692 
Fishery Bulletin 99(4) 
1/mL) collected from local lagoons. After two weeks of feed- 
ing, Artemia sp. brine shrimp (at approximately 10/mL) 
were added as well. Further details on larval rearing are 
described in Porter and Theilacker. 1 
Shrinkage experiments 
Larval walleye pollock shrinkage was examined for three 
age groups of larvae (8, 15, and 33 days after hatching) 
with treatment with preservative only (no net treatment), 
with 5-minute net treatment and then treatment with 
preservative, and 15-minute net treatment and then treat- 
ment with preservative. The preservatives used were 5% 
buffered formalin in 33%o seawater, and 95% ethanol 
buffered with 0.6 mM sodium carbonate, and 0.6 mM 
sodium bicarbonate. For each age group, 10 larvae were 
sampled for each treatment and preservative combina- 
tion. A pipette was used to place a live larva in seawater 
on a microscope slide and the larva was quickly recorded 
on videotape with a Panasonic 5100 digital video camera 
mounted on a dissecting microscope. Next, the larva was 
either placed directly into preservative or into the net- 
treatment apparatus, after which it was preserved. The 
net-treatment apparatus consisted of a small net sub- 
merged in a tank of 6°C seawater. A submersible pump 
was used to circulate seawater through the net to simulate 
a plankton tow (Theilacker, 1980). Water flow through the 
net was adjusted so that a larva would come in contact 
with the mesh as it might during a tow at sea. One month 
was chosen as the length of time walleye pollock larvae 
would remain in preservative before they were recorded 
on videotape. The length of time for standard length to 
stabilize in preservative varies depending on the species 
of fish. For herring ( Clupea harengus) larvae, standard 
length was stable after 30 days (Fox, 1996); for red sea 
bream ( Chrysophrys major) larvae, most of the shrinkage 
occurred during the first 2-3 days in presevative (Rosen- 
thal et al., 1978). Optimus version 5.0 image analysis 
software was used to measure standard length (SL), eye 
diameter, head length (tip of snout to pectoral fin), and 
body depth at the anus of both live and preserved larvae. 
Otoliths were dissected from the preserved larvae, and 
maximum diameter of the sagittae were measured at 
lOOOx magnification with an ocular micrometer. Some of 
the otoliths from larvae preserved in formalin had eroded 
edges and their diameters could not be measured reliably; 
therefore, otoliths from formalin-preserved larvae were 
not used in the shrinkage correction model. 
Myomere width measurements taken from formalin pre- 
served larvae were used in the shrinkage correction mod- 
el. The width of five consecutive myomeres located be- 
tween the two pigment bands on the tail were measured 
1 Porter, S. M., and G. H. Theilacker. 1996. Larval walleye pol- 
lock, Theragra chalcogramma, rearing techniques used at the 
Alaska Fisheries Science Center, Seattle, Washington. AFSC 
Processed Report 96-06, 26 p. US. Dep. Commerce, NOAA, 
NMFS, Alaska Fisheries Science Center, 7600 Sand Point Way 
NE, Seattle, WA, 98115. 
at 50x with an ocular micrometer and polarized light. For 
ethanol-preserved larvae, myomere measurements could 
not be made because the tissue had turned opaque during 
preservation. 
Data analysis 
For each preservative, data from all larval ages and 
shrinkage treatments were pooled and backward stepwise 
regression (SPSS, Inc., 1998) was used to determine which 
variables produced the best model for estimating live SL. 
The data were pooled because the regression is used to 
estimate live SL of larvae caught at sea and because 
the amount of time larvae spend in the sampling net is 
unknown; therefore a general relationship was sought. 
For each preservative, the growth rate of walleye pollock 
larvae between the first (8 days after hatching) and last 
sampling days (33 days after hatching) was determined 
by using linear regression. Only the first and last sam- 
pling days were used because there is a lag in growth of 
walleye pollock larvae that occurs for about a week dur- 
ing the transition from endogenous to exogenous feeding 
(Yamashita and Bailey, 1989). Dunnett’s test (Zar, 1996) 
was used to compare growth rates calculated with live SL 
(which was considered to be the control) against growth 
rates calculated from the shrinkage correction models. 
Results 
Shrinkage with ethanol 
Walleye pollock larvae shrank an average of 6% in 95% 
ethanol when all shrinkage treatments were pooled (Fig. 
1A). Larvae used in the shrinkage experiments ranged in 
live SL from 5.41 to 9.25 mm. The best model to correct 
for shrinkage contained preserved SL, body depth, and 
otolith diameter, r 2 = 0.90 (Table 1, Fig. IB). Results indi- 
cated that there was borderline multicollinearity in the 
data (the variance inflation factor for otolith diameter was 
4.79), probably due to correlation between preserved SL 
and otolith diameter. The r 2 with only preserved SL was 
0.82. The use of additional variables reduced the mean 
square error (MSE) by 44% (Table 1). 
Residuals from the best model increased as the per- 
centage of shrinkage increased, which suggests that the 
shrinkage correction model does not work well for larvae 
that shrink greatly. At largest values of shrinkage, 17% 
and 19%, the best model underestimated SL by 10%, but 
at 15% shrinkage the error was 3%. 
Shrinkage with formalin 
For larvae preserved in 5% formalin, the overall mean 
shrinkage was 10%. Larvae ranged in live SL from 5.61 
to 9.43 mm (Fig. 2A). When compared with the model 
using only preserved SL (r 2 =0.81, MSE=0.112; Table 1), 
the model containing both preserved SL and body depth at 
the anus produced the best model for estimating live SL 
(r 2 =0.88, MSE=0.073; Table 1, Fig. 2B). 
