FISHERY BULLETIN: VOL. 84, NO. 4 



variations in Hawaii, as well as evidence from the 

 Mariana Archipelago which shows that populations 

 of if. laevigatus are affected to a greater degree by 

 temporal factors than geographic ones (Polovina 7 ). 



Morphometries 



The results of performing functional regressions 

 (Ricker 1973) on the three linear size measurements 

 (CL, CW, TL) are given in Table 2. Estimates of 

 slope and intercept are provided for all possible per- 

 mutations of these variables. Separate analyses for 

 males (M), females without eggs (F0), and females 

 with eggs (FE) were not performed because all have 

 similar gross morphologies (but see King and Mof- 

 fitt 1984). As expected, the data were well described 

 with a linear fit. 



The relationship between weight and CL was ex- 

 amined by analysis of covariance (BMDP 1977) to 

 determine whether the M, F0, and FE subgroups 

 have different weight-length relationships. Results 

 showed all three were characterized by differing 

 slopes in the regression of log e (weight) on \og e (CL) 

 (F = 86.46, df = 2, 5912, P« 0.0001). Parameter 

 estimates with standard errors and other regression 

 statistics are presented in Table 3 for each of the 

 three subgroups. Note that the reduced r 2 of the 



7 J. J. Polovina, Southwest Fisheries Center Honolulu Labora- 

 tory, National Marine Fisheries Service, NOAA, 2570 Dole Street, 

 Honolulu, NI 96822-2396, pers. commun. June 1985. 



FE group is due to a substantial reduction in the 

 range of CL over which the data were fitted. The 

 results of performing functional regressions of log e 

 (weight) on log e (CL) are also given. 



During the analysis an anomalous bimodal distri- 

 bution of weight at length emerged. The bimodality 

 was not due to sexual class (M, F0, or FE) and 

 clearly diminished to a unimodal weight distribution 

 as CL increased from 15 to 40 mm. We have no ex- 

 planation for these data. 



Reproductive Biology 



The reproductive season of H. laevigatus was esti- 

 mated by plotting the percentage of ovigerous 

 females relative to total females against the month 

 sampled. For the data which overlapped 1983 and 

 1984 no interannual difference was evident (i.e., the 

 timing of reproduction was similar), so the data were 

 pooled by month between these years. The results 

 are presented in Figure 2 where the data have been 

 further aggregated into 2-mo "seasons". For each 

 the percentage of females bearing eggs is plotted 

 with its 95% confidence interval and associated sam- 

 ple size given above. 



The data show an increased incidence of ovigerous 

 females from August to February (>30% of 

 females). In particular over 50% of all sampled 

 females carried eggs from October to January. 

 Relatively few shrimp were caught with eggs dur- 

 ing the period from April to July (<10%). Moreover, 



Table 2. — Parameter estimates of functional regressions on linear size 

 measurements. All measurements in millimeters and all sample sizes n 

 = 5,920. 



Table 3. — Functional and predictive length-weight regressions for Heterocarpus 

 laevigatus. The natural logarithm of weight in grams is fitted to the natural logarithm 

 of carapace length in mm. The standard errors of the slope (b) and intercept (a) are 

 given by S b and S a respectively. 



Slope Intercept S b S a n r 2 



Males 



Females 

 without eggs 



Females 

 with eggs 



Predictive 

 Functional 



Predictive 

 Functional 



Predictive 

 Functional 



2.755 

 2.910 



2.605 

 2.745 



1.815 

 2.470 



- 6.809 

 -7.358 



-6.252 

 -6.757 



-2.986 

 -5.498 



0.0176 0.0629 2,788 0.8976 



0.0185 0.0671 2,202 0.8999 



0.0550 0.2114 928 0.5401 



918 



