Table of Contents -- < Backward -- Forward >
An important precursor to examining changes in the sport and commercial fisheries from 1958-61 to 1980-86 was to recognize, and when possible adjust for, differences in survey methods used. Substantial differences exist between methods used by Miller and Gotshall (1965), methods used in our restructured MRFSS estimates, and methods used for commercial landing statistics. The two recreational fishing surveys differed in fishing mode definitions, scope, methods for estimating effort and take, targeted species groups, and study-area boundaries. While both recreational fishery surveys are based on subsampling, commercial landing statistics are not (Oliphant et al. 1990).
To compare survey results by fishing mode, two of the modes used in the restructured MRFSS (jetty and breakwater, beach and bank) were combined to approximate the shore mode designation used by Miller and Gotshall (1965). The pier, skiff, party boat, and diver modes used by Miller and Gotshall (1965) are directly comparable to our pier and dock, PRB, CPFV, and spear modes, respectively.
Although Miller and Gotshall's (1965) sport survey began in June 1957, most of the summaries used for comparison to MRFSS spanned 1958 through 1961. Their survey was accomplished with considerably fewer resources than the MRFSS and as a result was much smaller and more field directed in design. Their survey was not intended to provide yearly estimates but instead provided 1958-61 average estimates. Miller and Gotshall (1965) described their effort and catch estimates as minimum values and did not provide statistical confidence limits.
It is likely that the 1958-61 survey underestimated fishing effort because it relied heavily on logbook data for estimation of CPFV effort, did not survey the substantial PRB fishery in San Francisco Bay, and did not survey several of the small piers. To compare total effort and catch between the two survey periods, we developed adjustments for the 1958-61 CPFV and PRB estimates as follows:
Logbook data provided by CPFV operators are mandated by state law but compliance is poor (P. Gregory, CDFG, pers. comm.). The compliance rate for submission of logs, for central California CPFV trips with an onboard CDFG sampler, ranged from 61% to 92% depending on the port and year (Reilly et al. 1993). Compliance without onboard samplers may be lower. Assuming our restructured MRFSS values are the best available estimates of actual effort and catch, CPFV logs reported averages of 51% of the actual CPFV effort and 65% of the actual CPFV catch in 1981-86 (Appendix C). Assuming CPFV log compliance has not changed between the two survey periods, we adjusted the 1958-61 CPFV effort and catch estimates accordingly. Future logbook compliance studies may provide a basis for the reader to make adjustments to increase the accuracy of the catch and effort estimates provided herein.
The other major adjustment was to correct for the omission of San Francisco Bay in the 1958-61 PRB estimates. In 1981-86, PRB fishing in the bay produced an annual average of 389,000 fish weighing 863,000 kg, or 81% by number and 90% by weight of the district PRB catch. Averaging the difference in the percentages, we assumed that San Francisco Bay fishing constituted 85% of the PRB effort in the district, or 47% of the total PRB effort in our northern and central California study area. Assuming the percentage had not changed over time, we adjusted the 1958-61 PRB effort and catch estimates accordingly. Since species composition of the PRB catch in San Francisco Bay differs greatly from species composition coastwide, we used the adjusted 1958-61 PRB data only for gross comparisons of total effort and catch. For purposes of historically comparing catches of individual taxa, we omitted the San Francisco Bay PRB catch from the 1981-86 estimates.
Several piers were not surveyed in 1958-61, but we did not have a rational basis to estimate effort at the unsurveyed piers. Miller and Gotshall (1965) stated that, had those piers been surveyed, pier fishing effort (estimated at 530,702 fishing days) might have exceeded shore fishing effort (estimated at 603,097 fishing days), a relatively minor difference. A fishing day is defined as one person fishing for all or part of one day.
We estimated direct expenditures on fishing activities from cost-per-trip data gathered in the 1981 northern California MRFSS survey. We assumed the cost of transportation in 1981 was $0.20 per mile, and we adjusted costs from 1981 dollars to 1992 dollars using consumer price index data supplied by the California Department of Finance.
Comparisons of sport and commercial catches of major taxa were based on commercial landings data for 1958-61 and 1981-86 (Marine Resources Operations 1960a, 1960b, 1961, 1963; Oliphant et al. 1990; CDFG unpublished data), our adjusted 1958-61 recreational catch estimates, and our 1981-86 recreational catch estimates.
