666 



Fishery Bulletin 91(4), 1993 



maturity (Hartnoll, 1978). We used MATURE1, the 

 computer method described by Somerton (1980) to fit 

 a pair of intersecting straight lines to plots of chela 

 height on carapace length. The method determines the 

 regression lines for crabs that are assumed to be im- 

 mature and mature, based on minimum observed size 

 at gonadal maturity and maximum size at gonadal 

 immaturity, respectively. Then it extends the lines to 

 the central area of the plot, where immature and ma- 

 ture crab are mixed, by iteratively assigning each point 

 to either line, then recalculating lines, until no points 

 switch lines on two successive iterations. On the basis 

 of this classification, the proportion of mature indi- 

 viduals was calculated for CL-size intervals of 5 mm 

 in males and of 4 mm in females and a logistic func- 

 tion was adjusted to provide 50% maturity (Wenner et 

 al., 1974). Measurements of partially regenerated 

 chelae were excluded from this analysis. Data for crabs 

 <50 mm CL were obtained from a study of juvenile 

 growth (Lovrich, 1991). 



Fecundity was defined as the number of eggs per 

 clutch. In the laboratory, pleopods with attached eggs 

 were removed from each female and preserved in buff- 

 ered 10% formalin in seawater. Later the eggs were 

 detached from the pleopods and the clutch was blotted 

 and weighed to the nearest 0.01 g (WC). Three sub- 

 samples were then weighed to the nearest 0.01 g (ws) 

 and eggs in each subsample were counted (ns). Fecun- 

 dity (F) was calculated as: 



F = H[(.WC*ns)/ws]/3 



Estimates of fecundity based on counts of three 

 subsamples did not vary by more than 5%. 



Egg characteristics, color, developmental stage, pres- 

 ence of chromatophores and appendage development 

 were determined monthly with the aid of a stereo- 

 scopic microscope. Fifteen P. granulosa females were 

 kept in a 600-L and 250-L tank in a controlled-envi- 

 ronment room with photoperiod, temperature, and sa- 

 linity adjusted to natural environmental variations 

 from April 1990 to May 1991. Crabs were fed ad libi- 

 tum with limpets, mussels, and fish 2-3 times a week. 

 A subsample of 15-20 eggs was taken from each egg 

 mass at weekly intervals at the beginning of the study 

 and at monthly intervals thereafter. Eggs from females 

 kept in the laboratory and eggs preserved in formalin 

 were separated, and the maximum diameter of each 

 egg was measured to the nearest 0.01mm with an 

 ocular micrometer. 



Ovaries were removed and weighed to the nearest 

 0.01 g. Two portions of the ovary were sampled: one 



was stored in buffered 10% formalin in seawater and 

 the other was fixed in Bouin's solution and later trans- 

 ferred to 70% ethanol. The latter portion was then 

 embedded in paraffin, sectioned at 8-10 urn and stained 

 with "one time" trichrome (Gabe, 1958). Ovarian de- 

 velopment was described in two ways: 1) the go- 

 nadosomatic index was the ratio weight of ovaries : 

 total body weight, multiplied by 100; 2) mean diam- 

 eter of oocytes was determined from measurements of 

 30 to 40 of the roundest oocytes from the periphery of 

 the formalin-preserved ovary by using an eyepiece mi- 

 crometer on a compound microscope. 



Ten to 12 males were dissected bimonthly to check 

 for the presence of spermatophores in the vas defer- 

 ens. The right vas deferens was stored in 10% forma- 

 lin and later pressed between a slide and coverslip to 

 detect spermatophores under a microscope. 



Data were log 10 -transformed to obtain normality and 

 to reduce heteroscedasticity. Simple correlations and 

 predictive regressions were then calculated for selected 

 life history traits. Outliers were detected by compar- 

 ing the values of the standardized residuals from the 

 regression line to Student-/ tabulated values. Slopes 

 and elevations of regressions were compared by analy- 

 ses of variance (ANOVA) and of covariance (ANCOVA), 

 respectively (Sokal and Rohlf, 1981). 



Discriminant analysis was used to determine the 

 existence of two groups in ovarian development data. 

 We then classified each crab into either group on the 

 basis of an objective decision rule (Morrison, 1976). 

 Variation in oocyte diameter and gonadosomatic index 

 were positively correlated with ovary size. Thus, one 

 of the assumptions of discriminant analysis was not 

 met since variance-covariance matrices were not sta- 

 tistically equal even though data were transformed. 

 The discriminant function was used to score individual 

 females. The Games and Howell test (Sokal and Rohlf, 

 1981) was then used to contrast groups identified by 

 the discriminant analysis. 



Results 



Gonadal and morphometric maturity 



Fifty percent gonadal maturity occurred at 50.2-mm CL 

 in males and at 60.6-mm CL (95% confidence limits: 

 58.3-62.9) in females (Fig. 2A). Even though the larg- 

 est females did not carry eggs, we assumed they were 

 mature because their ovaries were normally developed. 

 The smallest male with spermatophores was 49.1mm 

 CL, whereas the largest one without spermatophores 

 was 70.2 mm CL. The smallest female carrying eggs 

 was 59.7 mm CL and the largest one without eggs and 

 without developed ovaries was 75.1 mm CL. 



