NOAA logo JAS 1999 Quarterly Rpt. sidebar

Resource Ecology &
Fisheries Management

(Quarterly Report for Jul-Aug-Sep 1999)


Economists with the AFSC’s Socioeconomic Assessment Program assisted in preparing the NMFS Fishing Capacity Task Force draft report, which  was presented to the NMFS Science Board and participated in a fishing capacity training session held at Headquarters in Silver Spring, Maryland, in late September.  The NMFS Fishing Capacity Task Force has recommended data envelopment analysis (DEA) and stochastic frontier (SF) production functions as two of the primary quantitative methods of estimating technical fishing capacity.  Program staff were instrumental in identifying four critical issues with respect to assessing the use of these two methods to estimate fishing capacity:

  1. The circumstances for which it is appropriate to include efficiency as a component of capacity and the difficulty of doing so when inputs are heterogeneous.

  2. The ability of these methods to provide estimates of technical capacity.

  3. Their ability to deal with noisy data.

  4. The usefulness of the concept of technical capacity for fishery management purposes.

Although both methods can be used to estimate a production frontier that can in turn be used to estimate a type of capacity of individual vessels and fleets, their underlying assumptions and method of solving for the frontier are quite different. One substantial difference is how each model handles noisy data. An understanding of the implications of this difference is important because random variation is likely to exist in commercial fishery data.

Dan Holland and Todd Lee have completed the initial phase of a research project that explores the impacts of noisy data on fishing capacity estimates made with DEA and SF. Their research uses Monte Carlo simulations to investigate possible finite sample biases attributable to this type of noisy data.  The results of this research suggest that DEA may give highly biased results, and the bias increases as either the sample size or level of noise increases. The SF method is also shown to give biased estimates, but the average bias tends to be much smaller than that found with DEA. However, the simulations demonstrate that the bias of SF estimates can be bimodal when the sample size is “small” or the level of noise in the data is “large”; thus, possibly leading to a high occurrence of under- or overestimates of fishing capacity. Although these results are based on models that predict efficient output as opposed to maximum potential output for a given capital stock, they can be expected to carry over to estimates of technical capacity that assume full utilization of variable inputs.  The results of this study should not be interpreted as an indication of the general superiority of either DEA or SF.  While DEA may confound noise with inefficiency, SF may confound inefficiency with specification error. Other factors that this study does not address may impact the relative performance of the two methods.

Lee and Holland have prepared a paper on this research titled, “The Impact of Noisy Catch Data on Estimates of Fishing Capacity Derived From DEA and Stochastic Frontier Models:  A Monte Carlo Comparison” which will be submitted to Marine Resource Economics (MRE).  A working paper detailing some additional work with DEA models that is not discussed in the MRE paper has also been prepared.  Additional research along similar lines is planned.  This new research will focus on using SF and DEA models to estimate technical capacity with unconstrained use of variable inputs.  Lee and Holland also plan to investigate the factors that lead to negative and positive bias in SF capacity estimates and test the robustness of alternative distributional assumptions used with the SF model.

By Joe Terry.


Evaluation of Catch Sampling Methods

The Groundfish Forum (a fishing industry association) and the AFSC collaborated in research to examine species composition and length frequency sampling methods used by at-sea observers. The research was conducted under an experimental fishing permit approved by the North Pacific Fishery Management Council and NMFS in June 1999.  Fieldwork was completed in the Bering Sea aboard the factory trawler American No. 1 between 5 and 23 September 1999.  John Gauvin and John Henderschedt of Groundfish Forum coordinated the fieldwork with assistance from Craig Rose (RACE Division) and Sarah Gaichas (REFM Division).  The experimental design called for removing six 100-kg subsamples from each of 60 commercial sized (10-12 metric ton) trawl catches, as well as accounting for all production and discards by species group for each haul.  The length frequency sampling portion of the experiment required deck sorting of halibut to minimize mortality as well as collecting 120 length samples at specified intervals throughout each haul.  The excellent work and cooperation of the vessel’s captain, Mike Peterson, factory foreman, Young Ho Chu, and the entire crew made this labor-intensive experiment a success in the field.  Data analysis will continue throughout fall 1999, and a full report is expected by February 2000.

