300 



Fishery Bulletin 89(2), 1991 



Embryogenesis was examined by observing embryos 

 throughout the developmental process. The time to 

 eclosion, interbrood period, and the effect of tempera- 

 ture on embryogenesis were examined in crabs main- 

 tained in aquaria with running seawater at normal 

 oceanic temperatures. Crabs maintained in aquaria 

 were fed market squid or mackerel weekly or biweek- 

 ly. For statistical analyses, broods were grouped on the 

 basis of the embryogenic development of attached 

 eggs; i.e., embryos were in early (I— II), middle (III— 

 IV), or late (V-VII) stages of development, or near 

 hatching (VIII). Roman numerals refer to the develop- 

 mental stages of embryogenesis (EDS) of Shields and 

 Kuris (1988a) and Shields et al. (1990). 



The term fecundity is here defined as the total 

 number of live eggs carried by each female at any given 

 time during incubation. Fecundity per pleopod was 

 estimated as in Shields et al. (1990); it represents the 

 number of live eggs on the 2d left pleopod. In addition, 

 the actual fecundity per crab was determined for 12 

 crabs (96 pleopods). The fecundities of other crabs were 

 then estimated using the regression of fecundity/ 

 pleopod with the fecundity of the 12 crabs. 



Statistical analyses (ANOVA, ANCOVA, linear 

 regression, Sidak's inequality) were conducted with the 

 aid of SAS (1982). The log-transformation was used to 

 reduce differences in variance between groups. A value 

 of P<0.05 was accepted as significant. Data from all 

 of the 2d left pleopods were statistically independent. 

 Two statistics were used to minimize the influence of 

 outliers on the analysis of covariance of log fecundity 

 and log size between seasons, and between EDS 



N = 341 



<90 100 110 120 130 140 150 150+ 

 Carapace width 



Figure 1 



Size-frequency histogram of ovigerous Cancer anthonyi 

 by 10-mm increments. 



groups. An outlier was removed from the ANCOVA 

 and the subsequent regression analyses only if the 

 value of its studentized residual was ± 1.50, the value 

 of its Cooke's D influence statistic was greater than 

 0.006 (SAS 1982), and only if these statistics were con- 

 sistent between transformed and untransformed data 

 (5 outliers in early and middle embryogenesis). We sug- 

 gest that the 5 outliers were bearing a second or third 

 brood between molting and mating events, hence their 

 fecundity was low (see below). 



