Gunderson: Spatial patterns in the dynamics of slope rockfish stocks 
221 
Methods 
The Washington State Department of Fisheries car- 
ried out a series of trawl surveys during 1968-70 
aimed at monitoring the abundance and age compo- 
sition of Pacific ocean perch stocks. A 400-mesh east- 
ern otter trawl (28.7-m footrope) of uniform 3.5 -inch 
(8.9-cm) mesh and with a 1.5-inch (3.8-cm) codend 
liner was used, with roller gear attached to the 
footrope in the manner shown in Gunderson (1969). 
The net was fished with 1.5 m x 2.1 m steel “vee” 
doors, 18.3-m bridles, and 27.5-m sweeplines. The 
20.4-m research vessel Commando was used during 
each survey, which comprised a sampling design of 
two index sites (Fig. 1) and three depths (219 m, 293 
m, and 366 m) at each site. Four 45-min hauls were 
made at each of these site-depth strata, and an at- 
tempt was made to make each haul during a differ- 
ent part of the day (morning, mid-morning, afternoon, 
or evening). A total of 24 hauls were planned during 
each cruise, although this target was exceeded in 
1969 (Table 2). The trawl, bridles, sweeplines, and 
doors used in the 1992 survey were nearly identical 
to those used in the 1968-70 surveys although there 
were minor differences in the roller gear and in the 
mesh sizes used in the net (4 inch; 10.2 cm) and 
codend liner (1.25 inch; 3.2 cm). However, the 30.5-m 
research vessel Alaska was used during 1992, and 
comparisons of the distance covered during each 45-min 
haul showed that the average distance traveled by 
the Alaska was 11% greater than that covered by 
the Commando. Effort data for the 1992 survey were 
consequently adjusted upward by this amount. 
Sampling design and gear deployment in 1992 were 
similar to the 1968-70 protocol although it was not 
possible to complete the 24 hauls planned in the time 
allotted (Table 2). The trawl was monitored during 
fishing operations with a SCANMAR acoustic moni- 
toring system, and the mean horizontal spread (be- 
tween wingtips) was 13.9 m and the vertical open- 
ing was 3.0 m. 
Table 2 
Cruise dates and number of hauls at each depth for sur- 
veys of the northern Washington index sites, 1968-92. 
No. of hauls 
Cruise dates 
219 m 
293 m 366 m 
17 July-28 July 1968 
8 
8 8 
1 July-10 July 1969 
9 
10 9 
27 Sep-4 Oct 1970 
8 
8 8 
7 Oct- 10 Oct 1992 
7 
7 7 
An index of abundance for each survey was esti- 
mated by weighting each of the six site-depth strata 
separately (Eq. 5.1, Cochran, 1977): 
L 
y s t = 'Yj W h yh > 
h = 1 
where y st = 
Wh = 
y h = 
L = 
mean catch rate (kg/h); 
stratum weight (=1/6 for all strata); 
sample mean (kg/h) for stratum h; and 
number of strata (6). 
The variance of this index was estimated by using 
Equation 5.7 of Cochran (1977): 
h = 1 
n h = number of hauls in site-depth stra- 
tum h; and 
y h = catch rate (kg/h) for haul i. 
All rockfish were sorted and weighed by species 
during the surveys, and all other species were also 
sorted and weighed in 1992. Sex and length data were 
obtained for each catch of Pacific ocean perch. For 
large catches, sex and length data were obtained by 
sampling an equal portion of the first, middle, and 
last part of the sorted catch (Westrheim, 1967). All 
fish w'ere measured to the nearest cm (FL), and 
otoliths were extracted from random subsamples of 
the length-sex samples at each depth (Table 3). Ages 
were determined during 1968-70 by using surface 
readings of the otoliths, whereas broken and burned 
cross sections were used in 1992. Because compara- 
tive ageing experiments have shown that surface 
ages are biased for fish older than 17 (Tagart, 1984), 
all surface-aged fish older than this were pooled. Age- 
length keys were used to estimate age composition 
from length data, and separate keys were constructed 
for each cruise depth-sex stratum. Within a given 
stratum, size composition data for each haul were 
weighted by the catch per hour (catch per nmi in 
1992) prior to combining them. 
The age-length relation for Pacific ocean perch 
varies with depth so that there is an inverse relation 
between depth and apparent growth rate (Westrheim, 
1973; Gunderson, 1974). As a result, separate curves 
were fitted for the age-length data at each depth by 
using a nonlinear least-squares algorithm (EXCEL) 
to fit the data to the Bertalanffy growth model: 
