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

4.1 Introduction

California has a long history of commercial fishing, dating back to at least the 1860s. Groundfish became an important component of the commercial fishery soon after the first dragnet was towed along the bottom of San Francisco Bay by two lateen sail boats (paranzella trawl) in 1876. From that time to the present, groundfish have provided an abundant and consistent supply of fish to the California fishing industry. Groundfish landings in California averaged around 10 million lb annually from the mid-1920s through the 1940s. From 1980 to 1992, annual landings averaged nearly 90 million lb.

4.1.1 Geographical Overview

Major ports

From 1991 to 1992, 11 California ports (Fig. 4.1) received annual landings of groundfish that exceeded 1 million lb (Fig. 4.2, A and B). The percentage of the total groundfish catch for each of the major ports in 1992 was as follows: Crescent City (22%), Eureka (21%), Fort Bragg (9%), Morro Bay (9%), Bodega Bay (9%), San Francisco (8%), Princeton (5%), Moss Landing (5%), Avila (4%), Santa Barbara (2%), and Monterey (2%) (Fig. 4.2B). Twenty-six minor ports accounted for a total of 4% of the groundfish landings in 1992.

Major fishing grounds

Fishing for groundfish occurs almost continuously along the California coast from the Oregon border to the Mexico border; however, most groundfish are landed from Morro Bay northward. Trawl-caught groundfish accounted for roughly 77% of the total landings annually from 1991 to 1992 (Fig. 4.3, A and B). Line gear produced the next largest component (about 14%) during this time period, while set-net and trap landings were 5% and 1%, respectively.

In general, trawl gear is not used south of Point Conception; instead, groundfish are predominantly caught using line gear. North of Point Conception, trawl vessels fish at depths of 20-500 fathoms. The use of set nets is prohibited north of Point Reyes, and due to concerns for marine mammals and birds, these vessels are restricted from fishing in many nearshore areas south of Point Reyes. Traps and hook-and-line gear are used from the Oregon border to the Mexico border to catch rockfish and sablefish.

Principal species landed

In the Eureka INPFC (International North Pacific Fisheries Commission) area, the three major groundfish species landed in 1992 were Pacific hake (34%), Dover sole (22%), and thornyhead (15%) (Table 4.1). In the Monterey INPFC area, the three major species landed in 1992 were Dover sole (26%), sablefish (11%), and chilipepper (11%). In the Conception INPFC area, the three major species landed in 1992 were Dover sole (28%), unspecified rockfish (22%), and thornyhead (16%).

Port biologist locations

The marine management functions of the California Department of Fish and Game (CDFG) are accomplished at the local level through Marine Resources Division management units, with personnel stationed at various locations (Fig. 4.4). Each unit encompasses one or two counties and is staffed by one unit manager (associate biologist) and one or two assistant unit managers (biologists). Marine management units are located in Eureka, Fort Bragg, Bodega Bay, Menlo Park, Monterey, Morro Bay, Santa Barbara, Long Beach, and San Diego. The Groundfish Coordination Unit (GCU), located in Menlo Park, serves as a clearing house and repository for groundfish data.

In 1992 the Pacific States Marine Fisheries Commission (PSMFC) administered Pacific Fisheries Information Network (PacFIN) funds to support seven full-time and one part-time fisheries technicians to assist with port sampling duties, and a data technician and an associate programmer analyst to assist with data processing. Currently, the majority of California groundfish sampling activity is performed by PSMFC fisheries technicians. Supplemental sampling is performed by CDFG biologists and scientific aides.

Fisheries technicians ("port samplers") are assigned to major ports or port complexes (Table 4.2); however, some ports are never sampled, including several in the San Francisco Bay area. In 1992 there were no groundfish samples taken in southern California. Almost all samples are collected at processing plants adjacent to the docks where fish are unloaded.

4.1.2 History of Data Collection Systems

In California, systematic groundfish sampling began during the early 1960s. The focus during those years was primarily species of flatfish (Dover, English, and petrale sole), which together represented the major component of the groundfish landings. Additionally, a short-term study was conducted at several ports from 1962 to 1963 to determine the species compositions of the rockfish landings (Nitsos 1965).

Comprehensive sampling of rockfish landings did not begin until 1977, when the Cooperative Rockfish Survey was established. The survey is a fisheries monitoring program operated jointly by the CDFG and the National Marine Fisheries Service (NMFS). This cooperative sampling arrangement has remained virtually unchanged since its inception nearly two decades ago. However, major restructuring of the program occurred in 1992 in an effort to strengthen the collaborative role of the two agencies and to improve sampling statistics and program documentation.

The CDFG Landing Receipt System, which was the mandated system for recording commercial landings (see section 4.2), was started in 1917 and by 1930, "punch cards" and IBM tabulating machines were used to summarize the data. The current system uses a DEC/Alpha server located in Sacramento, with data entry performed by the Marine Fisheries Statistical Unit (MFSU) in Long Beach.

The trawl logbook, which is the mandated system for recording the catch locations, date, and amount of commercial catches, is the oldest CDFG logbook system. Trawl logbooks were introduced in 1933, as part of the official statistical system of the CDFG. At that time, the trawl logbook and landing receipt were combined on one form. This worked satisfactorily because wholesale houses owned the boats and gear and operated the fleet with paid crews. During the 1930s the dominant fishing method was sail-powered pair-trawlers towing paranzella nets. By 1944 the dominant fishing method was otter-board trawls towed by a single diesel-powered vessel, not necessarily owned by the company receiving the fish. This prompted the CDFG to design a new logbook, independent of the Landing Receipt System, to meet the needs of the developing fishery.

Through 1975 the logbook data were verified and edited by biologists, collated with corresponding landing receipts, and then forwarded monthly to the MFSU. The data were keypunched and summary reports were generated annually by April or May of the following year. With the need for greater resolution of catch by area data, the use of a computer system became necessary. A program developed in cooperation with NORFISH, a University of Washington Sea Grant program, was used to process the 1976-77 logbook data. In 1976, the enactment of the Magnuson Fishery Conservation and Management Act resulted in the formation of the Pacific Fishery Management Council (PFMC), and the formulation of the Pacific Coast Groundfish Fishery Management Plan (FMP). A provision of the FMP was to develop a uniform, coastwide trawl logbook. The PFMC provided a contract to the CDFG to develop a computer system that addressed the new logbook format (Goodrich 1988). This system has been used to process logbook data from 1978 to the present.

Documentation

This chapter represents an overview of the California groundfish sample collection and data processing systems. Other documents provide important additional information on subsets of the system and a historical perspective. The California Department of Fish and Game Operations Manual, sections 4370 through 4372, contains information on the authority for development of groundfish management, establishes and defines the position of the groundfish coordinator (Menlo Park), and reviews data collection procedures.