It is important to recognize that the MRFSS used a telephone survey to estimate effort for all modes, but the effort estimation methods of Miller and Gotshall (1965) were field directed and differed among modes. They applied aerial surveys and field angler counts in skiff and shore modes, while dive clubs were directly censused in the spear mode. Estimates derived through telephone survey in the MRFSS showed greatest variance between years for the spear mode (Albin et al. 1993) and comparisons to historic spear data should be interpreted cautiously. Single-year estimates derived from the restructured MRFSS data for rare and patchy fisheries, such as spear and shore-based surfperch fishing, are probably of poor quality.
The 1958-61 survey of Miller and Gotshall (1965) included salmon fishing from boats, but the MRFSS did not. To allow a complete comparison to historical sport and commercial fisheries, we provided salmon estimates (CDFG unpublished data) with the 1981-86 restructured MRFSS data. We estimated salmon catch by weight using calculated average weights of salmon incidentally caught by anglers interviewed in the MRFSS creel survey who were targeting other species. Commercial landing weights for salmon are reported for dressed (gutted, head-on) fish. We converted dressed weights to whole fish weights by multiplying by a factor of 1.15 (A. Baracco, CDFG, pers. comm.). We used the salmon catch-by-weight estimates mainly as a component of total catch by weight, and not to analyze trends in salmon fisheries.
Our restructured MRFSS districts and also the commercial port areas were very different from the subareas delineated by Miller and Gotshall (1965) (Figure 1). We did not attempt to assess historical changes by district or commercial port area. Instead we compared the entire northern and central California study area. Our northern and central California study area differed slightly from the Miller and Gotshall (1965) study area at the southern boundary; the former ended at the southern boundary of San Luis Obispo County but the latter extended 40 km south to Point Arguello (Figure 1).
The percentage of a given species in the total catch of all species along the coastline may reflect its latitudinal distribution and abundance over time. We therefore used 1980-86 MRFSS creel survey data (percent of the total sampled catch by number, by district and year) to construct distribution maps for selected species. To provide a larger frame of reference, we also included 1980-86 MRFSS creel survey data from Oregon and southern California; however, Oregon data were not available for 1980. Each map was based on catch data from over 230,000 angler bags from 12 defined districts from northern Oregon to southern California (Figure 2).
To rank the relative importance of species taken by district and mode, we developed an index of relative importance (IRI), similar to indices used in food habit studies (Pinkas et al. 1971). Our IRI incorporated three factors: percent frequency of occurrence in angler bags (i.e. the percentage of bags that contained one or more fish of a particular species), percent by number, and percent by weight. Our IRI does not necessarily indicate the desirability of the species. Unlike the Pinkas et al. (1971) IRI, which gave different weighting to occurrence, our index gave equal weight to all three components. The IRI was calculated as
IRI = Fi + Ni + Wi
where
= Percent frequency of occurrence in bags
Bin = Bag with species i in bags 1,2,...n
Bn = Bags 1,2,...n
= Percent by number
Nin = Number of fish of species i 1,2,...n
Nn = Number of all fish 1,2,...n
= Percent by Weight
Win = Weight of fish of species i 1,2,...n
Wn = Weight of all fish 1,2,...n.
In the creel survey, bags may or may not have contained fish and often contained the catch of more than one person.
Available data and time limited our detailed analysis to rockfish, lingcod, and surfperch. Our selection of those taxa was based on their importance to recreational fisheries and evidence of historic decline among surfperch. The selection of species within each taxa and the extent of analysis depended on importance in the catch and extent of available data. Sixteen rockfish species were analyzed for catch and distribution; the selection criterion used was an annual average landing minimum of 10,000 fish in 1981-86. Seven of those rockfish species were also selected for length-frequency analysis based on availability of 1980-86 MRFSS length data. Ten of the more common surfperches were analyzed for catch. Eight species were analyzed for distribution; the selection criterion used was a cumulative sample size minimum of 1000 fish in 1980-86. Barred surfperch and redtail surfperch, the two main surfperch species taken by sport and commercial fisheries in 1980-86, were analyzed in greater detail.
We analyzed length-frequency data to discern differences in stocks within the major species, either geographically or among different fisheries. Length-frequency data sources included the MRFSS, historic CDFG recreational fishery studies, and commercial fishery studies. Length frequencies were compared graphically using standardized histograms and statistical tests. Species whose length frequencies showed evidence of cohort dominance were subject to modal-progression analysis to identify differences in growth, survival, and patterns of recruitment by geographic area and fishery.
For most species, the length data collected were total lengths, measured from the tip of the lower jaw or end of the snout, whichever was terminal, to the tip of the longest caudal lobe with the caudal lobes pinched together (Miller and Lea 1972). For tuna-like fish with rigid concave caudal fins, fork lengths, measured from the tip of the snout to the middle of the fork in the tail, were taken.
