Link to NMFS Homepage Link to NOAA Homepage Keyboard Navigation Alaska Fisheries Science Center Homebanner

Resource Ecology & Fisheries Management Division

Development of  New Age-Structured Stock Assessment Model

Development of a new age-structured stock assessment model is underway in the Status of Stocks and Multispecies Assessment Program.  In addition to describing the dynamics of each age group in a stock on an annual time scale, as current models already do, the new model also describes the intra-annual and spatial dynamics of the stock.  The new model is based on the Kalman filter, in which the true system state (i.e., the number of fish in each age/area combination) at each time step is assumed to be a linear function of the true system state at the previous time step, modified by random natural variability.  Observations of the system state (e.g., official catch estimates, survey and commercial catch per unit effort (CPUE), etc.) at each time step are assumed to be a linear function of the true system state, modified by random observation error.  The likelihood function defined by the Kalman filter is especially useful because it reflects, in a statistically rigorous way, the change in precision that takes place as data are acquired over the course of the time series.  Other features of the new model include the following

1) Fish movement between areas is modeled as a trendless random walk, thus eliminating the need to estimate a matrix of migration rates which might not be constant.  In principle, the spatio-temporal scale of the model can be arbitrarily fine, potentially allowing assessment results to be presented on scales useful to studies of the dynamics of predator species such as the Steller sea lion.  By embedding the spatio-temporal dynamics of the stock in a full assessment model, random fluctuations in CPUE data can be distinguished more easily from true differences in abundance.

2) Selectivity, weight, and maturity at age are described by functional forms that assure a smooth transition between individually modeled age groups and the “age-plus” group (the older age groups that are pooled for convenience in most stock assessment models).  Both the abundance and average individual weight of fish in the age-plus group are described analytically as functions of the fishing mortality rate.  In its simplest form, where the age-plus group consists of all recruited fish, the new model collapses into “biomass dynamic” form.  Thus, the new model provides a rigorous interface between biomass dynamic and explicitly age-structured models.

3) Three measures of uncertainty are estimated:  the scale of random variability in future biological processes (e.g., future recruitment), imprecision in the estimate of current stock status, and imprecision in estimates of model parameters (e.g., the natural mortality rate).  These estimates of uncertainty are then incorporated into an optimization procedure that maps a given level of risk aversion into a recommended harvest level.  For any positive level of risk aversion, greater estimates of uncertainty result in a lower recommended harvest level, with the magnitude of the harvest reduction varying directly with the level of risk aversion.

4) A Bayesian approach is adopted in which prior distributions are specified for all model parameters.  One advantage of this approach is that the model should still function appropriately regardless of the size of the data set or the imprecision of the observations.  In principle, this could eliminate the need for the present “tier system” used in the Bering Sea-Aleutian Islands (BSAI) and Gulf of Alaska (GOA) Groundfish FMPs to define the harvest control rules for the overfishing level and acceptable biological catch, because all of the desired quantities can be estimated regardless of the amount of data available. The precision of those estimates is, of course, much less when data are extremely sparse or variable, but such imprecision is formally incorporated into the harvest recommendation process in the risk-averse manner described above.

5) The fishing mortality rate and spawning biomass associated with maximum sustainable yield (MSY) are explicit parameters of the model.  Most existing models do not attempt to estimate these quantities.  Other models estimate them in a separate maximization routine as functions of other model parameters such as those governing the stock-recruitment relationship. However, fishery biologists tend to have more prior knowledge of the MSY parameters than of the stock-recruitment parameters, so it is more convenient to treat the MSY parameters explicitly.

A simplified version of the model was developed, which was subjected to peer review through the Council of Independent Experts in July 2001.  The full model is expected to be ready for use by summer 2002.  It is anticipated that the model’s first application will be in the stock assessments for Pacific cod in the BSAI and GOA.

By Grant Thompson.

