2007 Agency Report to the Technical Subcommittee

NorthwestFisheriesScienceCenter

National Marine Fisheries Service

2007 Agency Report to the Technical Subcommittee

of the Canada-U.S. Groundfish Committee

April 2007

Review of Agency Groundfish Research, Assessments, and Management

A. Agency Overview

The Northwest Fisheries Science Center (NWFSC) provides scientific and technical support to the National Marine Fisheries Service (NMFS) for management and conservation of the Northwest region’s marine and anadromous resources. The Center conducts research in cooperation with other federal and state agencies and academic institutions. Five divisions, Conservation Biology, Environmental Conservation, Fish Ecology, Resource Enhancement and Utilization Technologies, and Fishery Resource Analysis and Monitoring, conduct applied research to resolve problems that threaten marine resources or that deter their use. The Center’s main facility and laboratories are located in Seattle. Other Center research facilities are located in Pasco, Big Beef Creek, Mukilteo, and Manchester, Washington; Newport, Hammond, and Clatskanie, Oregon; and Kodiak, Alaska.

The Fishery Resource Analysis and Monitoring Division (FRAMD) is the source focus for most of the research reported by the NWFSC to the Technical Subcommittee of the Canada-US Groundfish Committee. The FRAMD works in partnership with state and federal resource agencies, universities, and the groundfish industry to achieve a coordinated groundfish program for the West Coast.

FRAMD consists of a multi-disciplinary team with expertise in fishery biology and ecology, stock assessment, economics, mathematical modeling, statistics, computer science, and field sampling techniques. Members of this program are stationed at the NWFSC facilities in Seattle and in Newport, Oregon, with some Observer Program staff located in California. Together, they work to develop and provide scientific information necessary for managing West Coast marine fisheries and strive to provide useful and reliable stock assessment data with which fishery managers can set ecologically safe and economically valuable harvest levels. FRAM researchers develop models for managing multi-species fisheries; design programs to provide information on the extent and characteristics of bycatch in commercial fisheries, as they look at methods to reduce fisheries bycatch; characterize essential habitats for key groundfish species; investigate the design, feasibility, function, and value of marine protected areas; and employ advanced technologies for new assessments.

During 2006, FRAMD continued to: implement a West Coast observer program; build a survey program that will conduct West Coast groundfish acoustic and trawl surveys previously conducted by the AFSC; develop new technologies for surveying fish populations, particularly in untrawlable areas; and expand its stock assessment, economics, and ecosystem research. Significant progress continues in all programs.

For more information on FRAMD and groundfish investigations, contact the Division Director, Dr. M. Elizabeth Clarke at , (206) 860-3381.

Other Divisions at the NWFSC are:

The Conservation Biology Division is responsible for characterizing the major components of biodiversity in living marine resources, using the latest genetic and quantitative methods. It also has responsibility for identifying factors that pose risks to these components and the mechanisms that limit natural productivity. The Division’s multi-disciplinary approach draws on expertise in the fields of population genetics, population dynamics, and ecology.

The Environmental Conservation Division (ECD) conducts nationwide research on the effects of chemical pollution and harmful algal blooms on habitat quality and fisheries resources. ECD is also a leader in NMFS’ National Marine Mammal Health and Stranding Response Program’s bio-monitoring and quality assurances projects.

The Fish Ecology Division’s role is to understand the complex ecological linkages among important marine and anadromous fishery resources in the Pacific Northwest and their habitats. The Division particularly places emphasis on investigating the myriad biotic and abiotic factors that control growth, distribution, and survival of important species and on the processes driving population fluctuations.

The Resource Enhancement and Utilization Technologies Division draws together multi-disciplinary groups to address existing and developing challenges of captive rearing of salmon and other marine fish, improved hatchery practices, smolt quality, disease control, and developing technologies for full utilization of bycatch and fish processing waste.

For more information on NorthwestFisheriesScienceCenter programs, contact the Center Director, Dr. Usha Varanasi at , (206) 860-3200.

