Norcross et al.: Habitat models for juvenile pleuronectids 
509 
They were concentrated mainly in central, deep ar- 
eas of bays at depths of 80-120 m, 6.0-9.0°C, 31.5- 
33.5 psu, on mud or mixed mud substrates (Norcross 
et al. 4 ). High abundances of flathead sole were asso- 
ciated with deep stations, low temperatures, high 
salinities, low sand content, and high mud content; 
the highest rank correlations for flathead sole were 
obtained for depth and mud (Table 1). Flathead sole 
were predominantly collected in depths > 40 m, ex- 
cept on substrates with a high mud content (Fig. 2B). 
Pacific halibut composed 5% of the catch in 1991 
and 7% in 1992. During 1992, 627 age-0 halibut (22 — 
84 mm TL) were found in exposed sites at all loca- 
tions on the east and south sides of Kodiak Island 
(Fig. 1C). In northwestern Kodiak, halibut were col- 
lected only at the mouth of Uyak Bay. Age-0 halibut 
were found mainly at 10-70 m depth, 7.0-10.5°C, 
32.0-33.0 psu, on mixed sand substrates, outside of 
or within 7 km of bay mouths (Norcross et al., 1995). 
Pacific halibut abundances were positively correlated 
with temperature and sand content and negatively 
correlated with depth, distance from mouth of bay, 
and mud content in the substrate. The highest rank 
correlations were with sand and mud (Table 1). Un- 
like rock sole, halibut were seldom found in water 
deeper than 50 m. Halibut juveniles, like rock sole, 
were concentrated most often in shallow waters with 
sandy substrate, near or outside mouths of bays 
(Fig. 20. 
Yellowfin sole was very abundant in 1991, com- 
posing 28% of captured flatfishes, but this species 
represented only 4% of the 1992 total catch. Unlike 
the other three species examined, in which age-0 fish 
predominated, age-1 yellowfin sole (41-105 mm TL) 
were analyzed in both 1991 and 1992 because of the 
small number (n= 4) and size (15-20 mm TL) of age-0 
Table 1 
Spearman’s rank correlation coefficients between CPUE 
of four flatfish species and environmental parameters with 
1991 and 1992 data combined. * indicates significance at 
an overall 5 % confidence level. 
Parameter 
Rock 
sole 
Flathead 
sole 
Pacific 
halibut 
Yellowfin 
sole 
Depth 
-0.258* 
0.644* 
-0.284* 
-0.369* 
Distance 
-0.308* 
-0.074 
-0.314* 
0.204 
Temperature 
0.193 
-0.467* 
0.346* 
0.212 
Salinity 
-0.083 
0.246* 
-0.163 
-0.192 
Gravel 
-0.240* 
-0.219 
-0.078 
-0.168 
Sand 
0.583* 
-0.219 
0.449* 
0.113 
Mud 
-0.310* 
0.540* 
-0.417* 
0.188 
yellowfin sole collected during the second year. Dur- 
ing 1992, 268 age-1 yellowfin sole were collected at 
depths less than 40 m, mainly between 5 and 30 m. 
Age-1 yellowfin sole were found near the heads of 
bays, in warm (9.0-11.5°C) saline (31.0-33.5 psu) 
water (Norcross et al., 1995). They were collected on 
sandy mud, gravelly muddy sand, and muddy sand. 
Unlike rock sole and halibut, yellowfin sole were col- 
lected in the inner reaches of bays around Kodiak 
Island (Fig. ID). The only significant correlation be- 
tween yellowfin sole abundance and an environmen- 
tal variable was a negative rank correlation with 
depth (Table 1). Yellowfin sole were never found 
deeper than 50 m and were always on mixed substrate, 
i.e. not predominantly on one grain size (Fig. 2D). 
Linear discriminant function analysis for the com- 
bined 1991-92 data resulted in depth having the 
highest correlation with discriminant scores (canoni- 
cal loadings) for flathead sole and yellowfin sole 
(Table 2). Sand was most highly correlated with the 
discriminant scores for rock sole and Pacific halibut. 
For all species, except flathead sole, the three high- 
est canonical loadings were obtained for depth, tem- 
perature, and sand. In the case of flathead sole, mud 
was more highly correlated with the discriminant 
score than was sand. 
Sand was clearly a good predictor for rock sole pres- 
ence and was included in the habitat model for rock 
sole. Depth and temperature performed equally well 
in the discrimination owing to their high (negative) 
correlation. However, although rock sole abundance 
was significantly correlated with depth, the correla- 
tion with temperature was not significant. Therefore, 
sand and depth seemed to be the most important 
variables determining rock sole distribution (Fig. 2A). 
The three best predictor variables for flathead sole 
were depth, gravel, and mud. Of these, depth and 
mud resulted in the lowest total error rates. Because 
Table 2 
Canonical loadings from linear discriminant function 
analysis for combined 1991 and 1992 flatfish data. 
Parameter 
Rock 
sole 
Flathead 
sole 
Pacific 
halibut 
Yellowfin 
sole 
Depth 
-0.557 
-0.776 
-0.620 
-0.696 
Distance 
-0.379 
-0.011 
-0.501 
0.234 
Temperature 
0.474 
-0.597 
0.647 
0.545 
Salinity 
0.180 
-0.225 
-0.026 
-0.005 
Gravel 
-0.453 
0.321 
-0.249 
-0.377 
Sand 
0.783 
0.220 
0.655 
0.406 
Mud 
-0.391 
-0.624 
-0.473 
-0.099 