Documentation of sampling protocols are given in the California Cooperative Commercial Groundfish Survey Sampling Manual (GCU 1995) for the collection of rockfish and other biological samples. Sen (1984, 1986) provided an overview and statistical evaluation of the Cooperative Rockfish Survey as it existed in the early 1980s. Nitsos (1965) provided information on a sampling method that was used during the early 1960s.

The existing Landing Receipt System and a proposed, but unimplemented revision are detailed by Ernst and Young (1989) as part of their feasibility study report for improving the Marine Fisheries Statistical System. This report contains detailed information of the actual programs used in processing the data. The methods used to collect and compile the fishery-related data associated with the Marine Fisheries Statistical System for the years from 1936 to 1952 are presented in Conner et al. (1952).

4.1.3 Legal Authority to Collect Data

Obligations of fishers

California Fish and Game Code statute sections 8010, 8016, 8043, 8046, and 8047 grant authority to CDFG personnel to collect commercial fish landing information and requires fish receivers to maintain copies of landing receipts for four years. California Fish and Game Commission Division 1 (Title 14), and California Code of Regulations sections 176 (trawl fishing activity records), 174f (permit to use gill nets or trammel nets for commercial purposes), 180f (traps), and 120 (prawn and shrimp trawling) grant authority to CDFG staff to collect logbooks. Information collected by the CDFG is generally governed by the California Fish and Game Code section 8043 and the Information Practices Act as stated in the California Civil Code section 1798 et seq.

Noncompliance

Failure to comply with the above listed statutes results in the following penalties as listed in section 12002 (Punishment) of the California Fish and Game Code. "Unless otherwise provided, the punishment for a violation of this code, which is a misdemeanor, is a fine of not more than one thousand dollars ($1,000), imprisonment in the county jail for not more than six months, or both the fine and imprisonment."

4.2 Fish Ticket/Landing Receipt System

In California, official landing receipts (fish tickets) must be completed for all fish or shellfish purchased or received by commercial fish dealers (persons licensed as a fish receiver or multi-function fish business). Landing receipts are required for the following transactions: 1) upon receipt or purchase of fish from a commercial fisher not licensed as a fish receiver, and 2) upon receipt or purchase of fish from a commercial fisher licensed as a fish receiver, when the commercial fisher has not previously made out a landing receipt or if an original receipt has been voided. The principal purpose for requiring the landing information is to "gather and prepare data of the commercial fisheries, showing particularly the extent of the fisheries and the extent to which the various species abound" (California Fish and Game Code statute section 8010).

There are currently 14 types of landing receipts, of which 12 include categories for groundfish transactions. Examples of landing receipts with groundfish categories are presented in Figures A-3 and A-4. The fish dealer determines which landing receipt is used based on the fishing gear used and the market categories that compose the landing. The fish dealer is legally obligated to include the date, market category, landing weight, price, port, gear, area fished, vessel registration number and name, dealer number and name, and fisher license number and name on each landing receipt.

Reports that include information regarding catch by area fished from landing receipt data are usually based on the port where the fish were landed. There is a field on the landing receipt that specifies area caught (e.g., PSMFC area); however, in the past the accuracy of this information has been questioned. In recent years, editing at the MFSU has resulted in more accurate recording of area fished information.

California levies a tax on the value of fish landings, but the Landing Receipt System is not used directly for assessing the tax. Instead, taxation is based on monthly processor reports that are the responsibility of the Compliance and External Audits Branch (CEAB), which ensures that commercial fish businesses are remitting appropriate taxes to CDFG. Tax reports and landing receipt data are compared only during audits.

4.2.1 Market Categories

The market categories listed on the landing receipts (e.g., Figs. A-3 and A-4) represent individual species or groups of species. Market categories are defined by different sizes of fish, price, or mandated by federal/state regulations. There are currently 94 groundfish market categories officially recognized by CDFG, including 47 nominally single species rockfish market categories and 10 multi-species rockfish market categories.

The CDFG Technical Services Branch (TSB) in Sacramento is responsible for the assignment of three-digit market category codes. If a unit biologist, a marine supervisor, or the groundfish coordinator determines that a market category needs to be added or deleted, they must send a formal written request to TSB.

4.2.2 Processing System

Landing receipts are issued to the fish dealers by CDFG. Dealers are legally obligated to mail completed landing receipts to CDFG on or before the first and the 16th of each month, depending on the date of the landing transaction. The Department provides stamped and addressed envelopes for mailing landing receipts. All landing receipts were modified in 1993 to accommodate Optical Character Reader (OCR) technology that allows automated entry of landing receipt information. A flowchart that describes the overall processing system for landing receipts is presented in Figure 4.5.

Currently, all CDFG marine units edit landing receipts and then forward them to the MFSU in Long Beach. About 75% of the landing receipts have errors, with missing vessel registration numbers and invalid market category names being the most common. Other errors include missing or incorrect block (catch area), prices, gear codes, and dealer numbers. Because the unit biologists are generally familiar with vessels and their landings, they are able to identify and correct many of these errors.

At the MFSU, the landing receipts are processed using the OCR scanning technology. Considerable editing occurs at the OCR edit stations, including validation for missing fields, reference checks against lists of dealers, fishers, and vessel registration numbers and names. Additionally, a program checks to ensure catch weights and prices are within a specified range of values. Files created in the OCR process are stored and transmitted nightly via telephone links to the Alpha server in Sacramento. An extensive data validation process on the Alpha server is used to detect missing or invalid fields, including those not checked in the OCR process. A transaction report is created for updates and corrections. Corrections are processed manually using personal computer work stations.

A preliminary report of the catch by month and port for a calendar year is issued about six months after the end of the year. A final report (referred to as the Bulletin Tables) is printed and bound as a public document for release in October or November. The value of the catch is also included in the final report. If there are less than three dealers who buy fish in a given port, then the landings are reported within a larger port group to ensure confidentiality. The 1992 Bulletin Tables were the 52nd in a series that was first published in 1929 (CDFG 1992).

4.2.3 Groundfish Landings Not Covered by the Landing Receipt System

Some landings of groundfish in California are not accounted for in the current Landing Receipt System. Groundfish caught by the recreational fishery are not included in any CDFG data system at this time. Estimates of recreational catches are given in Marine Recreational Fishery Statistics Surveys prepared by the NMFS (e.g., USDOC 1992). The sampling procedures and estimation methods used to generate landing statistics for the marine recreational fishery in California are currently under re-evaluation. Estimates of recreational groundfish landings in 1989 were approximately 6,045 t, which was roughly 17% of the commercial groundfish catch for that year.