Atka Mackerel Tagging

AFSC scientists chartered a  fishing vessel from  29 July to  3 August for the purposes of tagging and releasing Atka mackerel in the Seguam Pass area of the Aleutian Islands, Alaska.  The cruise represents the third consecutive year of work by the AFSC on Atka mackerel in Seguam Pass and is part of an ongoing study to determine the efficacy of trawl exclusion zones to maintain prey availability for endangered Steller sea lions.  Trawl exclusion zones were established around sea lion rookeries as a precautionary measure to protect critical sea lion habitat, including local populations of prey such as Atka mackerel. Localized fishing may affect Atka mackerel abundance and distribution near sea lion rookeries.  T-bar tags (small brightly colored external plastic tags) were used to estimate Atka mackerel abundance and movement between areas open and closed to the fishery.

Approximately 1,000 Atka mackerel were tagged and released.  Fish will be recovered by the fishery in the open area and by a chartered vessel in the closed area.   Recoveries of fish tagged in 2001 will be used to estimate abundance and movement between open and closed areas.  Continuing recoveries of fish tagged in earlier years (1999 and 2000) will also contribute to these estimates.  Analyses of tag recoveries to date show that over the course of short 3-5 day fisheries in September, with catches less than 6,000 t of Atka mackerel, fish did not appear to move from inside the trawl exclusion zone to outside. Caution should be used in applying these results to other areas, each with resident Atka mackerel populations and fisheries of different size and distribution.

By Elizabeth Logerwell.

Trawl Data Retrospective Study

There has been considerable scientific research and discussion regarding decadal-scale variability in climate and circulation patterns over the North Pacific Ocean  and their possible effects on biological systems. Climate variation is important to AFSC research programs, both as a source of annual variability in stock size and as a suspected factor in the decline of the number of Steller sea lions.  We undertook this study to look at long-scale time trends in groundfish biomass in the BS and GOA and to see how these patterns are related to published series of changes in various climate indices.

The RACE survey database includes records of all research surveys in the BS and GOA since the NMFS survey programs began and includes some older data collected by the International Pacific Halibut Commission, the State of Alaska, and other agencies.  We are thus in a unique position to put together a long-term retrospective of bottom-trawl data.  Previous efforts to compare older surveys with new data were hampered because of changes in trawl gear used over the years.  A gear comparison study recently completed by the RACE Division provided information needed to prorate trawl catch volumes from the older nets on to a comparable basis with more recent surveys.  Using these results, REFM and RACE staff collaborated on the trawl data  restrospective study.

Time-series of bottom trawl catch, using a robust estimate of the median CPUE, were prepared for three study areas in the BS, three in the western GOA, and two in  the central GOA.  In order to look at patterns of groundfish biomass over space as well as time, “index sites” approximately 200 km by 200 km were used rather than basin-wide estimates.  Basic results of the study for both regions were presented in poster format at the American Fisheries Society meeting in August 2001.  A more detailed examination of the results will be presented at the October 2001 PICES meeting.

The most striking result of the study is a very obvious set of changes in biomass for a number of different species and taxa in the BS, beginning in approximately 1980. Large increases in biomass are seen for gadid fishes (walleye pollock and Pacific cod) and a number of flatfishes (rock sole, arrowtooth flounder, and halibut).  Steep increases also occur, however, in trawl catch of cartilaginous fishes (primarily rays) and benthic invertebrates (starfish, sponges, ascidians, and mollusks).  Many of the individual stock assessments had already noted an increased biomass in the late 1980s and ‘90s relative to the 1970s. This study, however, presents trends for all of the different species groups together and suggests that a system-wide restructuring of food webs in the EBS may have occurred.   Results from sites in the GOA do not show strong patterns over time, but show clearly the spatial variation in trawl catch from east to west across the gulf.

By Elizabeth Connors.


Stomachs collected totaled 6,781 from the EBS and 4,101 from the GOA. Laboratory analysis was performed on 1,496 groundfish stomachs from the EBS, 315 from the GOA, and 484 from the Washington-Oregon-California region. Two observers returned groundfish stomach samples during this quarter.

