Forage Fishes in the Bering Sea: Distribution, Species Associations, and Biomass Trends

Transcription

1 Dynamics of the Bering Sea CHAPTER 24 Forage Fishes in the Bering Sea: Distribution, Species Associations, and Biomass Trends Richard D. Brodeur, Matthew T. Wilson, and Gary E. Walters Alaska Fisheries Science Center, Seattle, Washington Igor V. Melnikov Pacific Research Institute of Fisheries and Oceanography (TINRO), Vladivostok, Russia Abstract Relatively little is known about distribution patterns and species associations of forage fishes in the Bering Sea despite their importance as major prey of many higher trophic level organisms, such as seabirds, marine mammals, and predatory fishes. In this study, we examined survey data on some dominant pelagic forage fishes (Pacific herring Clupea pallasi, capelin Mallotus villosus, and eulachon Thaleichthyes pacificus) and the juvenile stages of major commercial groundfish species (walleye pollock Theragra chalcogramma and Pacific cod Gadus macrocephalus). We analyzed two main data sets: (1) a 1987 Russian survey that covered most of the Bering Sea, and (2) National Marine Fisheries Service (NMFS) summer surveys ( ) in the eastern Bering Sea which sampled the same grid of stations each year. In the Russian survey, age-0 pollock had the highest biomass and were the most widely distributed forage fish, although jellyfish and age-2+ pollock dominated the biomass overall. Several geographically distinct assemblages were recognized in both the eastern and western Bering Sea. Age-0 pollock were associated with warmer bottom temperatures and capelin with colder bottom temperatures, compared with other species. Distributions of all species from the NMFS surveys during a cold year (1986) were more widespread and overlap among species was greater than during a warm year (1987). Herring showed the most dramatic fluctuations in their biomass index over 14 years of NMFS trawl surveys and was the dominant forage fish caught in most years, although when their biomass index was low, they were exceeded by age-1 pollock, eulachon, and capelin.

2 510 Brodeur et al. Forage Fishes in the Bering Sea Introduction The Bering Sea shelf is highly productive and contains some of the largest populations of fishes, crabs, marine mammals, and seabirds in the world. Although more than 300 species of fish are known to inhabit this shelf, the 20 most abundant species account for more than 98% of the total abundance of survey catches (Bakkala 1993). Our knowledge of the small non-commercial pelagic fishes and juvenile stages of commercially important demersal fishes is meager, in spite of the fact that these species are an important food base for many of the higher trophic levels, including fishes, birds, and mammals (Minerals Management Service 1987, Springer 1992, Alaska Sea Grant 1997). The limited long-term dietary information on these top-level predators (Hunt et al. 1996, Sinclair et al. 1996) and survey data (Naumenko et al. 1990, Fritz et al. 1993, Naumenko 1996) that do exist suggest that the abundance and distribution of many of these forage fishes have undergone dramatic changes through time, which may have important implications for the apex predators which depend upon these resources (Springer 1992, Alaska Sea Grant 1997). One of the important data gaps recognized by the National Research Council study of the Bering Sea ecosystem was that our understanding of forage species is inadequate and that there is almost no monitoring of the forage species upon which top predators rely (National Research Council 1996). Reliable estimates of how species biomass and distribution patterns change through time and how they are affected by environmental conditions are needed for effective management and as inputs to ecosystem modeling studies. Despite the importance of juvenile walleye pollock (Theragra chalcogramma) and other forage fishes in the Bering Sea ecosystem (Springer 1992, Livingston 1993, Brodeur et al. 1996), relatively little is known about the large-scale distribution patterns of these fish in the Bering Sea. This situation contrasts sharply with that in the Gulf of Alaska where the largescale distribution patterns of juvenile pollock and other small fishes are known (Walters et al. 1985, Hinckley et al. 1991, Bailey and Spring 1992, Brodeur et al. 1995, Brodeur and Wilson 1996, Wilson et al. 1996). Although these fishes are often captured incidentally in standard bottom trawl surveys such that an index of the relative size of the year class can be obtained (Walters 1989, Wyllie-Echeverria 1996, Hunt et al. 1996), directed surveys are often required to get more precise estimates of their abundance (Koehler et al. 1986, Wilson et al. 1996). This paper reports results from late-summer Russian surveys conducted in 1987 of the entire Bering Sea and analyzes distribution and species associations of the dominant forage fishes. We also present data on incidental catches of forage fishes in the eastern Bering Sea from multiyear ( ) surveys carried out by U.S. fisheries scientists and analyze interannual variation in the distribution patterns, species associations, and trends in biomass in the eastern Bering Sea, focusing on two adjacent

