McQuinn: Year-class twinning in sympatric spawning populations of Ctupea harengus 
133 
of an autumn-hatched cohort and that of the spring- 
hatched cohort of the following year, from hatching 
through metamorphosis (Fig. 1), it is apparent that 
at no time are these two cohorts in the larval stage 
at the same time, i.e. the autumn spawners have 
metamorphosed before the spring spawners of the 
following year have hatched. Autumn-spawned her- 
ring in the northwest Atlantic hatch from August to 
November, remain as larvae throughout the winter 
(lies and Sinclair, 1982), and metamorphose within 
a “window” between March and May (Sinclair and 
Tremblay, 1984 ). Larvae hatched in June or July from 
the spring-spawning event reach the required size 
for metamorphosis during September or October of 
their first year. Conditions affecting larval survival 
would therefore have to be favorable from Septem- 
ber of one year to June of the next, but unfavorable 
between July and August for Einarsson’s explana- 
tion to be credible. 
Einarsson (1952) speculated that a strong stand- 
ing stock of copepods in the autumn of one year may 
be correlated with enhanced copepod egg production 
the following spring, thus favoring larval herring 
survival over this extended period, although the data 
available to him did not show this correlation. In 
addition, these enhanced survival conditions would 
not only have to exist over a long, but nonetheless 
precise period of time (September-June), but would 
also have to be extremely widespread. Twinning oc- 
curs in most if not all sympatric herring populations 
in the northwest Atlantic (de Lafontaine et al., 1991) 
and sympatric spring- and autumn-spawners do not 
necessarily use the same breeding locations nor the 
same larval retention mechanisms, i.e. spring and 
autumn spawners along the west coast of Newfound- 
land (McQuinn and Lefebvre 4 ). 
The present study supports the alternative hypoth- 
esis that the twinning of year classes can be explained 
by the crossover or straying of a significant number 
of individuals from one seasonal-spawning popula- 
tion to the other. Our results also support the hy- 
pothesis of Jean 5 , Graham (1962) and Winters et al. 
(1986) that variable growth rates in the juvenile 
phase lead to this crossover. Results from several 
