566 
Fishery Bulletin 99(4) 
Table 5 
Sardine daily egg production (P Q ) from a conventional survey (1994), compared with a CUFES and DEPM survey (1997). 
1994 Conventional DEPM survey 
1997 CUFES and DEPM 
survey 
Area: 380,175 km 2 
Area: 174,096 km 2 
CalVET samples 
CalVET samples 
total 
684 
total 
141 
positive for eggs 
72 
positive for eggs 
102 
positive percent 
11% 
percent positive 
72% 
CUFES samples 
None 
CUFES samples 
total 
1227 
total positive 
687 
percent positive 
56% 
high-density stratum 
84% 
low-density stratum 
40% 
Daily egg production 
Daily egg production 
0.169/0.05 m 2 
Po 
2.57/0.05 m 2 
CV 
0.22 
CV 
0.27 
spawning biomass 
111,493 t 
spawning biomass 
379,940 t 
of eggs provided by the CUFES have not as yet been incor- 
porated into the DEPM design, but we believe in the long- 
term it will be possible to improve the accuracy of surveys 
by doing so, as well as possible to develope new insights 
into the processes involved in selection of spawning habi- 
tats by parent fish. 
Comparison of shipboard egg counts with preserved 
egg counts 
A key element of the allocation design was that the allo- 
cation of CalVET samples was based on near real-time 
counts and on the identification of eggs by CUFES opera- 
tors. After having been counted by CUFES operators at 
sea, the eggs were preserved in vials, and later recounted 
and identified in the laboratory by experts. We compared 
egg counts in the laboratory with those taken at sea from 
the 1997 survey to determine the reliability of shipboard 
counting and identification and to indicate the extent to 
which a difference affected the final estimate of the daily 
egg production. 
Within each stratum, we first computed an overall mean 
eggs/min for the laboratory and ship count by using a ra- 
tio estimator (where y=total number of eggs and x=the to- 
tal minutes for each transect summed over all transects) 
(Table 6). We also computed a ratio of eggs/min from the 
laboratory to that from the ship for each transect line and 
obtained a weighted ratio for each stratum, where weight 
was the duration for each transect within a stratum (Eq. 
7). The ratio for the entire survey area was a weighted av- 
erage of two ratios, one from each stratum, and weight was 
the area of each stratum, as in Equation 5. 
In the high-density stratum, the ratio from the labora- 
tory counts to the ship counts was 1.19:1 (CV=0.03). In the 
low-density stratum, the ratio from the laboratory to the 
ship was 1.22:1 (CV=0.05) (Table 6). The overall ratio for 
the entire survey area was 1.20:1 (CV=0.10); therefore the 
laboratory count was higher than the ship count by 20% 
(Table 6). 
Of a total 1227 collections, 687 were positive according 
to laboratory counts. There were 16 collections where ship 
counts were positive but counts in the laboratory were zero. 
Eggs of other species in those 16 CUFES collections were 
obviously misidentified as sardine eggs. (Table 6). A total of 
658 pairs had positive counts, out of which 130 pairs had 
equal positive counts. A total of 524 pairs had zero counts. 
Therefore there were 654 equal counts between laboratory 
and ship (524 zeros and 130 positive counts) and 573 pairs 
( = 1227-654) with a mismatch. The relationship between the 
total counts from the laboratory and the ship confirmed the 
undercount from the ship (Fig. 9). The variance of under- 
counts increased with the total counts from the laboratory. 
One outlier was a collection where the ship count was zero 
and the laboratory count was 341. Although the absolute 
undercounts increased with the total number of eggs, the 
percent of undercount decreased with the total egg count. 
Conversion coefficient between two strata (g) 
In the 1997 CUFES and DEPM survey, one of the primary 
functions of the CUFES collections was to provide a con- 
version coefficient ( q ) of egg density between two strata 
to convert the egg production in the high-density stratum 
(P 0 j) to the low-density stratum (P 02 ), be. P 02 =P 01 x q 
(Eq. 6). The conversion coefficients ( q ) computed from the 
laboratory and ship counts were similar: 0.213 (CV=0.44) 
from the laboratory and 0.211 (CV=0.43) from the ship 
counts. Therefore, the bias of using egg counts at sea to 
calibrate egg production per unit area in the low-density 
stratum would be small: if the ratio of 0.213:1 were used, 
the egg production for the entire survey area would be 
2.60/0.05 m 2 instead of 2.57/0.05 m. 2 
