Alvarez and Alemany: Birthdate analysis and its application to the study of recruitment of Sardina pilchardus 
191 
ery (see Table 1; Robles et al., 1992). Thus, no cor- 
rections for cumulative mortality were made because 
the corrected birthdate distribution would be quite 
similar to the uncorrected distribution (Methot, 
1983). The most significant aspect of these results is 
that they indicate a period of birthdates outside the 
main period of larval production in the Cantabrian Sea. 
The relationship between the estimated age and 
length of the sampled recruits from Vigo and La 
Coruna are shown in Figure 3. In both cases, the 
relationship was linear, and significant (ANOVA, 
P<0.000). The slopes (£=1.26,P>0.10, df=213) and the 
intercepts (£=0.96, P>0.20, df=214) were not signifi- 
cantly different. Thus, a regression from pooled data 
was fitted (ANOVA, P<0.000). 
The precision of ageing within each sample and 
within each 1-cm length range was assessed with the 
calculated coefficient of variation, CV (standard de- 
viation divided by the the mean estimated age). The 
precision was good (CV<20%), stabilizing at a level 
of about 5-10% as fish grew in length (Fig. 4). The 
within-sample CV did not show any trend and re- 
mained at values less than 10% across all ages (Fig. 
5). These results suggest that the intrinsic variabil- 
ity that may exist between otoliths of different fish 
of the same length range is low and that the preci- 
sion of the age estimates is not affected by age. 
To assess seasonal changes in the estimated 
birthdate distribution, we performed a test to com- 
pare birthdate distribution from early samples 
(June-September, n= 89) with a distribution calcu- 
lated from late samples (October-November, n = 128). 
There was no significant difference (Kolmogorov- 
Smirnov test, P>0.05), which indicated that the 
samples came from the same cohort and that mor- 
tality during the period was not age selective. 
Discussion 
The juvenile birthdate distribution, which shows that 
the 1992 recruits were winter spawned, does not 
support the hypothesis that sardine recruits in the 
Galician area are mainly spawned during spring in 
the Cantabrian Sea, as was suggested by a previous 
study on the birthdate distribution of juveniles in 
this area (Alvarez and Butler, 1992). In fact, the sur- 
viving juveniles observed in our study were spawned 
earlier than the time when the sampling cruises were 
carried out in 1992. Thus, it was not possible to draw 
any detailed conclusions on the relationship between 
larval survival and environmental factors from 
otolith data. This was a significant obstacle for the 
achievement of the objectives of SARP because the 
“within year” exercise relies on a comparison of the 
birthdate distribution of juveniles with environmen- 
tal conditions that occur during their larval develop- 
ment (Bakun et al. 10 ). 
10 Bakun, A., J. Alheit, and G. Kullenberg. 1991. The sardine- 
anchovy recruitment project (SARP): rationale, design and 
development. ICES Council Meeting 1991/L:43, 17 p. (mimeo). 