3 Dynamics of the Bering Sea 511 but highly contrasting years. Our objective was to determine how the geographic and interannual patterns of biomass distribution relate to variations in depth, temperature, and composition of co-occurring species. It is beyond the scope of this paper to present detailed information on the life history and ecology of the species examined, and the reader is referred to existing literature on forage fishes (e.g., papers in Minerals Management Service 1987; Alaska Sea Grant 1997). Methods Juvenile Fish Surveys We examined data from two surveys designed to determine the distribution of age-0 pollock in the Bering Sea. The first cruise took place aboard the Russian Pacific Research Fisheries Centre (PFRC) research vessel (R/V) Darwin during August and September Scientists from the Alaska Fisheries Science Center (AFSC) participated on this cruise and trawl collections were processed using standard AFSC sampling methodology. The main fishing gear was a pelagic rope trawl (77.4/212) with small mesh (10 mm) inserts extending from the codend up 15 m into the trawl. Horizontal opening (HO) of the trawl was measured as 40 m and vertical opening (VO) was m. Approximately m 3 of water was filtered in a typical 1 hour tow. A few stations were sampled with a 108/528 pelagic trawl (HO = 65 m, VO = m) and their catches were standardized relative to the main trawling gear based on differences in mouth area. Sampling was conducted on a predetermined grid of 149 stations over much of the eastern Bering Sea shelf and Aleutian Basin (Fig. 1) in a stepoblique fashion. A second cruise was conducted during August and September of 1987 aboard the R/V Gnevny. This survey consisted of 183 trawls carried out in the western Bering Sea shelf and basin using two different trawl types. The main gear used was the 108/528 trawl described previously, but at about one-third of the stations, a 118/620 trawl (HO = m, VO = 60 m) was used. The fish catch was identified to species and the invertebrate catch was classified only to major taxonomic categories (e.g., Cnidaria, Cephalopoda, Natantia). Pollock made up the majority of the fish biomass caught and was therefore subdivided by age group. Lengths were measured on a subsample of all pollock collected and individuals were assigned as age-0, age-1, and age-2+ based upon modes in the length-frequency distribution. The biomass caught from both gear types was standardized relative to the smaller trawl used on the R/V Darwin and converted to kg per hour trawled. Although the extensive coverage of the 1987 Russian surveys required two months to complete, no mortality or growth corrections were applied prior to making areal comparisons of fish density and size because of a lack of available information on these parameters for most species included here.

4 512 Brodeur et al. Forage Fishes in the Bering Sea Figure 1. Location of stations sampled during the R/V Gnevny (open circles) and R/V Darwin (closed circles) cruises during August and September Also shown are the 200 m and 3,000 m isobaths. To associate the distributions of forage fishes to the biotic community, we used a polythetic clustering technique (Clifford and Stephenson 1975) to detect existing natural associations among species and among stations. Species clusters were used to provide some indication as to how similar the distributions of the species were, whereas the station clusters were useful in determining the spatial extent of natural communities. Because rare taxa may adversely afftect the analyses, only those that occurred in >10% of the tows in each survey were included. Biomass estimates (kg/h) of the dominant species at each station was ln(x+1) transformed to reduce the sensitivity to large catches. From these logtransformed species-station matrices, dissimilarity indices were calculated using the QSK measure (Faith et al. 1987) and clusters were formed using a group-average fusion strategy (Clifford and Stephenson 1975). For each cruise, a station and species dendrogram was produced. Station groupings were made for various numbers of groups and these groups were plotted to find levels of dissimilarity that gave geographically cohesive groupings. Two-way tables (species station groups) were constructed to show how species were distributed within and among the station groups. Indices of constancy (Boesch 1977) were calculated for each taxon to see how distinct each grouping was. Constancy is defined as the ratio of the occurrences in a group to the total possible number of occurrences and ranges from 0 (taxon not found at any stations) to 1 (found at all stations).

5 Dynamics of the Bering Sea 513 Figure 2. Station locations for the standard Bering Sea bottom trawl survey grid occupied each summer since 1982 and the six areas used to stratify the sampling area by depth and location (northwest or southeast). The smaller boxes denote areas of increased sampling density. Finally, we examined bottom temperature and depth data for the R/V Darwin survey to determine how these variables relate to the observed fish distribution patterns. We statistically compared the cumulative distributions of select taxa and these variables following the methods described in Perry and Smith (1994) and Smith (1996). We compared the actual cumulative distribution of each taxon along with 1,000 randomizations of the data to the habitat available during each cruise. Retrospective Analysis of Eastern Bering Sea Forage Fishes We examined the catch of five species of forage fishes (age-1 pollock, age- 1 Pacific cod Gadus macrocephalus, eulachon Thaleichthyes pacificus, Pacific herring Clupea pallasi, and capelin Mallotus villosus) collected during the annual AFSC summer surveys in the eastern Bering Sea (Bakkala 1993). Virtually the same systematic sampling grid has been occupied since 1982 and consists of approximately 355 stations spaced 37 km apart for a total survey area of 465,000 km 2 (Fig. 2). The same gear (83/112 eastern bottom trawl; 2.3 m VO; 32 mm codend liner), towing method (0.5 hour tows at 3 knots), and processing methods were used throughout the sampling period (Bakkala 1993). The surveys were generally conducted from June 1

6 514 Brodeur et al. Forage Fishes in the Bering Sea Table 1. Dominant taxa collected during R/V Darwin survey in the eastern Bering Sea in 1987 used in cluster analysis. Shown are the frequency of occurrence in 149 tows and the mean biomass collected per hour towed. Freq. Mean Scientific name Common name occ. kg/h Cnidaria Jellyfish Theragra chalcogramma (age 2+) Walleye pollock Theragra chalcogramma (age 0) Walleye pollock Oncorhynchus keta Chum salmon Zaprora silenus Prowfish Cephalopoda Unidentified squid Gadus macrocephalus Pacific cod Pleurogrammus monopterygius Atka mackerel Clupea pallasi Pacific herring Mallotus villosus Capelin Stenobrachius leucopsarus Northern lampfish Theragra chalcogramma (age 1) Walleye pollock Pleuronectes asper Yellowfin sole Leuroglossus schmidti Northern smoothtongue Blepsias bilobus Crested sculpin 22 <0.1 Podothecus acipenserinus Sturgeon poacher Natantia Unidentified shrimp 20 <0.1 Oncorhynchus tshawytscha Chinook salmon to August 5 on NOAA research vessels and chartered fishing vessels. Although the trawl survey was designed to assess the abundance and distribution of bottom fishes, some pelagic species were captured mainly during the deployment and retrieval of the net. Thus, the abundances of the pelagic schooling species (capelin, eulachon, and herring) can only be considered a relative index of the total population abundance. We mapped the biomass distribution patterns (kg/ha) for each of the major forage fishes by year to look at long-term changes in spatial distribution. We were interested in examining differences in distributions for cold and warm years within a particular species. We chose to compare and contrast 1986 (cold year) and 1987 (warm year) which were quite different oceanographically. An added benefit to using the year 1987 is that we could then compare the AFSC distributions for that year with those of the R/V Darwin. We subsequently compared the catches of all five species for both years with the bottom depth and temperature at each station as described in the previous section to understand how these variables related to the observed distributions.

