CARBON FLUX TO SEABIRDS IN WATERS WITH DIFFERENT MIXING REGIMES IN THE SOUTHEASTERN BERING SEA

The southeastern Bering transfer-rates are over the middle (between combined


Introduction
The southeastern Bering Sea is a relatively wide continental shelf with low cross-shelf advection (Coachman and Chamell, 1979). Because advective flow is small, shelf waters become divided into four distinct mixing regimes separated by three fronts (Kinder and Schumacher, 1981).
An inner front (Schumacher et al., 1979) separates homogeneous coastal waters (less than 50 m depth) from stratified waters over the middle shelf (50 to 100 m depths). A middle front (Coachman and Charnell, 1979) separates a two-layer system on the middle shelf from a three-layer
Counts during 1978 and 1979 were made during PRO-BES cruises (Processes and Resources of the Bering Sea Ecosystem). Counts were made in a 90 ~ sector extending 300 m abeam from the ship while steaming at known speeds. Details on count procedure have been reported elsewhere (Hunt et al., 1980(Hunt et al., , 1981c. Possible bias by attraction of birds to ships was checked by observing birds from a helicopter in 1977. For ship-attracted species, a correction factor was computed from ship and helicopter counts, paired by location. Regions for analysis were delimited by bathymetry, with each region centering around the PROBES transect ( Fig. 1). Occupancy of each region by each species was defined as the average density for each month, multiplied by the number of days in the month, then summing these products over a period of maximum bird activity (April through August). The precision of this measure was taken as the average within-month variance, weighted by sample size. That is, the standard deviation was computed from the within-month sum of squares, rather than the total sum of squares.
Weights were obtained from birds collected in the southeastern Bering Sea. For non-breeding birds, daily food requirements were assumed to be 2.5 times the standard metabolic rate on the basis of work by MacMillen and Carpenter (1977) and by Weathers and Nagy (1980). Stan-dard metabolic rates (SMR) were obtained from the equation for non-passerine birds (Lasiewski and Dawson, 1967): where SMR = kcal d -1 and M= kg. The modified equation to account for diurnal activity (Aschoff and Pohl, 1970) was not used because seabirds were active at night. The following literature values were used to convert energy requirements to grams of carbon: 0.7 kcal assimilated kca1-1 ingested (Kendeigh et al., 1977); 5 kcal g-1 dry wt of fish (Nishiyama, 1977); and 0.4 g C g-1 dry wt of fish, squid and arthropods (Curl, 1962). A few estimates of daily food consumption by seabirds were available in the literature. These were compared to daily food (fish) requirements by using a conversion of 0.27 g dry wt g-1 wet wt (Wiens and Scott, 1975).
Behavioral observations suggested that density estimates for fulmar might be high due to attraction of birds to ships, including research vessels. The mean density for ship counts was 3.4 times that for helicopter counts. The difference was almost significant for a two-tailed test (paired t= 1.31, #=0.07). Since the variances involved were large and the sample size was small, it was prudent to compute the difference in means that could be detected with this sample size. The statistical test was sensitive enough to detect a 5-fold difference between ship and helicopter counts. To avoid Type II statistical error, we concluded that fulmar density estimated from ship counts fell within the range of 1 to 5 times the density estimated from helicopters. We reduced estimates of fulmar usage by a factor of 3.4 for the study areas.

Results
The common seabird species found in the Bering sea 20 showed considerable phenological variation (Table 1), which was related to differences in flight capacities. The al-]o cids, with small wings and short foraging ranges (Hunt etal., 1980(Hunt etal., , 1981b, decreased in numbers in June (Table 1) as the birds withdrew to the breeding colonies. This pattern was seen in 4 common alcids in the study area -Uria lomvia, Aethia pusilla, A. cristatella, and Cyclorrhyn-3o chuspsittacula.