Numbers of groundfish caught by Commercial Passenger Fishing Vessels (CPFV) are recorded on logbooks and entered into a separate CPFV system. The CPFV system is maintained independently and is not cross-referenced with the commercial fishery Landing Receipt System.

Transportation receipts are used by fishers when the catch is landed and moved to another site before being sold to a fish dealer. Transportation receipts are not currently entered into the data system, but are maintained in a permanent file system by the MFSU. Sometimes dealers fail to fill out a landing receipt when a transportation receipt has been filled out. These unrecorded landings are most likely minor relative to the total commercial landings represented by landing receipts; however, in some ports they can be significant. For example, in the San Francisco area unreported hook-and-line landings could be as high as 30% of the total reported commercial landings.

4.3 Logbook System

For the trawl fishery, CDFG uses the standardized Washington-Oregon-California Trawl Logbook. Trawl and CPFV logbooks are the only logbooks that are keypunched by the MFSU and included in the marine fisheries system on the Alpha server in Sacramento.

4.3.1 Groundfish Trawl

Collection

Typically, trawl logs are collected by the port samplers once a month or the vessel operators mail the logs to the nearest CDFG marine management unit. Vessel operators are required by law to file their logs for the preceding month by the 10th of every month.

Data entry and error checking

A flowchart that describes the overall processing system for logbooks is presented in Figure 4.5. Trawl vessel operators are legally required to complete a log for each trip. Logs are first edited by port samplers or unit biologists to assign each tow to a single 10 square-mile geographic "block" as defined by the CDFG block system that was established in 1933. The charting of the coastal waters of the state into fishing area blocks was completed concurrently with the designing of the first trawl logbook (Clark 1935). Assignment to a block is based on the Loran coordinates and depth at which a tow was made. Only the block data are keypunched; Loran coordinates and latitude and longitude are not currently captured.

After editing is complete, logbooks are accumulated by the GCU in Menlo Park, then mailed to the MFSU in Long Beach for data entry. Logbook data are keypunched twice in Long Beach and compared for verification. As of June 1995, annual processing of logbook data has been completed for 1978­92.

After all logbooks for one year have been keypunched and compared, final error checking is conducted using a program within the trawl logbook system. This program identifies fatal errors, such as invalid or missing trip return dates, block numbers, or vessel registration numbers, then generates an error report that is subsequently sent to the GCU for review. A manual search through the trawl logs is made to locate the source of the errors and to correct them. These corrections are sent back to the TSB, where they are keypunched into the existing database.

After the logbook database has received its final corrections, the TSB in Sacramento submits the data to an annual processing routine that matches logs with landing receipts. The market categories that have been matched successfully are then converted to one of 27 trawl market categories in the logbook database.

The logs and trawl landing receipts are matched by vessel registration number and trip return date. Because landing receipts are sometimes completed after the date of the landing, unmatched logs and unmatched tickets are reprocessed by subtracting one day from the landing receipt date and rerunning the matching process. After five processing cycles, all unmatched logs are deleted from the system and unmatched landing receipts are converted to "dummy" logs for compliance tracking and effort estimations. In addition, fishing trip summary records are created to be used primarily in a hail adjustment program.

For a given fishing trip, discrepancies between the logbook data and the associated landing receipt are addressed using a hail adjustment program that prorates the poundage from the receipt among the individual tows from the log. If the market category pounds listed on the landing receipt are greater than those on the corresponding trip summary record, the surplus pounds are distributed among trawl tows where that market category occurs. If market category pounds on the landing receipt are less than those on the trip summary record, the missing pounds are subtracted from trawl tows where that market category occurs. Market category landings found on a receipt but not in a logbook are distributed equally among trawl tows for that fishing trip. Species found in the logbooks but not present on the landing receipt are deleted.

The final logbook data file consists of adjusted logbook records (record "type 1") and dummy logbook records (record "type 2"). Reports are created by month, INPFC area, and block number. The logbook data file is downloaded to a Bernoulli cartridge and sent to the GCU for distribution. The block data from the trawl logbook database have never been merged with the landing receipt database to determine specific areas where catches were made. At this time, California does not analyze fishing effort based on the trawl logbook data.

Logbook coverage of the groundfish trawl fishery by port and quarter in 1992 is presented in Table 4.3. In general, the major ports, which are associated with relatively large amounts of landed groundfish (e.g., Crescent City and Eureka), were characterized by high logbook return rates, where at least 80% of the landing receipts had a corresponding log. The minor ports (e.g., San Pedro and Long Beach) were characterized by considerably lower logbook return rates than the major ports, where usually less than 10% of the landing receipts had a corresponding log. Approximately 70% of all trawl landing receipts processed in California in 1992 also had corresponding trawl logs.

4.3.2 Other Gears

Other CDFG groundfish logbooks include gill net, daily trap, daily sablefish trap, and Commercial Passenger Fishing Vessels (CPFV). In most cases, only limited information from these logbooks is entered into computers and can be retrieved electronically. Sablefish logbooks are forwarded to and edited by the GCU. The CPFV logbooks are mailed to Long Beach and are entered as CPFV landings. Although the need for a hook-and-line logbook has been recognized in recent years, current financial constraints have prevented its implementation.

Two other logbooks contain groundfish information. There is a shrimp logbook that includes a field for the amount of groundfish caught by the vessel; however, no formal analyses have been conducted to determine the extent to which this field is used. Regardless, data from these logbooks are not entered into the mainframe computer system, and thus, cannot be merged with either the trawl logbook information or the landing receipts. There is also a salmon logbook that includes a field for the vessel's rockfish bycatch; completion of this logbook is currently voluntary.

Future plans for processing commercial fishery data include an automated method for tracking logbook information and the use of OCR equipment for scanning logbooks. Data captured from each type of logbook may vary to some extent; however, common data can be standardized and criteria for matching these data with the landing receipts can be more clearly defined than the current evaluation methods. Plans to modify the processing system for logbooks have been developed to provide CDFG biologists with more detailed information than is currently available to assess the impact of commercial fishing on various fisheries.

4.4 Species-Composition Sampling

California's commercial groundfish sampling program, which includes species-composition sampling as well as biological sampling, is coordinated at the state level by the groundfish coordinator in Menlo Park and at the local level by marine unit biologists. Port samplers, stationed at various ports (Fig. 4.4), sample only rockfish market categories to determine the species compositions of the landed catch. Historically, most rockfish landings were caught using trawl gear in northern California (Morro Bay and north). Thus, the sampling program was largely designed to cover that gear and area. Only limited sampling has been conducted for the line (hook-and-line and longline) and set-net (set gill and trammel nets) fisheries throughout the state.

