NOAA-NMFS-NWFSC TM-31: Data Collection -- Groundfish (cont):

3.1 Introduction

The groundfish fishery is an important component of the commercial fisheries in Oregon. During the 1980s, the groundfish fishery surpassed the fisheries for salmon and shellfish in landed weight and was comparable to them in landed value (Fig. 3.1). During this decade, the rockfish species composed the largest segment of the groundfish landings, both in landed weight and landed value, followed by the flatfish species, and then the other groundfish species (Fig. 3.2). In 1991, the groundfish fleet in Oregon consisted of approximately 136 trawlers and a smaller number of longline, hook and line, and pot vessels (Silverthorne 1992).

Compared to the fisheries for salmon and halibut, the fishery for groundfish along the Pacific coast is a relatively recent development. According to Harry (1956) the Oregon fishery for groundfish was sporadic until the end of the 1930s, when a permanent otter-trawl fishery became established in response to increased demand for fish during World War II. During the period from 1940 to 1953 the fishery reached a peak in 1945, when 73 vessels harvested about 26 million lb of groundfish. Markets for groundfish declined after the war, and by 1953 the groundfish landings, delivered by 44 vessels, were only about 15 million lb, one third of which was used for animal food on commercial mink ranches.

Prior to 1979, groundfish landings in Oregon consisted primarily of flatfish species, particularly Dover, English, and petrale sole. The large landings of rockfish, which are typical of today's fishery, did not develop until the late 1970s. During the 1950s the flatfish species accounted for 58% by weight of the groundfish landings and the rockfish species accounted for 35%. During this decade the landings averaged 22.5 million lb annually. Similarly, during the 1960s the annual groundfish landings were 51% flatfish and 41% rockfish, and averaged 27.8 million lb per year. Even during the 1970s the landings were 52% flatfish and 30% rockfish, and averaged 28.5 million lb annually. Rockfish landings did not surpass flatfish landings until 1979, when the midwater trawl fishery developed for widow rockfish.

3.1.1 Geographical Overview

Major ports

The major ports for groundfish landings in Oregon are Astoria on the north coast, Newport on the central coast, and Coos Bay on the south coast (Fig. 3.3). These three ports accounted for 95% by weight and 91% by value of the reported groundfish landings in Oregon during 1991 through 1992 (Table 3.1). The ports of Garibaldi on the north coast and Brookings on the far south coast together accounted for an additional 4% of the weight and 7% of the value of the 1991 and 1992 total landings of groundfish.

Major fishing grounds

The groundfish vessels that land their catches in Oregon generally operate in the waters directly off Oregon or to the north off Washington, with some vessels from ports on the far south coast occasionally fishing off northern California. Although areas differ in productivity and in the species assemblages, and therefore attract differing amounts of fishing pressure, essentially all areas off the coast are fished to some degree (Fig. 3.4). Depending on the size of the fishing vessel and the species of interest, the vessels operate in the shallows near shore on out to depths of greater than 650 fathoms at a distance of 30 to 40 miles offshore. During the development of the fishery there has been a general tendency for the fleet to operate at greater and greater depths, fishing on stocks that were previously unexploited.

Principal species landed

In terms of the total weight of the harvest, the top five groundfish species landed in Oregon during 1991 through 1992 (in descending order) were Pacific hake, Dover sole, widow rockfish, sablefish, and thornyhead (Table 3.2); in terms of the total landed value, the five most important species were sablefish, Dover sole, thornyhead, Pacific hake, and widow rockfish. The relative importance of the different species was not uniform across all Oregon ports, however. In Astoria during 1991 through 1992, the top five species (in descending order by weight) were Pacific hake, Dover sole, arrowtooth flounder, yellowtail rockfish, and widow rockfish; in Newport, the top five species were Pacific hake, widow rockfish, Dover sole, thornyhead, and yellowtail rockfish; and in Coos Bay, the top five species were Dover sole, thornyhead, sablefish, widow rockfish, and canary rockfish.

The vast majority of the groundfish landed in Oregon are caught with trawl gear. During 1991 through 1992 midwater trawls accounted for 50% of the total landed weight of groundfish at Oregon ports, and bottom trawls accounted for an additional 46% (Table 3.3). The remaining landings were taken by line gear (2.5%), fish pots (0.8%), shrimp trawls (0.7%), and troll gear (0.1%). The large increase in the relative amount caught by midwater trawls, from 31% in 1991 to 62% in 1992, was due to the rapid development of the shore-based fishery for Pacific hake.

Port biologists and field staff

Maintaining communications with the fishers and the dealers is vitally important for the successful management of any commercial fishery. The field staff are the most visible and accessible components of the Oregon Department of Fish and Wildlife's (ODFW) Marine Finfish Program, and they are the primary instruments for collecting data and disseminating information to the fishing industry. The ODFW has local offices at the three major ports on the Oregon coast. These facilities are used by the four permanent groundfish port biologists: one stationed at Astoria, two at Newport, and one at Coos Bay (Charleston). Most of these port biologists have served in their present positions for more than 10 years. During the 1992 fishing season, ODFW employed four seasonal aides to assist the permanent port biologists. These temporary staff, who worked 5- or 7-month appointments, were based at the ports of Astoria, Newport, Coos Bay, and Brookings. Some of these seasonal aides functioned in the same capacity in previous years.

Port biologists involved in groundfish sampling in Oregon are responsible for collecting the following information: 1) Oregon Fish Receiving Tickets completed by the dealers, 2) trawl logbooks completed by the fishers, 3) species-composition samples from landings of rockfish, and 4) biological samples from selected groundfish species. The specific tasks associated with each of these responsibilities are further discussed later in this chapter.

In 1992 ODFW also instituted a special program for sampling the developing shore-based fishery for Pacific hake. Its primary function was to document the bycatch of regulated species. One of the port biologists in Newport was assigned to act as the overall program coordinator, and a seasonal aide was hired to serve as the crew chief for the six to eight samplers employed by the Pacific States Marine Fisheries Commission (PSMFC) to sample the Pacific hake fishery in Oregon. In addition to the field staff involved in the on-site sampling of groundfish landings, numerous other personnel serve in various roles in the data collection and processing systems used by ODFW to monitor the commercial groundfish fishery (Table 3.4).

3.1.2 History of Data Collection Systems

There have been few formal descriptions of the data systems used for collecting and processing information about Oregon's groundfish fishery. According to Harry (1956) there was little research on Oregon's otter-trawl fishery prior to World War II, but beginning in June 1941 fish processing companies were required for taxation purposes to record landing statistics in terms of pounds landed per delivery. These data were gathered and tabulated by staff of the Oregon Fish Commission. Beginning in 1943, as a check on the accuracy of company records, commercial fishers were required to record their landings in a "pass book" that was sent to the Fish Commission at the end of each fiscal year, as a condition of license renewal. Initially, the landing statistics were recorded and processed using hand-punch cards, but starting in April 1947, IBM punch-card machines were used to organize and tabulate the data. A program of scientific research that addressed the Oregon groundfish fishery began during the summer of 1947 and the first "market samples" from the landings were taken in January 1948.

Although Harry (1956) does not refer to any form of logbook system for collecting information from the groundfish fishery, in 1949 the Oregon Fish Commission was using logbooks for monitoring the soupfin shark fishery (Westrheim 1950), and trawl logbooks had become an integral part of the groundfish monitoring program at least by the late 1970s. During 1975 and 1976, staff of the Oregon Department of Fish and Wildlife, with assistance from Oregon State University (OSU), designed and implemented a groundfish and shrimp data processing system. The system utilized a mainframe computer at OSU to match trip-by-trip landing information (from fish tickets) with tow-by-tow trawl logbook data, thereby developing estimates of catch by geographic area (Johnson and Lukas 1976). This system was the forerunner of the system described below.