B. Multi-species Studies

2.) Stock Assessment

a) SS2 Stock Assessment Model Development

Stock Synthesis 2 (SS2) is an assessment model in the class termed integrated analysis. SS2 is built with a population sub-model operating by forward simulation. SS2 has an observation sub-model to estimate expected values for various types of data, and a statistical sub-model to characterize the data’s goodness of fit and to obtain best-fitting parameters with associated variance. It includes a rich feature set including age- and size-based population dynamics and the ability to specify observational phenomena, such as ageing imprecision. Model parameters can vary randomly or across time blocks or can be specified as functions of environmental data. SS2 includes routines to estimate MSY and exploitation levels that correspond to various standard fishery management targets. A user-selected harvest policy is used to conduct a forecast in the final phase of running the model. The model is coded in ADMB (Dave Fournier, Otter Research Ltd.). SS2 is now included in the NOAA Fisheries Assessment Toolbox ( incorporating a graphical user interface developed by Alan Seaver (NEFSC).

In 2005, SS2 was used to assess the status of about 20 groundfish stocks off the U.S. West Coast. At a workshop in November 2005, its general capabilities were compared to CASAL, Multifan-CL, and SCALA. There were subsequent exploratory applications of SS2 for Pacific tuna assessments. In January 2006, SS2 was introduced to CSIRO scientists in SE Australia and have been applied to six assessments to date. In the same year, SS2 was applied by AFSC scientists to the Pacific cod stock off Alaska. In August 2006, a workshop was held to review the model usage in 2005 and to plan enhancements for 2007. Finally, in 2007, SS2 was updated to version 2.00, which incorporated several enhancements including algorithms to define movement between assessment sub-areas and enhanced controls over processes for growth, selectivity, and recruitment.

For more information, please contact Dr. Richard Methot at

b) Stock Assessment Improvement Workshops

Between August and November 2006, FRAMD organized (or co-organized) three workshops intended to better prepare analysts for groundfish stock assessments to be conducted during 2007. These workshops included one focusing on general data and modeling issues, one on the use of pre-recruit survey data, and one on use of FRAMD’s bottom trawl survey data.

1)Data-modeling workshop

The West Coast Groundfish Data/Modeling Workshop was held August 8-10, 2006 at the NOAAWesternRegionalCenter in Seattle, Washington. The workshop was held to review available data sources for West Coast groundfish stock assessments and address a number of topics relating to the treatment of data in assessments and other modeling issues, including a review of the features and functionality of the SS2 modeling platform. Over 40 workshop participants included stock assessment scientists from NOAA Fisheries and State agencies, data managers, fishing industry representatives as well as members of the public. Workshop presentations and discussion addressed the following topics:

• Methods employed to construct age and length compositions during the 2005 assessments.

• Issues relating to historic age reading data, including the electronic archival of descriptive meta-data, and logistical protocols for the transfer of age structures and the communication of age determinations between agencies.

• General Linear Mixed Model (GLMM) theory, its application to survey data, in particular, the west coast groundfish bottom trawl surveys, including a preliminary analysis for canary rockfish.

• The Scientific and Statistical Committee’s Terms of Reference for Stock Assessments and Stock Assessment Review (STAR) panels, as well as issues discussed during the Groundfish Stock Assessment Review Workshop in January, 2006.

• Issues relating to the specification of sample sizes for compositional data in assessments, approaches for “tuning” input sizes based on model outputs;

• A review of some of the new features in Stock Synthesis 2 (SS2) and discussion of further additions that would be useful for 2007 assessments.

• A new tool for quickly summarizing the results of a Stock Synthesis 2 (SS2) model run. Using the free software “R” ( five SS2 output files are condensed into a short list of statistics and a number of plots with one function call. This enables quick and easy evaluation of all aspects of a model run during exploratory development. Authors can contact Ian at: , for a copy of the software or assistance in using it.

• An overview of approaches used to define stock structure and management units for West Coast groundfish, including use of genetics, demographic patterns, and management/assessment units. Results from age-structured simulations evaluating the effects of the spatial distribution of fishing effort on stock size and yield. Preliminary work to identify stock structure using commonly collected data (catch per unit effort; CPUE) and simple clustering techniques.