When the current sampling program began in 1977, the goal was to obtain samples from every market category contained within a sampled boat trip (Tomlinson 1977). However, this sample selection procedure proved to be impractical, given the logistical and time constraints associated with collecting information at the fish processing locales (sample sites). Sen (1984, 1986) demonstrated that species and age compositions could be estimated by treating the market categories separately and then summing the estimates over the market categories. The current sampling program does not require all of the market categories within a given boat trip to be sampled, and thus, there are generally no estimates produced that address the statistical properties of the individual boat trips.

The species-composition sampling program used in Oregon is also largely based on the design proposed by Sen (1984, 1986) and the following discussion is generally similar to the analogous section presented in the Oregon chapter (section 3.4). However, differences do exist between the two sampling approaches, dictated primarily by the unique fishery operations in each state; therefore, the sections are presented independently of one another.

California's species-composition sampling is based on a stratified, two-stage random sampling design. Port complex and quarter combinations are treated as strata, and boat trips within a stratum represent the first-stage sampling units. The boat trip landings are poststratified into market categories. Typically, two "clusters" (baskets of fish) of a fixed weight are subsampled within each market category at the second stage of sampling.

The current species-composition sampling program for rockfish is designed around the market category as the "domain of study" (Cochran 1977). Thus, the primary goal of the sampling design is to obtain reasonable estimates of the species compositions of the important market categories for a given port complex and quarter stratum. A port complex is a grouping of ports sampled by a CDFG unit, where each CDFG marine unit is responsible for sampling the major port for the area and other nearby minor ports.

There are 57 rockfish market categories that are currently defined, and each of these categories could potentially be landed by each gear type (trawl, line, set net, etc.). There are 47 nominal rockfish species market categories and 10 combined species market categories. Not all of the possible combinations of market category and gear type are routinely sampled; a number have never been sampled, and some have never been used.

Landings in 1992 for southern and northern California by gear type and market category show that relatively few market categories contributed to the total landings for each of the major types of gear (Table 4.4). For example, for northern California ports, approximately 73% (in weight) of the rockfish caught with trawl gears was landed within the thornyhead and unspecified rockfish market categories. Another roughly 18% of the rockfish catch taken by trawl gears and landed at northern California ports was associated with three other categories, namely the widow rockfish, chilipepper, and bocaccio market categories. The line fishery was also sampled fairly intensively in 1992. The line fishery of northern California used the most (28) market categories for landing rockfish catches. In 1992, no sampling for purposes of species-composition evaluations was conducted at southern California ports or for fisheries in northern California that employed gears other than trawls, set nets, or lines.

In annual analyses and reports, the minor market categories (i.e., categories that receive relatively few rockfish landings and usually no sampling) are typically combined with major market categories that have been sampled. Through 1992, the target sampling rates were set at four samples per major market category per month, but this goal was rarely met. Sampling effort is currently determined by a priority scheme that distributes the total time allocated to sampling among the major rockfish market categories for the trawl, set-net, and line fisheries (see section 4.4.2 for an example of a priority list). This sample allocation scheme is applied to each of the port complexes. The Pacific hake, Dover sole, lingcod, and sablefish fisheries are also subjected to a sampling schedule that is based on a priority list.

Among the rockfish, no market category is assumed to be completely "pure" (composed entirely of a single species). However, several market categories are assumed to consist largely of a single or small number of species, and for this reason they are analyzed somewhat differently than the other, "less pure" categories. For example, the widow rockfish market category is generally thought to be nearly pure. The thornyhead market category is thought to consist almost solely of the two species of thornyhead, longspine and shortspine thornyhead. The longspine and shortspine thornyhead market categories were added to the CDFG market category list in 1994. In January 1995, these two market categories were required to be landed as pure categories. In January 1991, bocaccio were required to be landed as a pure market category. When samples are available for individual market categories, including the pure categories, final landing estimates are based on the estimates of species composition determined from the samples and not on an assumption that the landings are pure.

4.4.1 Sampling Protocol

To estimate the species compositions of the landings, a number of boat trip samples are selected for each market category (domain of study). Port samplers choose particular trips to sample based on their own judgement, including the vessel's arrival time, information received from a dealer, or scheduling conflicts. Difficulties associated with obtaining strictly "random" samples from commercial fisheries are addressed in section 3.4.1.

Except in special circumstances described below, samples consist of clusters taken from a given market category contained within a boat trip. Clusters are subsamples (baskets of fish) that are a fixed weight, rather than a fixed number of fish. Generally two clusters are taken per sample, although occasionally samples consist of one or, more rarely, three clusters. Clusters are usually 50 ± 5 lb in size, but 25 ± 2 lb clusters are selected for some market categories that contain small fish, such as the thornyhead market category. In most cases, all of the clusters selected from a given market category have the same target size. However, samples from the nearshore hook-and-line fishery often consist of relatively small clusters (less than 25 lb in size), because total landing weights are frequently less than twice the target cluster size.

Clusters can be taken from the fillet tables, bins, or directly from boxes as they are unloaded from the vessel. The actual sampling site depends largely on where the fish are located in the landing or processing stage when the sample is taken. In general, two clusters are taken from different portions of the market category catch, in order to obtain a more representative description of the species composition. For example, if some of the landings were in a bin, and another portion was on a fillet table, one cluster would be selected from each area. If the entire catch was unloaded into a single bin, generally one cluster would be taken from the center of the bin and another selected from a corner. If more than one bin were available, then the two clusters generally would be taken from different bins.

In some cases, certain fish are selected from a landed market category and placed into groups differing in composition ("subsorts") before a sample can be obtained. Generally subsorts are based on a single species or a certain size range of a single species. In these cases, the sampler is expected to collect a sample from each subsort of the market category. When a market category consists of subsorts, time constraints often preclude a sampler from obtaining more than a single cluster per subsort.

All fish in each cluster are counted and identified to the species level. A combined weight is obtained for each species (all individuals pooled) within a cluster. When possible, the sex and length are recorded for every fish in each cluster.

A port sampling manual is available that describes in detail the selection procedures used by samplers to obtain species-composition samples from rockfish landings in California (GCU 1995). Newly employed samplers receive on-the-job training from experienced port samplers and unit biologists. Currently, all port samplers record information on standardized data sheets; however, historically, this was not the case. In prior years, the sampling data sheets were not uniform coastwide and some samplers used otolith storage (coin) envelopes to record information, while others used data sheets that were unique to their individual ports. Sampling is usually conducted during daylight hours on weekdays (Monday through Friday); however, sampling occasionally takes place at night and on weekend days.

