III. Conceptual Model and Scientific Approach

The high productivity of the Bering Sea which leads to the large biomasses of birds, mammals and fishes has long been an ecological enigma. The Bering Sea supports over 50 commercially important species and at least 50 species of marine mammals. The reasons for the consistently high productivity, even though the waters of the Bering Sea shelf are seasonally ice-covered and are cold and light limited for much of the year, have been investigated by several research programs such as PROBES, ISHTAR, BS FOCI, and marine mammal studies. The findings of PROBES (Processes and Resources of the Bering Sea Ecosystem) are particularly relevant to this proposed work, in that the emphasis was on the roles of physical oceanography and nutrient supply in regulating primary and secondary production on the southeast Bering Sea shelf (Continental Shelf Research, 5, No. 1/2, 1986). Primary productivity on the southeast Bering Sea shelf appears to be spatially variable and is highly episodic. Spring blooms associated with the ice edge and with thermal stratification are important, and the approximately 200 g C m-2 annual production remains available to the higher trophic levels, both pelagic and benthic. A limiting factor in the productivity of the southeastern Bering Sea shelf is the transport of nutrients onto the shelf.

Upper trophic level species of the SE Bering Sea can be divided into three fairly distinct species groups or trophic guilds based on characteristics of feeding (Livingston et al., 1994). The first guild consists of an outer shelf group of fish, mammals, and birds that consume small pelagic fish, primarily juvenile pollock, and euphausiids. The second guild is an inshore group of fish, crab and other epibenthic fauna that consume mainly benthic infauna. These two groups represent a biomass of about 8-10 million metric tons each. The third guild is a smaller (~1.2 million tons), more ubiquitous group, dominated by cod and skates that feed on crab and fish. Walleye pollock dominate the biomass of the outer shelf pelagic guild and the species diversity of the guild is very low (Figure 4). For the pelagic guild a biomass-weighted effective number of species, or species diversity index (Suchanek, 1994), was nearly constant (1.0 - 1.2) between 1979 and 1993. In contrast, the benthic guild had a species diversity near 5.1.

Figure 4. Biomass trends of the offshore pelagic consumer guild in the eastern Bering Sea from 1979-1993 (after Livingston et al., 1994). Adult pollock dominate. Birds and mammals are represented by the top dotted line.

Some ecologists speculate that a low species diversity within a guild could result in less year to year stability of the guild (Wilson et al., 1991, 1994). The instability results from interannual variations in the dominant species of the guild without compensation from other guild members. Given that walleye pollock represent a dominant species in a guild with low diversity, one might expect that pollock are a nodal or focal species in the outer shelf feeding guild and would play a central role in determining the general health of the Bering Sea ecosystem (Springer 1992; Livingston, 1993). If this hypothesis is true, then research that elucidates processes that influence interannual to decadal scale changes in the production of walleye pollock in the Bering Sea (Figure 5) provide information to the status of the outer shelf guild of the Bering Sea as well. This lack of ecological stability based on adult pollock and their primary food sources, juvenile pollock and euphausiids, is a major issue in understanding the present Bering Sea ecosystem.

Figure 5. The estimated recruitment for the 1964-1991 yearclasses is obtained from a catch-at-age model tuned to the age 3+ abundance observed in triennial surveys and an index of age 1 abundance in annual bottom trawl surveys. Both survey time series began in 1979. The projected abundance of the 1992-1994 yearclasses is estimated solely from the age 1 index.

Production of juvenile pollock is influenced by upper trophic level predation and by the spatial and temporal distribution of secondary productivity. We believe that density-dependent predation and environmental factors have influenced pollock recruitment since the 1960s. Contrast the low recruitment years 1986, 1987, 1988 in Figure 6 with the high recruitment years 1982, 1984 and 1989; all occurred at historical highs for population biomass. It is clearly necessary to move away from simple single species spawner-recruit models to view fisheries in an ecological context and to consider the full complexity of this approach (Ludwig et. al., 1993). Both the predator and food environment are important. These co-factors interact in unclear ways. Spatial overlap or lack of overlap is a clear issue.

