GUILLEMOT ET AL.: SEASONAL CYCLES IN CALIFORNIA ROCKFISH 



allows the use of standard least-squares regression 

 techniques, and stiihilizes variances (Bagenal 1978). 

 Thus, rather than using ratios of fat or gonad volume 

 to fish weight or length, which can be biased by allo- 

 metric relationships between the measurements 

 (Bagenal 1978; de Vlaming et al. 1982), we employed 

 empirically derived regression lines. The use of 

 power equations also allowed us to test hypotheses 

 with standard statistical procedures. Regressions 

 were calculated for each sex during each season. 

 Although most of the analyses were for adults, 

 seasonal regressions of visceral fat volume were also 

 computed for juveniles when samples were large. 



Differences between regressions of fat or gonad 

 volume were tested by analysis of covariance, using 

 the ratio: 



F = 



{N - 4) (SS,„^. - SSi - SS^) 

 2 (SS, + SSg) 



where SS, and SSo are the residual sums of squares 

 about the separate regressions, SS^.^.^, is the residual 

 sum of squares for a common regression, and A'' is 

 the number of specimens (Kleinbaum and Kupper 

 1978). When regressions differed significantly, the 

 direction of differences was assessed by examining 

 the positions of the regression lines and of calculated 

 volumes at mean fish length. We chose to display 

 only the calculated volumes at mean length, to 

 simplify the presentation of data. The mean of In 

 (total length) was determined for all specimens of a 

 sex and species included in the annual analysis. For 

 each season and sex, the estimated value of In 

 (volume) at the mean of In (length) (designated Y) 

 was calculated from the appropriate regression. We 

 intend these point values to be used only for making 

 comparisons within our data. The regression lines 

 themselves best represent fat or gonad volume, and 

 we present the parameters for these regressions. 

 See Guillemot (1982) for scatter diagrams of raw 

 data and regression plots. 



The estimated values of fat volume at mean length 

 were also used in an index of the relative amplitude 

 of fat cycles. This index was the antilog of the differ- 

 ence between estimated In (fat volume) in peak and 

 low seasons, and is equivalent to the ratio of peak 

 season -.low season fat volume. 



We compared fat cycles of males and females 

 within a species by qualitatively noting differences in 

 the timing and amplitude of fat cycles, and by statis- 

 tically comparing fat regressions during peak and 

 low seasons of fat volume. The latter comparisons 

 were carried out by analysis of covariance, as above. 



In comparing fat cycles among species, we noted 

 differences in the timing of fat cycles and differences 

 in the amplitude of cycles. We also assessed dif- 

 ferences in the absolute volume of fat stored by 

 applying analyses of covariance to the peak-season 

 fat regressions of the different species. 



RESULTS 



Seasonal Cycles of Fat and 

 Gonad Volume in Adults 



In most seasons, moderate to low correlations (0.5 

 and below) existed between In (visceral fat volume) 

 and In (total length), indicating a fair amount of scat- 

 ter about the regressions (Table 1). Lower correla- 

 tions occurred mainly in seasons of low fat volume 

 (cf. Fig. 1). The slopes of most regressions were not 

 significantly different from 3.0 (Table 1), indicating a 

 proportionate relationship between fat volume and 

 fish length. Slopes significantly smaller than 3 

 occurred in seasons of low fat volume (winter and 

 spring), and in general slopes from seasons of low fat 

 volume were smaller than those from seasons of 

 higher fat volume. During seasons of low fat volume, 

 fish of all sizes tended to have little fat, accounting in 

 part for some of the lower correlations. Some slopes 

 were significantly greater than 3 during seasons of 

 high fat content (5. entamelas females, 5. paucis- 

 pinis males, S. pinniger females, and S. Jlavidus 

 females), indicating disproportionately greater fat 

 content in larger fish. Correlations between In 

 (gonad volume) and In (total length) were generally 

 high (Table 2), and showed no marked relation with 

 season (Fig. 2). Below, we first examine the fat and 

 gonad cycles for adults of each species separately, 

 and then briefly compare the cycles of different 

 species. 



Sebastes entomelas 



Visceral fat content changed significantly during 

 the year in both male and female S. entomelaf; (Fig. 

 1). In females, where all seasonal regressions of In 

 (visceral fat volume) on In (total length) differed 

 significantly from each other, fat content increased 

 from a low in spring to a peak in fall, and then declin- 

 ed in winter (Fig. 1). In males, neither the spring and 

 winter nor the summer and fall fat regressions dif- 

 fered significantly (Fig. 1). Visceral fat volume in 

 males increased after spring to a peak spread 

 through summer and fall, and then declined in winter 

 (Fig. 1). During the fall, when both sexes had large 

 volumes of visceral fat, fat content of males barely 



301 



