398 



Fishery Bulletin 105(3) 



mature females allowed the investigation of the full 

 seasonal progression of ovary development. The propor- 

 tion of females in spawning condition was estimated for 

 each month by dividing the sum of the fish classified as 

 spawning by the sum of the fish classified as mature. 

 The length- and age-at-maturity estimates that were 

 chosen as representing the standard for each area were 

 based on specimens collected during the prespawning 

 or initial yearly spawning period, January 1999 and 

 2004 and February 2003. By January and February, all 

 specimens that would mature that year had matured 

 according to the results of seasonal sampling from this 

 study. The data from January of 1999 and 2004 were 

 combined to obtain length- and age-at-maturity esti- 

 mates with the lowest variance. The monthly length 

 and age composition of females in mature condition 

 were estimated by using standard length- and age-at- 

 maturity analysis. 



Maturity as a function of length was estimated by fit- 

 ting a logistic function to the maturity data with general- 

 ized linear modeling (Venables, 1997); for this procedure, 

 S-Plus software (vers. 2000 Professional release 3, Math- 



Soft Inc., Cambridge, MA) was used. The significance of 

 between-area, between-month, and between-classification 

 methods differences were tested by fitting the model of 

 maturity as a function of Ly, with a term distinguishing 

 area, month, and method and by recalculating without 

 each term. Significance of the area, month, and method 

 terms were determined by using analysis of deviance 

 (Venables, 1997). Length at 50%- maturity was also es- 

 timated by evaluating the fitted model at 50% maturity 

 and algebraically solving for length. The variance of 

 Ly,j.jj was estimated for each area, month, and method by 

 using bootstrapping (Efron and Tibshirani, 1993) based 

 on 200 resamplings, with replacement, of the maturity 

 and length data. Between-area, between-month, and 

 between-method differences in L^,^^ were then tested 

 with a 2-test (Sokal, 1969). With these same procedures, 

 maturity was described as a function of age, and spawn- 

 ing was estimated as a function of Lj,, age, month, and 

 ambient water temperature. 



With the use of S-Plus software, length-at-age was 

 described by the von Bertalanffy growth function, 

 which incorporated nonlinear least-squares fitted to 