Figure 6. Trajectory by year of the number of spawners and recruits of walleye pollock in the Bering Sea from 1964 to 1994.

Click on the image to see a larger version.

The dominance of pollock shown in Figure 4 suggests that predation by marine mammals and birds on juvenile pollock is overshadowed by cannibalism by older pollock. There is evidence that pollock year class survival may be enhanced by separation of juveniles from adults. About 64% of the above average 1982 year class was found in the middle shelf regime at age 1, whereas only 15% of the below average 1987 year class was located in the middle shelf (Figure 7). Both these year classes were produced from similar spawning stock sizes (Wespestad, 1994). Simulated drifter tracks started in the outer shelf region in April go to inner and middle shelf areas in 1982, but are retained in the outer shelf region in 1987 (Ingraham and Miyahara, 1988).

Changes in availability of juvenile pollock and other forage fish to upper trophic level predators may also be the result of variations in environmental conditions among various habitats in the eastern Bering Sea (Quinn and Niebauer, 1995). Temperature has a profound impact on time to hatch and growth rate of larval pollock, and influences the amount of energy available to the pelagic and benthic guilds. Mortality estimates from FOCI research suggest a decrease of larval abundance by a factor of twenty for a temperature decrease of 4 °C; growth rate for the prey of larval pollock is also temperature-dependent. In the middle domain (50-100 m water depth), a cold pool exists in the bottom layer as a remnant of previous year's ice cover. The heat content and horizontal extent# of this pool vary greatly each year with minimum temperature between -1.5 to 3.0 &176;C. Over the outer shelf (100-180 m water depth) the intrusion of warm slope waters limits extremes in temperature. Thus, while temperature variations could limit survival in the middle domain, the outer domain probably provides a more stable environment. Data suggests that cold waters resulted in the small 1976 year class (Bailey et al., 1986), and will similarly influence the 1995 year class. Analysis of water temperatures in the period after spawning indicate that the largest year-classes of pollock occurred during the first year of a warm period (Bulatov, 1989).

It is also possible that transport along and onto the shelf encourages juvenile pollock to migrate in a westerly direction passing through areas of particularly high primary, and presumably, secondary productivity; compare the location of age-0 juveniles (Figure 8) with that of age-1 (Figure 7). One high productivity area extends south from St. Matthews Island toward Cape Navarin (Sobolevski, et al., 1991). This region receives high concentrations of nutrients from the Bering Sea basin, in some cases probably via canyon upwelling. Juveniles that successfully transit the Bering Sea shelf westward can reach regions of high productivity.

We are now in a position to make the following assertions for a Southeast Bering Sea Regional Ecosystem Study:

Southeast Bering Sea Carrying Capacity will test the hypothesis that interannual ocean variability influences the availability of prey, growth rate, predation, and distribution of juvenile pollock and higher trophic level species. Although we already know that ocean variability can influence fisheries, what is not known is how these factors specifically co-occur in the Bering Sea. We use the phrase "carrying capacity " in a general context as to what limits the potential size of the pelagic guild. From the results of testing these hypotheses we will develop annual recruitment indices for pre-recruit pollock.

RESOURCES: This proposal requests $500K for a start-up year in FY1996 and $1.5 M per year for five years beginning in Fiscal Year 1997. We also will request 50 days of NOAA Ship Miller Freeman and 30 days Class I vessel time per year.

SCIENTIFIC APPROACH: The approach is interdisciplinary and balances time-series measurements, process studies and models, phased over the life of the project. Modification of the emphases and details suggested in these approaches will depend on the outcome of the workshop and will be presented in the implementation plan.

