Demer et al.: Seasonal migration of Sardmops sagax in the California Current Ecosystem 
59 
TS= -14.90xlog(TL)-13.21, for sardine; (3) 
TS- -12.15xlog(TL)-21.12, for anchovy; and (4) 
TS= -15.44xlog(TL)-7.75, for jack and (5) 
Pacific mackerel. 
where TL is in cm. These relationships were origi- 
nally estimated for anchovy ( Engraulis capensis), 
sardine ( Sardinops ocellatus-Sardinops sagax), and 
horse mackerel ( Trachurus trachurus ), on the basis 
of a combination of backscatter-versus-length and 
mass-versus-length measurements of in situ fish 
(Barange et ah, 1996). Because jack mackerel and 
Pacific mackerel have similar TS values (Pena, 2008), 
Equation 5 was used for both of these species. TL 
values of fish are derived from their measured SL 
or FL values by using linear relationships derived 
from measurements of CCE specimens: sardine, 
TL = 0. 3574 +1.149SL; anchovy, TL = 0. 2056 + 1. 1646SL; 
Pacific mackerel, TL = 0. 2994+1. 092TL; and jack mack- 
erel TL = 0.7295+1.078FL. 
Biomass and uncertainty estimation The s A values were 
converted to fish biomass density for the i species (p-; kg 
nmi -2 ) by using the following equation: 
Pi = 
4 k 10 (( ts .)M' 
( 6 ) 
Total biomass was calculated, by species, for strata 
having similar biomass densities and transect spac- 
ing. The mean biomass density of each stratum was 
calculated by a transect-length weighted average of the 
transect mean densities (Jolly and Hampton, 1990). 
During the summer 2008 survey, there was evidence 
of increasing biomass densities toward the coast, sug- 
gesting that the mean sardine biomass density calcu- 
lated for each transect did not account for the biomass 
in small coastal region between the end of the transects 
and the coastline. Therefore, a coastal stratum was cre- 
ated, and its mean biomass density was estimated as 
that measured in the transects from their inshore ends 
to 10 nmi offshore. 
The sampling variances and confidence intervals were 
estimated by using bootstrapping because it provides 
better statistical inference than do traditional methods 
for data with unknown statistical distributions and 
small sample sizes (Efron, 1981). The 95% confidence 
intervals for the mean biomass densities were esti- 
mated as the 0.025 and 0.975 quantiles of the distribu- 
tion of 1000 bootstrap survey-mean biomass densities. 
Coefficient of variation (CV) values were obtained by 
dividing the bootstrapped standard errors by the boot- 
strapped arithmetic means (Efron, 1981). Provided that 
statistical independence exists between the transects, 
bootstrap resampling of the transect means provides 
unbiased estimates of the variance for the survey mean, 
even for several levels of random variability nested 
(e.g., small-scale spatial sampling correlation or sparse 
trawl-derived TS estimation) at the intra-transect level 
(Williams, 2000). 
To evaluate the proportion of the sampling variance 
pertaining to species classification and TS estimation, 
the trawl samples with CPS were subjected to jackknife 
resampling. The jackknife procedure was performed by 
omitting one trawl sample per iteration. The variance 
was estimated by calculating the variance of the jack- 
knife means, corrected by the number of trawls in each 
stratum as per Efron and Tibshirani (1993). Each time 
a trawl was removed from the set, the biomass densi- 
ties of each target species in 100-m distance cells were 
recalculated, taking into consideration the new nearest- 
neighbor configuration (Fig. 2). 
Results 
During both the spring and summer surveys, the dis- 
tributions of echosounder- and trawl-sampled CPS 
were reasonably well matched (Fig. 2). Also, sardine 
were the most common species, in terms of their occur- 
rences in catches with CPS (Table 2). Excluding two 
large catches of anchovy, sardine were also the most 
abundant species in terms of total-catch mass. The 
next most abundant species in both surveys was jack 
mackerel. Anchovy and Pacific mackerel were caught in 
roughly the same proportions. The species-apportioned 
biomass densities (Fig. 3) reflect the distributions of 
sardine, and jack and Pacific mackerel in the trawl 
catches (Fig. 2). Too few trawl catches included anchovy 
and herring to allow evaluation of their distributions 
and abundances. 
During the spring survey, most of the sardine bio- 
mass was located off southern California (Fig. 3). The 
total biomass of sardine from San Diego to the Strait of 
Juan de Fuca, 0.751 Mt with a CV of 9.2%, compared to 
0.778 Mt from the 2010 assessment (Hill et ah, 2010), 
was estimated by summing the biomasses within each 
stratum (Fig. 3). The stock of jack mackerel was esti- 
mated to be 0.147 Mt with a CV of 28.4%. The stock of 
Pacific mackerel was estimated to be 0.018 Mt with a 
CV of 51.8%, compared to 0.275 Mt from the 2009 as- 
sessment (Crone et ah, 2009). 
During the summer survey, most of the sardine bio- 
mass was located in the northern portion of the study 
area, off Oregon and Washington, whereas jack mack- 
erel biomass was found mainly off central California. 
The biomass of Pacific mackerel was more scattered 
than sardine and jack mackerel (Fig. 3). The total bio- 
mass of sardine from San Diego to the Strait of Juan 
de Fuca, 0.801 Mt with a CV of 30.9%, compared to 
0.778 Mt from the 2010 assessment (Hill et al., 2010), 
was estimated by summing the biomasses within each 
stratum (Fig. 3). The stock of jack mackerel was esti- 
mated to be 0.448 Mt with a CV of 33.9%. The stock of 
Pacific mackerel was estimated to be 0.055 Mt with a 
CV of 38.9%, compared to 0.275 Mt estimated from an 
assessment of the entire stock extending south to Cabo 
San Lucas, Mexico (Crone et ah, 2009). 