7 Dynamics of the Bering Sea 515 Table 2. Dominant taxa collected during R/V Gnevny survey in the western Bering Sea in 1987 used in cluster analysis. Shown are the frequency of occurrence in 183 tows and the mean biomass caught per hour towed. Freq. Mean Scientific name Common name occ. kg/h Cnidaria Jellyfish Theragra chalcogramma (age 2+) Walleye pollock Oncorhynchus keta Chum salmon Theragra chalcogramma (age 0) Walleye pollock Cephalopoda Unidentified squid Oncorhynchus nerka Sockeye salmon Oncorhynchus tshawytscha Chinook salmon Mallotus villosus Capelin Clupea pallasi Pacific herring Aptocyclus ventricosus Smooth lumpsucker Theragra chalcogramma (age 1) Walleye pollock Lumpenus maculatus Daubed shanny 31 <0.1 Stenobrachius leucopsarus Northern lampfish Eumicrotremus sp. Unidentified lumpsucker 27 <0.1 Ammodytes hexapterus Pacific sand lance 26 <0.1 Leuroglossus schmidti Northern smoothtongue Lampetra tridentata Pacific lamprey 21 <0.1 Zaprora silenus Prowfish 21 <0.1 Natantia Unidentified shrimp Reinhardtius hippoglossoides Greenland halibut 19 <0.1 Blepsias bilobus Crested sculpin 19 <0.1 Hippoglossoides elassodon Flathead sole 18 <0.1 Hemilepidotus papilio Butterfly sculpin 18 <0.1 Results Forage Fish Distributions Out of 62 distinct taxa/age groups identified during the R/V Darwin survey in the eastern Bering Sea, 18 occurred in >10% of the hauls and were included in the cluster analysis (Table 1). Of these, the three age groups of pollock and scyphozoan jellyfish comprised 91.4% of the total mean biomass. Because of their relatively high individual weight, age-2+ pollock comprised nearly 55% of the total biomass of these dominant species. For the R/V Gnevny survey, 64 taxa/age groups were found and 23 were included in the clustering (Table 2). There was substantial overlap in

8 516 Brodeur et al. Forage Fishes in the Bering Sea the dominant groups in the two surveys in that 14 of the 18 commonly occurring groups in the R/V Darwin survey were also dominant in the R/V Gnevny survey. Age-2+ pollock dominated the catch in terms of weight (85.9% of total) and was second only to jellyfish in number of occurrences. Herring, the next most important species in weight, constituted only 4.7% of the total biomass. The distribution of age-0 pollock biomass was skewed toward the eastern Bering Sea shelf region with highest concentrations found around the Pribilof Islands and along the Alaska Peninsula (Fig. 3a). In the western Bering Sea, high biomass of age-0 pollock was found only in Karagin Bay. Age-1 pollock were found mainly on the outer eastern Bering Sea shelf, almost entirely north of the Pribilof Islands, with high biomasses found in only a few isolated areas of the Gulf of Anadyr and Bay of Olyutorsk in the western Bering Sea (Fig. 3b). Herring biomass showed a similar distribution to age-1 pollock but extended farther south to the Aleutian Islands in the east and Karagin Bay in the west (Fig. 3c). Capelin were distributed along the same latitudinal range as herring but tended to be found in the inner shelf regions on both sides (Fig. 3d). Species Associations Three major species groupings were identified using the dissimilarity index at the 0.65 level in the eastern Bering Sea (Fig. 4a). Five taxa were associated with no others at this level (Table 3). Station Group 1 consisted of mainly open ocean pelagic and mesopelagic nekton (Species Group 3) but also showed high constancy and biomass relationships with age-2+ pollock, jellyfish, and prowfish (Zaprora silenus) (Table 3). Station Group 2 occurred entirely on the shelf and was dominated by all age-0 pollock and jellyfish. Station Group 3 was dominated in terms of constancy and biomass by age-2+ pollock and jellyfish (Table 3). All three age groups of pollock and Pacific cod were important to the Station Group 4, which was confined to a relatively narrow part of the shelf north of the Pribilof Islands (Table 3, Fig. 4a). Station Groups 5 and 6 were confined to the northeast corner of the study area and were associated mainly with capelin, chinook salmon (Oncorhynchus tshawytscha), and age-0 pollock. Station Group 7 was found in the westernmost part of the study area and comprised mainly Atka mackerel (Pleurogrammus monopterygius) and jellyfish (Table 3, Fig. 4a). For the western Bering Sea, five species groups and four nonaligned taxa groups were identified using the dissimilarity index at the 0.75 level along with six station groups (Fig. 4b). Station Groups 2 and 4 contained only one station each (Table 4) and were not considered in our analysis. Station Group 1 occurred mainly on the shelf and showed high constancy and biomass estimates for the pelagic species groups (5 and 8) which included all the Pacific salmon species and pollock age-0 and 2+ classes (Table 4). In contrast, the relatively minor Station Group 3 was dominated by herring and age-1 pollock. Age-1 pollock and Species Group 5 (especial-

9 Dynamics of the Bering Sea 517 Figure 3. Distribution of age-0 pollock (a), age-1 pollock (b), herring (c), and capelin (d) from combined R/V Darwin and R/V Gnevny surveys conducted in 1987.