  1. Time Series. We will maintain a spring larval survey and autumn juvenile survey over the shelf and shelf break. We plan to maintain moored biophysical platforms in several habitats over a six-year period. They will track important changes in the ecosystem and the climate system forcing these changes, and provide data for model studies.
  2. Process and Retrospective Studies. We will ask such questions as: 1) how do horizontal and vertical physical dynamics influence the separation and overlap of aggregations of predators and prey, 2) what is the feeding and switching behavior of juvenile pollock and their predators, 3) what influences nutrient transport onto the shelf, and 4) how did pollock become established as the dominant species in the 1960s?
  3. Model-Based Research. We plan to implement 1) a three-dimensional physical model coupled to an individual-based model (IBM) for larval/juvenile processes and a trophic dynamics model, and 2) a spatially dependent model of pollock, their predators, and alternate food sources similar to the Multispecies Virtual Population Analysis (MSVPA) model. The former model will be used to support early life history studies and the later will support ecosystem stability studies and be used to assess alternate management strategies.
The following table provides a timeline for major components of the program. We have an investigative stage, a hypothesis development phase, and a synthesis stage.

Table 1 - Proposed Timeline

The following sections outline three subprojects that comprise SEBSCC: 1) characterization of five biophysical domains, 2) studies of juvenile pollock in the ecosystem, and 3) a physical/IBM model and a MSVPA model of the shelf and slope.

1. Biophysical Domains

The eastern shelf is partitioned into biophysical domains (Figure 1) that delineate habitats for many species. The habitats are differentiated along the shelf by latitude and, across the shelf, by hydrographic domains: the outer shelf with well-mixed upper and lower layers separated by continuous stratification (bottom depth between 100 and 180 m), the middle shelf with two layers (bottom depth between 50 and 100 m), and the vertically mixed inner shelf (bottom depth less than 50 m). Characteristics such as currents, temperature, dispersion, timing of the spring bloom, community structure of plankton, and carbon flux differ among domains. These differences influence hatching time, larval growth rate, and predator and prey availability, all affecting the survival of young pollock. Apparently separate aggregations of adult pollock are found southeast and northwest of the Pribilof Islands (Hinckley, 1987; Mulligan et al., 1989). Two regions, the Pribilof Islands and Unimak Island, are unique in having all three shelf domains within a limited area. Near the Pribilofs, extensive bird and marine mammal populations rely on young pollock for food. The Unimak region is different from adjacent shelf habitats because of the influence of both the Alaska Coastal Current and the North Aleutian Slope Flow.

The transport of new nutrients needed to support the lower trophic level productivity, and ultimately the juvenile pollock growth and survival, must flow from the deep basin onto the shelf. Temporal and spatial variations in such transport are poorly known, as are primary and secondary productivity during the late summer and early fall, at the time juveniles are present in the Bering Sea. The Pervenets Canyon along the central Bering Sea slope may be a particularly good source of nutrients as it differs from the other canyons along the shelf break by being wide with gently sloping walls. Most of the work done previously, particularly during the PROBES study, emphasized the early spring dynamics when only pollock larvae are present. Juvenile pollock in late summer feed on larger zooplankton and micronekton (Merati and Brodeur, 1995). Juvenile pollock undergo a diel vertical migration pattern that apparently follows that of euphausiids, their principal prey (Bailey, 1989). This pattern places them in the more productive surface waters at night when the risk of predation to visually-feeding marine birds and mammals is reduced.

We propose to contrast the temporal and spatial variability of the biological, nutrient and physical conditions between habitats. We will ask questions such as: What are the spatial distributions of larval, juvenile and adult pollock? What are the influences of sea ice and its conditioning of bottom temperature? What controls the variability of source waters in slope/shelf nutrient flux? What maintains the different stratification between the habitats?

Concept for a Field Program - We will sample the habitats in Figure 1. Observations will be made between April (ice permitting) and September. Sampling (Table 1) will include annual distribution surveys of spawning adults, larvae, and juvenile pollock; determination of primary and secondary productivity; assessment of prey and predators; measurement of the physical environment and nutrients; and study of important frontal regions which lie between habitats, to determine scales of variability. We will monitor seasonal predator food habits and energetics with index sites at the Pribilof Islands, outer shelf domains, and the middle shelf. From 1996 through 2001 we propose to maintain several moored biophysical platforms. Each moored platform will record meteorological (wind, temperature, irradiance, pressure), oceanographic (current, salinity, temperature), and biological (chlorophyll) data. A subset of sites will have ADCPs to measure ocean currents and indicate zooplankton biomass. Both shipboard surveys and platforms will continue time series begun in 1995 as part of BS FOCI. Other chemical-water column measurements will include 13C and 13N isotope content of phytoplankton, zooplankton, fishes and other higher animals. The 13C and 13N of zooplankton collected over deep basin pelagic waters appear to be lower than those collected from on shelf and continental slope waters (D. Schell, pers. comm.) The elevated ratios on the shelf may be the result of a vigorous supply of deep nutrients onto shelf waters via upwelling, and this may offer a useful signal to determine the importance of deep basin nutrients to shelf productivity.