Canadian Workshop on Objectives and Indicators For Ecosystem-based Management

The Canadian Department of Fisheries and Oceans (DFO) sponsored the Canadian National Workshop on Objectives and Indicators for Ecosystem Based Management from 27 February-2 March 2001 in Sidney, British Columbia.  The workshop was designed to identify ecosystem-level objectives, with associated indicators and reference points that could be used in managing ocean activities.  Participants included DFO scientists, fisheries managers, ocean managers, and habitat managers.  Invited experts from other Canadian government departments and academia along with scientists from other nations attended the workshop.  Alaska Fisheries Science Center scientist Pat Livingston was invited to attend and relate information pertaining to Alaska ecosystem-based fisheries management and development of the ecosystem considerations document that accompanies Alaskan groundfish stock assessment advice to fishery managers.  The full report of the workshop by Jamieson and O’Boyle can be obtained on the World Wide Web at

By Pat Livingston.

Measuring Capacity, Utilization, and Economic Performance

Results of recent research by Ron Felthoven were presented to the North Pacific Fishery Management Council’s Groundfish Plan Teams.  The methods of analysis and results are discussed in a working paper titled, “Effects of the American Fisheries Act on the Harvesting Capacity, Capacity Utilization, and Technical Efficiency of Catcher-Processors.”  Some general results from this analysis are 1) the vessels that were decommissioned as a result of the American Fisheries Act (AFA) were on average less technically efficient than AFA-eligible vessels; and 2) for the fishing companies that voluntarily idled AFA-eligible vessels in 1999 or 2000, the vessels idled typically had exhibited the lowest historical levels of technical harvesting efficiency or capacity utilization of the AFA-eligible vessels in each fishing company and lacked meal plants.

Felthoven collaborated with Rita Curtis (NMFS Office of Science and Technology) on “The Effects of Temporal Aggregation in Fishery Supply Models.”  The paper identifies an assumption commonly made about production technologies that may give rise to biased estimates.  The assumption concerns the ability to aggregate a vessel’s daily fishing production cycle over individual fishing days to the trip level.  To conform with temporal aggregation requires that at given output prices, production over the course of a trip is characterized by no search or learning effects and no stock effects (fish are distributed uniformly over time).  Furthermore, if vessels fish more than one site, temporal aggregation requires that fish be uniformly distributed across fishing sites.  Temporal aggregation would then fail, for example, if fishers employ multiple fishing strategies during a trip, such as fishing offshore for cod and then move inshore to a different ecological niche to harvest yellowtail flounder because stock distributions are not uniform across sites.

In an application to the Hawaii longline fishery, aggregation consistency was rejected for all models, and tests on the technology using aggregated data were found to be biased.  In particular, results obtained using aggregated data suggest that production is joint and that fishermen are either not able to target specific species or have implausible targeting strategies.  In contrast, daily results identify plausible targeting strategies and suggest that several species, including swordfish, have separate production functions.  Importantly, this implies that swordfish, which comprises a high percentage of fishery catch and revenue, can be managed independently.

Foregone Catch and Gross Product Value in 2000 Due to Steller Sea Lion Protection

Estimated annual total catch of groundfish in the BSAI and GOA groundfish fisheries ranged from about 1.6 to 2.1 million t in 1995-99 and averaged 1.9 million t.  Total catch in 2000 was almost 1.8 million t or about 7.4 percent less than the 5-year average.  Estimated annual retained catch (total catch - discarded catch) ranged from about 1.5 to 1.8 million t and averaged 1.7 million t.  The retained catch in 2000 was about 3.2 percent less than the 5-year average.

The estimated annual gross product value of the groundfish catch after primary processing ranged from $1.1 to $1.4 billion in 1995-99.  Despite lower than average catch in 2000, the gross product value exceeded $1.3 billion and was almost 10 percent greater than the 5-year average of $1.2 billion.

