562 
Fishery Bulletin 99(4) 
ranging from 1 to 54 minutes with a mean of 
30 minutes and median of 15 minutes (Table 
1). Only samples from CalVETs were used to 
model the sardine embryonic mortality curve, 
thereby avoiding a possible stage-specific bias 
with the CUFES. 
Spacing of transect lines was also deter- 
mined to some extent by samples from the 
CUFES. The scheduled spacing of lines was 40 
nmi apart but if a high density of sardine eggs 
was encountered, an additional transect line 
was added between the two scheduled lines. As 
a result of this procedure, two lines were add- 
ed (lines 4 and 8) in the 1997 DEPM survey 
(Fig. 2). 
To estimate P 0 , the survey area was then 
stratified into two strata: a high-density stra- 
tum where the density of eggs taken by the 
CUFES was at least 2 eggs/min; and a low- 
density stratum (Fig. 2) where only CalVET 
samples were taken and where the density of 
eggs was less than 2 eggs per minute. Within 
each stratum, the transect line was selected 
as the sampling unit both for CalVET samples 
and for CUFES samples. Because lines were 
spaced at intervals greater than 12 nmi, statistical pro- 
cedures based on uncorrelated data were used (correla- 
tion should not be ignored if the distances are less than 
12 nmi). We describe below the two different estimation 
methods used to estimate P 0 : one for the high-density stra- 
tum and the other for the low-density stratum; we com- 
bined the two methods to estimate P 0 combined for the to- 
tal survey area. 
We used transects as the primary sampling units and 
applied conventional statistical methods to construct the 
mortality curve from CalVET samples (Armstrong, 1988). 
CalVETs, the secondary sampling units, were taken for 
the first ten transects, except transect number eight. The 
P 0 was estimated for each of two strata by using different 
methods, and an weighted average was obtained for the 
whole survey. 
Table 3 
Daily sardine egg and yolksac larval production and their 
standard error in the high-density stratum in 1997. 
Eggs 
Yolksac larvae 
Age (day) 
Daily production/ 
0.05 m 2 
Standard 
error 
0.32 
3.29 
1.32 
0.76 
4.48 
2.10 
1.17 
4.32 
2.05 
1.71 
6.12 
2.88 
1.92 
3.21 
1.60 
2.52 
1.86 
0.66 
5.62 
0.65 
0.16 
Egg production in the high-density stratum (/* 01 ) 
Ages of staged sardine eggs (Fig. 7) were assigned accord- 
ing to a temperature-dependent time-to-stage model (Lo 
et al., 1996). Sardine eggs were grouped in half-day cat- 
egories excluding those eggs that were newly spawned 
during the first three hours. An average eggs-per-tow and 
an average age were obtained for each half-day category 
within each transect. A weighted average number of eggs 
in each half-day category was obtained where the weight 
is duration of the CUFES sampling interval for each tran- 
sect line (Eq. 4, Table 3). 
p t 
Zj 171 - 
var (P t ) - 
n / (n- l)^ 1 mf(P l - P, ) 2 
(4) 
where P lt = the eggs or yolksac larvae/0.05 m 2 ; and 
m i = the total CUFES sampling time (minutes) 
for the ith transect for i = 1,..., 7, 9, and 10. 
For a weighted nonlinear regression, the weight is l/SE(p f ). 
The yolksac larval stage includes larvae from time of 
hatching to the time they fonn functional jaws, i.e. larvae 
<6 mm live length or 5 mm captured length (Zweifel and 
Lasker, 1976). As with the egg data, the average number 
of yolksac larvae per tow was calculated for each transect 
and the overall mean catch of yolksac larvae was comput- 
ed as a weighted average. Yolksac larva production was 
the mean catch of larvae <5.00 mmi (captured size) per 
CalVET divided by its temperature-dependent stage dura- 
tion (Zweifer and Lasker, 1976). The age of yolksac larvae 
was estimated by the average of minimum age and maxi- 
mum age, i.e. hatching time + age of forming functional 
jaw)/2 (Lo et al., 1996) (Table 3). 