3.1.3 Legal Authority to Collect Data

Obligations of fishers

Authority to collect fisheries information is granted in general terms specified by the Oregon Fish and Wildlife Code, which contains rules established by the Oregon Legislature. However, most of these rules are only indirectly associated with data collection activities in the marine environment. Rules directly related to marine fisheries are found in the Oregon Administrative Rules (OAR), which are established by the Oregon Fish and Wildlife Commission. The Director of the Oregon Department of Fish and Wildlife, acting on behalf of the Commission, has authority to establish emergency rules, when doing so is deemed to be in the best interests of the resource.

Division 4 of the OAR (Commercial Fisheries Other Than Salmon or Shellfish) contains several rules that facilitate data collection activities for "Ocean Food Fish Fisheries." OAR 635-04-018 defines the general intent of the rules regarding these data collection activities. OAR 635-04-029 requires fishers to sort their catch according to identified "management categories," which are defined as those species or groups of species that have a trip limit, quota, or harvest guideline. OAR 635-06-210 requires that landings reported on Oregon Fish Receiving Tickets be sorted by individual species; however, in practice, sorting is only enforced at the management category level rather than at the species level. Because of recent changes in the tax structure, this rule now requires that the species or groups of species also must be recorded according to the ex-vessel prices paid to the fisher.

Three rules facilitate the collection of data and are used to ascertain where at-sea commercial catches were made. OAR 635-04-050 establishes that logbooks must be maintained aboard each vessel that harvests ocean food fish, which includes fish caught with groundfish trawls, pots, and longlines, as well as shrimp trawls. Currently, small vessels that use miscellaneous gears (e.g., jigs, vertical longlines, and trolls) are not required to keep logbooks. A set of rules regarding offshore fisheries (OAR 635-04-125, 130, 135, 140, and 145) establishes, via a mandatory declaration procedure, the identification of vessels that fish outside the U.S. Exclusive Economic Zone and land their catches at Oregon ports.

Noncompliance

There have been few problems associated with the collection of mandatory data, with the exception of the late 1970s and early 1980s, when many fishers refused to maintain logbooks. An informal education program regarding the legal requirements of maintaining logbooks and visits from Oregon State Police officers eliminated most of these problems. Although citations may be issued for noncompliance, the need for this has been rare. In general, problems have occurred due to inadequate communication regarding the need for the information, or when individual buyers or fishers have been upset by recent management actions.

3.2 Fish Ticket/Landing Receipt System

Any fish dealer who purchases groundfish from a commercial fisher is required by law to complete an Oregon Fish Receiving Ticket indicating the weight and value of the fish purchased. An example of the fish tickets used in 1992 is given in Figure A-2. The fish tickets provide important information about harvest levels, species composition, and economic value of the fishery. Also, Oregon uses the fish tickets to levy a tax on the value of the reported landings.

3.2.1 Market Categories

In consultation with other ODFW staff, the fisheries information supervisor and the groundfish technical supervisor determine the physical format of the ticket and the official market categories under which the fish must be landed. The existence of different market categories is controlled in part by the marketplace, e.g., different species or sizes of fish may command different prices. However, in some cases the Pacific Fishery Management Council (PFMC) imposes the market categories on the fishery because of the need to maintain control over the landings of regulated species. For example, in the 1970s all of the rockfish species landed in Oregon were recorded in a single market category. However, special market categories have since been established for certain species (e.g., Pacific ocean perch, widow rockfish, yellowtail rockfish, and thornyhead) because the PFMC established landing quotas and required that the fishers sort these species from other rockfish species prior to landing their catches.

In 1992 there were a total of 89 official market categories used by ODFW. Each market category is designated as an individual species or group of species. As of 1992 in Oregon, only landings of rockfish market categories were sampled for determining species compositions (see section 3.4). Some of the market categories were sampled as part of the biological sampling program (see section 3.5).

Dealers are allowed to write in additional, unrecognized market categories on the fish tickets. For example, some dealers might itemize the landings of yelloweye rockfish separately from the "large rockfish" market category because they pay the fishers a higher price for this species than for the other species landed within the large rockfish category. When this occurs, the local port biologist corrects the tickets so that they conform with the officially recognized market categories.

Because of their ability to obtain higher prices for certain species, fishers now sort their landings of rockfish to a much greater extent than in earlier years. This is particularly true for fish caught by hook and line that are sold as whole fish. The occurrence of mixed-species landings that are sometimes sorted, but not always, greatly complicates the sampling program. A system in which there are only a few categories has the advantage of compactness and simplicity, but with such a system important information may be lost regarding the composition of the landings.

A groundfish market category has never been deleted from the Oregon system. The decision to delete a category would be made by the fisheries information supervisor and the groundfish technical supervisor.

3.2.2 Processing System

The system for collecting, entering, and verifying the landing information on a fish ticket is fairly complicated, involving several steps and numerous people (Fig. 3.5). The process begins with the purchase of fish by a dealer and the completion of a fish ticket. The dealer sends one copy of each completed multi-part ticket to the ODFW office in Portland. Dealers are legally obligated to send their completed tickets to the Portland office within five working days of the purchase of any fish. Another copy of the ticket is collected by the local port biologist during one of his regular visits to the fish dealers. In some cases, dealers who are remote from a port biologist send copies of their tickets to the local ODFW office. The ODFW routinely provides the dealers with pre-addressed business reply envelopes for mailing in the completed tickets. The two copies of the ticket are processed and keypunched independently, one copy in Newport and one in Portland. The two resulting data files are compared and discrepancies are resolved. Through 1992 an additional check on the consistency of the ticket data was provided by an ODFW auditor in Portland, who identified discrepancies between the total landings reported by each dealer each month and the total landings from the corresponding individual tickets sent to the Portland office. However, starting in 1993, the commercial fish auditor was assigned other duties.

Although it is generally true that a fish ticket represents landings from only one fishing trip, it is important to recognize that a fishing trip can result in more than one ticket. This can occur for several reasons: 1) dealers are legally required to create separate tickets for trips with multiple gear types (e.g., for the bottom trawl and midwater trawl portions of trips that used both types of fishing gear), 2) dealers create separate tickets for landings that are over trip limits if instructed to do so by the Oregon State Police, and 3) a fishing vessel may deliver to more than one dealer. Because of the Oregon ad-valorem tax on groundfish purchases, each delivery must have a separate ticket.

Data entry and error checking

Every fish ticket that involves the sale of groundfish is examined by the local port biologist for missing or inconsistent information. The port biologists check for errors, such as inappropriate gear codes (e.g., 30,000 lb of widow rockfish caught with a bottom trawl rather than a midwater trawl) or an incorrect market category (e.g., rockfish that were reported as "large rockfish," but which the port biologist knows were "small rockfish"). The port biologist revises the ticket as required.

When a port biologist finds an error, he fills out an error correction form, which indicates the required change and the reasons for it. Copies of the error correction form are given both to the data entry staff in the Newport office and to the fisheries information supervisor in Portland.