A combination of factors, including regulatory, market, and biological changes, contributed to the lower than average catch and the higher than average gross product value in 2000.  Some of those changes offset the gross product value foregone due to the restrictions that were in place in 2000 to provide additional protection for Steller sea lions.  It is estimated that the foregone gross product value resulting from the restrictions in place in 2000 may have exceeded $100 million.  This is more than 7 percent of the estimated gross product value of groundfish for 2000.  The cost data necessary to estimate the effects on product value net of harvesting and processing costs are not available.

The foregone gross product value has a variety of economic effects.  They include the following: 1) decreased employment and income for those involved in harvesting and processing groundfish; 2) decreased expenditures on goods and services by the harvesting and processing sectors; 3) decreased expenditures on goods and services by those who receive income from the harvesting, processing and support sectors; and 4) decreased tax revenues in fishing communities.  Given the importance of the groundfish fishery to some fishing communities, the direct, indirect, and induced income effects of decreased groundfish catch and gross product value can be significant.  However, estimates of these effects are not available.

A variety of restrictions have been placed on the Alaska groundfish fisheries in order to provide additional protection for Steller sea lions and their habitat.  These restrictions closed the AI pollock fishery and forced fishermen to change the timing and location of their fishing operations in other groundfish fisheries.  The closure directly reduced catch and the gross value of the resulting seafood products.  By preventing fishermen from fishing at their preferred times and locations, the other restrictions adversely affected harvesting costs, vessel safety, product quality, product mix, and bycatch rates.  These effects have in turn decreased catch and gross product value while increasing the harvesting and processing costs per unit of catch. Estimates are provided below for five of the major elements of the foregone catch and gross product value.

The estimates of foregone gross product value do not account for the product price increases that probably occurred as the result of the foregone catch.  This source of an upward bias in the estimates of foregone gross product value is offset to an unknown extent by the failure to fully account for the decreases in gross product value per unit of catch that occurred due to the sea lion restrictions.   Estimates of the increases in harvesting and processing costs and decreases in vessel safety are not available.

Prior to the closure of the AI pollock fishery, usually 95-100 percent of the annual total allowable catch (TAC) (quota) was caught.  If the pollock fishery had not been closed, the TAC probably would have been set equal to the allowable biological catch (ABC) of 23,800 t, and catch in the pollock fishery during the roe season probably would have been about 22,000 t with a gross product value of about $24 million in 2000.

The TAC apportionment for the roe season was reduced from 45 percent of the annual pollock TAC to 40 percent to provide additional protection for sea lions during the winter.  With a TAC of 1,139,000 t in 2000, this shifted about 57,000 t of catch from the roe season to the nonroe season and decreased product value by $26 million because gross product value per metric ton of catch was about $460 more during the roe season.

To the extent that pollock roe recovery rates are higher inside the Steller Sea Lion Conservation Area (SCA), the restrictions on the amount of pollock TAC that can be taken inside the SCA decrease gross product value per metric ton of catch.  Based on the percent of the pollock catch that occurred in the SCA in 1998, it is estimated that this limit redistributed 166,000 t of pollock catch from inside to outside the SCA.  The At-Sea Processors Association estimated that the difference between the gross product value per metric ton of catch for these two areas was $235 in 2000.  Based on that estimate and the 166,000-t redistribution of catch between the two areas, the estimated foregone gross product value was $39 million. NMFS has not yet verified this estimate.

The limits on the percent of Atka mackerel TAC that can be taken inside the SCA and the injunction-imposed trawl closure of Steller sea lion critical habitat prevented the Atka mackerel TACs from being fully utilized in the central and western Aleutian Islands in 2000.  Based on catch data from 1995-99, it is estimated that 100 percent of the central area TAC and 95 percent of the western area TAC would have been taken in 2000 in the absence of these restrictions.  This compares to actual utilization of TAC of 90 percent and 35 percent in the central and western areas, respectively.  Due to the high bycatch rates for rockfish outside the SCA, NMFS closed the western area Atka mackerel fishery well before the Atka mackerel TAC was taken. The 2000 TACs were 24,700 t and 29,700 t; therefore, the catches foregone were about 2,470 t and 17,820 t by area for a total of 20,290 t.  The associated loss of gross product value was about $11 million.  However, the early closure of the Atka mackerel fishery resulted in more effort, catch,  and gross product value in the yellowfin sole fishery.  It is estimated that about half of the foregone gross product value in the Atka mackerel fishery was offset by increased product value in the yellowfin sole fishery.  The net effect of the early closure of the Atka mackerel fishery was, therefore, about a $5.5 million reduction in gross product value.