4.4.2 A Hypothetical Example

In 1993, a new sample allocation plan was implemented that redefined the sampling tasks of the port samplers. The plan encompassed the species-composition sampling program for rockfish, as well as the sablefish, Dover sole, and lingcod sampling programs. Prior to 1993, these four programs had sampling goals that were independently established, and, in effect, competed with one another for a sampler's time. This proved to be a generally unsuccessful sampling approach because one or more of the programs usually received inadequate sampling attention. The goals of the new plan were to: 1) effectively apportion sampling effort among the sablefish, Dover sole, and lingcod programs, and among market categories for rockfish, and 2) ensure that the sampling effort within each of the programs was distributed over an extended period of time so that sample information would be available throughout the fishing season.

The new plan established sampling itineraries based on a "priority list." That is, a port sampler is required to first collect samples at the top of the list (high priority), completing each sampling task before proceeding to other tasks at the bottom of the list (low priority). At the beginning of a new quarter, the port sampler starts from the beginning of the priority list again. Currently, the priority lists reflect the specific characteristics of the individual port complexes. The priority lists are constructed once a year at the Annual Cooperative Survey Meeting. The sampling goals are determined by analyzing the landings and sampling effort of the previous year, and evaluating how well the sampler met the goals outlined in the priority list of the previous year. The priority lists also reflect any current directives established by the Groundfish Management Team (GMT) to meet the needs of future stock assessments, NMFS statisticians and biologists, the groundfish coordinator, and other GCU staff. The GCU has the final decision and responsibility for informing the port samplers of the new priority list.

The variability in fishery operations between ports plays a significant role in what samples are taken and how the priority list is used. The following priority list illustrates the approach used in California to select commercial fishery samples. This list, in descending order of importance, is applicable to a port complex and quarter (samples denote individual boat trips). Age structures (fin rays from lingcod and otoliths from other species) are generally collected from selected species for up to three cycles through the list during a quarter.

1. Trawl: unspecified rockfish market category - 3 samples.

2. Trawl: widow rockfish market category - 1 sample.

3. Trawl: bocaccio market category - 1 sample.

4. Trawl: red rockfish (group) market category - 1 sample.

5. Trawl: sablefish market category - 1 sample.

6. Trawl: Dover sole market category - 1 sample.

7. Trawl: lingcod market category - 1 sample.

8. Trawl: chilipepper market category - 1 sample.

9. Trawl: thornyhead market category - 1 sample.

10. Line: unspecified rockfish market category - 2 samples.

11. Set net: unspecified rockfish market category - 2 samples.

12. Line: gopher rockfish (group) market category - 1 sample.

13. Set net: small rockfish (group) market category - 1 sample.

The following example demonstrates how a port sampler might use the priority list to decide which market categories to sample.

A trawl boat arrives at a dealer and unloads the following market categories: unspecified rockfish, bocaccio, lingcod, and chilipepper. Following the priority list presented above, the sampler first obtains a sample from the unspecified rockfish market category. The sampler selects a 50 ± 5 lb cluster from one of the bins that the unspecified rockfish catch has been unloaded into using protocols discussed in section 4.4.1. The sampler then sorts the cluster into individual species and records their aggregate weights. Length, sex, and stage of maturity are recorded for each fish in the subsample (cluster). Otoliths are removed from those species that are included in a biological sampling program for rockfish (see section 4.5). A second cluster is collected from a different bin of unspecified rockfish, if possible, and processed in the same manner as the first cluster. Sample information collected in the field is written on standardized data sheets, or in the case where length frequency boards are used, the data are transcribed to the data sheets from the boards after the sampler has returned to the office.

If time permits, the sampler will next sample the bocaccio market category (number 3 on the priority list). That is, two similar-sized clusters are selected from the bocaccio market category using the same sampling protocols that are discussed above for the unspecified rockfish market category. Because bocaccio is currently included in the biological sampling program, the otoliths from each sampled specimen are collected in addition to the length, sex, and maturity information that is routinely recorded.

Given that the sampler still has adequate time to collect samples, a third market category from this boat trip would be sampled. Lingcod (number 7 on the priority list) are currently included in a federally coordinated sampling program that involves the cooperation of all three Pacific coast states (Washington, Oregon, and California). The federal sampling program for lingcod is discussed further in section 5.2. The sampling program for lingcod is generally similar to the stratified, two-stage designs used in California and Oregon to obtain information regarding the species compositions of rockfish landings. Given that at least 200 lb of lingcod were landed, the sampler selects four 50 ± 5 lb clusters. The sampler is instructed to obtain the four total clusters from as many different bins as possible. The length and sex of each fish in the sampled clusters are recorded and dorsal fin rays from the specimens are collected.

Again, if time permits, the chilipepper market category (number 8 on the priority list) will be sampled; however, fish operations currently used by dealers in some areas do not typically provide a port sampler enough time to sample more than three market categories per boat trip. The chilipepper market category is sampled in the same manner as the unspecified rockfish and bocaccio market categories.

4.4.3 Processing System

Because age-composition data (see section 4.5) and species-composition data are routinely processed together, the discussion presented in this section describes the processing system for both types of information. Species-composition data, including fish length information, are entered directly into personal computers by port samplers at marine units. Prior to 1992, data entry was done using a FORTRAN program that directly stored data in a binary form. From 1992 to 1993, the data entry component of this program was replaced by customized database software that was more "user friendly" than the original program. Beginning in 1994, the entire FORTRAN program was rewritten in QuickBASIC. The format used in the final end-of-year data file has remained constant since 1977 to maintain a consistent file structure for the species-composition data that are collected annually. Every month port samplers send the data files for the current quarter to the GCU. Data sheets are collected by the GCU for permanent storage.

When age data become available, they are appended to these data files. Personnel at the NMFS (Southwest Fisheries Science Center, Tiburon Laboratory) enter all data associated with ageing studies conducted at their laboratory. Age data generated from other studies are entered by personnel from the GCU. Data entry of all length and age information for Dover sole is done by the Dover sole age reader located in Menlo Park.

The analysis procedures for summarizing species- and age-composition information and for generating landing statistics require a master data set. Because age data are entered at two locations, there are often two different data files, one containing ages entered by the NMFS staff and another that was prepared by the GCU personnel. Annually, a routine in the data entry program is used to merge the two data files of age information together. The merged file is referred to as the "master" data set of age information and is stored at Menlo Park and the Tiburon Laboratory. The current processing system for species and age-composition data reflects a design that provides information in a format(s) that can be utilized by a diverse group of fishery researchers and managers.