The port biologist is responsible for filling in the fishing area code (PSMFC area) on the ticket, using either logbook information, if available, or the default fishing area code for that port. The fishing area data from the tickets are used to apportion the landings to fishing areas for preliminary reports to PacFIN. A more complete accounting of catches by area is provided in an end-of-the-year report that makes full use of all available logbook data to designate which fishing areas the landings came from.

Entry of the fish ticket data in Newport is done directly onto the mainframe computer in Portland and occurs after the same ticket information has been entered by staff at the Portland office. As the data technician enters the data in Newport, the data are checked immediately against the corresponding information in Portland and the operator in Newport is alerted to any discrepancies. If the data technician in Newport determines that data have been incorrectly entered, the discrepancy is resolved later by the fisheries information supervisor after consultation with the port biologist and possibly with the fish dealer.

Data processing in Portland

Books of fish ticket forms are issued to fish dealers by clerical staff in the Portland office. Every fish ticket is preprinted with a unique serial number, and the clerical staff in the Portland office maintains a complete list of all outstanding serial numbers. When tickets are received in the Portland office, the ticket numbers are cross-checked against the complete list. Dealers are warned when they return tickets out of sequence and are required to account for all of the tickets they have been issued.

Each completed fish ticket is examined by clerical staff members in Portland, who are responsible for filling in appropriate codes for any unofficial market categories that the dealer may have written on the form. They leave blank the "area fished" field, which is completed by the data technician in Newport. The coded tickets are then keypunched by the data entry staff during the evening, usually on the same day they were received in the mail.

During data entry the following data elements from each ticket are cross-checked against preexisting information: the commercial license number, the boat number, the dealer number, the date, and the port, gear, and species codes. Also, the values entered by market category for the pounds and prices are checked against valid ranges. Error reports are generated for each evening's entries and are investigated and corrected the following day by the fisheries information supervisor and his clerical staff.

Typically, one week elapses between the date of a landing and the initial data entry in Portland of the ticket that accounts for that landing. About 95% of the tickets are entered within two weeks of the actual date of the landings. The main objective for the initial data entry in Portland is to get the information onto the database as quickly as possible. Although some routine error checking occurs during this initial stage, the more complete validation of the data does not occur until the data technician in Newport enters the information a second time. This stage of the process does not occur until the month following the landing, after the port biologists have examined and coded the tickets. Rectifying an error after it is discovered may take one month or until the final end-of-the-year reporting, in May of the following year. However, most errors are corrected shortly after the second entry in Newport.

3.2.3 Groundfish Landings Not Covered by the Fish Ticket System

In Oregon, the vast majority of the groundfish catches are taken by the commercial fishery and are reported on fish tickets. However, there are active fleets of charter and private boats that catch groundfish for recreation and personal consumption. These recreational catches are not included in the Fish Ticket System, but the recreational landings are minor relative to the commercial landings (Table 3.5). Using interview data and boat information that are maintained separately from the commercial fishery database, staff from the Marine Recreational Fishery Program routinely estimate the catches of groundfish by the recreational fishery in Oregon. The recreational fishery in Oregon tends to target nearshore groundfish species that are not commercially important, the main exceptions being halibut, yellowtail rockfish, canary rockfish, and lingcod.

Other potential sources of landings that may go unreported include fish sold directly to consumers and fish taken home by the crews of commercial vessels ("home pack"). However, in Oregon it is illegal for fishers to sell fish directly to restaurants or consumers unless they possess a "limited fish seller's license," and unreported direct sales of groundfish are believed to occur infrequently. Also, fishers in Oregon are legally required to report any home pack on the fish tickets, and the amount of unreported home pack is believed to be small.

3.3 Logbook System

The ODFW issues logbooks to groundfish vessels to obtain detailed information about fishing locations and fishing effort (hours of fishing). Because most of the groundfish landed in Oregon are caught in the trawl fishery, this section addresses the collection protocols and data processing associated with the standardized Washington-Oregon-California Trawl Logbook. During 1991 through 1992, only 4% of the reported groundfish landings in Oregon were caught by fishing gears other than bottom or midwater groundfish trawls (Table 3.3).

The ODFW has two trawl logbook database systems. One is maintained by the Marine Finfish Program and is used to assist in the routine management of the fishery. The other is maintained by the Marine Habitat Program for use with its Geographic Information System (GIS). The GIS and associated database are described in detail in Starr and Saelens (1987), Starr and Saelens (1989), and Fox et al. (1992). Both databases are derived from the same standard trawl logbook data.

3.3.1 Groundfish Trawl

Collection

Logbooks for trawlers from a given port are collected routinely by local port biologists, whenever they are working on the docks and in contact with vessel captains. Sometimes, however, captains mail their logbooks directly to the local ODFW office; the port biologists provide pre-addressed and stamped envelopes for this purpose. Data entry for all Oregon trawl logbooks occurs at the ODFW office in Newport. The entry is done directly onto the mainframe computer in Portland.

Data entry and error checking

The local port biologist examines every logbook for completeness and consistency. The process includes checking the logbook for incorrect temporal sequencing of the tows or inappropriate dates or times, and filling in the following items: 1) the ticket number(s) corresponding to each trip, 2) missing depths based on the tow locations and the depths indicated on the nautical charts, and 3) missing target species based on the most prevalent species hailed.

The port biologist assigns each logbook a code of 1, 2, or 3 depending on its degree of completeness. The criteria used to make this determination are fully described in Wood (1992a). If a logbook has only partial information on the tow locations or hailed weights (the captain's estimates of the weights of the retained catches), the port biologist will attempt to get the missing information from an interview with the captain. If the missing information cannot be obtained, the logbook will be assigned a code 2 if only hailed weights are missing, or a code 3 if tow locations are missing. Logbooks that are assigned a code 3 are excluded from further processing.

The data entry program in Newport is not a "double-entry system"; the data are entered only once. However, invalid dates, ticket numbers, boat numbers, port codes, fishing block codes, and species codes are "trapped" during entry. Also, if tow locations are reported as Loran coordinates, the microsecond values must fall within valid ranges; if tow locations are reported as latitude and longitude, they must fall within generally valid ranges. Tow-specific information is not directly verified, but a program is run routinely that adjusts the hailed weights in the logbooks on the basis of the landings reported on the fish tickets on a trip-specific basis. Hailed weights of market categories that appear on the logbook but not on the ticket are identified. Discrepancies that are 500 lb or greater are investigated and resolved by the port biologist; smaller catches are ignored. For landings of species that appear on the ticket but not on the logbook, the program apportions the landings to all tows (see section 3.6.1).

Typically, the processing of a logbook by the port biologist and the subsequent data entry occur within two months of the corresponding landing date. Final processing of the logbook data does not take place until the end-of-the-year reporting in May. In 1991 and 1992, there were valid trawl logbooks for roughly 83% of the deliveries of groundfish by trawlers (Table 3.6).

3.3.2 Other Gears

In addition to the standardized trawl logbook used coastwide by the states of Washington, Oregon, and California, special logbooks in Oregon are used 1) in the shrimp fishery, 2) by vessels that fish with fixed gear (e.g., longlines and pots), and 3) by vessels that fish with miscellaneous gears (e.g., hook-and-line and vertical longline). Fishers are legally required to complete and turn in a trawl logbook, a shrimp logbook, or a fixed-gear logbook for every fishing trip in which they participate in any of these fisheries, but this regulation is not strictly enforced. As of 1992, completion of the miscellaneous gear logbook was done on a voluntary basis.

