DeMartini: Annual variations in reproductive traits of Senphus politus 



I I 



Cone 1989). I therefore limited my evalu- 

 ation of robustness to comparisons of 

 estimates of ordinary least-squares re- 

 gression parameters (Cone 1989). 



The relationship between wet and dry 

 body weights might change throughout 

 the spawning season (Love 1970). There- 

 fore, I evaluated seasonal changes in dry 

 somatic weight using 57 females collected 

 at the start of (April-May, n = 32) and im- 

 mediately following (August-September, 

 n = 25) the 1985 spawning season. After 

 ovarectomy, fish were frozen in air-tight 

 "zip-lock" bags. Fish were then thawed, 

 and each entire fish was macerated and 

 individually oven-dried to a constant 

 weight at 120°C for 24-32 hours. 



Statistical analyses 



I used nonparametric analysis of variance 

 (Kruskal-Wallis One-way ANOVA) to 

 compare the body lengths and somatic 

 weights of females among years. Year 

 was evaluated as a fixed-effect class vari- 

 able, because I was interested in evalu- 

 ating potential differences among a pre- 

 established series of years. Analysis of 

 covariance (ANCOVA) was used when- 

 ever possible to evaluate the effects of 

 sampling year on batch fecundity, and on 

 relative gonadal condition, after adjust- 

 ment for year differences in somatic weight. A two- 

 way Model I ANCOVA was used to evaluate sub- 

 seasonal (approximately bimonthly) influences of egg 

 diameter among years for females of differing body 

 lengths. Dry egg weight was related to egg diameter 

 by parametric regression. Dry somatic weight was 

 regressed on wet weight for sample fish collected at 

 the beginning and at the end of the 1985 spawning 

 season; regressions were then compared using 

 ANCOVA with body length as the covariate. Computa- 

 tions were made using the GLM, REGRESS and 

 TTEST software procedures of the Professional Data- 

 base Analysis System (PRODAS; Conceptual Software 

 Inc. 1986). 



Results 



Variations in female body size and CPUE 



The size composition of the nearshore, adult female 

 queenfish stock differed among years. Mean female 

 length and weight were significantly lower in 1985 and 

 especially 1984 (Kruskal-Wallis one-way ANOVA; both 



Table 2 



Summary statistics for adult female queenfish Seriphus politus used in analyses 

 of reproductive variables and condition. Gonadal condition (RGI), batch fecun- 

 dity, and egg diameter variables are least-square means (adjusted for annual 

 differences in the body-size covariate). All variables, including fecundity, refer 

 to sample females only. See Methods for explanations of the covariate used 

 to adjust particular variables. One standard error is given in parentheses below 

 each mean. 



Estimates ( ± SE) of the exponent "b" in the power equation, gonad weight 

 (G) = aW b , were 1.185 ± 0.092 (in 1979), 1.162 ± 0.053 (1980), 1.232 ± 

 0.124 (1984), 1.280 ± 0.102 (1985), and 1.011 ± 0.102 (1986). Estimates did 

 not differ among years, and were consistently greater than one (ANCOVA; In 

 somatic W x Yr interaction: F 4383 = 0.87, P = 0.48; pooled slope = 1.1810. 

 Batch fecundity estimates were back-calculated from the means and SE's of 

 In-transformed data, multiplied by the correction factor of Sprugel (1983). 

 Using the largest size-class of oocytes (ripe, hydrated-state) present in Gilson's- 

 fixed ovaries (see Methods). 



P<0.001). Large females (> 165mm SL, chosen be- 

 cause they were relatively rare during 1984-86) in fact 

 were nearly absent in 1984, when overall mean female 

 abundance was at an estimated low (Table 1). 



Females used in analyses of reproduction and condi- 

 tion also differed in mean body size (both somatic 

 weight and length) among sample years (Kruskal- 

 Wallis ANOVA, both P< 0.001; Table 2). 



Dry vs. wet somatic weight 



The dry somatic weight of female queenfish aver- 

 aged 24% of wet weight for fish collected at the begin- 

 ning and at the end of the 1985 spawning season. 

 Body length obviously influenced dry weight; sub- 

 season, however, had no significant effect on dry 

 weight (ANCOVA of effects of body length and sub- 

 season on dry weight: length effect— F 154 = 346, 

 P<0.0001; subseason effect— F 154 = 0.42, P = 0.52). 

 Wet weight, therefore, could be used as an accurate 

 proxy for dry weight throughout the queenfish spawn- 

 ing season, once adjusted for variations in female body 

 size. 



