Lo et al.: Application of the continuous egg sampler to estimation of the daily egg production of Sardinops sagax 
559 
Distance (nmi, km) Distance (nmi, km) 
0.25 
0.20 
0.15 
o.io 
0.05 
0.0 
0 6 
0.4 
0.2 
0.0 
Figure 3 
Variogram of residuals of In (sardine e.ggs/minute+1) for age groups: 4-27 h (A), 28-51 h (B), and 
52-75 h (C), and total eggs for four directions (D). Degree 0 is the direction perpendicular to the 
coastal line, degree 90 is the direction along the coastal line during leg 2 of cruise 9603. On the 
x-axis, the inner ticks are in km and the outer ticks are in nmi. 
Paired samples were taken wherever high abundance 
of eggs appeared in samples collected with the CUFES. 
We obtained egg/minute = 0.73 eggs/0. 05m 2 (CV=0.16) 
(Fig. 5). This means that for one egg observed from a Cal- 
VET, one would expect to see, on average, 0.73 eggs/min. 
Or for one egg/min from a CUFES, one would expect to 
see 1.5 eggs/tow. A striking difference existed between the 
data from the 1996 pilot survey and the full survey car- 
ried out in 1997. In 1997, the catch ratio of eggs/minute to 
eggs/tows was 0.25 (CV=0.08) from 110 pairs of CalVET 
and CUFES of which at least one sample was positive 
(Fig. 5). This means that one egg/tow from a CalVET 
tow was equivalent to approximately 0.25 egg/min from a 
CUFES, or one egg/minute from the CUFES was equiva- 
lent to 4 eggs/tow from a CalVET sample. The ephemeral 
nature of such conversion coefficients was not known to us 
when we developed the design for the 1997 survey. How- 
ever, the variance associated with the direct 1996 conver- 
sions was a strong incentive to reduce the effects of direct 
conversions in the design of the 1997 survey and in the 
calculation of biomass. 
To determine if the CUFES provides an unbiased sam- 
ple of all sardine eggs stages, we compared the distribu- 
tions of developmental stages between the two samplers 
taken in 91 paired CUFES and CalVET samples during 
leg 1 of the 1996 survey (Fig. 1). A 2 x 11 contingency ta- 
ble was constructed for total counts of each of 11 stages 
of eggs collected with the CalVET and the CUFES. A chi- 
square statistic was computed to test the null hypothesis 
that the distribution of stages was independent of the 
samplers. 
The chi-square (% 2 ) analysis showed that the distribu- 
tion of stages was not the same between the two samplers 
(% 2 =188.47, df =10, P-value <0.01) (Table 2). The differ- 
ence was primarily due to eggs of stages I, III, V, and VI. 
The CUFES caught only two stage-I eggs; therefore we de- 
cided to run a % 2 test with stage-I and stage-II eggs col- 
lapsed into one group. A similar conclusion was reached 
for the later case <x 2 =160.31, df =9, P-value <0.01). A large 
X 2 value indicates that staged sardine eggs in the upper 3 
m may not be representative of the sardine egg stages in 
the whole-water-column and that the egg production com- 
puted from the CUFES survey would be biased. 
Design for the 1997 CUFES and DERM survey 
The foregoing analysis of the 1996 survey data indicated 
that adequate correlation exists in overall egg abundance 
between the CUFES and the full-water-column egg sam- 
ples, but a direct conversion of CUFES data to full-water- 
column abundance would add considerable variance to an 
estimate of daily egg production. The foregoing analysis 
