DeMartim et al Fecundity comparisons of Panuhrus margmatus 



F. = F.  ( 



w. 



where F s = number of eggs in subsample, W s = weight 

 of egg subsample, and W, = total weight of eggs. Some 

 frozen-thawed egg masses were fixed in 4% formalde- 

 hyde for 1 month to harden eggs prior to weighing and 

 counting. A single subsample was used to characterize 

 the fecundity of each "before" specimen. Three repli- 

 cate subsamples were used to estimate the sampling 

 error of "after" fecundity determinations; the three 

 pooled subsamples provided the best estimate of "af- 

 ter" fecundity. Total eggs were counted for one of the 

 "after" specimens to gauge the accuracy of the weigh- 

 ing and counting procedures. 



Egg sizes were estimated to complement the fecun- 

 dity data. For a subset of both "before" and "after" 

 Stage- 1 specimens, a minimum of 25 eggs per female 

 were randomly chosen and measured (random axis, at 

 50 X ) using a dissecting microscope with calibrated eye- 

 piece micrometer. 



Total egg production is the product of the number of 

 eggs produced per spawning (brood size) and the num- 

 ber of spawnings. For females above threshold body 

 sizes at onset of egg production at each of the sites 

 during the two time-periods (Polovina 1989), we in- 

 dexed spawning frequency based on the relative fre- 

 quencies of berried (to total) females present in his- 

 torical catch data of the Honolulu Laboratory. We used 

 records of catches made at Maro Reef and Necker Is- 

 land on summertime cruises during years within pre- 

 and postexploitation periods when sufficient data were 

 available (1977, 1988-91). 



Statistical analysis 



Analysis of covariance (ANCOVA, SAS Proc GLM; SAS 

 1985) was used to compare mean fecundities between 

 sampling periods; CL was used as a covariate to adjust 

 for potential body-size differences between periods. As 

 justified, central tendencies in fecundity were compared 

 between periods ("before," "after") using least-square 

 means (LSM) and their standard errors (SEM). Period 

 and site (Maro Reef, Necker Island) were evaluated as 

 class variables. Student's i-test, with degrees of freedom 

 adjusted (as necessary) by Satterthwaite's approxima- 

 tion for unequal variances (Bailey 1981), was used to 

 compare indices of spawning frequency between periods. 



Results 



Size-fecundity relationships 



Paired CL and fecundity data were available for 54 

 spiny lobster from Maro Reef. At Necker Island, there 



were 67 analogous data pairs (Appendix A). Over 90% 

 of the "before" specimens had Stage- 1 eggs, and Stage- 

 2 eggs were equally distributed among specimens from 

 the two sites. Incidence of Stage-2 eggs appeared higher 

 in the "after" samples from Maro Reef (8/24=33%) 

 than in the analogous samples from Necker Island 

 (3/32=10%). No lobsters with Stage-3 eggs were col- 

 lected during either time-period. The coefficient of 

 variation [CV=(SD/.r)-100] of the triplicate "after" fe- 

 cundity estimates was about 2%. The accuracy of the 

 mean of the two weighings of an entire egg mass was 

 within 4% of a total count. 



At Maro, slopes were indistinguishable between pe- 

 riods, regardless of whether an obvious outlier (whose 

 residual deviated 8.5% from its predicted value) was 

 included (In CL x period interaction: F 150 =0.06, 

 P=0.81) or was deleted from the analysis (F 149 <0.01, 

 P>0.99). Slopes also were indistinguishable between 

 the two periods at Necker Island (F 163 =1.17, P=0.28). 



Intercepts did not differ between periods at Maro 

 Reef (F 150 =0.22, P=0.64), but the period (intercept) ef- 

 fect at Necker Island was significant (F 164 = 10.17, 

 P=0.002). Greater size-specific fecundity in the "after" 

 period at Necker Island persisted, even if the five most- 

 extreme "before" values ( noted by the arrows in Fig. 1 ) 

 were deleted and the analysis rerun (F 159 =4.40, 

 P<0.05). Using all available data, the power (1 minus 

 Type-II error) of the test for period differences at Necker 

 Island was 84%, for a critical Type-I error of 5% 

 (a 2 =0.05). 



CL significantly influenced fecundity at both Maro 

 Reef (Fj 50=70.6, P<0.001; Fig. 1) and at Necker Island 

 (F 164 =215.6, P<0.001; Fig. 1). After adjustment for pe- 

 riod differences in CL, the fecundity of lobsters at 

 Necker Island was an estimated 16±9% greater dur- 

 ing the "after" versus the "before" period (LSM±SEM 

 of InF = 12.224±0.034 and 12.072±0.033, respectively). 

 Unlike the case at Necker Island, mean fecundity at 

 Maro Reef differed only by <3% between the "before" 

 and "after" periods (LSM±SEM = 12.680±0.041 and 

 12.651± 0.046, respectively). 



Egg size 



The median egg diameters of 22 females collected from 

 Maro Reef and Necker Island during the "before" pe- 

 riod were 0.58-0.69 mm. The analogous data for 53 

 "after" females were 0.61-0.73 mm, with a grand me- 

 dian of 0.66 mm. 



Slopes of female body size/egg size (median diam- 

 eter) relations were indistinguishable between sites 

 (CL X site interaction: F 151 =0.55, P=0.46). Intercepts 

 also were indistinguishable: egg size was uninfluenced 

 by site (F 152 =2.75, P=0.10). Carapace length had no 



