The Alaska Fisheries Science Center website is now part of the NOAA Fisheries website.
Some information may not be up to date. Join us at our new location,
Please contact with any questions.

link to AFSC home page
Mobile users can use the Site Map to access the principal pages

link to AFSC home page link to NMFS home page link to NOAA home page

Recruitment Processes: Modeling and Analysis Project

The modeling and analysis project takes information produced by the Recruitment Processes teams, and data from other sources to develop computer simulations to investigate the influence of various factors on the success of spawning (year class strength). These models start with a conceptual view of what processes are important and how they act to affect survival of larvae. Numerical models are then produced with the factors and linkages indicated in the conceptual models. Simulations using physical data from various years can be run and compared with the actual year class strength. The models can also be run with hypothetical physical conditions to see their influence on year class strength.

Current studies:

Estimates of Natural Mortality from Life History Data

A new model based on the published Alverson and Carney (AC) model for estimating the instantaneous rate of natural mortality M was reformulated to include β and t0 parameters. catch composition by gear typeThe new model Zhang and Megrey (ZM) model can be used for the estimation of M instead of the AC model since von Bertalanffy and allometric growth parameters are readily available for most exploited fish stocks. Using 91 published life history parameters from FishBase, it was determined from the new formulation that the ratio relating the maximum age (tmax) to the age at maximum biomass (tmb) for pelagic and demersal species were significantly different from 0.38, the value originally proposed by Alverson and Carney. The ratios for two ecological groups, pelagic and demersal species, were 0.302 and 0.440, respectively. These values were significantly different between the two ecological groups. We examined the sensitivity and bias in M from the new formulation relative to the AC model which assumed that ß = 3.0 and t0 = 0. Estimates of M from the AC model are most sensitive to the assumption that growth starts at t0 = 0 when growth rates are high, and to the ß and tmb assumption. The performance of the revised ZM model was evaluated, comparing calculated M values from the AC and ZM models based on a paired sample t-test. Results of the two statistical analyses showed that the ZM model produced values of M closer to published estimates compared to the AC model. Thus, the ratio of tmb/tmax for specific ecological groups should be used rather than using the Alverson and Carney’s constant 0.38. Analyses of exploited stock dynamics might be conducted using the possible range of M instead of the constant value. The range could be estimated from the ZM model using the mean ratio of tmb/tmax ± standard deviation to get the tmb for each subgroup or by explicit variance calculations.

Comparative Ecosystem Analyses

The biophysical and trophic characteristics of the Bering and Barents Sea, two high latitude sub-arctic ecosystems that support major commercial fisheries and in which sea ice plays a dominant role in shaping the ecosystem and annual production, are compared. The eastern Bering Sea and the Barents Sea share a number of common biophysical characteristics. For example, both are seasonally ice-covered, high-latitude shelf seas, dependent on advection for heat and for replenishment of nutrients on their shelves, and with ecosystems dominated by a single species of gadoid fish. At the same time, they differ in important respects (see comparison tables). In the Barents Sea, advection of Atlantic water is important for zooplankton vital to the Barents Sea productivity. Advection of zooplankton is not as important for the ecosystems of the southeastern Bering Sea, where high levels of diatom production can support production of small, neritic zooplankton. In the Barents Sea, cod are the dominant gadoid, and juvenile and older fish depend on capelin and other forage fish to repackage the energy available in copepods. In contrast, the dominant fish in the eastern Bering Sea is the walleye pollock, juveniles and adults of which consume zooplankton directly. The southeastern Bering Sea supports considerably larger fish stocks than the Barents. In part, this may reflect the greater depth of much of the Barents Sea compared to the shallow shelf of the southeastern Bering. However, walleye pollock is estimated to occupy a trophic level of 3.3 as compared to cod in the Barents Sea at a trophic level of 4.3. This difference alone could have a major impact on the abilities of these seas to support a large biomass of gadoids. In both seas, climate-forced variability in advection and sea ice cover have the potential to have majors effects on the productivity of these sub-arctic seas. In the Bering Sea, the size and location of pools of cold bottom waters on the shelf may influence the likelihood of predation of juvenile pollock.

Conceptual Modeling

The working conceptual model for Bering Sea walleye pollock (Figure 1) is adapted from a similar model developed for Gulf of Alaska walleye pollock. The model can be termed a switch or survival gauntlet model in that it represents successive conditions or switches that must be realized for the fish to survive. Each switch has a conditional probability of being set for survival or mortality. The probability is subject to spatial and temporal variability. For example, a "hatch switch" could be dependent on water temperature that varies in space and time. Switches can act on individuals, cohorts, or populations. This model has a type of dynamic termed supply dependent, multiple-life-stage control (Bailey et al. 1996).

