PARSONS and TUCKER: FECUNDITY OF NORTHERN SHRIMP 



Table 1 .—Regression equations for fecundity (F) vs. length (L) for Pandalus borealis in the Northwest Atlantic. 



Temp. 

 Date N Regression equation r 2 °C 



Sample 



and +5.65%. The difference between the total 

 number of eggs counted and recounted was only 

 - 0.22% of the initial count. A recount of eggs from 

 one female indicated a difference of -9.38%. It is 

 possible that, in this case, some of the eggs were 

 inadvertently lost between counts. 



Parameters for the relationship between number 

 of eggs and carapace length for each sample were 

 determined by linear regression using log-log (base 

 10) transformation. Some data sets were compared 

 by analysis of covariance, assuming homoscedas- 

 ticity. All statistical analyses were performed using 

 the REG (regression) and GLM (general linear 

 models) procedures of SAS (Statistical Analysis 

 System). 



It must be stressed that samples were obtained 

 opportunistically and not according to a predeter- 

 mined sampling design. Consequently, the statistical 

 analyses were performed based on a practical ap- 

 proach rather than attempting methods for which 

 strict sampling procedures are required. It was anti- 

 cipated that differences in fecundity-length relation- 

 ships could be due to seasonal, annual, and areal 

 effects. Our data only permitted simple compari- 

 sons, investigating each factor separately. 



Bottom temperatures at most sample locations 

 were recorded to the nearest 0.1 °C using either 

 manual or expendable bathythermographs. 



RESULTS 



The parameters of the fecundity-carapace length 

 relationships for all 15 samples are given in Table 



Figure 1.— Positions of stations in the northwest Atlantic where 

 northern shrimp fecundity samples were collected. 



1. Data and the fitted line for each sample are 

 displayed in Figure 2. Coefficients of determination 

 ranged from 0.31 to 0.78 and all relationships were 

 significant (differences from zero slope were highly 

 significant). Intercepts for the log transformed data 

 were less than zero in all but one case. Slopes ranged 

 from 2.4 to 3.9 except for the sample with positive 

 intercept (1.5). 



Only two samples were available (Hawke Chan- 

 nel, August and November 1974) for comparison of 

 fecundity between seasons. Analysis of covariance 

 on the log of both variables indicated a highly sig- 

 nificant difference in slopes (Table 2). The data 

 showed that larger females (>24 mm), on average, 

 carried more eggs in August whereas smaller 

 females showed higher fecundity in November (Fig. 

 3). 



Samples from specific areas and seasons were 

 compared to determine similarities or differences 

 between years. Five simple comparisons were possi- 

 ble: St. Mary's Bay - March 1971 vs. February 1974, 

 Hawke Channel - August 1974 vs. September 1975, 

 Fortune Bay - March 1978 vs. March 1979, Cart- 

 wright Channel - September 1978 vs. September 

 1982, and Hopedale Channel - September 1978 vs. 

 September 1982. 



No significant differences in either the rate of in- 

 crease in fecundity with increasing size (slope) or 

 mean number of eggs produced (intercept) were 

 detected between years in three of the five areas 

 compared (Table 2). These were St. Mary's Bay, 

 1971 and 1974; Hawke Channel, 1974 and 1975; and 

 Fortune Bay, 1978 and 1979 (Fig. 4a, b, and c, 

 respectively). Samples from Cartwright Channel 

 from September 1978 and 1982 showed a significant 

 difference in slopes at a = 0.05 (Fig. 4d) whereas 

 samples from the Hopedale Channel for the same 



551 