• A summary of the model parameter priors used in 2005 assessments, along with presentations on the use of prediction intervals for natural mortality (M) and a method for calculating priors for the Beverton-Holt stock-recruitment steepness parameter.

2)Pre-recruit survey workshop

In 1983 theSouthwest Fisheries Science Center (SWFSC) initiated a mid-water trawl survey to collect data on, among other things, the abundance and distribution of young-of-the-year (pre-recruit) groundfish, including especially rockfishes of the genus Sebastes. Through 2003 this survey was narrowly focused in an area off the coast of central California from lat. 36º30’–38º20’ N. In 2001 a new pre-recruit survey conducted cooperatively by the NWFSC and the Pacific Whiting Conservation Cooperative (PWCC) was initiated, with the primary intent of monitoring young-of-the-year Pacific whiting abundance. The initial coverage of this survey ranged from lat. 35º00’ N (just south of Morro Bay CA) to lat. 45º00’ N (just north of Newport OR). Beginning in 2004, the geographic extent of both surveys was expanded, so that by 2005, the combined area of both surveys covered the entire U. S. west coast, from the Canadian to Mexican borders (lat. 33º00’– 48º00’ N).

A workshop focusing on the integration of data from these two pre-recruit surveys in west coast groundfish stock assessments was held September 13-15, 2006 at the SWFSC facility in Santa Cruz, CA. The workshop was jointly organized by the SWFSC and the NWFSC and was attended by over 20 people, including individuals involved in conducting both pre-recruit surveys, stock assessment scientists, and the public. Twelve scheduled presentations were organized into three sessions:

Session 1. Developing a Coast-wide Survey of Groundfish Pre-Recruit Abundance

Session 2: Incorporating Pre-Recruit Indices in Stock Assessments

Session 3: Case Studies

Throughout the course of the workshop’s discussions, several findings and suggestions for future surveys, research, and/or applications were broadly supported by participants. With respect to whethersurvey data collected by the SWFSC and NWFSC-PWCC surveyscan be combined into coast-wide indicesof pre-recruit abundance for Pacific whiting and rockfish, areas of general agreement included the following:

• For species that are distributed exclusively or predominantly north of Point Conception, data from the 2001-06 combined surveys provide acceptablespatial coverage for creating a coast-wide index. The combined spatial coverage during 2004-06 is reasonable for all species, including those with substantial catches taken south of Point Conception. However, the spatial coverage of the SWFSC survey during the 1983-2000 period is largely inadequate to index pre-recruit abundance for most species, particularly where coast-wide assessment areas are used in population modeling. The core SWFSC survey area appears to represent the preponderance of the distribution of a few species reasonably well (e.g., chilipepper),but may also prove useful in region-specific modeling for other stocks that have a more coast-wide distribution (e.g., widow rockfish).

• Comparison of methods and patterns in catch rates currently indicate that the SWFSC and PWCC/NWFSC surveys are sufficiently similar that data from the two surveys can be combined to form a single pre-recruit index over the area covered. However, detailed and more rigorous statistical comparisons of paired trawl observations should continue, and thetwo surveys should continue to be executed withsubstantial spatial overlap.

• Existing data from (time-separated) replicate tows in the SWFSC data should beanalyzed to assess the potential magnitude of variance and bias effects associated with varying numbers of replicate tows.

• AlternativeGeneral Linear Model (GLM)formulations should be explored for developing pre-recruit abundance indices.In particular, the potential benefits of replacing sampling stations with broader latitudinal and depth zones and introducing interaction terms, should be examined. Additionally, mixed-model (GLMM) forms should also be explored, for example, by treating calendar day as a random effect.

• As more data become availableand the development ofregional ROMS (Regional Ocean Model System) or otheroceanographic models progresses,their outputs may help in identifying the manner in which meso-scale ocean variability affects the abundance and distribution of young-of-the-year groundfish. A better understanding of these relationshipsmay facilitate improvements in pre-recruit survey design or interpretation of results.