10 518 Brodeur et al. Forage Fishes in the Bering Sea Figure 4. (a) Distribution of station groups resulting from numerical classification of biomass of the dominant species during the R/V Darwin cruise. Also shown is the species dendrogram up to the 0.65 dissimilarity level (inset). The species groups are given in Table 3. (b) Distribution of station groups resulting from numerical classification of biomass of the dominant species during the R/V Gnevny cruise. Also shown is the species dendrogram up to the 0.75 dissimilarity level (inset). The species groups are given in Table 4.

11 Dynamics of the Bering Sea 519 Table 3. Constancy (C) and mean biomass (B) of species in station groups for R/V Darwin 1987 data. Biomass is expressed as kg per hour towed. Station group & 6 7 Species Number of hauls group Taxon C B C B C B C B C B C B 1 Pleurogrammus monopterygius 2 Zaprora silenus Oncorhynchus keta Leuroglossus schmidti < Cephalopoda < < Stenobrachius leucopsarus Oncorhynchus tshawytscha Clupea pallasi Theragra chalcogramma , Theragra chalcogramma Pleuronectes asper Theragra chalcogramma Cnidaria Podothecus acipenserinus Gadus macrocephalus 0.1 < Blepsias bilobus Mallotus villosus Natantia 0.1 < < <

12 520 Brodeur et al. Forage Fishes in the Bering Sea Table 4. Constancy (C) and mean biomass (B) of species in station groups for R/V Gnevny 1987 data. Biomass is expressed as kg per hour towed. Station group Species Number of hauls group Taxon C B C B C B C B C B C B 1 Ammodytes hexapterus < < Clupea pallasi Lumpenus maculatus 0.2 < < <0.1 3 Hippoglossoides elassodon 0.1 < < < <0.1 Theragra chalcogramma Reinhardtius hippoglossoides 0.2 < < <0.1 Hemilepidotus papilio 0.2 < < < <0.1 4 Natantia < Mallotus villosus < < Theragra chalcogramma < < Cnidaria Blepsias bilobus < < Oncorhynchus tshawytscha Eumicrotremus sp. 0.2 < < < <0.1 7 Zaprora silenus <0.1 8 Aptocyclus ventricosus 0.1 < Theragra chalcogramma Cephalopoda Lampetra tridentata <0.1 Oncorhynchus keta Oncorhynchus nerka < Stenobrachius leucopsarus Leuroglossus schmidti

13 Dynamics of the Bering Sea 521 ly capelin and jellyfish) were important to the inshore Station Group 5 (Table 4, Fig. 4b). Finally, Station Group 6 comprised mainly age-2+ pollock and the mesopelagic Species Group 9 (Table 4). Retrospective Analysis of Eastern Bering Sea Forage Fishes The distribution of bottom temperatures showed dramatic differences between 1986 and 1987 (Figs. 5a and 5b). During 1986, a cold pool of <2 C bottom water penetrated almost to the Alaska Peninsula (56 N), whereas in 1987, it extended only a short distance south of 60 N. Age-1 pollock showed a much broader distribution during the cold year of 1986 compared with 1987 when they were relatively sparse and confined mainly to the middle shelf region (Figs. 5c and 5d). Age-1 Pacific cod exhibited a less dramatic change in their distribution than pollock but were found more on the middle shelf during 1986 compared with the more inner shelf distribution of 1987 (Figs. 5e and 5f). Herring were substantially more widely distributed in 1986 and were found well offshore compared with 1987 (Figs. 5g and 5h). Capelin were found mainly in the northeast part of the study area but extended farther south and west in 1986 than in 1987 (Figs. 5i and 5j). Finally, eulachon were restricted almost entirely to the southern part of the shelf (south of 57 N) but their distribution was also more widespread in 1986 than in 1987 (Figs. 5k and 5l). As can be seen in these figures, there was some overlap in the distribution ranges of these five forage fish species. We compiled the number of species occurrences at each station for the same two years (Fig. 6) and observed that the overlap was higher in a colder year (1986; Fig. 6a), when the species distributions are expanded, than in a warmer year (1987; Fig. 6b). For all 14 years combined, age-1 pollock were the most widespread in their distribution and most likely to be found when only one species was present (i.e., single species found at 34% of the sites, of which 70% of these were age-1 pollock; Fig. 7) due to their wider distribution than the other species (Fig. 6). Eulachon were found only in a restricted area in the southern part of the Bering Sea and had a distribution different from the other species. Thus, eulachon have a low probability of being found with the remaining species. The mean probability for the series that none of the species was present was around 14% while the mean availability of all five being found at the same station was 1.6%. Substantial interannual variability was evident in the biomass index for the five forage species in the AFSC bottom trawl collection series (Fig. 8a). Herring showed the most variability and also made up the majority of biomass during most years. Age-1 pollock were generally of secondary importance and showed relatively little change over the period compared to herring; however, their biomass estimates exceeded that of herring during years of low herring biomass (e.g., 1982, 1986, 1990). Capelin was relatively important only during one year (1993), whereas the biomass index of age-1 Pacific cod and eulachon represented a small fraction of the

14 522 Brodeur et al. Forage Fishes in the Bering Sea Figure 5. Distribution of bottom temperature and standardized catches of the dominant forage fishes from a cold year (1986) and a warm year (1987) collected in AFSC summer surveys of the Bering Sea.

15 Dynamics of the Bering Sea 523 Figure 5. (Continued.)

16 524 Brodeur et al. Forage Fishes in the Bering Sea Figure 5. (Continued.) Distribution of bottom temperature and standardized catches of the dominant forage fishes from a cold year (1986) and a warm year (1987) collected in AFSC summer surveys of the Bering Sea.

