FISHERY BULLETIN: VOL. 84, NO. 2 



Table 9.— Comparison of school length measurements 

 (m) derived from echo sounder versus sector-scanning 

 sonar data collected aboard the FV Ocean Leader, 14 

 March-7 April 1982. 



assumption is that the variability has a strong ran- 

 dom component and catch per unit effort values are 

 consequently unbiased. The same situation may well 

 be true here, in which case an important component 

 of the survey design would be multiple replication 

 to obtain good estimates of both long- and short-term 

 variance 



Burnham et al. (1980) reported that good results 

 from line transect surveys require observation of a 

 minimum of about 40 objects per replicate Fitting 

 the observed perpendicular sighting distances to a 

 detection function becomes less reliable with a 

 smaller number of objects. Widow rockfish abun- 

 dance is now low on all major grounds and the 

 recommended minimum number of schools was not 

 observed during any single replicate in the 1982 

 survey, but by pooling replicates a sufficient data- 



base was constructed. Sample sizes could be in- 

 creased through more intense sampling. A time- 

 stratified analysis of the data would be desirable 

 to define within-night variability, but this would 

 place even further demands on a sampling pro- 

 gram. 



Surveys of the type used for widow rockfish must 

 cover the geographic range of the species of interest 

 more thoroughly than most other survey methods. 

 The dynamic behavior of widow rockfish suggests 

 that the survey method should cover large areas in 

 a relatively short time in order to survey a given 

 fishing ground at least once during the night. Be- 

 cause of day-to-day variability, surveys should include 

 sampling each area during several nights over a 1- 

 or 2-wk period. Most areas containing fishable widow 

 rockfish concentrations have probably been iden- 

 tified and there are a limited number of these 

 grounds (probably 12-20); nearly all are character- 

 ized by ridges or rises on the outer continental shelf 

 or upper slope and are relatively small in area. In- 

 tensive sampling of widow rockfish, therefore, is 

 more feasible than for most other groundfish species 

 inhabiting less well-defined areas. 



Because widow rockfish schools are continually 

 forming and breaking up, there may be a significant 

 portion of the population which is not schooling at 

 any given time and is therefore not susceptible to 

 these survey techniques. This project did not answer 

 whether this is so, but nothing was found to suggest 

 that widow rockfish are significantly detectable by 

 trawl or hydroacoustic surveys in any form other 

 than midwater schools. Until more is learned about 

 the proportion of the stock occurring as schools, 

 surveys must be considered as yielding minimum 

 biomass estimates. Clark and Mangel (1979) pro- 

 posed a study of rates of school formation and disper- 

 sal to explain and evaluate a similar relationship be- 

 tween overall stock size and the proportion of a 

 yellowfin tuna stock occurring as schools. Such a 

 technique should receive further consideration in this 

 situation, but present low widow rockfish school 

 abundance (schools/km 2 ) and lack of a consistent 

 pattern of school formation and dispersal would 

 probably make its application in widow rockfish 

 assessment difficult. This question is analogous to 

 that of defining catchability coefficients (i.a, what 

 proportion of those fish in the path of a net are ac- 

 tually captured) for quantitative trawl surveys. 

 Changes in relative abundance can be monitored by 

 such surveys without knowing the catchability if one 

 assumes that the available proportion of the popula- 

 tion is constant. 



Results of other analyses of widow rockfish be- 



308 