3.4 Species-Composition Sampling

Often when fish are landed in Oregon, particularly rockfish, they are not sorted by species, but instead are sold as a mixture. To estimate the species compositions of these landings and in particular, to ascertain accurately the landings of those species that are regulated with annual quotas and trip limits, the port biologists routinely take samples from the rockfish market categories that appear on the fish tickets, namely widow rockfish, yellowtail rockfish, Pacific ocean perch, thornyhead, and the miscellaneous categories "small rockfish" and "large rockfish." All other market categories that appear on the fish tickets are assumed to represent single species, although there are no sampling programs to confirm their purity.

As of 1992, large and small thornyhead were often presorted by the fishers and were sampled separately. In this case, the sample data from these subcategories were combined to provide a single category to match the fish ticket market category.

Beginning in 1991, the ODFW has conducted a small-scale sampling program for groundfish landed by the shrimp fishery in addition to the routine monitoring of rockfish landings made by the groundfish fishery. Typically, the shrimp dealers report the groundfish bycatch simply as "rockfish," with no further specification concerning the appropriate market categories. In 1992 the port biologists were instructed to take approximately four samples per month in each major port. In the early 1980s, there was a similar but more intensive sampling program that addressed the groundfish landings made by the shrimp fishery.

The sampling supervisor and the groundfish technical supervisor coordinate the rockfish species-composition sampling in Oregon. In the late autumn of each year, these supervisors determine the sampling priorities for the upcoming year. The staff associated with the groundfish sampling program in Oregon has monthly or bimonthly meetings to discuss "quality control" issues related to sampling and data processing. Prior to 1992, the groundfish sampling program personnel in Newport also held weekly quality control meetings to discuss sampling and data processing problems.

The port biologists, with assistance from seasonal aides, are responsible for the sampling programs within their individual ports. They prepare monthly summaries of their sampling activities, which the groundfish technical supervisor reviews to identify deficiencies in the level of sampling. The sampling itinerary of a port biologist is largely self-supervised. Since 1989, for each rockfish market category, port biologists have been instructed to obtain 5 boat trip samples per 100 t of the category landed, with the exception of the widow rockfish market category (2 samples per 100 t landed). From 1989 to 1992, port biologists usually achieved and surpassed these recommended sampling rates. The number of species-composition samples collected annually since 1981 is given in Table 3.7.

Since 1989, Oregon has employed a stratified, two-stage random sampling plan. Combinations of port and quarter (a year partitioned into four, 3-month blocks) are treated as strata, and boat trips (primary sampling units) within a stratum are selected at the first stage of the design. Landings are poststratified into market categories, and for each category a number of boat trips are selected based on the recommended sampling rates presented above. Not all of the market categories landed by a boat are necessarily sampled. At least two baskets with fixed weights of fish are subsampled within each market category; the baskets (secondary sampling units) are selected at the second stage of the design. As of 1992, samples were taken on the basis of a fixed weight, rather than a fixed number of fish. However, in years prior to 1989, fixed-number samples were taken.

3.4.1 Sampling Protocol

To assess the species composition of a rockfish market category, a port biologist selects fish from the landed market category and determines the percent composition by weight of the different species contained in the sample. Port biologists arbitrarily select boat trips at the first stage of the design; there are no strictly random procedures involved. They follow the sampling rates presented above, making adjustments based on their own judgement and advice from the sampling supervisor and groundfish technical supervisor.

The sampling rates required at the second stage of the design are as follows: 1) two to four 25-lb baskets are taken for the widow rockfish and thornyhead market categories, 2) two to four 50-lb baskets are selected for the yellowtail rockfish and Pacific ocean perch market categories, (3) four to six 25-lb baskets are chosen for the "small rockfish" market category, and (4) four to six 50-lb baskets are selected for the "large rockfish" market category. Currently, a typical basket of fish contains from 10 to 50 fish, depending on the market category of interest.

At the fish processing facilities, the fish are removed from the hulls of the vessels and placed in "totes," which are bins made of plastic, wire, or wood, containing from 800 to 1700 lb of fish when full. The totes are then either immediately transported to processing rooms (via forklifts, conveyor belts, or vacuums) or placed in a temporary cold storage room within the facility. The port biologists sample the totes while they are en route to the processing rooms, usually as the vessel is being unloaded, or while the totes are in cold storage. Because most ports have several fish processing facilities, each with different methods for handling fish, there is no single dockside sampling technique that can be applied to all of the facilities. The port biologists are instructed to select baskets of fish from totes separated over the entire unloading time of a vessel, e.g., a basket of fish from one of the first totes unloaded and a basket from later in the unloading operation. The individual fish selected for each basket are taken from one corner of a tote, starting at the top and working to the bottom. The port biologists try not to account consciously for sizes or species of fish selected. They record the aggregate weights for each species contained in a basket.

Ultimately, each sample consists of two to six baskets of fish (secondary sampling units) selected from a market category (poststratification unit) within a boat trip (primary sampling unit). Detailed instructions for taking and processing a species-composition sample are given in Wood (1992b).

In commercial fishery sampling programs it is often very difficult to obtain random samples (Tomlinson 1971). Because it is not possible to predict accurately when a boat that has completed a fishing trip will arrive at a particular fish processing facility, there is no way of constructing a sampling frame and thus, no efficient method for randomly choosing a trip. The issue of nonrandom sampling in commercial fisheries, including those in Oregon, is most often dealt with by assuming that boats arrive at a fish processing facility in a random manner and any selection thereof will produce samples that can be treated as random units.

With the exception of the Pacific hake samplers, the ODFW port biologists usually sample during weekdays and daylight hours. Generally they do not sample at night or on weekend days. Except for Pacific hake, deliveries of groundfish that are made at night or on weekends may be available for sampling during standard hours, because the totes sometimes remain at the plants for a day or two before they are processed.

In general, the ports in Oregon are sampled in proportion to the total amount of fish landed there. In 1991 and 1992, the three major ports, where samples are routinely collected, accounted for more than 90% of the total commercial landings of rockfish (Table 3.2). Because of the unpredictable and infrequent nature of the groundfish landings at the smaller ports (e.g., Depoe Bay), very few samples are taken there.

3.4.2 A Hypothetical Example

The following example illustrates the sampling process and some of the choices that the port biologist must make. A boat trip arrives at a fish processing facility with the following catches: 1) 10,000 lb of fish in the widow rockfish market category, which are unloaded into approximately 10 totes, 2) 4,000 lb of fish in the yellowtail rockfish market category, which are unloaded into approximately 4 totes, and 3) 2,000 lb of fish in the "large rockfish" market category, which are unloaded into approximately 2 totes.

A port biologist might select this boat trip if additional samples were needed to meet the recommended sampling rates for any of the market categories being landed (widow rockfish, yellowtail rockfish, and "large rockfish"). That is, any combination of the three categories could be sampled, depending primarily on whether the port biologist was ahead of, or behind schedule in terms of the target sampling rates. Regardless of which market categories were sampled, the port biologists would use the same process for subsampling fish within the categories.

Suppose, for example, a port biologist needed more samples for two of the three market categories, namely the widow rockfish and "large rockfish" market categories. He would obtain two 25-lb baskets of fish (subsamples) from the widow rockfish market category, using the procedures discussed previously (see section 3.4.1). He would attempt to obtain a subsample from different locations of the vessel's hold, e.g., by collecting one sample early during the unloading operation and another towards the end of the operation. If the boat had unloaded its catches before he arrived at the processing facility, then he would select each basket from a different tote; however, because the unloading sequence was unknown, he would arbitrarily choose which totes to sample. He would use a similar protocol to sample the "large rockfish" market category; however, now four to six 50-lb baskets of fish would be selected from two available totes, with two to three baskets being subsampled per tote. If samples are needed and time permits, port biologists in Oregon often sample more than one market category of a boat trip.

