Table of Contents -- < Backward -- Forward >
MRFSS data for 1980-86 were the main source of recent recreational fishery data examined in this report. Recreational salmon fishery statistics and commercial fish catch statistics were also incorporated from other sources (Pearson and Ralston 1990; CDFG unpublished data). Recreational effort and catch estimates for 1958-61 were from Miller and Gotshall (1965). Length-frequency data for years prior to 1980 were from the CDFG field surveys conducted pursuant to the studies of Miller and Gotshall (1965) and Miller and Geibel (1973).
The MRFSS had two major components: a telephone survey which estimated total effort and a field-based creel survey which determined catch per unit of effort (CPUE), species composition, and length frequency (U.S. Department of Commerce 1987). The telephone survey was random and restricted to coastal counties (counties within 40 kilometers of the coast, including San Francisco Bay), where most people who engage in marine sport fishing live. The creel survey was randomly stratified by mode, gear, and relative fishing effort at each fishing site. Sampling effort remained relatively constant throughout the 1980-86 period. Due to telephone survey quality problems in 1980, we used the creel survey data for that year, but did not produce total effort and catch estimates.
For 1981-86, effort and catch estimates were recalculated based on the MRFSS methods (U.S. Department of Commerce 1987) to allow comparison to historical data and provide a more useful baseline for future work. The recalculations produced smaller study areas, corrected oversights in the federal expansion methods, and defined catch in a format comparable to Miller and Gotshall (1965).
Federal MRFSS data reports divided California into northern and southern subregions along the boundary between Monterey and San Luis Obispo counties (U.S. Department of Commerce 1987). The federal reports calculated total effort and catch for each subregion by pairing the creel survey data for each subregion with the telephone survey effort estimates from the coastal counties in the subregion. Calculation of effort and catch by smaller geographic units would allow greater flexibility in examining latitudinal differences in total catch, species composition, length frequency, and CPUE. We therefore modified the effort and catch calculation methods to divide the northern California subregion into four coastal county districts, each consisting of one or more counties; San Luis Obispo County was included as a fifth district (Albin et al. 1993) (Figure 1). To further refine our analysis of boat mode fisheries, boat mode estimates for district 2 were separated into Mendocino County (mainly Noyo Harbor) and Sonoma County (mainly Bodega Bay). Our methods paired creel survey data for each district with telephone survey estimates of effort in the district originating from the northern California subregion. Trip estimates for the San Luis Obispo district did not include trips originating from the southern California subregion and are therefore probably underestimated. Another drawback of our method was the frequent lack of telephone survey effort data for a district, although it was known from the creel survey data that fishing effort did occur there. In those instances the six-year average effort for the district was applied.
The methods used and the results published in federal MRFSS data reports had several deficiencies. To ensure that large population counties would not monopolize the finite number of telephone survey assignments available, the MRFSS telephone survey assignments were intentionally biased towards small population counties by assigning the number of telephone contacts to a coastal county during each two-month period using the square root of the number of censused households (U.S. Department of Commerce 1987). The flaw in the federal MRFSS data reports was the subsequent failure to down-weight the number of trips reported by residents of small counties to reflect their true proportion of the county population. In a 1988-89 MRFSS-based study of red abalone Haliotis rufescens, Karpov (1991, 1992) found that small coastal counties had disproportionately high participation in abalone fishing. He down-weighted trip estimates to adjust for the square root assignment of telephone intercepts and found effort had been exaggerated by 28% and 32% in the two years examined. We used similar methods to downweight MRFSS trip estimates for 1981 to 1986.
In California the MRFSS excluded fishing for salmon aboard PRBs and CPFVs. Those fisheries were subject to a separate survey conducted by CDFG. However the MRFSS creel survey did collect salmon angler county-of-origin data without catch information (in intercepts described as "salmon short-forms"). In the federal data reports, salmon angler county- or state-of-origin data were combined with nonsalmon creel survey data to determine relative fishing effort from people living outside the coastal county telephone census population (L. Ponseggi, CIC Research, pers. comm.). We excluded all salmon short-form data to avoid a potential bias of salmon anglers having a county- or state-of-origin composition different from others interviewed in the creel survey.
Fishing trips on CPFVs targeting striped bass in San Francisco Bay were excluded in the MRFSS telephone survey, but were not excluded from the creel survey. Our estimates of striped bass catch should therefore not be used to analyze the striped bass fishery. We used the striped bass estimates mainly as a component of total catch.
Our editing of creel survey data found coding errors that had not been corrected in the federal data reports. Surf smelt, a species caught in large numbers by net, was frequently erroneously assigned to hook-and-line gear, causing overestimation of catch by number in the federal reports.
Fishing with nets for surf smelt and night smelt from northern California beaches occurs when the fish are running, but fishing activity is especially patchy in occurrence and was poorly sampled due to the MRFSS's stratified random design. We therefore excluded the net fishery from our analysis.
Unless otherwise specified, the catch type used in this report is the landed catch (the catch that is utilized in some fashion). Landed catch is essentially the same as used by Miller and Gotshall (1965). It includes fish landed whole and available for identification by the surveyor, and also fish caught and filleted, given away, or used for bait or other purposes. Landed catch excludes fish that were caught and subsequently released alive or discarded.
Fish were not weighed directly. The weights of measured fish were computed from known length-weight relationships. A portion of the landed catch was unavailable to the sampler for measurement (e.g. fish filleted at sea or used for bait). The sampler identified those fish based on information from the person being interviewed, and identification to a general taxon (e.g. rockfish) was frequently the case. The weights of fish unavailable for measurement were assumed to be equal to the average weight of fish of the same taxon, or of a more specific taxon from the same mode, gear, district, area, year, and quarter (quarters were January-March, April-June, July-September, or October-December). When substitute weight data from the same quarter were unavailable, average weight data from other quarters of the same year were used.
After the weights of landed but unavailable fish were estimated, the average weight per fish for each combination of taxonomic group, year, mode-gear combination and district-area combination was calculated by dividing the total weight of all fish by the total number of fish. Then, estimates of catch by weight for each combination were calculated by multiplying the catch-by-number estimate by the average weight per fish.