By Sarah Gaichas.

Pilot Study of  CPUE Variability for Atka Mackerel

Members of the RACE and REFM Divisions, NMML, and the Office of Fisheries Information Systems designed, organized, and participated in a preliminary pilot study of small-scale Atka mackerel distribution and abundances in the Aleutian Islands region.  The study was initiated because of concerns regarding the efficacy of trawl exclusion zones around Steller sea lion rookeries and haulouts to maintain prey populations for the endangered Steller sea lion. The pilot study is just one of several recent collaborative efforts within the Center requiring the collection of disparate types of data necessary to manage fisheries to reduce Steller sea lion/fishery interactions.

Data collected during the study will be used to design a larger-scale experiment for detecting the effects of fishing on prey distributions and the efficacy of trawl exclusion zones.  Trawl exclusion zones were established around sea lion rookeries as a precautionary measure to protect  critical sea lion habitat, including local populations of such prey as Atka mackerel.  Localized fishing may affect Atka mackerel abundance and distribution near sea lion rookeries.  The larger project is intended to assess such effects. To this end, AFSC scientists are considering two types of fishery-independent methods to measure changes in abundance and distribution of Atka mackerel:

  1. Bottom trawl surveys before and after a fishery.  Potential impact of the fishery will be judged by the change in catch per unit effort (CPUE) where effort is the area swept by the trawl.

  2. Mark and recapture (tagging) experiments to estimate abundance and movement between areas open and closed to the Atka mackerel fishery.

This pilot study will help determine design parameters and feasibility for these two methods.  Trawl CPUEs may provide an adequate index of change in abundance.  The preliminary study tested survey strategies appropriate for estimating change in CPUE.  The data collected will be used to estimate within-station and between-station variances and assess the sample size required for the larger experiment to detect effects of the local fishery on Atka mackerel abundance near sea lion rookeries.

Initial research on the feasibility of tagging Atka mackerel and on mortality of tagged fish was also conducted as part of this study. The initial tagging work will allow us to estimate tagging mortality, daily tagging production, and methodologies for tagging aboard a commercial vessel.  Results from this work will help determine the feasibility and design of future mark-recapture research for estimating abundance and movement of Atka mackerel between areas open and closed to the commercial fishery.

The vessel factory trawler Seafreeze Alaska, owned and operated by U.S. Seafoods LP, was chartered for 16 days to conduct Atka mackerel research in the vicinity of Seguam Pass in the Aleutian Islands region.  The charter began on 4 August in Dutch Harbor and ended in the same port on 19 August.

A total of 40 randomly-selected stations spread amongst the three strata were sampled twice.  This constituted the 40 paired stations for CPUE variance determination.  In addition, four stations (two within the buffer zone and two outside of it) were sampled every 3-4 hours over a 24-36 hour period to determine diel variations in Atka mackerel abundance at defined locations related to tides, light, and current. A total of 140 hauls were taken during the 12 days of trawl operations from 6 to 17 August.

A total of approximately 2,500 tagged Atka mackerel were released during the cruise.  Most of the fish (approximately 1,700) were tagged and released opportunistically during trawl operations during 6-17 August.  About 10% of the tagged fish were double tagged to estimate tag shedding rates.  During the last 12 hours on station,  all time was devoted to tagging.  Approximately 800 mackerel were tagged and released in the area currently open to the fishery.

The B-season Atka mackerel fishery began on 1 September and was closed on 8 September 1999 when the total allowable catch (TAC) was reached.  During the fishery, six tagged Atka mackerel were caught by the fishing vessel SeaFisher during operations south of Seguam Island.  Four of the tags were recovered, while two tags were observed by the processing crew but not subsequently recovered.  The four tagged fish recovered were each released outside the trawl exclusion zone and recaptured 4-7 nmi away from their release point; each was recaptured 15-22 days after release.

The tagging mortality experiment was conducted during the 12 days of trawl operations 6-17 August.   Twenty fish were placed in each of four tanks; 10 of the fish were tagged (2 double-tagged) and 10 were untagged.  By the end of the 12-day experiment, only three fish had died, one untagged fish and two tagged fish, resulting in a 97.5% survival rate for untagged fish and 95% survival rate for tagged fish.  Observations of internal injuries and scale/skin loss suggest that the three fish died not from the tagging procedure, but the effects of initial trawl capture.  On 17 August, all untagged fish in the experiment were tagged and released, along with all tagged fish.