2. Juvenile Productivity

We hypothesize that survival of juvenile pollock is increased when predator-prey overlap is decreased via the horizontal and vertical separation of juvenile pollock from their upper trophic level predators, particularly adult pollock, marine mammals, and birds. Bottom-up effects are seen if changes decrease the availability of juvenile pollock to top level predators with limited foraging ranges or depths. Several key questions result from these hypotheses:

We presently have a limited understanding of how pollock eggs spawned elsewhere in the Bering Sea end up as juveniles near and to the northwest of the Pribilof Islands (Dell'Arciprete, 1992). Satellite-tracked drifters placed on the southeast Bering shelf are generally transported toward the northwest and some have become entrapped around the Pribilofs. After being transported as larvae, the juvenile pollock may remain to capitalize on enhanced feeding conditions, despite the increased risk to predation. Juvenile pollock residence time on the southeast Bering Sea shelf is not known, but it appears that by the second year of life, food production in the region is insufficient or unsuitable and the juveniles move, partly by swimming and partly via currents to a more northwesterly location (Figures 7).

Mortality estimates are used to examine the role of environmental conditions on larval survival, to construct life history models, and to evaluate hypotheses concerning the relationship of larval size and survival rates. We believe that combining otolith (Bailey and Macklin, 1994) methods to determine cohorts, with Lagrangian methods of marking a patch of larvae and monitoring changes in abundance as it drifts, offers an accurate estimate of mortality (Talbot, 1977; Yoklavich and Bailey, 1990; Hill, 1991). Field results indicate that retention mechanisms operate to maintain larval patches (Hinckley et al., 1993).

We hypothesize that the unique physical and biological conditions associated with the frontal regions around and to the northwest of the Pribilof Islands provide a rich nursery habitat for juvenile pollock. Further, we hypothesize the importance of intermittent advection of nutrient-reach deep Bering basin slope water onto the shelf leading to intense new summer production. To test this, we will compare the abundance, size composition, growth, and condition of juvenile pollock at these fronts compared with those on either side of the fronts.

A factor in the survival of age-0 pollock in the Bering Sea may be the high abundance of large medusae in the summer. There is some evidence that the largest medusae are capable of feeding on small age-0 pollock (Hamner, 1983). At the very least, the high biomass and prodigious feeding capability of these medusae make them potential competitors with juvenile pollock and other small fishes for the available food. There is also some evidence that juvenile pollock may derive benefit by associating commensally with these abundant medusae, obtaining shelter from predators and possibly food from these hosts (Van Hyning and Cooney 1974, Hamner 1983).

Research Activities

Retrospective Analyses

Bio-Physical Oceanography

Juvenile Pollock Distribution, Mortality, Feeding, Growth and Condition

Role of Gelatinous Zooplankton

3. Bering Sea Modeling

We propose the development of a coupled, spatially explicit, biophysical model of the eastern Bering Sea, driven by wind, tides, upwelling, insolation and temperature, to explore the growth and mortality of young pollock from the various spawning sites through the juvenile stage, and the mixing among these subpopulations (Figure 9). The spatially explicit nature of this model will allow for realistic interaction with the ecosystem by the incorporation of density dependent predation and spatially variable food sources, both of which can be badly misrepresented in spatially aggregated models.

Figure 9. Schematic of a coupled physical/IBM model for SEBSCC. Components are a 3D fine grid shelf/slope circulation model, an individual based biological model (IBM) and a nutrient-phytoplankton/zooplankton lower trophic level model. Results of this model will be coupled to an upper trophic level energetics model (MSVPA).