switch model figureIn our working conceptual model, most mortality takes place in the juvenile life stage, although mortality during the larval life stage could be important in some years. Two factors have been suggested as important mechanisms in regulating eastern Bering Sea pollock year-class strength. These are predation, primarily cannibalism of juvenile pollock by adult pollock (Laevastu and Favorite 1988), and environmental factors (Quinn and Niebauer 1995). Predation on pollock is greatest in the first and second year of life, and cannibalism has been shown to be a significant source of predatory mortality (Livingston 1991). It has been suggested that the intensity of cannibalism is primarily a function of the degree of spatial separation of adults and juveniles (Wyllie-Echeverria 1995, Wespestad et al. in prep.). Environmental factors enter our model through their impact on spatial overlap. Wind drift influences the distribution of animals prior to the juvenile life stage. The strength of vertical stratification of the water column during the juvenile life stage likely is also important. Our hypothesis is that strong year classes result when planktonic stages are transported shoreward and away from adults by near surface currents in spring (warm years). In cold years, the associated winds reduce the affect of this mechanism, and juvenile utilization of inshore regions is more limited. This results in similar distribution patterns of adults and juveniles, potential for more cannibalism, hence a weak year class. Even when adult and juvenile life stages coincide, stratification of the water column can effectively separate adults from juveniles. This "switch" works in opposition to the wind-drift switch. Wind mixing (turbulence) can influence year-class strength by affecting the ability of first feeding larvae to feed successfully. Its influence is mainly restricted to the yolk sac larvae and first feeding life stages. Climatic factors, such as atmospheric circulation dynamics that determine frequency and trajectory of storms, wind direction and intensity, ice extent and water temperature, can affect all life stages of pollock.

The spawner-recruit relationship for eastern Bering Sea pollock (Figure 2) reflects a moderate density dependence between the spawning stocks and recruitment, with reduced survival of recruits at high levels of adult abundance (Wespestad and Quinn 1997). Figure 2 adds evidence supporting our working conceptual model. Cannibalism is presumed to be the mechanism underlying density dependence. The spawner-recruit relationship indicates that several points are well above and below the fitted relationship. It has been shown that most of the points above the line (i.e., 1978, 1982, 1989) are associated with warm years. Warm years are characterized by strong shoreward wind drift, subsequent high spatial separation between adults and juveniles, low rates of cannibalism, and good recruitment. In these years, density-dependent (cannibalism) mechanisms are not in effect. Those data points close to the line are associated with cold years (Fritz et al. 1993). Cold years are characterized by weak shoreward wind drift and subsequent low spatial separation between adults and juveniles, high rates of cannibalism, and poor to average recruitment. In cold years, the cold pool temperature tends to alter distributions of both adult and juveniles tending to enhance coincidence. Thus, in cold years, density-dependent (cannibalism) mechanisms are important. Niebauer and Quinn (1995) also found a correlation with temperature that they suggested results from variation in the intensity of the Aleutian Low. They also found that the best fit occurred with a one year lag, which suggested that environmental effects were exerted, not in the first few months of life, but rather at later juvenile stages.

The current conceptual model focuses on cannibalism and the physical processes that enhance or deter cannibalism. Other biophysical processes important to pollock recruitment (i.e., ice and spring bloom dynamics) are under examination but not yet included in the conceptual model.

OceanGIS - Integration of Java and GIS for Visualization and Analysis of Marine Data

Ideally, scientists should be able to format, explore, analyze and visualize data in a simple, powerful and fast application that would seamlessly integrate georeferenced data from a variety of data sources into a powerful intuitive visualization. Geographic information systems (GIS) provide a high level of functionality for spatial analyses, but are not yet able to provide the extended functionality needed to create a truly "scientific GIS". Java can be used to program scientific calculations and analyses, but it isn't inherently spatial. VRML provides the ability to visualize scientific data and to allow the user to interact with the data by rotating, zooming and panning, but you cannot easily query VRML objects. Recent developments in GIS and in Java can be exploited to produce a prototype of this kind of integrated scientific system.

In this project we used a combination of Java/Java3D and a commercial GIS package (ArcGIS) to create a prototype for a scientific GIS. We combine the spatial tools exposed through the ArcEngine API with the analytical capabilities of algorithms written in Java with the complex visualization capabilities of Java3D. Modules from each of these technologies are combined to create innovative tools to allow users to import data, perform spatial and scientific analyses and output the results as visualizations for further examination.

The OceanGIS application demonstrates the integration of advanced visualization techniques with standard GIS techniques, and moves the user from thinking in a two-dimensional plane to a more interactive three and four-dimensional world. Additionally, the ability to connect to distributed data servers and project and visualize data on the fly opens this application to a new group of users, including emergency managers and scientific modellers who must respond to disaster management scenarios rapidly.

Future plans for OceanGIS include the development of marine-science related tools, including towed-instrument (fence-line) rendering, three dimensional statistical tools such as optimal interpolation, and the porting of standard hydrographic algorithms to the GIS application.

This project was funded by the NOAA High Performance Computing and Communications (HPCC) program. To download and test OceanGIS please go to

Recent Publications, Poster Presentations, Reports & Activities

  • ZHANG, C.-I., and B.A. MEGREY. 2006. A revised Alverson and Carney model for estimating the instantaneous rate of natural mortality. Transactions of the American Fisheries Society. 135 (3): 620-633.
  • HUNT, G.L., Jr., and B.A. MEGREY. 2005. Comparison of the biophysical and trophic characteristics of the Bering and Barents Sea. ICES Journal of Marine Science 62(7): 1245-1255.

To view and print these documents, you must install Adobe Reader freeware.  Adobe also offers free tools for the visually disabled.

            | Home | Site Map | Contact Us | FOIA | Privacy | Disclaimer | | Accessibility | Print |           doc logo