With respect tothe modeling non-linearity in early life history processes, and specifically whether a power transformation should be used,areas of participant agreement included:

• Substantial density-dependent compensatory mortality can occur following measurement of pre-recruit abundance at the ontogenetic stage sampled by the surveys (e.g., 100-d). If compensation is substantial, then non-linearity will be introduced in the relationship between “pre-recruit” and “recruit” abundance.

• When non-linear transformation of an index is considered, the transformation should be conducted internally within the stock assessment model as an explicit part of the estimation procedure.

• The new SS2 option which allows specification of an expectation of density-independent pre-recruit abundance may remove the need for transformation. Comparative work to evaluate this issue should be performed, with a good candidate being southern widow rockfish.

• It is important to evaluate the degree to which non-linear transformation of pre-recruit survey indices is confounded with tuning to the model’s RMSE. Transformation and variance inflation should be conducted jointly.

Aside from their use in short-term forecasts of impending recruitment, pre-recruit surveys have the potential to provide significant insights into ecosystem dynamics, including:

• monitoring of epipelagic micronekton species diversity

• sensitivity of sampled taxa to high-frequency environmental variation

• monitoring of “small” forage species for use in trophic models

• potential for early detection of regime shifts (e.g., indicator species)

• providing information that may be useful in retrospective studies

• sampling is consistent with the ocean observing system (OOS) framework.

3)Bottom trawl survey workshop

The NWFSC Bottom Trawl Survey Workshop was held October 31 – November 2, 2006 at the NOAAWesternRegionalCenter in Seattle, Washington. The goal of the bottom trawl survey workshop was to provide stock assessment authors with guidance regarding the incorporation of data from the NWFSC West Coast groundfish bottom trawl survey into stock assessments, particularly those that will be conducted in 2007. The roughly 20 participants included stock assessment scientists, trawl survey personnel, fishing industry representatives as well as members of the public.

The NWFSC bottom trawl survey was initiated in 1998, covering depths from 100-700 fathoms. The survey was expanded in 2003 to include depths ranging from 30-100 fathoms. While data from the 100-700 fathom range have been included in prior assessments, data from the expanded coverage of shallower depths have not.

Primary workshop objectives and areas of general agreement are summarized below:

Objective 1. Review survey protocols and data collected by the NMFS West Coast groundfish bottom trawl surveys: AFSC & NWFSC triennial shelf surveys (1977-2004), NWFSC slope survey (1998-2002), and NWFSC “expanded” shelf-slope survey (2003-2006).

Participants generally agreed that substantial differences exist between the triennial shelf and NWFSC shelf-slope “expanded” survey protocols and gear (Table 1.). Some of these differences such as towing speed and duration, size and type of the nets and footropes, and selection of tow locations, may contribute to changes in catchability and selectivity. It is difficult to disentangle the effects of the various changes in protocol and gear on catchability and selectivity. However, it appears the effects differ among species.

Objective 2. Evaluate methods for including AFSC and NWFSC survey time series in “full” stock assessments.

Workshop participants generally agreed that the triennial and NWFSC Shelf-slope “expanded” surveys are different time series and should be included separately in assessments, based on the analyses presented for canary rockfish, English sole, darkblotched rockfish and arrowtooth flounder. This conclusion was based on the fundamental differences in survey protocols and performance. Stock assessment authors may explore alternative use or combinations of the surveys. Alternative options for including the NWFSC “expanded” survey are outlined below.

Two principal options were discussed for using the NWFSC “expanded” survey data in assessments for species which primarily occur on the slope (e.g. darkblotched rockfish). The first option is to continue using the NMFS conducted slope survey time series (>100 fm) as in previous assessments, and add data from the recently sampled shelf depths (<100 fm) of the NWFSC “expanded” survey as a new and separate time series. Although this approach may represent a viable interim method for including the new data, as it preserves a longer continuous slope time series, participants generally thought this method is not a long term solution. As the length of the “expanded” survey time series increases, information and statistical power may be lost by separating data from the “expanded” survey into two concurrent time series Also, since the NWFSC survey was not extended to the Mexican border until 2003, a slope time series from 1998 to 2006 would reflect very different geographic coverage. In order to maintain geographic consistency over this time period, data from the Conception INPFC area would need to be excluded from the derivation of a survey index.