768 
Fishery Bulletin 95(4), 1997 
lar results to those using the Ricker model (Table 2; 
Fig. 9). We conclude that our results are robust in 
relation to the method used to estimate a . 
Discussion 
Perhaps remarkably, our study has revealed that the 
allometric (cross species) approximate inverse pro- 
portionality between r m and age-at-maturity holds 
within the species Atlantic cod. Despite the much 
narrower range of r m used in our single-species com- 
parison (in contrast to the wide variation in r m found 
in cross-species studies), the relation between r m and 
age-at-maturity prevails over other influences. The 
expectation that cod populations at the northern and 
southern extremes of their range should show higher 
resilience (r m ), because of greater susceptibility to 
environmental change (Myers, 1991), is not realized. 
In a similar vein, Roff (1984) suggested that early 
maturity in fish species may arise through r-selec- 
tion (in response to extreme environmental variabil- 
ity); our findings show that age-at-maturity appears 
to be chiefly explained by ambient temperature. 
Although we have found a clear relation between 
r m and temperature, this was not necessarily ex- 
pected a priori because mortality is positively related 
to temperature in comparative studies (Pauly, 1980). 
That temperature dependent (egg to adult) mortal- 
ity can offset the effect of temperature-dependent 
growth is emphasized by the temperature indepen- 
dence of a . More specifically, a depends on both fe- 
cundity and mortality. Fecundity, being 
growth dependent (Roff, 1984) increases with 
temperature; however, mortality also in- 
creases with temperature and counteracts the 
influence of temperature-dependent growth, 
leaving a temperature independent. (Note 
that many empirical studies of life histories 
of fish have found that somatic growth rate 
and survival covary [Beverton and Holt, 1959; 
Pauly, 1980; Myers and Doyle, 1983; Hut- 
chings, 1993]). 
The relation between r m and temperature 
is presumably a metabolic effect, in that fish 
growth is strongly influenced by temperature 
(Taylor, 1958; Pauly, 1980), and implies that 
a fish in a warm environment will reach the 
required size for maturity at an early age, 
which tends to increase r . Although Birch 
(1948) has noted, for insect populations, that 
higher r m values do not necessarily corre- 
spond to higher temperatures, investigators 
such as Hennemann (1983) and McNab 
(1980) have suggested that r m should be 
closely related to metabolic rate, which is 
closely linked to temperature. Certainly our 
presentation corroborates this proposed par- 
allel between metabolism and r m . 
The importance of the determination of r m 
has long been known (Lewontin, 1965); how- 
ever, it is certainly not always the case that 
age-at-maturity is the dominant factor. For 
species for which the production of replace- 
ment adults at low population density, a , is 
relatively low (e.g. for many mammals and 
birds), then changes in a or adult survival 
will have large effects on r m (Fig. 1). How- 
ever, for cod, and perhaps many fish, a is 
relatively large (e.g. around 4). In this case, 
the effects of reasonable changes in adult 
Figure 4 
Estimates of slope at the origin ( a ) and approximate 95% confidence 
limits for 20 cod populations in the North Atlantic estimated from 
the Ricker model. 
