136 



Fishery Bulletin 101(1) 



In this application a term for fishing mortality is assumeci 

 to be unnecessary because there is no commercial or recre- 

 ational harv'est of shortbelly rockfish. The model was fitted 

 by a weighted linear regression of log^.lA^j,) on T, where the 

 statistical weights were derived from expansions of the 

 aged subsamples to the full trawl catches. Results showed 

 a declining trend in abundance with age (Table 2, Fig. 4), 

 and an estimated adult natural mortality rate of 0.26/yr 



Like the weight-length relationship, fecundity is typi- 

 cally related to fish size with the power function (Bagenal 

 and Braum, 1968 1, which is linearized by logarithmic 

 transformation, i.e. 



F = 0W'\ 

 \og^,(F) = log,(0) + Sx log,,(W), 



where /■' = individual annual Iccuiulilv i larvae/female); 



log., I 



W = female weight (g); and 

 and 5 = fitted parameters. 



In our study fecundity estimates were gathered from 531 

 females taken during the trawl surveys and the transformed 

 data were fitted by simple regression (Fig. 5, Table 2). 

 There was considerable residual variability in the fecun- 

 dity at weight relationship (r2=0.65; a%^^. =0.29715) and, 

 in this instance, the addition of a bias-correction term 

 had a noticeable effect on back-transformed predictions of 

 fecundity at weight on the arithmetic scale. Although the 

 distribution of regi'ession residuals deviated significantly 

 from normal (P=0.0001i, due in large part to negative 

 skewness, the overall fit was deemed adequate. Predic- 

 tions of fecundity at weight were ~25-35'^( less than the 

 results presented in Lenarz ( 1980), although his equation 

 is based on only the 10 data points provided in I'hillips 

 11964). 