By Lowell Fritz.


The Ocean Surface Current Simulations (OSCURS) model is available to the public on a Live Access Server via the REFM Division’s webpages at The interactive numerical model allows users to pick their own input by: clicking a start-point on the graphic chart of the North Pacific Ocean and Bering Sea; selecting any start-day from January 1980 to July 1999; and selecting a duration (the number of days to drift). Based on the input, the model generates a chart that appears in about 20 seconds showing the vectors of daily movement strung together in a trajectory that gives the net drift trajectory from the start-point.

Development of OSCURS was motivated by the need in fisheries research for indices that describe variability in ocean surface currents.  These synthetic data, derived through empirical modeling and calibration, provide insight that far exceeds their accuracy limitations.  OSCURS daily surface current vector fields are computed using empirical functions on a 90-km oceanwide grid based on daily sea level pressures (1946-97).  In addition, long-term mean geostrophic currents have been incorporated.  The model was tuned to reproduce trajectories of satellite-tracked drifters with shallow drogues from the eastern North Pacific.

The internet application provides two options for generating output; 1) a color chart of the North Pacific Ocean and Bering Sea with the trajectory in red; or 2) an ASCII data file of daily latitude-longitudes of surface water movement.  Trajectories replicate satellite drifter movements quite well on time-scales of a few months.  Up to 1-year-long trajectories can be produced, but their absolute accuracy diminishes with time.  Currently, trajectories after July 1999 simply use the long-term mean pressure fields (but this will be updated as new pressure field data become available).

Repeating the runs from the same location in different years can give a time series of surface current patterns.  This serves as one of the main purposes of OSCURS for comparison with fisheries data.

Our most recent experiment compares 90-day drift starting 1 March from Unimak Pass each year from 1970 to 1999 with year- class strength of walleye pollock.  There is some indication that on-shelf, northeastward drift of eggs and larvae after spawning favors greater recruitment, whereas weak or northward drift does not.

By Jim Ingraham and Jim Ianelli.


Activities and Research Results

Stomachs collected totaled 6,430 for the eastern Bering Sea and 3,040 for the Gulf of Alaska. Laboratory analysis was performed on 1,157 groundfish stomachs from the eastern Bering Sea and 1,236 from the Aleutian Islands region. Three observers returned with 159 stomach samples collected from the eastern Bering Sea.

Diet Overlap of Some West Coast Groundfish

The stomach contents of several species of groundfish were collected off California, Oregon, Washington, and British Columbia during standard NMFS surveys in 1989, 1991, and 1992.  Details of the analysis of our findings from these samples were recently published in Buckley et al. (1999).  One part of the analysis was the estimation of the amount of diet overlap (calculated from diet composition in terms of weight) among the groundfish species that were sampled.  Pacific whiting (Merluccius productus), arrowtooth flounder (Atheresthes stomias) and juvenile sablefish (Anoplopoma fimbria) consumed mostly euphausiids and other euphausiid consumers.  Adult sablefish, shortspine thornyhead (Sebastolobus alascanus), longspine thornyhead (S. altivelis), Pacific grenadier (Coryphaenoides acrolepis) and giant grenadier (Albatrossia pectoralis) were omnivorous, preying on a wide variety of fishes, crustaceans and other invertebrates, and scavenging on offal. Most of the prey were benthic or epibenthic.  Two small-mouthed flatfishes, Dover sole (Microstomus pacificus) and deepsea sole (Embassichthys bathybius), consumed mainly polychaete worms and brittle stars.  Spatial trends in main prey categories consumed by these groundfish species are also shown in the report.

Diets of Groundfish in the Gulf of Alaska in 1990, 1993, and 1996.

A total of 13,928 stomachs from 13 species were analyzed to describe the food habits of the major groundfish species in the Gulf of Alaska in 1990, 1993, and 1996. The analysis emphasized predation on commercially important fish, crab, and shrimp.