The remaining species belong to the Procellariidae (tube-noses) and Laridae (gulls), families with well devel-'E 2o oped capacities for gliding and soaring. These species, with long foraging ranges, increased in density during the sum-lo met ( Table 1). Two of the procellariids (Puffinus griseus and P. tenuirostris) are non-breeding species moving into the Bering Sea after breeding in Australia and New Zealand. The two other procellariids (Fulmarus glacialis and Oceanodroma furcata) and the two larids (Rissa tridactyla and R. brevirostris) move onto the shelf to breed during the summer. Standing stock of seabirds was computed from weights shown in Table 2. Differences in standing stock of seabirds among shelf and slope regions depended on month (Fig. 2). Biomass rose sharply during the summer in the outer shelf and slope regions, while rising less sharply in the middle shelf region (Fig. 2). The sharper rise on the outer shelf was due to a seasonal influx of surface-feeding species (Fulrnarus glacialis, Oceanodroma furcata, Rissa tridactyla, and R. brevirostris) after May (cf. Table 1). This influx of surface-foraging species is related to the well-developed pelagic food web on the outer shelf (Hunt and Schneider, in preparation). The pattern for subsurface 30 foraging species was more complex. Alcids (Uria lomvia, Aethia pusilla, A. cristatella, and Cyclorrhynchus psittacula) left the middle and outer shelf regions in May ( Ta-20 ble 1), to be replaced by shearwaters (Puffinus griseus and P. tenuirostris). The observed pattern of change in standing Io stock between shelf regions was due to a seasonal influx of surface-feeding species, combined with seasonal replacements of subsurface-foraging species. The middle shelf sup-.~ o. ported about half the seabird biomass found on the outer and o slope regions. The average monthly biomass for the middle ~ Io shelf was 11.6 kg km -2, compared to 21.2 and 22.5 kg km -2 for the outer and slope regions, respectively (Fig. 2).
2o Food requirements of seabirds are not directly proportional to biomass differences because of the allometric relation between body weight and food requirements (Ta-30 ble 2). When the allometric factor is taken into account, food requirements were about equal for outer shelf and slope waters (Fig. 3). Food requirements were 37% lower on the middle shelf than the outer shelf (Fig. 3). This difference was due primarily to the lack of surface-feeding species on the middle shelf.
The food requirement of subsurface-foraging species was about equal in the middle and outer shelf regions, and only slightly reduced over the continental slope (Fig. 3).   Table 13 of Hunt et al. (1980) Thus, the waters over middle and outer shelf differed more in the pathways of energy transfer than in aggregate energy transfer to seabirds. The reduced carbon flux to zooplankton in waters over the middle shelf was associated with reduced energy transfer to surface-feeding birds. There was no replacement of subsurface-foraging birds by surfaceforaging birds in the outer shelf domain. Substantial carbon flux to seabirds occurred over the middle shelf region, despite a considerable transfer of carbon to benthic food webs.

Discussion
The differences we found in food requirements between one region and the next are independent of the formula used to compute food requirements. For seabirds, at least four techniques are available (Table 3). The most frequently used method for seabirds (Wiens and Scott, 1975) gave values that were 20% higher than ours (Table 3). A more recent method, using weight losses in starved seabirds (Jouventin and Mougin, 1981), gave values that were 60% higher than ours. Two other techniques ("daily energy budget" of Kendeigh etal., 1977, with interpolation to 5 ~ and "daily energy expenditure" of King, 1974) also gave values higher than ours. Such alternative methods, when applied to our data, produced higher values, although the differences among regions remained the same on a proportional basis, giving equal rates for the outer shelf and slope and a 36 to 40% reduction in rate for the middle shelf (Table 3).
Our estimates of aggregate trophic transfer to seabirds must be considered minimum values. The estimates presented here do not include energy transfer to young or eggs, since there was no way to distinguish breeding and non-breeding birds at sea. The estimates were also limited to a period of 153 d during the summer. Food transfer between September and March must amount to some appreciable fraction of the summer consumption by seabirds.
Several species (Fulmarus glacialis, PufjTnus griseus, P. tenuirostris) remain abundant into October. Large gulls (Larus glaucescens) were abundant along the outer shelf in October, 1980, and may move offshore in the fall.