Small sample sizes in the MRFSS length data compelled us to combine length-frequency data from adjacent districts to form two larger areas, northern California (districts 1, 2, and 3) and central California (districts 4 and 5; Figure 1). Inclusion of the San Francisco district in northern California was logical because Cordell Bank, the major offshore reef fished by boats from Sonoma County, is also fished from San Francisco. Length-frequency counts for adjacent districts were not weighted by landings; instead the sample size by district was assumed to represent size of the landings. That assumption is valid for MRFSS data since MRFSS sampling effort was generally allocated in proportion to an area's fishing effort (Karpov 1987; U.S. Department of Commerce 1987).
When necessary, length-frequency data for PRBs and CPFVs were combined to provide a sample size sufficient for resolution of annual recruitment patterns. Although sizes of fish taken by PRBs were usually smaller than those taken by CPFVs, frequency-distribution patterns were similar.
Commercial trawl length-frequency data were assigned to the same approximate northern and central California areas by combining samples from coincident port areas (Figure 1). Again, port-area data were not weighted by size of landings. Commercial length-frequency data were utilized for bocaccio, chilipepper, canary rockfish, and yellowtail rockfish (CDFG unpublished data).
Lengths were grouped into 10-mm or 20-mm intervals depending on the species maximum length reported in Miller and Lea (1972), as used by Karpov (1987). Thus 10-mm intervals were applied to blue rockfish, chilipepper, brown rockfish, black rockfish, barred surfperch, and redtail surfperch, and 20-mm intervals were applied to canary rockfish, bocaccio, yellowtail rockfish, lingcod, and Pacific mackerel.
Length-frequency distributional differences were tested using the Kolmogorov-Smirnov (KS) test. Differences in lengths between areas and years were compared using analysis of variance (ANOVA). For lingcod, ANOVA was followed by a Scheffe multiple-comparison test for paired comparisons. Regression analysis and analysis of covariance (ANCOVA) were used to compare decreases in length of yellowtail rockfish and canary rockfish for commercial trawl catches from 1978 to 1989 and MRFSS from 1980 to 1986. Significance for all tests was accepted at a = 0.05.
Modal-progression analysis of dominant cohorts was possible for four species: blue rockfish, yellowtail rockfish, bocaccio, and lingcod. The analysis was used to compare growth rates and mortality rates, and to interpret recruitment patterns in northern versus central California. Age-at-length data were available and were based on otoliths for bocaccio, chilipepper, and lingcod, and on scales or tag-and-recapture studies for blue rockfish (Miller and Geibel 1973; Rogers and Bence 1992; Bence and Rogers 1993). For bocaccio and chilipepper, age-at-size data were used to assign a cohort birth year to dominant length-frequency modes for the 1980-86 period.
A longer time series of length-frequency data was available for lingcod and blue rockfish (1959-86). Ford-Walford analysis was applied to the modal growth increments for both species to determine growth rates and age at size. For both species modal-progression analysis was possible only in central California. In northern California, blue rockfish did not show dominant cohorts and lingcod had too few samples. As with bocaccio and chilipepper, age-at-size data were used to identify birth year for the dominant cohorts, assess year-class strength, and estimate age and size at recruitment to the fishery.
The Ford-Walford analysis involved applying Ford's equation,
where
lt+1 = the length the following year
L= the asymptotic length (intersection of the regression and a 45° line)
e-K = Ford's growth coefficient and slope of the regression line,
to modal progressions in yearly pairs to generate parameters of a Von Bertalanffy growth model (Ricker 1975). Using these parameters age at size was approximated using the relation
described by Gulland (1969). In our analysis t0 is assumed to equal zero. Age-at-length approximations were then used to estimate an approximate birth year for each dominant mode from the 1959-86 sample data.
We used 1980-86 MRFSS creel survey data for single-angler bags to examine the potential effects of lowering recreational bag limits of rockfish, lingcod, redtail surfperch, and barred surfperch. Potential effects on the number of fish harvested were estimated by plotting the frequency of number of fish per bag, and determining percent of total catch at hypothetical new bag limits. At each hypothetical limit, we assumed bags containing numbers of fish greater than the hypothetical limit would contain the hypothetical limit.
All species sampled by the MRFSS, excluding salmon, were examined graphically for annual changes in percent of catch (both released and kept) by coastal county district in Oregon and California for 1980 through 1986. The data were plotted on the same type of maps used in the section on rockfish, lingcod, and surfperch to describe distribution through availability to sport fisheries. We looked for shifts in availability during the 1982-83 ENSO years to the north or south of the 1980-81 and 1984-86 distribution.