17 Dynamics of the Bering Sea 525 Figure 6. Number of forage species occurring at each sampling site from the (a) 1986 and (b) 1987 AFSC bottom trawl surveys. total biomass index of these five species during all years (Fig. 8a). There was no indication of any synchrony in the peak biomass index of these forage fishes, nor was there any apparent relation of the peaks to either mean surface or bottom temperature measured during each survey (Fig. 8). Forage Fish Distribution Relative to Depth and Temperature We compared the cumulative distributions of habitat variables sampled during the survey with those of the dominant species collected during the 1987 R/V Darwin survey (Fig. 9). Because of the large number of deepwater stations which do not appear to be good habitat for most of these shelf species (Fig. 3), we included only the 75 stations that were <200 m in depth for this analysis. Greater than 60% of the capelin were collected at depths <50 m and this species showed a significant response to depth

18 526 Brodeur et al. Forage Fishes in the Bering Sea Figure 7. Probability of a species being present at a site as a function of the number of total species present based on averages for the AFSC bottom trawl surveys. Group size probabilities are shown for each grouping.

19 Dynamics of the Bering Sea 527 Figure 8. Log biomass of the dominant forage fishes collected in AFSC bottom trawl surveys from 1982 to 1995 and annual averages of surface and bottom temperatures recorded throughout the survey area for the same period.

20 528 Brodeur et al. Forage Fishes in the Bering Sea Figure 9. Cumulative distribution functions for the observed (habitat) and the four species examined from the R/V Darwin surveys for (a) depth and (b) temperature. See Table 5 for statistical comparison of the distributions.

21 Dynamics of the Bering Sea 529 Table 5. Results of Kolmorogorov-Smirnov test for statistical differences in catchweighted cumulative frequency of each species and the cumulative frequency of habitat (temperature or bottom depth) for R/V Darwin trawl survey data for Species Depth Temperature Age-0 pollock Age-1 pollock Pacific herring Capelin The P-values given are the proportion of the 1,000 test-statistic values that were greater than or equal to the observed test statistic. Figure 10. Cumulative distribution functions for the observed (habitat) and the five species examined from the AFSC bottom trawl surveys for both years by depth (a and b) and temperature (c and d). See Table 6 for statistical comparison of the distributions.

22 530 Brodeur et al. Forage Fishes in the Bering Sea Table 6. Results of Kolmorogorov-Smirnov test for statistical differences in catch-weighted cumulative frequency of each species and the cumulative frequency of habitat (temperature or bottom depth) for AFSC bottom trawl survey data for 1986 and Species Depth Temperature Depth Temperature Age-1 pollock Age-1 Pacific cod Pacific herring Capelin Eulachon The P-values given are the proportion of the 1,000 test-statistic values that were greater than or equal to the observed test statistic. (Fig. 9a; Table 5). Capelin were significantly (P = 0.002) associated with colder temperatures in the northern part of the study area, whereas age-0 pollock were associated (P = 0.009) with warmer temperatures than those sampled overall (Fig. 9b; Table 5). None of the remaining species were significantly associated with either bottom temperature or depth. Based upon the 1986 and 1987 AFSC bottom trawl surveys, most species were found over a relatively narrow depth range compared to what was sampled in the survey (Figs. 10a and 10b). Pollock and eulachon tended to be found deeper than the average depth of the survey during both years, whereas Pacific cod and capelin were found significantly shallower than the depth distribution sampled (Table 6). Herring were found shallower in 1986 but not significantly different from their depth distribution in 1987 (Fig. 10, Table 6). The bottom temperature occupation curves of most species were not significantly different from those found in the environment with the exception of eulachon which was found only in the warmer temperatures at the southern part of the sampling area (Fig. 10c; Table 6). Herring was the only species to show interannual differences in distribution with respect to temperature, with 1987 showing a distribution in cooler waters compared with 1986 (Fig. 10d). Discussion This study presents the first examination of the large-scale distribution patterns of several prominent forage fish species throughout the Bering Sea. We also present new data on the association of these forage fishes with other taxa that may compete with these forage fishes and, in some cases, prey upon them (e.g., age-2+ pollock). Although other forage species were collected during these surveys (e.g., Pacific sand lance Ammodytes

23 Dynamics of the Bering Sea 531 hexapterus, rainbow smelt Osmerus mordax, Atka mackerel, and Arctic cod Boreogadus saida), these other species were either not adequately sampled by our gear or did not occur over a broad enough range of the sampling area to be included here. Pacific salmon (Oncorhynchus spp.) were commonly caught and occurred in high biomass in the Russian surveys, but their distribution was more oceanic and they are generally not utilized as forage fish except in rare circumstances. Also, the data for the AFSC bottom trawl surveys included only age-1 fish for pollock and cod since age-0 fish were not captured due to the relatively large mesh used in the trawls, although the importance of age-0 juveniles in the diets of seabirds and predatory fishes of the Bering Sea is clearly recognized (Livingston 1993). Many of these forage fishes form layered aggregations and structured schools which facilitates capture by predators but may introduce some bias in estimates of population size in trawl surveys. Most also undergo seasonal horizontal and diel vertical migrations which further complicate quantitative sampling. Some of the interspecific differences in distribution may be related to spawning habits in that some species (e.g., eulachon, herring, capelin) spawn in freshwater, intertidal or nearshore areas, whereas others (pollock, Pacific cod) spawn in relatively deep water off the continental shelf (Wespestad 1987, Fritz et al. 1993). Interannual differences in abundance of the species that undergo spawning migrations may result merely from temperature differences altering the time of spawning relative to the survey. Because of the substantial number of stations and wide geographic area covered in these surveys, the distributions portrayed here are not synoptic and may actually mask smaller-scale migrations occurring during the surveys. Differences in observed distributions between two adjacent but thermally contrasting years in the AFSC bottom trawl data suggests that temperature is important in affecting the horizontal extent and overlap in distribution of these forage fishes. The consistency among species is somewhat surprising because some species (e.g., herring and capelin) are typically more pelagic than others (e.g., Pacific cod) but all were responsive to the variability in bottom temperature. However, we were unable to determine temperature effects on the vertical distribution or absolute biomass of these species due to the lack of depth-specific sampling and incomplete coverage of their distribution area. For example, what may appear to be a decline in fish populations during the warm year of 1987 may actually be a shift in the distribution to the region north of that sampled during the annual surveys (e.g., near St. Lawrence Island and in the Gulf of Anadyr), where high biomass of age-1 pollock, herring, and capelin was found in the 1987 Russian surveys (Fig. 4). The distributions of these species are important by themselves, but what makes them most critical for their role as forage is their proximity to marine mammal and seabird rearing or feeding sites. Although capable of foraging at sea for extended periods, the practical foraging radius of many