3.4.3 Processing System

Data entry and error checking

Entry of all species-composition data occurs at the ODFW office in Newport. The entry is done directly onto the mainframe computer in Portland. Data entry is done using a "double-entry system" in which the operator enters the data once and later re-enters the same data. Discrepancies between the two sets of data are identified and resolved during the second entry. For example, invalid dates and unrecognized port, gear, market category, and species codes are trapped, and entries for ticket numbers must fall within a valid range. The program does not check the landed weights on the species-composition form against the landed weights reported on the corresponding ticket.

Sampling coverage of the fishery

The ODFW port biologists usually meet or exceed their target sampling rates for rockfish species compositions. Based on the sampling rate criterion in 1991 there should have been a total of 635 samples taken, but port biologists actually collected a total of 1,012 samples (Table 3.8A); in 1992 there should have been a total of 703 samples taken, but a total of 1,924 were collected (Table 3.8B). Nevertheless, even these levels of sampling were insufficient to provide certain details about the species compositions of the rockfish landings. For example, in 1992 there were potentially 120 aggregate combinations for which sampling might be required (six market categories within 20 port/quarter strata). However, rockfish were landed in only 114 of these aggregate combinations. Eleven of the combinations received no sampling (e.g., the yellowtail rockfish market category within the Garibaldi/4th quarter port/quarter stratum) and 20 had only one sample collected (e.g., the thornyhead market category within the Newport/3rd quarter port/quarter stratum) (Table 3.8B). However, according to the target sampling rates, 16 (80%) of the aggregate combinations where just one sample was collected actually required one sample or less to be taken. There was a similar pattern to the sampling coverage for 1991 (Table 3.8A). General issues relevant to sampling coverage and intensity are further discussed in chapter 7.

3.5 Biological Sampling

In addition to providing information on the total quantity and species composition of the commercial groundfish landings, the ODFW port biologists routinely collect detailed information on the biological characteristics of some of the main species, particularly those that are regulated or heavily exploited. In Oregon the samples that are collected to provide biological information are referred to as "market samples." Estimates of the age, size, maturity, and sex compositions of the landings are used in the stock assessments prepared by the Groundfish Management Team (GMT) of the PFMC. The age-composition data in particular are vital to catch-at-age assessment methods such as Virtual Population Analysis (Gulland 1965), Cohort Analysis (Pope 1972), and Stock Synthesis (Methot 1990).

The mix of species that is sampled for biological information varies from year to year. For 1992 the ODFW port biologists were instructed to sample the following species: English sole, petrale sole, Dover sole, arrowtooth flounder, canary rockfish, yellowtail rockfish, widow rockfish, darkblotched rockfish, shortspine and longspine thornyhead, lingcod, and Pacific grenadier. Also, in 1992 ODFW staff from the Marine Recreational Fishery Program collected biological sample data for black and blue rockfish landed by the recreational charter boat fishery. The numbers of biological samples taken annually by ODFW staff since 1981 are given in Table 3.9.

One of the major tasks of the biological sampling program is to collect the physical structures that are used for determining the ages of the fish. For many of the commercial groundfish species, age can be measured by counting annual marks (rings) on the otoliths. After the structures are read, they usually are retained in storage. A list of the age structures stored at the ODFW facility in Newport is given in Table 3.10. Some additional otoliths are available from odd specimens collected during groundfish cruises. Also, several hundred Pacific grenadier otoliths collected by the ODFW port biologist in Coos Bay are stored at the Department of Oceanography, Oregon State University, in Corvallis.

Because it requires considerable training and skill to read the age structures from some species, staff from different institutions specialize in particular species. As a consequence, processing of the structures collected by the ODFW port biologists occurs at various sites along the U.S. Pacific coast. For example, structures from widow rockfish and lingcod are sent to the NMFS laboratory in Tiburon, California for processing. Staff at the ODFW facility in Newport specialize in English sole, petrale sole, Dover sole, sablefish, canary rockfish, and black rockfish.

The biological sampling program in Oregon is based on the same general framework as the species-composition sampling program for landings of rockfish (see section 3.4). The sampling supervisor and groundfish technical supervisor coordinate the statewide sampling activities, and each year in the late autumn they decide the sampling priorities for the following year. Any sampling or data processing problems that arise during the season are discussed amongst the staff at monthly quality control meetings.

Because data from the biological samples are used primarily for stock assessments, the decisions regarding which species to sample and appropriate target levels for sampling are based largely on suggestions by the GMT. However, the port biologists, with assistance from the seasonal port samplers, are responsible for the sampling programs within their individual ports. They prepare monthly summaries of their sampling activities, which the groundfish technical supervisor reviews to identify any deficiencies in the level of sampling. Target sampling rates for 1992 are presented in Table 3.11.

3.5.1 Sampling Protocol

Biological sampling consists of selecting a quantity of a particular species from the market category in which the species is most often landed. If a market category is assigned to a single species, then all biological samples for that species are selected from its market category. This sampling design is generally effective in Oregon because the species for which biological samples are required are primarily landed within their own market categories. For example, in 1991, over 95% of the total poundage of yellowtail rockfish was landed within the yellowtail rockfish market category. In 1992, 10 of the 12 species included in the biological sampling program were sampled from their own market categories (Table 3.11). Those species that have not been assigned their own market categories are sampled from the categories in which they are most often landed. For example, in 1992 canary rockfish and darkblotched rockfish, which were not assigned their own market category, were sampled from the "large rockfish" market category and the "small rockfish" market category, respectively.

In general, for each fish in the sample, the port biologists are instructed to measure and record its length, sex, and to collect physical structures for age determination. For some species, the state of maturity (reproductive stage) is also routinely recorded. Port biologists attempt to perform all of these sampling duties whenever possible, but logistical and time constraints occasionally preclude a sampler from conducting all of these tasks on a given sample. The port biologists choose species (market categories) and boat trips to sample based on their own judgment and advice from the sampling supervisor and groundfish technical supervisor.

All biological samples collected from landings of rockfish are obtained while the port biologists conduct species-composition sampling. For example, if port biologists are conducting species-composition sampling from landings designated as widow rockfish and they need more biological information on this species, they may decide to measure the lengths, remove the otoliths, and record the maturity information from the individual fish that compose the two 25-lb basket subsamples. When the port biologists sample those species other than rockfish, they follow the same basic procedures.

The protocol used to select subsamples (baskets of fish) is similar to that used in species-composition sampling, described earlier in section 3.4.1. Depending on the species of interest, two or four, 25-lb or 50-lb baskets of fish are selected from totes from each boat trip sampled. At the fish processing facility the port biologist chooses a tote to sample and then selects individual fish for a basket from one corner of the tote, starting at the top and working down to the bottom, trying not to account consciously for sizes of fish selected. Baskets for most species are based on weight. Because some species do not occur in "pure" market categories (e.g., darkblotched rockfish), the port biologist may have to re-sample the tote and select more of the target species to get a complete basket sample. To maintain the statistical integrity of the two data sets, any additional fish selected to secure a complete biological sample are not incorporated in the species-composition sample. Details regarding sample sizes and port-by-port sampling priorities for 1992 are given in Table 3.11.