Arrowtooth flounder, Pacific halibut, sablefish, Pacific cod, and pollock were the main predators that consumed fish.  The main predators that fed on Tanner crabs were Pacific halibut and Pacific cod.  Pollock, shortspine thornyhead, shortraker rockfish, flathead sole, and rougheye rockfish were the main consumers of pandalid shrimp.  Pacific ocean perch, northern rockfish, dusky rockfish, and Atka mackerel fed mainly on zooplankton such as euphausiids and calanoid copepods.

refmfig1.jpg (43170 bytes)
Figure 1.  Percent weight of dominant prey items in the diet of major groundfish species in the Gulf of Alaska in summer of 1990, 1993, and 1996.
(prey key:  pan = pandalid shrimp,  cap = capelin,  plk = pollock)

A significant finding was the degree of predation on pandalid shrimp, capelin, and pollock by the dominant groundfish species in the Gulf of Alaska (Figure 1 above).  Pollock were the dominant prey fish every year and were consumed mainly by arrowtooth flounder, Pacific halibut, sablefish, and Pacific cod.  Pollock cannibalism, which accounted for only 2% of the diet by weight in 1990 and 1% in 1993, increased to 10% in 1996.  Other forage fish such as Pacific herring, capelin, eulachon, Pacific sand lance, and Atka mackerel can be categorized as the next most important prey fishes in the diet of groundfish.  All predators sampled, except northern rockfish and Atka mackerel, preyed on pandalid shrimp.

Multispecies Forecasting of the Effects of Fishing

For several years, the groundfish plan teams of the North Pacific Fishery Management Council have expressed specific ecosystem concerns with regard to the effects of fishing on species composition.  In particular, the plan teams have noted that large differences exist in the harvest rates of groundfish species off Alaska.  Some groundfish (such as walleye pollock, cod, sablefish, and rockfish) are harvested at or close to the recommended allowable biological catch level while other species (such as some flatfish) are harvested at substantially lower levels.  The plan team has requested analysis of the long-term implications of disproportionate harvest rates.

In order to address this concern, we have used results from a multispecies virtual population analysis (MSVPA) model for the eastern Bering Sea, which has been updated to reflect information from the 1998 stock assessments.  The outputs from the MSVPA model have been used to forecast the possible long-term multispecies effects of harvesting groundfish.

A system of eight species was defined for the eastern Bering Sea and modeled in the multispecies VPA and the resulting multispecies forecast.  Four species played the role of both predator and prey including walleye pollock, Pacific cod (Gadus macrocephalus), Greenland turbot (Reinhardtius hippoglossoides) and yellowfin sole (Pleuronectes asper). Rock sole (Lepidopsetta bilineata) and Pacific herring (Clupea harengus pallasi) were considered only as prey.  Finally, arrowtooth flounder (Atheresthes stomias) and northern fur seal (Callorhinus ursinus) were considered ‘other predators.’ Other predators are considered external predators within the MSVPA because their populations are not estimated within the MSVPA; instead, they are provided externally from other sources. For the forecasting of eastern Bering Sea, it was assumed that the northern fur seal population remained constant. For arrowtooth flounder, we followed the same assumptions on recruitment and fishing mortality as the rest of the species.  Figure 2 below shows the biomass flow for the system defined for the eastern Bering Sea.

refmfig2.jpg (41058 bytes)

Figure 2.  Biomass flow of the system defined for the eastern Bering Sea.

Deterministic multispecies and single species forecast models were set up to analyze and contrast the equilibrium dynamics of the system under three regimes of fishing mortality. The first regime evaluates the status quo removal rates of groundfish (Fref - calculated as the average of the last 3 years (1996-98) of the fishing mortality estimates from the MSVPA models). The second fishing regime examined the effects of fishing at rates that produced the allowable biological catch levels recommended in the 1998 stock assessments (FABC).  The final scenario examined the effects of no fishing on the multispecies complex.  Long-term equilibrium forecasts of these scenarios were made using a multispecies model that took predation interactions into account and single species models that did not include predation interactions.  Results from the multispecies and single species forecasts predict almost the same trends for the relative changes of yield biomass, total biomass, and spawning biomass under the FABC regime relative to the Fref case with the exception of rock sole. For rock sole, the multispecies model predicts a larger increase in yield and a smaller decrease in total biomass than predicted by the single species forecast, with the difference presumably caused by predation interactions (particularly with cod) not taken into account in the single species model.