24 532 Brodeur et al. Forage Fishes in the Bering Sea land-based predators may be quite limited (Schneider and Hunt 1984). Access to one or more species of forage fishes within a reasonable distance of a nest or rookery may be imperative if the offspring are to survive to maturity. The presence of two or more species at the same location not only increases the biomass available for a generalist feeder, but also improves the probability of finding suitable prey for the specialist for whom some food items may be preferred over other more abundant ones. This may be due to higher fat content, protein content, or total caloric value in some of the species (Payne et al. 1997). Many of the dominant mammal and seabird predators on these species have undergone dramatic shifts in their abundance (National Research Council 1996; Hunt and Byrd, chapter 28, this volume), with several species currently listed as endangered. Reproductive success in seabirds nesting on the Pribilof Islands has been shown to be strongly linked to the availability of appropriate forage fishes (Decker et al. 1995). Merrick et al. (1997) have found that population declines in Steller sea lions (Eumetopias jubatus) are less severe at rookeries where a higher diversity of prey is available than those where only one or two species are available. The abundance of these major forage fishes fluctuated over the period studied and no long-term trends were evident. It should be noted, however, that our data were collected entirely after the major regime shift which occurred around (National Research Council 1996). Although some trawl surveys were conducted in the 1970s in the eastern Bering Sea, the data are not directly comparable to what we present here due to differences in gear, survey design, and areal coverage. Naumenko (1996) analyzed 36 years ( ) of Russian pelagic trawl data from the western Bering Sea and found four distinct periods based upon ichthyofaunal communities which were related to environmental conditions. Beginning in 1975, herring, capelin, and other smelts showed substantial declines in abundance and pollock dominated the catches. This pattern held through the late 1980s, but in the 1990s there has been a resurgence in the pelagic species, particularly herring and capelin, in the western Bering Sea and Okhotsk Sea (Naumenko 1996, Shuntov et al. 1996). Although systematic survey data are less complete for the eastern Bering Sea, these trends in forage fish abundances are reflected in fisheries bycatch (Fritz et al. 1993) and seabird diet (Hunt et al. 1996) data. Our analysis of the relationship between these forage fishes and two easily measured habitat variables is an admittedly simple approach to a complex ecological situation. For example, bottom temperature and depth are often correlated and a more powerful approach may be to examine both variables at the same time using a bivariate distribution (Perry and Smith 1994). However, our finding that most of these species do not appear to actively select a particular temperature range was surprising, considering that the distributions of these forage fishes appeared to vary so

25 Dynamics of the Bering Sea 533 much between warm and cold years. Although we presented distributions only from two adjacent years which showed the greatest change in bottom temperature in the 14-year time series (Fig. 8) and also a substantial difference in sea ice and cold pool extent (Wyllie-Echeverria 1996), other years not presented showed similar patterns in warm and cold years. Most taxa were associated with, or excluded from, a particular bottom depth range, particularly for the U.S. bottom trawl survey data, where there is some segregation of the dominant forage fish by depth strata. The mechanism underlying why depth, or some factor related to it, is important to these fishes needs to be more fully explored. Acknowledgments This study could not have been undertaken without the diligent and conscientious efforts of the many U.S. and Russian scientists who collected the data during the many cruises included in our analysis. We thank Kathy Mier for her assistance in the clustering analysis. Steve Syrjala assisted with the statistical comparison between the fish biomass and environmental data. We thank Dr. Vladimir Radchenko for assistance in data collection and preparation. Art Kendall, Alan Springer, Lowell Fritz, and an anonymous referee provided helpful comments on previous drafts of the manuscript. Funding for this study was provided by the Coastal Ocean Program, Bering Sea Fisheries Oceanography Coordinated Investigations (BS FOCI). This is FOCI Contribution No. B294. References Alaska Sea Grant Forage fishes in marine ecosystems. Proceedings of the International Symposium on the Role of Forage Fishes in Marine Ecosystems. University of Alaska Sea Grant, AK-SG-97-01, Fairbanks. 816 pp. Bailey, K.M., and S.M. Spring Comparison of larval, age-0 juvenile and age-2 recruit abundance indices of walleye pollock, Theragra chalcogramma, in the western Gulf of Alaska. ICES Journal of Marine Science 49: Bakkala, R.G Structure and historical changes in the groundfish complex of the eastern Bering Sea. NOAA Technical Report NMFS pp. Boesch, D.F Application of numerical classification in ecological investigations of water pollution. Environmental Protection Agency Ecological Research Series, EPA-600/ pp. Brodeur, R.D., M.S. Busby, and M.T. Wilson Summer distribution of late-larval and early juvenile walleye pollock (Theragra chalcogramma) and associated species in the western Gulf of Alaska. Fisheries Bulletin, U.S. 93: Brodeur, R.D., and M.T. Wilson A review of the distribution, ecology and population dynamics of age-0 walleye pollock in the Gulf of Alaska. Fisheries Oceanography 5:

26 534 Brodeur et al. Forage Fishes in the Bering Sea Brodeur, R.D., P.A. Livingston, T.R. Loughlin, and A.B. Hollowed (eds.) Ecology of juvenile walleye pollock, Theragra chalcogramma. NOAA Technical Report NMFS pp. Clifford, H.J., and W. Stephenson An introduction to numerical classification. Academic Press, Inc., New York. 229 pp. Decker, M.B., G.L. Hunt Jr., and G.V. Byrd The relationships among sea-surface temperature, the abundance of juvenile walleye pollock (Theragra chalcogramma), and the reproductive performance and diet of seabirds at the Pribilof Islands, southeastern Bering Sea. In: R.J. Beamish (ed.), Climate change and northern fish populations. Canadian Special Publication of Fisheries and Aquatic Sciences 121, pp Faith, D.P., P.R. Minchin, and L. Belbin Compositional dissimilarity as a robust measure of ecological distance. Vegetatio 69: Fritz, L.W., V.G. Wespestad, and J.S. Collie Distribution and abundance trends of forage fishes in the Bering Sea and Gulf of Alaska. In: Is it food? Addressing marine mammal and seabird declines: Workshop summary. University of Alaska Sea Grant, AK-SG-93-01, Fairbanks, pp Hinckley, S., K.M. Bailey, S.J. Picquelle, J.D. Schumacher, and P.J. Stabeno Transport, distribution, and abundance of larval and juvenile walleye pollock (Theragra chalcogramma) in the western Gulf of Alaska. Canadian Journal of Fisheries and Aquatic Sciences 48: Hunt, G.L., Jr., M.B. Decker, and A. Kitaysky Fluctuations in the Bering Sea ecosystem as reflected in the reproductive ecology and diets of kittiwakes on the Pribilof Islands, 1975 to In: S.P.R. Greenstreet and M.L. Tasker (eds.), Aquatic predators and their prey. Fishing News Books, Oxford, pp Koehler, P.A., P.C.F. Hurley, P. Perley, and J.D. Neilson Juvenile fish surveys on the Scotian shelf: Implications for year-class size assessment. Journal du Conseil International pour l Exploration de la Mer 43: Livingston, P.A Importance of predation by groundfish, marine mammals and birds on walleye pollock Theragra chalcogramma and Pacific herring Clupea pallasi in the eastern Bering Sea. Marine Ecology Progress Series 102: Merrick, R.L., M.K. Chumbley, and G.V. Byrd Diet diversity of Steller sea lions (Eumetopias jubatus) and their population decline in Alaska: A potential relationship. Canadian Journal of Fisheries and Aquatic Sciences 54: Minerals Management Service Forage fishes of the southeastern Bering Sea. OCS MMS pp. National Research Council The Bering Sea Ecosystem: Report of the Committee on the Bering Sea Ecosystem, National Research Council. National Academy Press, Washington, DC. 324 pp. Naumenko, N.I Long-term fluctuations in the ichthyofauna of the western Bering Sea. In: O.A. Mathisen and K.O. Coyle (eds.), Ecology of the Bering Sea: A review of Russian literature. University of Alaska Sea Grant, AK-SG-96-01, Fairbanks, pp

27 Dynamics of the Bering Sea 535 Naumenko, N.I., P.A. Balykin, E.A. Naumenko, and E.R. Shaginyan Year-toyear variation of stocks and community structure of western Bering Sea pelagic fishes. In: International Symposium on Bering Sea Fisheries, Khavarovsk, U.S.S.R, pp Payne, S.A., B.A. Johnson, and R.S. Otto Proximate analysis of some northeastern Pacific forage fish species. In: Forage fishes in marine ecosystems. University of Alaska Sea Grant, AK-SG-97-01, Fairbanks, pp Perry, R.I., and S.J. Smith Identifying habitat associations of marine fishes using survey data and application to the northwest Atlantic. Canadian Journal of Fisheries and Aquatic Sciences 51: Schneider, D., and G.L. Hunt Jr A comparison of seabird diets and foraging distribution around the Pribilof Islands, Alaska. In: D.N. Nettleship, G.A. Sanger, and P.F. Springer (eds.), Marine birds: Their feeding ecology and commercial fisheries relationships. Canadian Wildlife Service Special Publication, Ottawa, pp (Available from Canadian Wildlife Service, 400 Laurier Ave. W., Ottawa, ON, K1A 0H3, Canada.) Shuntov, V.P., E.P. Dulepova, V.I. Radchenko, V.V. Lapko New data about communities of plankton and nekton of the far eastern seas in connection with climate-oceanological reorganization. Fisheries Oceanography 5: Sinclair, E.H., G.A. Antonelis, B.W. Robson, R.R. Ream, and T.R. Loughlin Northern fur seal, Callorhinus ursinus, predation on juvenile walleye pollock, Theragra chalcogramma. In: R.D. Brodeur, P.A. Livingston, T.R. Loughlin, and A.B. Hollowed (eds.), Ecology of juvenile walleye pollock, Theragra chalcogramma. NOAA Technical Report NMFS 126, pp Smith, S.J Analysis of data from bottom trawl surveys. In: H. Lassen (ed.), Assessment of groundfish stocks based on bottom trawl survey results. NAFO Scientific Council Studies, Northwest Atlantic Fisheries Organization 28: Springer, A.M A review: Walleye pollock in the North Pacific how much difference do they really make? Fisheries Oceanography 1: Walters, G.E The value of pre-recruit abundance estimates from resource assessment surveys in forecasting year class strength of commercially important species in the eastern Bering Sea. In: S. Sundby (ed.), Year class variations as determined from pre-recruit investigations. Institute of Marine Research, Bergen, Norway, pp Walters, G.E., G.B. Smith, P.A. Raymore Jr., and W. Hirschberger Studies on the distribution and abundance of juvenile groundfish in the northwest Gulf of Alaska, : Part II, Biological characteristics in the extended region. NOAA Technical Memo NMFS F/NWC pp. (Available from National Marine Fisheries Service, Scientific Publications Office, 7600 Sand Point Way N.E., Seattle, WA, ) Wespestad, V.G Population dynamics of Pacific herring (Clupea pallasi), capelin (Mallotus villosus), and other coastal pelagic fishes in the eastern Bering Sea. In: Forage fishes of the southeastern Bering Sea. Minerals Management Service OCS MMS , pp

