498 
Fishery Bulletin 95(3), 1997 
ging studies (Wakabayashi, 1989) indicated only lim- 
ited movements of yellowfin sole between northwest 
and southeast areas as they migrate inshore, and 
Kashkina (1965) supported a two-stock concept, cit- 
ing differences in egg-stage advancement and abun- 
dance between northern and southern regions of the 
eastern Bering Sea. Despite the lack of genetic evi- 
dence supporting the coexistence of two independent 
Table 3 
Factorial analysis of variance of male and female yellowfin sole length (cm) at age (years), by area (northwest and southeast) and 
bottom depth (<30, 30-49, and >50 m), pooled across years 1982-94. Note that nonsignificant (P>0.05) main effects were retained 
because they were included in one or more significant interaction terms. SS = sum of squares. 
Source 
df 
SS 
Mean square 
E-value 
P > F 
Males 
Age 
29 
68,007.6 
2,345.1 
360.0 
0.0001 
Area 
1 
295.5 
295.5 
45.4 
0.0001 
Depth 
2 
5.7 
2.8 
0.4 
0.6457 
Age x depth 
49 
1,906.9 
38.9 
6.0 
0.0001 
Age x area 
24 
160.4 
6.7 
1.0 
0.4268 
Area x depth 
2 
51.8 
25.9 
4.0 
0.0188 
Age x area x depth 
34 
366.9 
10.8 
1.7 
0.0098 
Error 
3,749 
24,421.2 
6.5 
Females 
Age 
29 
152,746.0 
5,267.1 
667.9 
0.0001 
Area 
1 
408.1 
408.1 
51.8 
0.0001 
Depth 
2 
2.1 
1.1 
0.1 
0.8738 
Age x depth 
49 
3,219.2 
65.7 
8.3 
0.0001 
Age x area 
24 
411.2 
17.1 
2.2 
0.0008 
Area x depth 
2 
44.7 
22.3 
2.8 
0.0589 
Age x area x depth 
44 
593.1 
13.5 
1.7 
0.0025 
Error 
5,057 
39,880.8 
7.9 
Table 4 
Logistic regression coefficients for the equation MAT=l/( 1+e - '^ + P' L + a area + S D + AL D) ) relating female yellowfin maturity status to 
fish length (L), binary variable area (northwest or southeast), and continuous variable depth ( D ). L xD denotes the interaction 
between length and depth. SE= standard error of the estimate. L x D coefficients were considered nonsignificant (P>0.05) and 
were therefore removed from the model. Remaining coefficients and estimates result from model runs with L xD removed (under- 
lined values). 
Year n 
Variable 
Coefficient 
Symbol 
Estimate 
SE 
P > chi-square 
1992 7 1,478 
Length 
P 
-0.82 
0.047 
0.0001 
Depth 
8 
0.046 
0.0048 
0.0001 
LxD 
A 
0.0046 
0.0024 
0.057 
Intercept 
P 
22.22 
1.36 
0.0001 
1993 1,028 
Length 
P 
-0.83 
0.063 
0.0001 
Area 
a 
2.29 
0.30 
0.0001 
Depth 
8 
0.042 
0.0078 
0.0001 
LxD 
A 
-0.0065 
0.0035 
0.063 
Intercept 
P 
21.70 
1.81 
0.0001 
1994 291 
Length 
P 
-0.69 
0.085 
0.0001 
Area 
a 
0.91 
0.46 
0.049 
Depth 
8 
0.042 
0.012 
0.0004 
LxD 
A 
0.0046 
0.0048 
0.34 
Intercept 
P 
17.92 
2.52 
0.0001 
1 Southeast area only. Small sample 
size in the northwest area (n=107) limited 
a comparison by area. 
