prey of harp seal in these areas during late autumn (Chapskii 1961).
The consumption of other gadoids, mainly cod, was estimated to increase from a range of ca. ---600

Months
approxi mately 2 million tonnes in 1995 (Anonymous 1996).The total consumption of gadoids is most likely larger, since the group denoted "other fish" also includes gadoids.In June-Ju ly harp seals are di stributed in open waters in the central and the southern part of the Barents Sea, in pack-ice waters from Novaja Zemlja in the east, along the pack-ice belt to the north, and in Svalbard waters (Haug et at. 1994).
However, since 1988 the stock size has gradually increased to a level comparable to that in the early 1960s, and the stock size was estimated to be about 2.8 million tonnes in 1994 (Anonymous 1994).Thus, increasing quantities of immature herring (O-group and recruits up to 4 years old) have been present in the southern Barents Sea (Anonymous 1993), and it is possible that immature herring may have been one of the most impOltant winter prey species for harp seals since 1989 (Nilssen 1995).The estimated millU- al consumption of herring in this study increased from a level of ca.212,400 -318,1 00 tonnes, depending on model input parameters when capelin was abundant, to a level of ca.392,500 -610,000 tonnes when the capelin stock was at a low level.These annual consumption estimates were lower than the estimated consumption by minke whales (ca.633,000 tom1es) (Folkow et al. this volume).The total herring fishery was approximately 900,000 tom1es in 1995 and the spawning stock size was estimated to be ca.3.9 million tonnes (ICES 1996).
The total annual food consumption of Barents Sea harp seals, as estimated to be within the range of ca.3.35 -5.05 million tonnes for 2.22 million seals in the present study, seems reasonable compared with the estimate by Nord0Y et al. (1995a) who calculated the total food consumption to be approximately 1 million tonnes for 600,000 seals (1+).Markussen and 0ritsland (1991) estimated the total food consumption, based on restricted knowledge about the feeding habits at that time, to vary from 1.4 to 4.4 million tonnes per year for 1 million harp seals in the Barents Sea.These estimates seem to be too high.In eastern Canada, the total prey consumption by 4.8 million harp seals was estimated to be 6.9 million tonnes in 1994 (Stenson et al. 1997), which seems reasonable with the lower range of the estimates in the present study.
The food consumption estimates are sensitive to the model assumptions.The most critical parameter for the total consumption estimates examined in the model , was the choice of multiplier for prediction of field metabolic rate from basal metabolic rate (FMR = a .BMR).If "a" was increased from 2 to 3, the food consumption estimate increased by approximately 41 %.Stenson et al. (1997) estimated an increased food consumption of 25% when they increased "a" from 2 to 2.5.The consumption estimates based on the lowest FMR ("a" = 2) in our study, seem to be most similar to estimates based on monitored energy expenditures of immature harp seals in captivity throughout the year (Nord0y et al. 1995a).The consumption estimates showed little sensivity to variations in the procedure for calculating basal metabolic rate (BMR) in the model (annual average BMR or monthly average BMR).
A contributing factor to the decline in harp seal body mass during spring may be seasonal variations in the energy content of prey species.Some of the prey species consumed by the seals display very low energy densities at this time of the year (Henderson et al. 1984;Martensson et al. 1996).Variations in energy densities, influenced by season, as well as size, geography and year, have also been observed for important prey species of marine vertebrate predators such as harp seals in the northwest Atlantic (Lawson et al. 1998).Seasonal changes in feeding intensity have also been observed in captive harp seals.Lager et al. (1994) reported that food intake in subadults fed ad libitum was generally low during moult (May-June).This was followed by a pronounced increase in feeding in the period July-October.During the period October-April, the food intake of captive seals was observed to be relatively stable.Thus, the decline in mass observed during the moult is probably due to reduced food intake, in addition to consumption of prey that are not particularly energy-rich.
Results from the present study suggest that Barents Sea harp seals consume substantial amounts of commercially exploited fish species, such as capelin, herring and cod.Data on the energy requirements of harp seals throughout the year, and increased knowledge of their feeding habits during summer and early winter, are essential in order to improve our estimates of the food consumption of this seal stock.Satellite tracking and remote sensing -a method which has been successful applied in studies of hooded seals (Cystophora cristata) (Folkow et al. 1996) -has also been applied to study Barents Sea harp seals (e.g.Nord0Y et al.MSI995b).The results are now being analysed and will improve our knowledge of harp seal activity and distribution patterns.Regular collection surveys are also required in order to monitor" changes in feeding habits and consumption of harp seals in the Barents Sea, in particular when the capelin stock has recovered.

Table 2 :
Average blubber masses' (kg) of harp seals, fo r each length group (cm) and momh, based on data collected in 1990-1996 in the Barems Sea.

Table 3 :
Average tOlal body masses' (kg) of harp seals, for each length group (cm) and month, based on data collected in 1990-1996 in the Barents Sea.Annual mean masses for each length group (cm) are also given.M Ol/fils'Linear interpolati ons between months (in italics).

Table 4 :
Relative harp seal diet composition (percentage wet weight) based on bulk biomass, with capelin abundant in the Barents Sea ecosystem,

Table 5 :
Relative harp seal diet composition (percentage wet weight) based on bulk biomass, with a depleted capelin stock in the Barents Sea ecosystem.

Table 6 :
Estimated consumption (i n 1,000 tonnes) of various fish species by the Barents Sea harp seal stock under various assumptions concemi ng basal metabolic rate (BMR) and the field metabolic rate (FMR), when capel in was abundant in the Barent Sea ecosystem Va.=BMR was predicted based on the observed monthly variations in the mean body mass for each length group.Mean=BMR was predicted based on the annual mean body mass for each length group.Percentage cont ribut ions of each prey group are given in paranthesis.

Table 7 :
Estimated consumption (i n 1,000 tonnes) of various fish species by the Barents Sea harp seal stock under various assumptions concerning basal metabolic rate (BMR) and the field metabolic rate (FMR), when capelin was depleted in the Barent Sea ecosystem Va.=BMR was predicted based on the observed monthly vari at ions in the mean body mass for each length group.Mean=BMR was predicted based on the annual mean body mass for each length group.Percentage contributions of each prey gro up are given in paranthesis.

Table 8 :
Estimated total consumption of prey by the Bare nts Sea harp seals based on the 1999 stock size estimates (based on upper and lower 95% confidence limits and adjusted for a pup mortality rate of 0.3) and assuming that capelin is abundant in the Bare nts Sea.Field metabolic rate (FM R) corresponds to two times their estimated basal metabolic rate (BMR).