28 536 Brodeur et al. Forage Fishes in the Bering Sea Wilson, M.T., R.D. Brodeur, and S. Hinckley Distribution and abundance of age-0 walleye pollock, Theragra chalcogramma, in the western Gulf of Alaska during September In: R.D. Brodeur, P.A. Livingston, T.R. Loughlin, and A.B. Hollowed (eds.), Ecology of juvenile walleye pollock, Theragra chalcogramma. NOAA Technical Report NMFS 126, pp Wyllie-Echeverria, T.W A relationship between the distribution of one year old pollock and sea ice characteristics. In: R.D. Brodeur, P.A. Livingston, T.R. Loughlin, and A.B. Hollowed (eds.), Ecology of juvenile walleye pollock. NOAA Technical Report NMFS 126, pp

29 Dynamics of the Bering Sea CHAPTER 25 Ecology of Groundfishes in the Eastern Bering Sea, with Emphasis on Food Habits Kei-ichi Mito Seikai National Fisheries Research Institute, Ishigaki, Japan Akira Nishimura and Takashi Yanagimoto National Research Institute of Far Seas Fisheries, Shimizu, Japan Abstract The food habits of groundfishes in the Bering Sea are presented based on data collected by the National Research Institute of Far Seas Fisheries of Japan. The ecology and biology of important species caught in fisheries also are described. Juvenile walleye pollock (Theragra chalcogramma), the most abundant species, preyed primarily on zooplankton such as euphausiids and copepods. Zooplankton was also the primary prey of Pacific ocean perch (Sebastes alutus) and Pacific herring (Clupea pallasi). Adult pollock preyed mainly on small pollock in the eastern Bering Sea, but on euphausiids and copepods in the Aleutian Basin. Pacific cod (Gadus macrocephalus) also exhibited an ontogenetic diet shift from shrimp and Tanner crabs to pollock with increasing cod size. Large-mouthed piscivorous flatfishes such as arrowtooth flounder (Atheresthes stomias), Greenland turbot (Reinhardtius hippoglossoides), and Pacific halibut (Hippoglossus stenolepis) preyed primarily on pollock, while smaller-mouthed flatfishes such as flathead sole (Hippoglossoides elassodon), rock sole (Lepidopsetta bilineata), and yellowfin sole (Limanda aspera) preyed on benthic invertebrates. In addition, many other species such as sculpins and skates preyed on juvenile pollock. Juvenile pollock occupy a key position in the Bering Sea ecosystem by transmitting energy from zooplankton to large-sized fishes.

30 538 Mito et al. Groundfish Ecology in the Eastern Bering Sea Introduction The parts of the continental shelf and the basin considered in this chapter are located in the eastern and the middle-western parts of the Bering Sea, respectively. Japanese fishing boats caught groundfishes in waters of the eastern continental shelf of the Bering Sea from the latter half of 1950s. The current catch of groundfishes in this area is by U.S. fishing vessels. Total catch of groundfishes on the eastern Bering Sea shelf and the Aleutian Basin is large, 2-3 million metric tons per year. Over 300 species of fish among 36 families have been recorded in the Bering Sea. However, the number of target species for fisheries and the number of species having large biomass is few. Ecological and biological studies on groundfishes in the Bering Sea in recent times have been conducted principally by scientists from Japan, Russia, and the United States. Surveys continue to be conducted by the National Research Institute of Far Seas Fisheries (NRIFSF) of Japan and the Alaska Fisheries Science Center (AFSC) of the United States every year. Surveys of groundfish resources and their results up to 1986 are summarized by Bakkala (1993). In this paper, results of studies on food habits of groundfishes conducted by NRIFSF are summarized. Also, characteristics of their ecology and biology are described for commercially important species. Food habit studies are important for analyzing interspecific relationships. Clarification of these relationships enhances our understanding of food web structure and the ecosystem. These results become the basic information for future multispecies population dynamics models. Materials and Methods Data on Food Habits Contents of fish stomachs were sampled from groundfish resource surveys conducted by NRIFSF in the Bering Sea, including data analyzed by Mito (1977). Both data sets were included in the ecosystem model development work which the Fisheries Agency of Japan conducted in These results of food habits are reported by Mito (1990a,b,c) and Mito et al. (1994a,b, 1996a,b,c). Data on Ecology and Fisheries Biology To clarify ecological and biological characteristics of important species, we relied on results from Ikeda (1980) and Bakkala (1993). Furthermore, we referred to the stock assessment documents published in recent years by the AFSC.