Individual Based Models (IBM) have been successful in the OPEN, Georges Bank and Shelikof Strait FOCI programs (Werner et al., 1994; Hinckley et al., 1995; Hermann and Stabeno, 1995). We propose including lower trophic level interaction (NPZ). The IBM follows the trajectories through space of individual fish using the hydrodynamic model flow fields. The IBM is a probabilistic and mechanistic model which includes development, behavior, feeding, bioenergetics, growth and mortality for each life stage. Processes are driven by physical factors (temperature, salinity and turbulence) derived from the hydrodynamic model, and by prey levels derived from the NPZ model. The addition of the NPZ model will help address issues of match/mismatch of larval feeding to the timing of the spring phytoplankton bloom. Predation pressure exerted by other species will be incorporated into the IBM. Circulation of the Bering Sea shelf and slope should be calculated using an eddy resolving, free surface, hydrostatic primitive equation model which accurately resolves mixed layers and shelf break processes. Model hindcasts can include assimilation of data from cruises, the biophysical platforms and drifters.

We will ascertain the most important parameters that govern the year-to-year variability of food and predation relative to physical processes and the location of shelf fronts. We intend to establish links between success of the juvenile pollock, interannual variability, and spatial and temporal shifts in forcing. Combined observations and simulations will suggest which pollock spawning sites yield lowest mortality for present and projected physical forcing. Sensitivity analyses will yield insights into the fundamental predictability of pollock dynamics in this region.

We will use the output of the IBM/trophodynamic model to parameterize a spatially explicit model of upper-trophic level predators and their predation on juvenile pollock and alternate sources of prey. This second model, known as MSVPA (multispecies virtual population analysis) will include pollock and major predators of pollock and will be used to address ecological stability issues and assess the influence of various management schemes on long-term species abundances. Questions that we can ask are: What is the importance of juvenile pollock versus euphausiids in the diet of the pelagic guild based on different food switching scenarios? Can we quantify predation pressure on juvenile pollock? What is the stability of the present ecosystem? We will make use of the facilities at the University of Alaska Arctic Region Supercomputer Center. As the causal mechanisms are determined as a basis for survival, statistical models (Megrey et al., 1995) will be used to assess the robustness of the SEBSCC indices to biological and physical environmental fluctuations.

4. Leverage and Collaboration

Southeast Bering Sea Carrying Capacity will be a highly leveraged program. It plans to work collaboratively with ongoing research by other National Marine Fisheries Service (NMFS) programs examining pollock resources and ecology of the Bering Sea (fishery acoustics group, stock assessment group, and Marine Mammal Protection Act Studies), programs at the University and State of Alaska, EPA, Shelikof Strait FOCI, Japan Far Seas Fisheries Laboratory, Ocean Research Institute of Tokyo University, Faculty of Fisheries, Hokkaido University, the Japanese Marine Science and Technology Center, Tokai University in Sapporo, Tohoku National Fisheries Institute, Korean Ocean Research and Development Institute and the Institute of Marine Biology, Far East Branch of the Russian Academy of Sciences. We will also coordinate with the inhabitants of St. Paul Island. We will promote collaborative research with the ONR, NSF, and NASA. Marine mammalogists from the AFSC, ornithologists from the University of California-Irvine, and bioacousticians from the Southwest Fisheries Science Center (NMFS) and Scripps Institute of Oceanography are collaborating on ecosystem studies as part of the current BS FOCI project. An example of existing leverage is that Japanese researchers are providing facilities for BS FOCI aboard two cruises in the early summer of 1995; future collaborative work is being planned for SEBSCC in 1997. When combined with NOAA cruises, this will allow several larval cohorts to be followed through their period of maximum mortality. Japanese researchers (JAMSTEC) also are cooperating with University of Alaska scientists in research on the northern Bering Sea and Chukchi Seas in consort with Russian participants, and are providing financial support for ship time. Southeast Bering Sea Carrying Capacity will be considered a component in the PICES-GLOBEC Climate Change and Carrying Capacity (CCCC) Program.

Continue to Application to Management - Pathway to a New Product
Return to Table of Contents
Return to Theme Page