Prior to 1991, biological samples for all species except widow rockfish were composed of fixed numbers of fish rather than fixed weights of fish. Biological sampling that incorporated subsamples of a fixed weight began in 1989 for widow rockfish.

For each basket subsample, the fork length of every fish is measured to the nearest centimeter. For most groundfish species, the sex and reproductive stage of each fish are determined by external examination or by internal investigation of the gonads. Males are separated from females, and the total numbers and weights by sex are recorded. During certain years, individual fish are also weighed to establish a length-weight relationship. Age structures (otoliths for most species, fin rays for lingcod, and interopercles for English sole) are removed from the individual fish and placed in distinct containers. Depending on the species of interest, either the left, right, or both otoliths are collected. When sampling is completed at the processing facilities, the collected data and age structures are returned to the laboratory for further processing. The completed data forms and age structures are sent to the Newport office approximately one month after the sample date. Detailed instructions for collecting and processing biological samples are given in Wood (1992b).

Biological samples in Oregon are collected mostly from trawl landings, but in previous years sablefish samples have also been collected from landings by pot and longline fishers. The GMT recommended that the landings of lingcod be sampled in place of sablefish in 1992.

If a dealer refuses to allow a port biologist to remove otoliths or otherwise mutilate a fish to determine its sex or state of maturity, the port biologist will discuss the situation with the dealer and try to arrange some acceptable method for obtaining the information. In any case, the port biologist will collect the length information.

3.5.2 Age Determination Methods

Age determinations for selected species of groundfish have been conducted by ODFW age specialists ("age readers") since the early 1970s. The ODFW has appointed a representative to the Committee of Age Reading Experts (CARE) since 1983, when CARE was first established by the Technical Subcommittee of the Canada/United States Groundfish Committee. Prior to 1992, when ODFW formally established a Fish Ageing Unit at the Newport office and hired a full-time age specialist, routine age readings were primarily performed by staff biologists who were involved in preparing stock assessments for regulated species of groundfish.

Over the years, ODFW personnel have performed age determinations for various species of groundfish including Dover sole, English sole, petrale sole, sablefish, Pacific ocean perch, canary rockfish, black rockfish, yelloweye rockfish, blue rockfish, and shortspine thornyhead. For some of these species, such as Dover sole, scales were the first structures used to determine the age of a fish specimen. However, by 1980, otoliths largely replaced scales as the preferred structures for age reading, with the exception of interopercles, which are currently the structures used to determine the age of English sole specimens. Prior to 1980, ages were determined predominantly using the "thin section method" (by counting annual marks (rings) on otolith sections mounted on microscope slides) and the "surface method" (by counting rings on whole otoliths immersed in an alcohol solution). In 1980, a break and burn technique (Christensen 1964) replaced other methods of age determination for virtually all species analyzed by ODFW personnel. The break and burn procedure has been demonstrated as an effective technique for processing large numbers of otolith specimens for various species of groundfish (Chilton and Beamish 1982) and is currently the most widely employed age reading method used by management agencies responsible for groundfish fisheries off the Pacific coast of the United States and Canada.

No formal validation studies have been conducted to assess the accuracy of age determination techniques for groundfish species analyzed at the Fish Ageing Unit in Newport. However, workshops conducted by the ODFW, WDFW, and Canada DFO have generated several unpublished reports that generally support the break and burn method as a valid technique that can be used to identify annual rings on otoliths of groundfish species.

3.5.3 Processing System

Data entry and error checking

Data entry for all biological sample data occurs at the ODFW office in Newport. The entry is done on a personal computer using a "single-entry" system, with visual inspection of a verification screen. The data entry program does only cursory error checking, such as trapping invalid characters, and doing simple range-checks on the length and age data. The data are not cross-checked against the corresponding fish tickets or logbooks.

Sampling coverage of the fishery

The ODFW port biologists often meet or exceed their target sampling rates, but the level of coverage is more sporadic than the coverage for species-composition sampling. From 1991 through 1992, for example, widow rockfish was the only rockfish species for which the target sampling rates were met consistently (Table 3.12). The targets for yellowtail rockfish were achieved in only 4 of the 12 port and quarter strata during 1991, and in 5 of the strata during 1992. The targets for canary rockfish were achieved in only 3 of the 12 port and quarter strata in 1991, and in 7 of the strata during 1992.

3.6 Estimating Derived Quantities

The fishery data collected by the various ODFW programs are used to estimate quantities that are relevant to the assessment and management of the fish stocks. The quality of the estimates depends both on how the sample data are collected and processed, and on how they are then combined into final estimates. This section documents some of the manipulations and calculations that are involved in producing the final estimates of the groundfish landings and the characteristics of those landings.

3.6.1 Landings by Area

Often fish tickets do not contain complete and accurate information concerning the locations of the catches. This occurs because the port biologists are not always cognizant of where the fleet is operating, and because the boats sometimes fish on several grounds. Although fishing boats in Oregon tend to operate near the port in which they land their catch, the landings by port give only a rough representation of the spatial distribution of the catches. However, detailed information is available from the logbooks.

For those fish tickets for which corresponding logbooks are available, the reported landings are apportioned to the PSMFC area where the tows were made (commonly referred to as the area fished variable). Then, total landings by area fished are estimated using the partial statistics on landings by area fished (from the fish tickets that had corresponding logbooks) along with the reported landings for which no corresponding logbooks are available (Fig. 3.5). All calculations are performed on the central computer system at the Portland office.

The following algorithm illustrates the general procedures used to estimate landings by PSMFC area.

1. Fish tickets are matched with logbooks.
   

2. For those tickets with logbooks, the tow-by-tow hails are adjusted so that for each trip the total hails by market category conform to the landings by market category reported on the corresponding fish tickets.
   

1. When a market category is reported on a fish ticket but is missing from the corresponding logbook, adjusted hails are generated using the total hails by tow of all market categories to prorate the missing landing to individual tows, provided the tow-by-tow hails are reported on the logbook. If the hails are not reported but the tow durations are, then the adjusted hails are based on the tow durations.
   
2. When a market category is reported on a logbook but does not appear on the corresponding fish ticket, an error report is generated and the ODFW staff contact the owner or captain of the vessel to reconcile the discrepancy.
   
3. The adjusted hails from all of the logbooks are tabulated to create a proration table of catch by market category, month, port, gear, and area fished. The proration table consists of percentages that individual PSMFC areas contributed to the total landings of each market category by month, port, and gear.
   

1. The following hypothetical example illustrates the general form of an entry in a proration table for a particular market category. The percentage of the thornyhead market category caught using bottom trawl gear and landed in Newport during March 1992 was 60% from PSMFC area 2B, 35% from PSMFC area 2C, and 5% from PSMFC area 3A.
   

2. This proration table is submitted to PacFIN, where the actual proration of fish tickets to area fished is performed.
   

3. Logbooks collected in a given year may not be processed until as long as six months into the following year and thus, the monthly proration table is often constructed from the available fish ticket information, rather than from the logbook data. As the logbook information becomes available, the monthly proration tables are reconstructed and submitted to PacFIN. The monthly proration tables for a given year generally receive final updates by June of the following year.
   
4. Individual fish ticket records, rather than data summaries, are submitted to PacFIN, where they are tabulated to estimate landings by market category, month, port, and gear.
   

5. At PacFIN, the proration table from (3) is applied to the results from (4) to estimate the total landings by market category, month, port, gear, and area fished.
   