Smaller Atka mackerel with a lower value per metric ton of catch account for a larger percent of the catch outside the SCA in the central and western areas.  In addition to being a factor that prevented the TACs from being fully utilized, this decreased the gross product value per metric ton of Atka mackerel catch.  The data required to estimate the effect on gross product value are not available.

The seasonal apportionments of the GOA pollock TACs and the seasonal rollover rules, the limits on the percent of the TACs that can be taken inside Steller sea lion critical habitat, and the injunction-imposed trawl closure of critical habitat prevented the pollock TACs from being fully utilized in the central and western Gulf of Alaska in 2000.  Based on catch data from 1995-99, it is estimated that 100 percent of the central and western area TACs would have been taken in 2000 in the absence of these restrictions.  In 2000, 21,435 t (over 23%) of the TACs were not caught.  The gross product value foregone probably exceeded $8 million.

By Joe Terry.

Age and Growth Program

Estimated production figures for 1 January to 30 September 2001

Flathead sole 455
Northern rock sole 996
Yellowfin sole 1,487
Walleye pollock 12,681
Sablefish 3,829
Atka mackerel 1,989
Pacific whiting 1,913
Pacific ocean perch 2,762
Northern rockfish 1,240
Light dusky rockfish 186

Total production figures were 27,538 with 5,047 test ages and 170 examined and determined to be unageable.

U.S. North Pacific Groundfish Observer

During the third quarter of 2001, 206 observers were trained, briefed, and equipped for deployment to fishing and   processing vessels and shoreside plants in the Gulf of Alaska, Bering Sea, and Aleutian Islands. They sampled aboard 228 fishing and processing vessels and at 16 shoreside processing plants.  These observers were trained or briefed in various locations.  The University of Alaska Anchorage (UAA) Observer Training Center  trained 32 new observers and another 102 observers with prior experience were briefed at this site.  The AFSC Observer Program in Seattle briefed 49 observers who had prior experience. In addition, 23 observers were excused from briefing, because they had just completed a cruise successfully and were returning immediately to the field.  The third quarter 2001 observer workforce thus comprised 16 percent  new observers and 84 percent experienced observers.  The Observer Program conducted a total of 120 debriefings.  One debriefing was held in Dutch Harbor, one in Kodiak, 51 in Anchorage, and 67 were held in Seattle.

Groundfish Program Third Quarter Highlights

The North Pacific Groundfish Observer Program hosted a contracting workshop from 30 July through 3 August 2001.  The workshop was conducted by staff from the NMFS National Observer Program in Silver Spring, Maryland.  The objective of the workshop was to bring together contracting specialists and representatives from all NMFS observer programs in order to develop contracting standards and improve contract management.

Workshop participants included contracting specialists from NOAA’s Western and Eastern Administrative Support Centers, officers from three NMFS regional offices, a contracting specialist from the ADF&G, and other NMFS observer program staff who are involved in writing statements of work or evaluating bids for contracts.

The emphasis of the workshop was on development of performance-based contracts that address the primary goals of an observer program.  These goals were identified as:  observer coverage, data quality, observer well-being, and observer program integrity.  A template for a statement of work with measurable performance standards was drafted by the workshop participants.

By Bob Maier.



quarterly Jul-Sept 2001 sidebar Contents


Auke Bay Lab

National Marine
Mammal Lab

RACE Division

REFM Division


Quarterly Index

Quarterly Home