Allen et al.: Seafood consumption rates among recreational anglers 



599 



trips to pier-and-jetty, party-boat, and private-boat 

 sites on two week days and two weekend days per 

 month in summer and on one week day and one week- 

 end day per month in nonsummer. 



To minimize sampling bias, we used a stratified- 

 random design to schedule and conduct the survey 

 trips. Fishing modes, regions, sites, and sampling 

 times were selected randomly to maximize spatial 

 coverage of the Bay. For each month, we selected a 

 sampling sequence at random for four regions and 

 sites within each region, using a random numbers 

 table; we surveyed each site in the selected sequence. 

 If a site could not be sampled, we sampled the next 

 site on the list. Sampling was conducted without re- 

 placement within a month and with replacement 

 between months. 



We chose survey-trip times at random for the three 

 major fishing modes. Interviewers surveyed pier-and- 

 jetty anglers in the morning (0800-1200 h), after- 

 noon (1200-1600 h), or evening (1600-2000 h) with 

 roving surveys. They surveyed party-boat anglers on 

 half-day boats within the study area (full-day boats 

 generally fished outside the area) in the morning 

 (0700-1200 h) or afternoon (1230-1730 h). Private- 

 boat anglers were interviewed at boat launches or 

 hoists in the morning (1000-1400 h) or afternoon 

 (1400-1800 h). Interviewers conducted 1-h beach 

 survey trips before pier-and-jetty surveys and con- 

 ducted 2-h rocky intertidal surveys on randomly se- 

 lected afternoons ( 1300-1700 h) during low tides. 



Interviewers tried to interview all anglers at a site. 

 However, if too many anglers were present, they sys- 

 tematically selected every second or third angler. 

 Interviewers censused (counted and characterized) 

 anglers at each site and asked all (or a subset of the 

 anglers) questions from a specially designed ques- 

 tionnaire. All interviewers spoke English; at least 

 one interviewer per survey spoke Spanish; and oth- 

 ers spoke Vietnamese, Chinese, or Pilipino. They in- 

 terviewed a designated household head if two or more 

 individuals from a household were present. Inter- 

 viewers asked anglers 35 questions regarding their 

 background, their fishing history, the types of fish 

 they had eaten, their consumption habits, their meth- 

 ods of preparing fish, their awareness of health risk 

 warnings, and their response to warnings (Table 1). 

 Anglers chose the ethnic-group category best repre- 

 senting their background from a list used in the most 

 recent United States census report (USBC, 1990). 



Interviewers identified to species any fish pos- 

 sessed by anglers and measured its fork length to 

 the nearest centimeter. They showed photographs of 

 important species in order to enable anglers to iden- 

 tify other species consumed during the previous 

 month. The species included bocaccio (Sebastes 



paucispinis), barred sand bass (Paralabrax nebu- 

 lifer), kelp bass (Paralabrax clathratus), white 

 croaker, queenfish (Seriph us politus), California cor- 

 bina (Menticirrhus undulatus), chub (=Pacific) mack- 

 erel (Scomber japonicus), and California halibut 

 (Paralichthys californieus). Interviewers also carried 

 a balsa-wood model of a 150-g generic fish fillet (based 

 on a fillet of that size obtained from a supermarket 

 [the USEPA standard meal size; USEPA, 1989]). They 

 asked anglers to estimate their meal size for each 

 species (in hand or in photographs) in relation to the 

 fillet model. The fillet model gave the angler a three- 

 dimensional image of the standard meal size. 



Data analysis 



Questionnaire responses were numerically coded and 

 entered into a computer database for analysis. We 

 calculated consumption rates by two methods. The 

 primary method used the angler's estimates of meal 

 size based on a 150-g fillet model. The other method 

 used consumable-portion sizes of fish possessed by 

 an angler. For the latter method, we converted 

 lengths offish measured in the field to total and con- 

 sumable-portion weights. These were estimated for 

 each fish from weight-length regressions 7 and from 

 consumable-portion information. 8 In both methods, 

 we multiplied the amount consumed per meal or the 

 consumable portion by the number of times an an- 

 gler consumed that species during the previous four 

 weeks. This gave a monthly consumption rate (in 

 grams per month). We divided this rate by 28 days 

 to get daily consumption rates (in grams per day). 

 For the consumable-portion method, we divided the 

 consumable portion by the number of consumers in 

 the household. We did not do this for the fillet-model 

 estimates because we obtained only information on 

 the angler's meal size. 



Consumption rate data were summarized by para- 

 metric (means, standard deviations, and 95% confi- 

 dence limits) and nonparametric statistics (median 

 and upper decile or 90th percentiles). We tested for 

 consumption-rate differences among ethnic and in- 

 come groups using a Kruskal-Wallis one-way analy- 

 sis of variance on ranked data. We tested for differ- 

 ences between fillet-model and consumable-portion 

 estimates of consumption rates using a Wilcoxon's 

 sign-rank test or paired f-test, as appropriate. 



7 S. J. Crooke, Calif. Dep. Fish Game, Long Beach, CA. provided 

 weight-length relationship information. 



8 W. Jacobson, U.S. Dep. Commer., Natl. Ocean. Atmos. Admin., 

 Natl. Mar. Fish. Serv., Los Angeles Reg.. Long Beach, CA, pro- 

 vided consumable-portion information. 