Error checking of data occurs at many levels. Port samplers are responsible for checking their data entries for accuracy. The current version of the data entry program checks for a number of common errors during the entry process, such as species codes that are not within a valid range, invalid species lengths, invalid market categories, and characters that are entered in fields where numbers are required. Any errors that are identified by the program must be corrected before data entry can continue. In addition, the sample data are extensively checked for missing values and other errors by the GCU. For the species-composition sample data that are sent to PacFIN, there is a BASIC program on the Alpha server that checks for errors, such as blank fields, duplicate records, and invalid species codes.

4.5 Biological Sampling

In general, the biological information collected from California groundfish landings consists of sex, size (length), age, and maturity data. The sampling design used to collect biological data from landings of rockfish and Dover sole is an extension of that used to collect rockfish species-composition samples (see section 4.4). Biological sampling for sablefish, lingcod, and Pacific hake is coordinated through a federal program and is described in chapter 5.

It is important to note that biological sampling requires considerably more time to administer in the field than species-composition sampling, and can result in samples with incomplete information. For example, time- and dealer-imposed constraints often preclude samplers from performing the following tasks for each biological sample: 1) determining the sex and stage of maturity of each fish in the sample, 2) measuring the length of each fish in the sample, and 3) collecting the age structures from each fish in the sample.

4.5.1 Sampling Protocol

Rockfish

Rockfish species-composition sampling includes collecting sex and size (length) information for all species when possible. Maturity and age data are currently collected for bank rockfish, bocaccio, chilipepper, darkblotched rockfish, widow rockfish, and yellowtail rockfish. Only species composition and size data can be collected when fish are to be sold whole and a dealer will not allow them to be mutilated for purposes of extracting otoliths and determining sex and stage of maturity.

Currently, biological sampling rates for rockfish are determined for each market category (i.e., for those species presented above) within individual port complex and quarter combinations. Examples of biological sampling rates are included in the priority list presented in section 4.4.2. The number of age structures that have been collected from targeted species of rockfish in California since 1955 are presented in Table 4.5. Prior to 1993, otoliths were collected whenever a sampler had adequate time to remove the structures while conducting a species-composition sample. However, beginning in 1993, the number of samples in which otoliths were removed from the sampled fish was reduced so that the number of species-composition samples could be increased.

Port samplers clean the otoliths and store them in coin envelopes. Once a month, the otoliths are shipped to the GCU in Menlo Park. A technician sorts the otoliths and sends the yellowtail and widow rockfish samples to the NMFS (Tiburon Laboratory) for age analysis. Otoliths from chilipepper and bocaccio are analyzed by the GCU. After age data have been recorded for the otoliths, the structures are permanently stored at the facility where they were analyzed. No age analysis has been conducted for otolith specimens from bank and darkblotched rockfish, and these age structures are stored at the Tiburon Laboratory. Over the years, the commercial and recreational fishery sampling programs conducted in California have generated a large number of otoliths of various species, many of which have not yet been examined; these otoliths are stored at Menlo Park and at the Tiburon Laboratory.

Prior to 1984, surface ages were recorded from whole otoliths of widow rockfish; however, this method proved to be unreliable for older fish and it was replaced by a break and burn technique that is currently used to prepare otoliths from widow rockfish for ageing. Analysis procedures involve the enumeration of annual marks. Starting in 1992, only partial samples (one cluster per sample) of widow rockfish were processed for purposes of age determination. All otolith samples from widow rockfish have been analyzed through 1994 (Table 4.5).

Otoliths from yellowtail rockfish are also prepared and examined using the break and burn technique. Prior to 1983, surface ages were recorded from whole otoliths of yellowtail rockfish. Ages have been determined for all otolith specimens from yellowtail rockfish through 1994 (Table 4.5).

There is a backlog of otolith specimens from bocaccio and chilipepper to be analyzed due to staffing limitations in recent years. However, an age reader was hired in 1993 to work specifically on these reserves of unexamined otoliths. Ages have been determined for all chilipepper otolith samples through 1990; samples taken since 1985 have been aged using every other fish in a sample. Otoliths from chilipepper sampled after 1982 have been prepared and analyzed using the break and burn method. Prior to 1982, only surface ages were recorded from whole otoliths of chilipepper. Ages have not been determined for bocaccio otolith specimens collected after 1985. All otoliths from bocaccio have been examined whole; however, the validity of the results from some of these analyses has been questioned by fishery researchers, in particular the analysis conducted in 1985 (Bence and Hightower 1990).

Dover sole

Beginning in 1990, the only species of flatfish to be routinely sampled for purposes of age determination was Dover sole. However, since 1955, relatively small numbers of English, petrale, and rex sole have been sampled as part of the biological sampling program in California (Table 4.5). The Dover sole market category is assumed to be pure (composed only of Dover sole) and thus, it is not sampled for purposes of species-composition determination.

Historically, landings of Dover sole were primarily sampled at the ports of Morro Bay, Fort Bragg, and Eureka. These samples of Dover sole provided sufficient biological data for each INPFC area. From 1960 to 1990, biological samples collected from landings of flatfish consisted of 50 randomly selected fish per boat trip sample. Length and sex data were recorded for all 50 fish that were sampled; however, otoliths were removed from only the first 25 fish.

In 1991, CDFG adopted the two-stage sampling design that is based on selecting clusters (subsamples) of a fixed weight, rather than a fixed number of fish. Current samples of Dover sole consist of two 25 ± 2 lb clusters, where length and sex data are recorded and otoliths are collected for all of the fish in each cluster.

Otoliths from Dover sole are analyzed by different age readers based on the INPFC area (i.e., port) where the samples were collected. The break and burn technique is currently used to prepare and analyze otolith specimens from Dover sole. In the past, surface ages were determined from whole otoliths of Dover sole; however, most of these samples have been re-analyzed using the break and burn method. All otolith samples from Dover sole collected within the Eureka INPFC area have been analyzed through 1990 and also for 1994. In the Monterey INPFC area for the years 1984-89 and 1994, the only otolith samples from Dover sole that have been examined are those that were collected at ports within the Fort Bragg complex. Ages have been determined for only those Dover sole collected in 1985 for the Conception INPFC area. Starting in 1993, otolith samples collected from both the Monterey and Conception INPFC areas were assigned to a single age reader in Menlo Park.

4.5.2 Processing System

Procedures used to enter and process biological sample data are presented in section 4.4.3.