In the trawl logbooks the fishers can report their tow locations either as Loran coordinates, as latitude and longitude, or as a "block" number. If tow locations are reported as Loran coordinates, the coordinates are converted to latitude and longitude values by the systems analyst at the Portland office. Tows are assigned to particular geographic regions of interest, such as PSMFC area, on the basis of the starting positions.

3.6.2 Rockfish Landings by Species

Species-composition sample data are merged with the fish ticket information to estimate rockfish landings by individual species, rather than by reported market categories. The computations are performed on a mainframe computer at the Portland office. Also, ODFW provides monthly estimates to PacFIN of the species compositions (in percent) for the rockfish market categories by port, gear, and PSMFC area fished. The estimated percentages are based on the weight that a species contributed to the total landings of a given market category for a specified port, month, gear, and area fished. For example, the percentage breakdown by species for the yellowtail rockfish market category might be 83% yellowtail rockfish, 10% widow rockfish, 5% redstripe rockfish, and 2% sharpchin rockfish for the landings in Astoria during March 1992 that were caught using bottom trawl gear from PSMFC area 3A.

Species-composition estimates are generated for the following types of gear: 1) bottom trawl gear, which includes roller and sole gears, 2) midwater trawl gear, 3) longline gear, and 4) hook-and-line gear. Because there are rarely enough samples collected in a given month to determine estimates of species composition for individual market categories by port, gear, and area fished, the monthly estimates are based on sample data collected over a series of months (quarters). For example, the estimates of species compositions for March are based on sample data collected from January through March.

Because port biologists sample in an opportunistic fashion, obtaining boat trip samples when and where possible during their daily work schedule, in some instances there may be no sample data available for a particular port or quarter. When this occurs, sample data from adjacent ports or quarters are substituted for the missing values according to the following scheme.

1. If there are no sample data for landings of a particular market category made at a minor port, then comparable samples (with similar quarter, gear, and area fished characteristics) from the nearest major port are used. To perform these reassignment tasks, Astoria, Newport, Coos Bay, and Brookings are defined as the major ports in Oregon.
   

2. If after step (1) there are no suitable sample data, then comparable samples are used from the next closest major port to the south.
   

3. If after step (2) there are no applicable sample data, then samples are used from the preceding quarter for the nearest major port.
   

4. Finally, if after step (3) there are no applicable sample data, then samples are used from the succeeding quarter for the nearest major port.
   
</0l>

The following formulae are used to estimate the weight, and subsequent percent contribution, of a given species in the landings. As stated above, estimates of species composition are calculated separately for market category, gear, and area fished combinations, but for simplicity the subscripts for these variables have been dropped. The estimated mean weight of species y per pound of a market category of interest of boat trip k within a port i and quarter j stratum is calculated as,


(3.1)

where y_ijkl is the weight of species y in basket l, w_ijkl is the weight of basket l, and m_ijk is the number of baskets selected from the sampled market category of boat trip k. Note that y_ijkis equivalent to the proportion by weight that species y contributed to the landings of this market category for this boat trip.


The ratio estimate for the mean weight of species y per pound of a market category of interest across all boat trips within a port i and quarter j stratum is calculated as,


(3.2)

where W_ijk is the total weight of all species in the sampled market category of boat trip k and n_ij is the number of boat trips sampled. Equation 3.2 is a weighted estimator, where samples from boat trips with market category landings that are large are given more weight in estimating y_ij.. than boat trips with relatively small landings. Values for Wijk are determined from the fish ticket information.


The percent of the landings (in weight) that species y constitutes for a market category of interest within a port complex i and quarter j stratum is simply,


(3.3)

The ODFW does not routinely generate variance estimates associated with the estimates above, but the redefined PacFIN system is capable of accepting coefficient of variation estimates, along with the estimates of species composition. Further details of the calculations, including appropriate variance estimators, are given in Sen (1986) and Crone (1992a, 1992b, 1995).


The estimates of species composition for each market category by port, quarter, gear, and area fished are submitted monthly to PacFIN in Seattle. At PacFIN, the estimated percents derived above are applied to aggregated fish ticket information to partition the landings (in pounds) by reported market categories into landings (in pounds) by individual species. Landing estimates generated by PacFIN are routinely revised throughout a year as fish ticket information is updated and species-composition sample data accumulate.


3.6.3 Catch-at-Age


Although age-composition data (catch-at-age) play a central role in almost all current stock assessments, ODFW staff do not routinely publish estimated age or length frequency distributions for the landings of groundfish in Oregon. Instead they respond to specific requests for information by different researchers. Because there is no standard, universally acceptable procedure for estimating catch-at-age or catch-at-length, several methods have been developed for expanding the sample data and for combining different samples. It is left to the researchers using the data to determine which method best suits their requirements.

The age-composition estimates are based on the data collected in the biological sampling program (see section 3.5). The methods used to calculate age compositions of individual species are fundamentally similar to the methods used to derive species compositions (see section 3.6.2). The primary differences are: 1) age-composition estimates require the landings to be partitioned into age groups for a particular species of interest, whereas species-composition sampling requires the landings to be partitioned into the species observed within a market category of interest, and 2) age-composition estimates are based on the number of fish landed, whereas species-composition estimates are based on the weight of fish landed.


Because the market category is still maintained as the poststratification unit in the biological sampling program and because the species incorporated in the program are only sampled from their own respective market categories, estimates of age composition for a particular species reflect the characteristics of the total landings of the corresponding market categories. However, for those species involved in the biological sampling program, the vast majority of the landings occur within their own market categories (i.e., the market categories are "pure") and thus, the final estimates generally reflect the total landings of the species of interest. For the major species of rockfish, a species-composition sample exists for every biological sample; consequently, it is possible to estimate the total landings of the rockfish species of interest within the sampled market category and then use this estimate as a weighting variable in the procedures presented below.


Additionally, although the biological sampling program is based on a two-stage design, the current procedures for estimating age composition ignore the second stage of sampling (baskets of fish), primarily to simplify analysis, and in effect, are based on single samples of fish from each boat trip. Also, because of the relatively small number of boat trips that are sampled, it is generally not practical to calculate age-composition estimates by port, quarter, gear, and area fished. Instead, most researchers combine biological samples across all ports and generate statistics for annual landings.


The formulae used to estimate the age (or length) composition of the landings for individual species are based on one of the following general methods (1-4). The calculations are performed using Pascal programs at the Newport office.


1. For the species of interest, the number of fish of age group y is estimated as,
   
   
   (3.4)
   
   Equation 3.4 generates an estimate of the total number of fish of age y in the total landings of the corresponding market category, W_k is the total weight of the market category within boat trip k, w_kl is the weight of basket l subsampled from the market category within boat trip k, y_kl is the number of fish of age y in basket l subsampled from the market category within boat trip k, n is the number of boat trips sampled, N is the total number of boat trips, and m_k is the number of baskets sampled from the market category within boat trip k. The ratio W_k / w_k is a weighting variable that expands the subsample values (y_k) to total values for each sampled boat trip, and the ratio W / W_k expands the values for the individual boat trips to total values across all trips.
   

2. The second method for estimating age (or length) composition is identical to the first (eq. 3.4) except that the w_k values are not the observed basket weights, but instead are estimated from the observed lengths of the fish in the basket subsamples using appropriate weight-length relationships. The current estimation program has standard weight-length coefficients for widow rockfish, canary rockfish, and Dover sole. The values for the total weight of the baskets subsampled from each sampled boat trip are estimated as,
   
   (3.5)
   
   where L_klf is the length of fish f (f = 1,. . . , F) in basket l subsampled from the corresponding market category of boat trip k, and and ß are coefficients determined empirically using regression techniques.
   
