Weber and McClatchie: Effect of environmental conditions on the distribution of Scomber /aponicus 
87 
[Merluccius productus] eggs; Watson 2 ). Samples of eggs 
from CalCOFI tows collected during 1988-2009 were re- 
sorted to identify those of Pacific mackerel. Of the latter 
about 98% were captured from April through September. 
Larvae were collected with 1.0-m-diameter bridled 
ring nets until 1977 and 0.71-m-diameter bridleless 
bongo nets thereafter. Nets consisting of 0.55-mm-mesh 
silk were towed obliquely at an angle of approximately 
45° from 140 m depth to the surface for all samples 
until 1969. In 1969 the net was changed to 0.505-mm- 
mesh nylon and the beginning tow depth was increased 
to 210 m, as described by Smith and Richardson (1977) 
and Ohman and Smith (1995). Oceanographic data 
used to develop predictor variables for the model were 
dynamic height (referenced to 0/500 decibars) and water 
temperature. These data were measured or calculated 
from bottle casts and conductivity-temperature-depth 
sensor (CTD) casts at each station. Variables were in- 
terpolated to the nearest 10 m for depths of 0-100 m 
and at 125, 150, 200, 250, 300, 400, 500 m. Detailed 
sampling protocols for the CalCOFI bottle and CTD 
samples are described by Lynn et al. (1982). 
Catch data collected from the commercial boats in the 
recreational fishery by the California Department of 
Fish and Game (CDFG) were used to estimate relative 
annual abundance of Pacific mackerel. Captains of com- 
mercial passenger fishing vessels have been required 
to provide logs of fishing effort and catch to the CDFG 
since 1936 (cf., Hill and Schneider 3 ). These data have 
been used to develop an index of abundance, known 
as the commercial-passenger-fishing-vessel or CPFV 
index, which represents data standardized by using a 
A-generalized-linear model approach (Stefansson, 1996) 
to account for potential changes in catchability associ- 
ated with the CPFV fleet over time (Crone et ah, 2009). 
Index values were provided by P. Crone. 4 
Larval densities 
We estimated larval Pacific mackerel densities from net 
tows on the basis of volume of water sieved by the tow 
(Smith and Richardson, 1977). Estimates were then 
corrected for extrusion of very small larvae through the 
net mesh, and for avoidance of the net by larger larvae, 
by using the method described by Lo et al. (2009). Esti- 
mates for larvae less than or equal to 3 mm in length in 
each tow were divided by 0.28 to correct for extrusion. To 
correct for net avoidance, estimates were multiplied by 
a capture coefficient ( R L h ), which varied as a function 
of diel period (. h; i.e., hour) and fish length (L): 
2 Watson, W. 2011. Personal commun. NOAA Southwest 
Fisheries Science Center, 8604 La Jolla Shores Drive, La 
Jolla, CA 92107. 
3 Hill, K. T. and N. Schneider. 1999. Historical logbook 
databases from California’s commercial passenger fishing 
vessel (partyboat) fishery, 1936-1997. Scripps Inst. Ocean. 
Ref. Series 99-19, 65 p. Univ. Calif. San Diego, CA. 
4 Crone, P. 2011. Personal commun. NOAA Southwest 
Fisheries Science Center, 8604 La Jolla Shores Drive, La 
Jolla, CA 92107. 
L.h ~ 
+ 
1 -Dl 
2 
2k* h \ 
24 y 
( 1 ) 
where D L = the noon/night catch ratio for length L cal- 
culated as 
D, = 2.7exp(-0.39L). (2) 
Most captured larvae were in the range of 3-20 mm 
long, or were aged to be about 0-20 d (Lo et al., 2009). 
Because most larvae captured were estimated to be 
only a few days old and had poor swimming ability, we 
assumed their distribution was directly related to the 
distribution of spawning adults. 
Model development 
We used six initial predictor variables to model the pres- 
ence of Pacific mackerel larvae. They were mean water 
temperature (°C), mixed-layer depth (m), an index of geo- 
strophic flow, the log of volume displaced by zooplankton 
captured in nets (mL/1000 m 3 filtered), the CPFV index 
of Pacific mackerel stock size for the previous year, and 
day of the year. Temperature was entered as a predictor 
of the physiological suitability of the habitat. Zooplank- 
ton displacement volume of the habitat and indicator of 
the water mass in which fish were located, was entered 
as an index of the standing crop of available food. Large 
jellyfish and tunicates whose individual volume was 
greater than 5 mL were excluded from zooplankton 
samples (Kramer et ah, 1972). However, zooplankton 
samples were not specifically sorted into prey items and 
predators. Mixed-layer depth was used as an indicator 
of stratification of the water column, and geostrophic 
flow as a measure of horizontal current strength, both 
of which also potentially affected production and food 
availability (Mantyla et al., 2008). 
The index of geostrophic flow was calculated on the 
basis of a fitted surface in dynamic height for each year, 
which was estimated by a method similar to that used 
to fit digital elevation maps to terrestrial slope data. 
First a surface was fitted by using the “loess” function 
(Cleveland and Grosse, 1991) in the R programming 
environment, vers. 2.12.0 (R Development Core Team, 
2011). Geostrophic flow occurs perpendicular to the 
slope in dynamic height because of the Coriolis effect. 
Therefore the index of geostrophic flow was calculated 
as the slope of a line on the loess-estimated surface 
that extended for 10 km on each side of a sampling 
location in the direction of maximum slope, with flow 
direction perpendicular to this line. For points located 
on the outer edge of the surface, only the 5-km line 
that was located within the bounds of the surface was 
used. Visual inspection of plots indicated that the index 
matched contours in dynamic height well and thus pro- 
vided a reasonable proxy for geostrophic flow. 
Two blocking variables that were not related to the 
physical quality of the habitat were included as poten- 
tial predictors of larval abundance. Day of year was 
used to account for changes in larval abundance asso- 