The standard conversion factor of 78.3 kcal kg -1 d -1 (Lasiewski and Dawson, 1967) may be low for high-latitude seabirds, based on the fact that high-latitude species tend to fall above rather than below expected values. For Uria lomvia the observed metabolic rate was 159 kcal d -1, (Johnson and West, 1975) compared to the expected value  (Kendeigh et al., 1977) of 83.6 kcal d -~ (Lasiewski and Dawson, 1967). The observed metabolic rate of Oceanodromafurcata was 16 kcal d -~ (Iversen and Krog, 1972), compared to the expected value of 10.9 kcal d -~. Evidence from seabirds in captivity also suggests that the values we report are closer to minimum values than to average values. Food consumption by a captive murre (Uria aalge), with minimum effort expended for foraging, was 28% of its body wt d -~ (Sanford and Harris, 1967). The allometric value, using a conversion of 2.5 times basal rate, is 35% of body wt d-L A conversion factor of 2.5 times basal rate would appear to be the minimum for active, unrestrained birds. Ebbinge et al. (1975) summarized studies indicating that wild birds require energy at 2 to 4 times basal rates in order to move and feed. For seabirds, Jouventin and Mougin (1981) reported weight losses that amounted, on a caloric basis, to 4 times basal costs, assuming no water losses during their measurements.
The lack of any independently verified estimate of daily food consumption by a bird at sea is the most serious limitation on our estimates of carbon flux at sea. One solution to this problem may lie in examining digestion rates and the proportion of recently fed birds out of a large collection made at sea, a modification of the technique used by Bajkov (1935) for fish, and applied to breeding seabirds by Uspenski (1956) andBelopol'skii (1957).
It is instructive to compare estimates of trophic transfer obtained in this study with those previously calculated for this and other areas. A value of 50 mg C m -2 for the 153 d season was taken as the lower limit for carbon flux to birds on the outer shelf. This was converted to grams (wet weight) of food (0.5 g food m -2 for the 153 d season) for comparison to other studies. Resutts of other studies were similarly converted, using literature values listed in the "Materials and Methods" section. Favorite etal. (1977) used a trophic transfer of 2.5 g food m -2 yr -~ to birds in their model for the entire southeastern Bering Sea; Hunt etal. (1981 a) estimated the rate to be 0.7 to 1.4 g food m -2 yr -1 over the same area. These two estimates included shallow coastal waters (50 m or less), which support large aggregations of shearwaters not found in the present study area. For the area around a breeding colony at a similar latitude in the North Sea, Furness (1978) obtained a value of 1.9 g food m -2 yr -1. These values are rather similar when compared to values from non-shelf ecosystems. For the eastern north Pacific Ocean, Sanger (1972) obtained a value of 0.09 g food m -2 yr -1. For the upwelling system off the Oregon coast (USA), Wiens and Scott (1975) obtained values of 8 g m 2 yr-1. For the Peruvian Upwelling, Schaeffer (1970) reported values ranging from 11 to 45 g m -2 yr -1, depending on the year.
For the Bering Sea, we compared seabird food requirements to primary production, under the assumption of limited carbon transfer between mixing regimes. Minimum carbon flux to birds during summer, away from breeding colonies, was 0.03% of primary production over the middle shelf, and 0.05% over the outer shelf and slope waters. Annual carbon flux to birds will be higher, since some species remain on the shelf until November, and other species move into the area during the winter. Carbon flux to birds in the vicinity (less than 100 km) of the Pribilof breeding colonies is at least an order a magnitude higher, based on bird densities around these colonies (Hunt et al., 1980). Comparison with a more familiar marine organism, a commercially important fish, may perhaps provide some sense of scale. The walleye pollock Theragra chalcogramma supports a major fishery on the outer shelf of the southeastern Bering Sea. Juvenile pollock (2 to 20 cm) are a major prey item for birds in this area , while adult pollock (20 to 60 cm) have roughly the same mass as birds (0.06 to 1.5 kg) and take some of the same prey (Smith, 1981). Consumption of pollock by birds has been estimated at 1.5 • 105 metric tons yr -1 .
Seabirds, because they are ubiquitous and easily measured apex predators, are valuable for examining the transfer efficiency of marine food webs. For regions of similar primary productivity, such as the North Sea and the southeastern Bering Sea (Coachman and Walsh, 1981), one would predict similar standing stocks of seabirds with similar food requirements, unless transfer efficiency were much greater in one ecosystem than the other. Seabirds should prove to be useful in examining mechanisms contributing to high transfer efficiency, as well as in identifying areas with high transfer efficiency.