4.6 Estimating Derived Quantities

The procedures described below are currently under review and could be changed in upcoming years, and not all of these potential changes in the analytical methods will necessarily be made retroactively. Users of landing estimates should check with their data source for details of the procedures used to generate the specific estimates they have requested.

4.6.1 Landings by Area

Landing estimates are calculated for each port complex and not on an area-fished basis, such as PSMFC areas. Consequently, it is not necessary to use logbook data to apportion landings to fishing areas. It appears reasonable to determine estimates by port complex in California because the vessels are relatively small and tend to range only a short distance from their home ports. In any case, until recently, there was a considerable backlog of keypunching of logbook data, so it was not feasible to use the data to derive current landing estimates.

4.6.2 Annual Estimates

Rockfish

Annual estimates of trawl-gear landings, and age and length compositions, have been routinely generated on a species-by-species basis for northern California (Morro Bay and north). Similar estimates are now generated for set-net and line gear landings, including southern California, in cases when sampling was adequate.

The calculation of the estimates requires the sample-composition data and landing receipt information for each market category landed within each port complex and quarter stratum. In practice, there are often market categories and port complex and quarter strata for which no samples were taken. Thus, the first step in the processing of the data is to reassign landings to other quarters or port complexes, and/or to combine market categories. Ultimately, these procedures generate a reduced set of port complex and quarter strata and possibly market categories, with the necessary data for each.

The method used to combine and reassign the landings depends on the type of market category. Market categories considered to be relatively pure (e.g., widow rockfish, bocaccio, and thornyhead) are generally not combined with other market categories. When possible, the landings for a quarter that were not sampled are assigned to an adjacent quarter that had applicable samples. If no samples were obtained from a port complex during a year, then the landings for the market category are assigned to the nearest port complex for which samples were taken, with the caveat that landings are not reassigned across INPFC area boundaries except when absolutely necessary. When there are two candidate quarters or port complexes for reassignment, the landings are reassigned to the candidate with the most samples.

Some efforts are also made to not combine minor single species market categories with other market categories, under the assumption that these categories would not be recorded unless they were somewhat distinct. For example, if there were samples for these market categories (minor single species market categories, such as vermilion rockfish) in a given port complex but not for every quarter of the year, then the landings from the unsampled quarters would be reassigned to quarters that had been sampled, instead of categories with other rockfish market categories that had been sampled. If reassigning the landings for these market categories to nearby quarters is not possible, then the sample information is sometimes combined with the unspecified rockfish market category for that port complex.

The method currently used to reassign the landings of the mixed species categories of rockfish is also unique. If samples were taken for a given mixed species market category within a port complex and quarter stratum, then the landings for that category and stratum are treated separately; otherwise, the landings are combined with the unspecified rockfish market category for that stratum. In the case that the unspecified rockfish category is not sampled within a given stratum, then these combined landings are reassigned to a quarter that was sampled, and if no quarter was sampled, to the nearest port complex that was sampled. In some prior years, mixed species market categories were combined based on the judgements of the individual port biologists regarding the similarity of the categories at their port.

Annual reports are generated using a list of the sample data files, specific sort groups (market categories that are combined and evaluated as though they were a single category), and the landings that apply to the specified sort groups within each port complex and quarter stratum. For each port complex and quarter analyzed, landing estimates (in number) are determined for individual species by length class and sex, and by age class when age data are available. Total landings (in number and weight) are also estimated for the individual species for each port complex and quarter stratum. In these analyses, lengths are rounded to the nearest centimeter, and unknown information (i.e., missing values) for sex, length, and age is a recognized classification, with estimates being calculated for the number of fish landed in each of these classes. No attempt is made to reassign the unknown sex, length, or age information to other classes, although users of the data frequently need to do this.

The software first calculates estimates for each of the specified sort groups, and then adds these values together and reports the total quantity for all of the sort groups combined; however, the totals are based on only those market categories that were included in a sort group. Reports for individual sort groups are not generated.

For a given port complex i and quarter j, let the estimated quantity be designated as V_ij. Calculations of V_ij are done separately for specified sort groups and gear combinations, but these subscripts have been dropped for simplicity. The V_ij could be estimates of the total number of fish landed of a particular species, the weight landed for that species, the number of 8-year-old females of that species, or any other sex/length or sex/age combination. The quantity V_ij is found by first calculating an average quantity per pound of landings and then multiplying this value by the total pounds landed.

The average quantity per pound, vij, is calculated as a ratio estimate. Let the quantity observed for a single cluster l on trip k be v_ijkl (say the number of yellowtail rockfish observed in an individual cluster). The average quantity per pound on trip k can be estimated as,

(4.1)

where w_ijkl is the weight of cluster l selected on trip k in port complex i and quarter j.

The average quantity per pound on trip k (v_ijk ) multiplied by the total pounds of the sort group on trip k (L_ijk) provides an estimate of the total quantity for the trip. Values for L_ijk are determined from landing receipt information. The ratio estimator of the average quantity per pound over all trips is then,

(4.2)

which is simply the ratio of the sum of the sampled quantities to the sum of the landing weights of the sort group for all of the trips sampled within the port complex and quarter stratum.

The estimate of the total quantity of the species for the specified sort group within port complex i and quarter j can now be calculated as the average quantity per pound over all trips (v_ij) multiplied by the total reported landing weight of the specified sort group within the port complex and quarter stratum (W_ij),

(4.3)

Values for W_ij are determined from landing receipt information. Note that the values for W_ij are almost always greater than the sum of L_ijk, because W_ij is the sum of the landing weights of the specified sort group for both sampled and unsampled trips.

No standard reports on variances are generated. Estimates of variances have been occasionally generated following procedures outlined by Sen (1986).

The information on age, size, and sex compositions of the catch comes from the same samples used to estimate species composition. Thus, numbers-at-age or size, by sex, are calculated using essentially the same algorithms that are used to estimate total number or weight landed.

Dover sole

Species-composition analysis is not done on Dover sole because these landings are assumed to be pure. Annual estimates of age and length compositions for Dover sole are determined separately from the estimates for the rockfish species, but the two sets of analyses use basically the same programs. However, the problems associated with rockfish analyses that are due to lack of sample data do not occur for Dover sole analyses, because estimates for Dover sole are generated by INPFC areas and not by port complexes. That is, landings of Dover sole are never reassigned across INPFC area boundaries and sufficient levels of sampling are obtained from each INPFC area. Estimates are generated on a request-only basis and for the trawl gear only.