   

3. The third method is essentially the same as the first (eq. 3.4), however, the numbers of fish at age (y_k) are not weighted by the size of the market category landings (W_k / w_k). Thus, the estimator is simply,
   
   (3.6)
   
   
4. The fourth method is identical to the third (eq. 3.6) except that values for w_k are estimated using equation (3.5).
   

If there are large differences between boat trips, both in terms of age (or length) compositions and total landings of the market categories sampled, then there can be considerable differences between the results from (1) versus (3), and from (2) versus (4). Also, if there are changes in the weight-length relationship associated with a particular species, then there can be large differences between the results from (1) versus (2), and from (3) versus (4).


Variance estimates associated with the estimators above are not routinely calculated by ODFW. However, formulae for calculating estimates of age composition and their variances that fully accommodate the two stages of the sampling design are given in Sen (1986) and Crone (1993, 1995).


3.6.4 Other Estimates


In addition to the fish ticket data, the proration tables for estimating landings by area (section 3.6.1), and the proration tables for estimating rockfish landings by species (section 3.6.2), the ODFW data processing staff periodically provide the PacFIN office with lists of licensed commercial fishing vessels, with tables for translating various codes (e.g., market categories, gear types, ports, etc.), and with estimates of the number of hours of trawling (the amount of time that trawl nets were dragged through the water). The estimates of trawl hours by month, port, gear type, and PSMFC area are derived by 1) summing the trawl hours reported in the logbooks, and then 2) multiplying the total logbook hours by the ratio of the total pounds landed from the fish tickets divided by the total adjusted hailed pounds from the logbooks (see section 3.6.1). The expansions are needed to account for landings for which logbook data are unavailable.


3.7 Acknowledgments


We thank Stan Allen, Dave Douglas, Mark Freeman, Gary Hettman, Mike Hosie, Dave Judkins, Jerry Lukas, and Claire Wood for their technical support and helpful criticisms and comments on earlier drafts of the chapter.


3.8 Citations


Chilton, D. E., and R. J. Beamish. 1982. Age determination methods for fishes studied by the groundfish program at the Pacific Biological Station. Can. Spec. Publ. Fish. Aquat. Sci. 60:49-52.


Christensen, J. M. 1964. Burning of otoliths, a technique for age determination of soles and other fish. Journal du Conseil, Conseil International pour l'Exploration de la Mer 29:73-81.


Crone, P. R. 1992a. Sampling design and statistical considerations for the commercial rockfish fishery of Oregon, Part 1 - Species composition. Report to the Oregon Department of Fish and Wildlife, Contract 91-54, 97 p. (Available from Oregon Department of Fish and Wildlife, 2040 SE Marine Science Drive, Newport, OR 97365.)


Crone, P. R. 1992b. Sampling design and statistical considerations for the commercial rockfish fishery of Oregon, Part 1 - Species composition, Report Addendum. Report to the Oregon Department of Fish and Wildlife, Contract 91-54, 50 p. (Available from Oregon Department of Fish and Wildlife, 2040 SE Marine Science Drive, Newport, OR 97365.)


Crone, P. R. 1993. Sampling design and statistical considerations for the commercial groundfish fishery of Oregon, Part 2 - Age composition. Report to the Oregon Department of Fish and Wildlife, Contract 92-43, 40 p. (Available from Oregon Department of Fish and Wildlife, 2040 SE Marine Science Drive, Newport, OR 97365.)


Crone, P. R. 1995. Sampling design and statistical considerations for the commercial groundfish fishery of Oregon. Can. J. Fish. Aquat. Sci. 52:716-732.


Fox, D. S., M. R. Saelens, M. Long, B. Bond, and A. Merems. 1992. Fishery databases: description, documentation and narrative for the State of Oregon. Report to the U.S. Department of the Interior, Contract MMS 92-0016, 94 p. (Available from U.S. Department of the Interior, Minerals Management Service, Pacific Outer Continental Shelf Region, Camarillo, CA 93011.)


Gulland, J. A. 1965. Estimation of mortality rates. Annex to Arctic Fisheries Working Group Report (meeting in Hamburg, Jan. 1965). ICES C.M. Gadoid Fish Comm. Doc. No. 3.

9 p. (mimeo).


Harry, G. 1956. Analysis and history of the Oregon otter-trawl fishery. Ph.D. Thesis, Univ. Washington, Seattle, WA, 328 p.


Johnson, B. A., and J. Lukas. 1976. Groundfish and shrimp data processing system. Unpubl. manuscr., 257 p. (Available from Oregon Department of Fish and Wildlife, 2040 SE Marine Science Drive, Newport, OR 97365.)


Lukas, J., and C. Carter. 1987. Pounds and value of commercially caught fish and shellfish landed in Oregon, 1987. Unpubl. manuscr., 163 p. (Available from Oregon Department of Fish and Wildlife, 2501 SW 1st Avenue, P.O. Box 59, Portland, OR 97207.)

Lukas, J., and C. Carter. 1992. Pounds and value of commercially caught fish and shellfish landed in Oregon, 1992. Unpubl. manuscr., 119 p. (Available from Oregon Department of Fish and Wildlife, 2501 SW 1st Avenue, P.O. Box 59, Portland, OR 97207.)


Methot, R. D. 1990. Synthesis Model: an adaptable framework for analysis of diverse stock assessment data. Int. North Pac. Fish. Comm. Bull. 50:259-277.


Pope, J. G. 1972. An investigation of the accuracy of virtual population analysis using cohort analysis. Int. Comm. Northwest Atl. Fish. Res. Bull. 9:65-74.


Sen, A. R. 1986. Methodological problems in sampling commercial rockfish landings. Fish. Bull., U.S. 84:409-421.


Silverthorne, W. 1992. Economic status of the Washington, Oregon, and California groundfish fisheries. In Pacific Fishery Management Council, Status of the Pacific coast groundfish fishery through 1992 and recommended acceptable biological catches for 1993: Stock assessment and fishery evaluation, Appendix H, 13 p. (Available from Pacific Fishery Management Council, 2130 SW Fifth Avenue, Suite 224, Portland, OR 97201.)


Starr, R. M., and M. R. Saelens. 1987. Identification of important marine habitat. Unpubl. manuscr., 51 p. (Available from Oregon Department of Fish and Wildlife, 2040 SE Marine Science Drive, Newport, OR 97365.)


Starr, R. M., and M. R. Saelens. 1989. Fishery resource mapping. Report to the National Coastal Resources Research and Development Institute, 31 p. (Available from Oregon Department of Fish and Wildlife, 2040 SE Marine Science Drive, Newport, OR 97365.)


Tomlinson, P. K. 1971. Some sampling problems in fishery work. Biometrics 27:631-641.


Westrheim, S. J. 1950. The 1949 soupfin shark fishery of Oregon. Fish Commission of Oregon, Research Briefs 3(1):39-49.


Wood, C. (editor). 1992a. 1992 Oregon groundfish fishticket and logbook processing manual. Unpubl. manuscr., 98 p. (Available from Oregon Department of Fish and Wildlife, 2040 SE Marine Science Drive, Newport, OR 97365.)


Wood, C. (editor). 1992b. 1992 Oregon groundfish sampling and lab workup manual. Unpubl. manuscr., 82 p. (Available from Oregon Department of Fish and Wildlife, 2040 SE Marine Science Drive, Newport, OR 97365.)





Table of Contents