Resource Ecology & Fisheries Management  (cont.)

Research on Incorporating Environmental Data Within Stock Assessments:
Bering Sea Pollock

by Jim Ianelli

A large body of research on changes in the physical environment of the North Pacific Ocean  is ongoing at the AFSC  in collaboration with oceanographers at several institutions.  Within the Center, scientists have pursued application of the Ocean Surface Current Simulations (OSCURS)  model to describe changes in the Bering Sea that may have affected early-life conditions of walleye pollock.  The OSCURS model represents a drift-simulator that uses sea-level pressure data to predict surface current movements.  The observed pressure data are used to derive wind characteristics and obtain measures of drift from arbitrary locations in the ocean.  Arsenev (1967) presented drift patterns for the Bering Sea based on limited drift observations from Soviet research vessels during the 1960s.  Direct observation of drift has been shown to be consistent with the magnitude and type of pattern expected based on simulations from the OSCURS model.

To enhance the description of Arsenev, we conducted OSCURS model runs from each month over a grid of points throughout the eastern Bering Sea from 1960 to 2000.  Computing the monthly average over these years, a “climatology” of surface currents indicates strong seasonal shifts (see Figure 1).  The degree to which these seasonal patterns affect pollock abundance distribution and survival is an ongoing research project at the AFSC in collaboration with other climate and oceanographic research groups.   In addition to describing the general patterns of surface currents within the Bering Sea, these analyses provide the ability to scrutinize the degree of interannual variability in surface advection patterns that may affect larval-adult separation and cannibalism.  For example, examining the current patterns for April in different years gives some indication of the kind of interannual variability in current patterns (see Figure 2).  Given alternative hypotheses on the importance of different spawning distributions, these patterns provide insight into factors that may lead to high survival levels for eggs and larvae.  Advection in the months subsequent to peak spawning (e.g., April) may also provide a good indication of movement of eggs and larvae into prime nursery areas.  To date, implementation of an advection model within the stock assessment model has had relatively little impact on values critical for harvest management regulations but has explained a large part of the interannual variability in pollock recruitment.

Figure 3.  Mean summer bottom temperatures used to model bottom trawl survey pollock catchability, 1982-2000.  Triangles represent years classified as "warm", squares as "intermediate", and circles as "cold" temperature years.  (Note: these were normalized to have mean zero for use in the model).

In addition to evaluating the overall geographic concentrations of pollock over time, these data were further broken down by length and age.  In general, smaller fish are more common in the northern areas of the eastern Bering Sea with apparent movement towards the south and east as the pollock become larger.  These patterns are also revealed when one computes the centers of abundance based on age-specific CPUE (catch-per-unit-effort) data.  This is done by simply computing the CPUE-weighted average location for specific ages.  Because bottom temperature has long been considered important in the distribution of pollock on the shelf (e.g., the Southeast Bering Sea Carrying Capacity),  we pooled over years into three categories: cold (< 2°C ), intermediate (2° - 3°C ), and warm ( >= 3°C) based on the mean bottom temperature (Figure 3).  (See related report “Spatial Distribution and Ontogenetic Movement of Walleye Pollock in the Eastern Bering Sea” in this section.) The average locations for warm years are farther on-shelf than for cold years indicating a broader dispersal onto the shelf in warmer years.  The average locations for intermediate years were not depicted here, but were most similar to the cold years.  The mean centers of distribution in both warm and cold years have very similar patterns with age.  Younger fish are found to the north and northwestern regions and as they age, the centers of distribution move south and southeasterly.

Figure 4.  EBS pollock weighted (by number) average location by ages 1-8, 1982-2000.  Lower left line represents the average from "cold" years while the upper right line represents average location during "warm" years.  Triangles represent the centers of survey operations in each year.

Bottom temperature data collected during NMFS summer bottom-trawl surveys were used within the stock assessment model.  Based on age-specific estimates of centers-of-distribution, it appears that temperature affects the distribution of pollock on the shelf (Figure 4).  It therefore seems likely that temperature may affect the availability of the stock to the survey.  That is, temperature may affect the proportion of the stock that is within or outside of the standard survey area.  We therefore evaluate this potential as an effect on the survey catchability in year t base on temperature Tt as:

     qt = μq + βqTt

Where μq is the mean catchability and βq represents the slope parameter.  The time series of temperature (Figure 3 above) is used in a model alternate for this year's assessment.

Figure 5.  Estimated relationship between pollock bottom-trawl survey catchability and bottom temperature

Results suggest that there is a slight negative relationship between bottom temperatures and survey catchability (slope -0.631, with standard error 0.363).  Based on this relationship, survey catchability tends to be lower at warmer temperatures and slightly higher at colder temperatures (Figure 5 below).  In other words, in cold years pollock appear to be more available to the survey gear than in warm years.

Additional research investigating the mechanism for the apparent effect of bottom temperature on survey catchability/availability is ongoing.  One hypothesis is that during colder years, pollock are more prevalent on the bottom than in warm years.  Alternatively, their overall distribution may be different (i.e., fall further outside of the standard survey area during warmer years).

Next >>>

AFSC Quarterly
Research Reports
Oct-Dec 2001