Estimating the Harvest of Salmon by the Marine Sport Fishery in Puget Sound: Evaluation and Recommendations

The Washington Department of Fisheries (WDF) annually estimates the number of salmon harvested by the marine recreational fishery in Puget Sound using the Salmon Punch Card System (SPCS). Anglers fishing for salmon are required to have a state-issued catch record and to record the date and location of any salmon that they harvest during a calendar year. A random sample of all punch cards issued is used to estimate the salmon harvest. WDF and the twenty Treaty Tribes of Western Washington conducted a joint four-year study to assess the accuracy of the estimates from the SPCS. This study used access-site creel surveys to estimate salmon harvest independently of the SPCS.

The creel surveys estimated the number of salmon harvested by the sport fishery in a catch area during a one-month period. Sixteen of these area-month cells were sampled during each year of the study providing a total of 64 estimates of punch card bias. Each of the nine catch areas in Puget Sound was surveyed four to nine times during the study. Bias estimates for the 64 areamonth cells surveyed ranged from 0.34 to 11.41, but most of the extremes (the low and high estimates of bias) occurred in cells with very small harvests. Therefore, for the final estimates of bias only area-month cells with estimated harvests of 500 or more salmon were used. A major assumption of the study was that the creel survey provided unbiased estimates of the salmon harvest by the sport fishery.

Estimates of bias for Area 05 were significantly different from the estimates for Areas 06 through 13 combined, but no significant differences were found within Areas 06 through 13. There were no significant differences in bias between seasons, summer and winter, or among years. Therefore, the data from all four years of the study were combined to estimate bias adjustment factors for two geographic strata: (1) Area 05 and (2) Areas 06-13 combined. The biases estimated for these two strata were 0.99 and 1.46, respectively.

This bias has two probable sources: (1) non-response bias and (2) recall error bias. We recommend that the Salmon Punch Card System continue to be used to estimate the number of salmon harvested by the marine recreational fishery in Puget Sound, but the return rate of in-sample punch cards must be increased to a minimum of 70% to provide acceptable estimates which require no bias adjustment. WDF should investigate methods of increasing the response rate through: improvements in data control; increased information and education efforts; and, if necessary, by instituting angler incentives. We project that with a 70% minimum response rate and supplementary surveys, the Salmon Punch Card System would supply harvest estimates with 10% or less bias. Once this goal has been reached and verified, neither bias adjustment to the harvest estimates nor creel surveys for estimating bias would be necessary.