Under the no-fishing regime, much larger differences were seen in biomass levels relative to the Fref case in the single species forecasts than in the multispecies forecasts.  The single species forecasts predicted much larger increases in total biomass of groundfish populations under a no-fishing regime than in the multispecies no-fishing regime, which included predation interactions.  Although a simple assumption of constant recruitment was implemented, the multispecies forecasting model provided valuable information on the possible consequences of fishing regimes and highlighted the importance of including predation. In the future, more realistic assumptions on recruitment will be added to the model.

By Pat Livingston.


During the third quarter of 1999, 241 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 region.  They sampled aboard 227 fishing and processing vessels and at 16 shoreside processing plants.  These observers were trained or briefed in various locations.  The AFSC Observer Program in Seattle trained 51 first-time observers and briefed 45 observers who had prior experience, while the University of Alaska Anchorage (UAA) Observer Training Center briefed 85 observers and trained another 32 first-time observers. At the Observer Program’s field offices in Dutch Harbor and Kodiak, 14 more observers were briefed and 14 were excused from briefing because they had just completed a cruise successfully and were returning immediately to the field.  The third quarter 1999 observer workforce thus comprised 34% new observers and 66% experienced observers.

The Observer Program conducted a total of 72 debriefings during the third quarter of 1999.  Three debriefings were held in Dutch Harbor, 18 in Anchorage, and 51 were held in Seattle.  No debriefings are conducted at the UAA.

During the third quarter of 1999, implementation of an expanded Community Development Quota (CDQ) program and implementation of provisions of the recently enacted American Fisheries Act (AFA) continued.  The CDQ program was developed for the purpose of allocating fishery resources to eligible Western Alaska communities to provide the means for starting or supporting commercial fishery activities that would result in ongoing regionally-based, commercial fishery or related businesses.  The program was initiated in 1992 with walleye pollock and expanded to include fixed-gear halibut and sablefish in 1995.  In 1998, it was further expanded to include multiple species (MS) of groundfish and crab (MSCDQ).  As of 1999, NMFS is responsible for monitoring the groundfish (including pollock and sablefish) and halibut CDQs, and the state of Alaska is responsible for monitoring crab CDQs.

The AFA, enacted by Congress in late 1998, made changes to the pollock fishery in the Bering Sea and Aleutian Islands.  These changes included reallocation of fish among industry segments, provided for the formation of fishing cooperatives, and increased observer coverage levels on some components of the fleet.  The offshore component of the fleet organized a fishing cooperative this year and has been receiving increased, mandatory observer coverage.  More recently, the Observer Program has been involved in implementation of aspects of the AFA related to shoreside pollock.  The shoreside component has proven to be more complex than the offshore component and will involve possible NMFS regulatory actions and a changing role for the observer.

MSCDQ and AFA catch accounting for offshore processors is based entirely on data collected by observers and, unlike the open access fisheries where observer data is used to manage a fleetwide quota, industry participants in the MSCDQ and AFA fisheries require individual accounting of fish harvested in each haul or set. This change has required great effort on the part of the Observer Program to develop special selection criteria and training requirements for observers, develop new sampling strategies and regulations to enhance the observer’s working environment, and implement changes to the data collection and data management software systems.

An extensive, independent review of the Observer Program began during the third quarter of 1999.  The review is being carried out by Marine Resources Assessment Group (MRAG) Americas, Inc., an independent consulting firm, which provides professional advice and services for the management of marine fisheries throughout the world.  Their final report is due sometime next year.

By Bob Maier.


Flathead sole


Rock sole


Rex sole


Alaska plaice


Northern rock sole


Yellowfin sole


Arrowtooth flounder


Walleye pollock




Atka mackerel


Pacific whiting


Pacific ocean perch


Northern rockfish


Production figures from 1 January to 30 September 1999 are shown to the right. Total production figures were 20,324 with 7,278 test ages and 190 examined and determined to be unageable.

By Dan Kimura.