JONES: DIFFERENCES IN LARVAL I IKRHING GROWTH 



cantly different (P < 0.01)- early-hatched larvae 

 grew faster than late-hatched larvae. 



1978-79 Study 



A frequency plot of hatching dates under the 

 assumption of daily increment deposition for larvae 

 sampled in the Sheepscot estuary is shown in Figure 

 5. 



Analysis (Table 5) showed that the data were 

 better fit with two lines {P < 0.01) and that the 

 slopes were different (P < 0.01). The change in 

 length was 0.21 mm/increment and 0.18 mm/incre- 



ment for early- and late-hatched larvae, respectively 

 (Fig. 6). However, the results should be interpreted 

 with the knowledge that Bartletf s test showed the 

 variances to be heterogeneous. This could have been 

 caused by actual heterogeneity of variances, or by 

 nonnormality in the data. The F tests used in these 

 analyses assumed equal variances between the hatch 

 classifications. Cochran (1947) reported, however, 

 that lack of homogeneity would decrease the power 

 of an F test to discern true differences when they 

 did, in fact, occur. Since differences were statisti- 

 cally significant, not meeting this assumption did not 

 hinder analysis (the use of various transformations 



Table 5.— Regression analysis of 1978-79 Gulf of Maine Atlantic herring data. (Data fro 



send and Graham 1981.) 



im Town- 



Data were classified into early- and late-hatched larvae. These two groups were compared 

 by fitting ordinary least squares regression lines to 1) all the data within the two classifications, 

 and 2) using only lengths from larvae with 60 or fewer increment counts. Slopes and intercepts 

 were compared between early vs. late for each group. 



60 -n 



50 



E 



E 40H 



I 



O 



m 



30- 



20- 



)0 - 



LARVAL HERRING GROWTH 



1978-1979 



A A, 



W/ : 



TIME OF HATCH 



EARLY • 

 LATE A 



T 



20 



40 



80 100 



T 



60 80 100 120 140 160 



INCREMENT COUNTS 



~1 

 180 



Figure 6. - Regression plot of length-at-otolith increment count for Atlantic herring. Complete data set represented. Data from Townsend 



and Graham (1981). 



295 



