Results and Discussion 



Curvilinear size-fecundity relationships de- 

 rived from log-log (base 10) regression analysis for 

 spring-caught (April- June) samples are presented 

 for three areas in Placentia Bay, along with the 

 same relationships obtained from reanalysis of the 

 data presented by Squires (1970) and Squires et al. 

 (1974) for two Newfoundland west coast areas 

 (Figure 2, Table 1). The methodology of Squires 

 (1970) and Squires et al. (1974) was similar to that 

 used here except that the estimates were deter- 

 mined volumetrically instead of by weighing. 

 These estimates varied from actual counts by 

 <2% which is within the range of error deter- 

 mined in this study For this reason the following 

 statistical comparisons between the Placentia Bay 

 and west coast samples are considered valid. 



An Fmax test on the log-transformed data dem- 

 onstrated significantly different (P<0.01) re- 

 sidual variances among the five samples. Various 

 combinations of the log-log relationships were 

 compared by analysis of covariance. In all com- 

 parisons between relationships for Placentia Bay 

 and the comparison between the two relationships 

 for the west coast, residual variances were similar. 

 However, in all comparisons between one of the 

 Placentia Bay and one of the west coast relation- 

 ships, residual variances were significantly differ- 

 ent (Table 2). Two of the four sets of relationships 



Table l. — Regression equations from which the curvilinear 

 relationships shown in Figure 1 were derived. 



Table 2. — Results of analyses of covariance of size-fecundity 

 relationships presented in Figure L 



70 



75 



BO 



85 



90 95 100 105 110 

 CARAPACE LENGTH (MMI 



115 



120 125 130 



FIGURE 2.— Carapace length-fecundity relationships for 

 American lobsters from three areas in Placentia Bay, Newfound- 

 land and two areas on the west coast of Newfoundland. Dashed 

 lines indicate extrapolations beyond the data. 



with similar residual variances had significantly 

 different slopes, the other two had similar slopes 

 but significantly different means. There was wide 

 variation in fecundity at size and the samples dif- 

 fered in size composition (Table 3). Significant dif- 

 ferences in these relationships may result in large 

 part from differences in sample size and size com- 

 position. 



Samples with at least six specimens in the same 

 5 mm size group were compared by analysis of 

 variance. There is some size-related variation in 

 fecundity within each 5 mm size group. This could 

 confound a comparison if, in the samples being 

 compared, there is clustering of specimens at op- 

 posite ends of the size range. Upon examination it 

 was found that for each sample compared in Table 

 4 the specimens were fairly well distributed 

 throughout the 5 mm size range and it is assumed 

 that size-related variation does not invalidate 



798 



