.lONKS: DIFKKKKNCKSIN LAIUAL HKKKINC OKOWl'H 



would have had to appear in the plankton in the mid- 

 dle of the summer (Fig. 3). However, newly hatched 

 larvae are not found in significant numbers in the 

 plankton before September (Boyar et al. 1973; Col- 

 ton et al. 1979). It is far more plausible that larvae 

 hatched early in the season, when growing condi- 

 tions are more nearly optimal (Sherman and Honey 

 1971; Cohen and Lough 1983), deposit increments 

 with close to daily periodicity. Hence, in order for 

 this hypothesis to be true, late-hatched larvae would 

 have to deposit increments at a rate greater than 1 

 increment/d. There is no evidence in the literature to 

 support this for larval herring. 



Difference in population growth rates within a 

 spawning season could also result from a shift in size- 

 specific mortality during the season. The observed 

 differences in growth rate could result if early-hatch- 

 ed larvae have higher cumulative mortalities for 

 slower growing individuals, while late-hatched larvae 

 have higher mortalities for faster growing indivi- 

 duals. Progressively, fewer and fewer of the selec- 

 tively predated larvae would be seen in older ages. 

 This would result in differences in population growth 

 rates that are not apparent for individuals within the 

 population. 



Although differential mortality cannot be dis- 

 missed with the available data, the most plausible 

 explanation for the differences in length-at-incre- 

 ment count is an actual difference in larval growth 

 rate over the spawning season. Such differences in 

 population growth rate can be important for larval 

 herring survival. Since greater time spent in the lar- 

 val stage is thought to be related to increased mor- 

 tality, it is interesting to note that an early-hatched 

 larva from the 1978 study would require, on the 

 average, 80 d to reach 30 mm, compared with 88 d 

 for a late-hatched larva. For the 1976 study, it would 

 take, on the average 63 d for an early-hatched larva 

 to reach 30 mm compared with 157 d for a late- 

 hatched larva to reach this size. 



It has been shown that in both years, late-hatched 

 larvae are larger than early-hatched larvae at the 

 time of first increment formation. This could result 

 from larger eggs being produced in the winter 

 (Cushing 1967), or from different growth rates from 

 hatch to the age of larvae covered in this study. 

 Without further evidence of differences in egg size 

 or actual growth rates between hatch and the age 

 these studies began, neither hypothesis can be sup- 

 ported. 



Differences in growth rate within the spawning 

 season can contribute to error when using an age- 

 length key to age larvae. For a given length, samples 

 containing early-hatched larvae would yield different 



ages than samples containing late-hatched larvae. 

 For the 1978-79 study (under the assumption of daily 

 increment deposition), a 25 mm larva would average 

 60 increments for early-hatched larvae versus 56 for 

 late-hatched larvae. For the 1976-77 study a larva of 

 this length would average 47 versus 102 increments, 

 respectively. This additional variation should be 

 taken into consideration when using age-length keys 

 for larvae. 



Differences in growth during the spawning season 

 might be due to changes in the environment when a 

 species of fish spawns over a protracted time period, 

 such as Atlantic herring which spawns from late 

 August through November (Boyar et al. 1973; Col- 

 ton et al. 1979). Early in the season copepods, the 

 main food for larval herring (Sherman and Honey 

 1971; Cohen and Lough 1983), are more abundant 

 than late in the spawning season (Sherman et al. 

 1983). Temperatures average 12°-16°C early in the 

 season and < 8°C later in the season (Colton 1968; 

 Colton and Stoddard 1972). Day length and 

 metabolic demand may also vary over the spawning 

 season. Alternately, differences in growth between 

 larvae hatched early and late in the season could be 

 the result of genetic differences if early and late 

 spawners are from different stocks. 



ACKNOWLEDGEMENTS 



I thank R. G. Lough, D. Townsend, and J. J. 

 Graham for providing their data; and B. E. Skud, R. 

 G. Lough, J. J. Graham, and K. R. Hinga for their 

 suggestions and review of this manuscript. 



LITERATURE CITED 



Beverton, R. J. H., AND S. J. Holt. 



19.54. On the dynamics of exploited fish populations. Fish. 

 Invest. Minist. Agric, Fish., Food (G.B.) Ser. II, Vol. XIX, 

 533 p. 

 Boyar. H. C, R. R. Marak, F. E. Perkins, and R. A. Clifford. 

 1973. Seasonal distribution and growth of larval herring 

 {Clupea harengns L.) in the Georges Bank-Gulf of Maine area 

 from 1962 to 1970. J. Cons. Int. Explor. Mer 35:36-51. 

 Cochran, W. G. 



1947. Some consequences when the assumptions for the ana- 

 lysis of variance are not satisfied. Biometrics 3:22-38. 

 Cohen, R. E., and R. G. Lough. 



1983. Prey field of larval herring Clupea harengus on a Con- 

 tinental Shelf spawning area. Mar. Ecol. Prog. Ser. 10:211- 

 222. 

 Colton, J. B., Jr. 



1968. Recent trends in subsurface temperatures in the Gulf of 

 Maine and contiguous waters. J. Fish. Res. Board Can. 25: 

 2427-2437. 

 Colton, J. B., Jr., and R. R. Stoddard. 



1972. Average monthly sea water temperatures, Nova Scotia 

 to Long Island, 1940-1959. Ser. Atlas Mar. Environ., Am. 



297 



