FISHERY BULLETIN: VOL. 76, NO. 2 



classified estimates of stock contribution are ob- 

 tained at the end of this stage. However, these 

 estimates are then used in the second stage as 

 prior probabilities that specimens come from the 

 three stocks. For example, the as-classified esti- 

 mate of Hudson stock contribution obtained at the 

 end of the first stage was used at the beginning of 

 the second stage as our best guess of the proportion 

 of specimens in the sample that originate from the 

 Hudson. These prior probabilities are then used to 

 weight the decision to classify each specimen into 

 one of the stocks. Similarly, the proportion of 

 specimens classified into each stock in the second 

 stage were used as priors in the third stage. The 

 procedure was carried out for nine stages. 



The effectiveness of adjusted and iterative esti- 

 mates in reducing bias in the as-classified esti- 

 mate due to misclassification was investigated in a 

 simulation study. Discriminant functions from 

 the cross-validation study were used to classify a 

 subset of specimens from the independent half of 

 the spawning-stock collections, and each of the 

 three types of estimates of relative percentage 

 were obtained and compared with the known stock 

 percentage. For percentages of Hudson stock rang- 

 ing from to 907f , the difference between each 

 estimate of Hudson percentage and the known 

 percentage of Hudson specimens in the subsample 

 was obtained as a measure of bias in the estimate. 



Collection, Processing, and Analysis of 

 Atlantic and Hudson River Specimens 



Assessment of the relative contribution of vari- 

 ous stocks of striped bass to the Atlantic coastal 

 fishery required a stratified sampling design that 



provided samples from the entire coastal fishery 

 and considered the migratory nature of striped 

 bass; therefore a geographically and temporally 

 stratified sampling design was used. The geo- 

 graphical stratification consisted of 10 strata from 

 southern Maine to Cape Hatteras, with 2 to 4 

 substrata within each stratum to compensate for 

 variations in stock composition within the 

 stratum ( Figure 1 ). The Rhode Island stratum was 

 not subdivided because of its small size. Tempo- 

 rally, the year was divided into six 2-mo periods to 

 obtain estimates of stock composition by stratum 

 throughout the year. 



Collections of striped bass from the coastal 

 fishery were obtained primarily from sport and 

 commercial fishermen; however, in areas where 

 adequate sport and commercial fisheries did not 

 exist, study personnel used haul seines and gill 

 nets to collect specimens. Collections were limited 

 to striped bass caught during the same day (i.e., 

 within 24 h) to assure freshness. In many in- 

 stances the entire catch was used, but due to the 

 size of some catches, a random sample propor- 

 tional to the number of small ( <550 mm), medium 

 (550-850), and large (>850) striped bass caught 

 was obtained. 



Oceanic and overwintering specimens were pro- 

 cessed in the same manner as spawning-stock 

 specimens. Two replicates of 10 counts and mea- 

 surements were taken from each specimen, and 

 scale samples were obtained for subsequent age 

 and growth rate determinations in the laboratory. 

 A total of 2,737 oceanic specimens with a complete 

 set of meristic, morphometric, and scale charac- 

 ters were processed (Table 1). Additionally, 79 

 striped bass overwintering in Croton Bay on the 



Table l . — Number of striped bass with complete character sets' collected by spatial stratum and period from Atlantic coastal 



fishery in 1975. 



Spatial 

 stratum 



8 

 9 



10 

 Total 



Locality 



Legal/ 

 subiegai^ 



S Maine-N Mass 

 S. Mass. 

 Rhode Is. 

 E. Long Is. Sound 



W Long Is Sound 



E Long Is S. Shore 



W Long Is S Shore 



N.J, 



Del.-Md -N Va. 



S Va.-N C. 



Jan. -Feb. Mar.-Apr. May-June July-Aug. Sept -Oct. Nov. -Dec. Total 



'Measurements and counts taken on all variables used in the character set 



^Sublegal-sized striped bass (< 406.5 mm FL) from New York waters (strata 4 to 7) were analyzed separately. 



338 