Annual reports are generated using a list of the sample data files, the Dover sole sort group, and the landings that apply to the Dover sole sort group within each INPFC area and quarter combination. For each INPFC area and quarter analyzed, landing estimates (in number) are determined for Dover sole by length class and sex, and by age class when age data are available. In these analyses, lengths are rounded to the nearest centimeter. Since sex and age information is determined for all samples of Dover sole prior to processing the annual reports, estimates for missing values are not needed.

4.6.3 Monthly Estimates to PacFIN

In a separate process from the annual estimates, estimated landings (in weight) for individual species are generated monthly. First, monthly estimates of the species composition of the landings are generated by the GCU based on the sample data. These estimates are forwarded to Sacramento, where they are processed together with data on landings for each market category.

The species-composition reports generated in Menlo Park consist of separate records for each market category, port complex, gear type, and month. Species compositions are reported separately for: bottom trawl (beam trawl, otter trawl, balloon trawl, roller gear trawl, and Danish seine); midwater trawl gear; longline gear; and hook-and-line gear (vertical hook-and-line, troll, and rod-and-reel). All gill- and set-net landings are combined and reported to PacFIN as a single category. Other gear types are generally unsampled and are not expected to contribute significantly to the total landings of rockfish.

Species compositions are generated for a gear group, not individual gear; therefore, the species compositions for midwater and bottom trawls are identical, and similarly, longline and hook-and-line species compositions are identical. Monthly estimates are based on sample data collected over a series of months. For months other than January and December, three months of data are used. For example, the species composition for July would be based on samples from June, July, and August. January estimates are based on January and February data, and December estimates on November and December data. Multiple months of data are used because sampling intensity is generally too low within a given month to provide reliable estimates of the landings.

Species compositions are calculated separately for each port complex. Port complexes are not combined as is done in the calculation of the annual estimates. Species compositions are also calculated separately for each market category, which are not combined together into sort groups as is done in the calculation of the annual estimates. However, in past years the chilipepper/bocaccio market category was combined with the unspecified rockfish market category, but because bocaccio are now generally reported in their own category, this combination of market categories is no longer applicable. Species-composition reports are not prepared for all landing categories (see below), because not all market categories are sampled for each gear type.

Species compositions are reported as the percent of the landings (in weight) that each species constituted. These percentages are based on the estimators described above for deriving the annual estimates. First, for a given market category sampled on trip k within port complex i and time period j (a 2- or 3-month time block), the average weight per pound of a species is calculated using equation 4.1. The weight of the species per pound of the market category over all trips is then calculated in the same manner as the annual estimate using equation 4.2.

The estimated percent of the landings (in weight) that a species constitutes for the specified market category within the port complex and time period combination is simply,

(4.4)

In the reports, the percentages are rounded to the nearest integer, and species with percentages less than 0.5 (i.e., those that would be rounded to zero) are not reported. The percentages for these unreported species are summed together, rounded to the nearest integer, and are reported as "unknown" fish if the rounded sum is 1% or greater.

The estimated species compositions are applied to the reported landings (for a corresponding year, month, port complex, gear, and rockfish market category) by the PacFIN Coordinator in Sacramento. Generally, when a species composition is not generated, landings are categorized as unknown rockfish. If the unsampled category is the unspecified rockfish market category, landings are left in this category. If unsampled landings are in the widow rockfish, thornyhead, bocaccio, or Pacific ocean perch market categories, they are treated as pure landings that contain only their respective species. The PacFIN data are updated as landing reports become more complete and are replaced with the results from the annual estimates when these become available.

4.7 Acknowledgments

We would like to thank Bill Lenarz, Frank Henry, the staff at the MFSU in Long Beach, Connie Ryan, Bernice Hammer, John Geibel, Larry Quirollo, Bob Leos, Becky Ota, Rachael Miller, and Cat Talbot for their help, without which this chapter would never have been completed.

4.8 Citations

Bence, J. R., and J. E. Hightower. 1990. Status of bocaccio in the Conception/Monterey/Eureka INPFC areas in 1990. In Pacific Fishery Management Council, Status of the Pacific coast groundfish fishery through 1990 and recommended acceptable biological catches for 1991: Stock assessment and fishery evaluation, Volume 2, Appendix J, p. J1-J65. (Available from Pacific Fishery Management Council, 2130 SW Fifth Avenue, Suite 224, Portland, Oregon 97201.)

California Department of Fish and Game (CDFG). 1992. Final Bulletin Tables No. 52. Unpubl. manuscr., 200 p. (Available from California Department of Fish and Game, 411 Burgess Drive, Menlo Park, CA 94025.)

Clark, G. H. 1935. Logs on California trawlers. California Department of Fish and Game Fisheries Bulletin 44:37-43. (Available from California Department of Fish and Game, 411 Burgess Drive, Menlo Park, CA 94025.)

Cochran, W. G. 1977. Sampling techniques, 3rd edition. John Wiley & Sons, New York, 428 p.

Conner, G., and nine coauthors. 1952. The commercial fish catch of California for the year 1950 with a description of methods used in collecting and compiling the statistics. California Department of Fish and Game Fisheries Bulletin 86:1-120. (Available from California Department of Fish and Game, 411 Burgess Drive, Menlo Park, CA 94025.)

Ernst and Young. 1989. Feasibility study report for the marine fisheries statistical system. Unpubl. manuscr., 105 p. (Available from California Department of Fish and Game, 411 Burgess Drive, Menlo Park, CA 94025.)

Goodrich, M. 1988. California Department of Fish and Game information systems section, trawler log system. Unpubl. manuscr., 180 p. (Available from California Department of Fish and Game, 411 Burgess Drive, Menlo Park, CA 94025.)

Groundfish Coordination Unit (GCU). 1995. California cooperative commercial groundfish survey sampling manual. Unpubl. manuscr., 72 p. (Available from California Department of Fish and Game, 411 Burgess Drive, Menlo Park, CA 94025.)

Nitsos, R. J. 1965. Pacific Marine Fisheries Commission (PMFC) Annual Reports, Nos. 16-17, p. 55-60. (Available from California Department of Fish and Game, 411 Burgess Drive, Menlo Park, CA 94025.)

Sen, A. R. 1984. Sampling commercial rockfish landings in California. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SWFC-45, 95 p.

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

Tomlinson, P. K. 1977. Rockfish sampling plan. Unpubl. manuscr., 10 p. (Available from National Oceanic and Atmospheric Administration, National Marine Fisheries Service, Southwest Fisheries Science Center, 8604 La Jolla Shores Dr., La Jolla, CA 92038.)

U.S. Department of Commerce (USDOC). 1992. Marine recreational fishery statistics survey, Pacific Coast, 1987-1989. Current Fishery Statistics, No. 9205. Washington, D.C., 367 p.

Table of Contents