496 
Fishery Bulletin 95(3), 1 997 
females. Preferred models were estimated by a least- 
squares backward stepwise procedure that sequen- 
tially removed the highest-order nonsignifcant fac- 
tor until only significant terms remained. Nonsig- 
nificant (P>0.05) main effects were retained if they 
were included in one or more significant interaction 
terms. Residual plots were examined to test assump- 
tions of homoscedasticity and normality of error 
terms. Least squares estimates of model coefficients 
were obtained by means of the Statistical Analysis 
System procedure GLM (SAS Institute, 1989). 
Length at maturity Logistic regression was used 
to assess the effects of area and bottom depth on the 
Tabie 1 
Number of male and female yellowfin sole sampled for age 
and length by the Alaska Fisheries Science Center during 
resource assessment surveys conducted in northwest and 
southeast areas of the eastern Bering Sea from 1982 to 
1994. 
Bottom depth (m) 
Northwest Southeast 
Year 
<30 
30-49 
>50 
<30 
30-49 
>50 
Males 
1982 
26 
58 
26 
65 
97 
45 
1983 
0 
52 
85 
56 
87 
27 
1984 
14 
102 
31 
31 
109 
42 
1985 
28 
99 
36 
26 
65 
84 
1986 
0 
100 
42 
39 
74 
66 
1987 
133 
30 
13 
38 
111 
31 
1988 
44 
33 
13 
27 
28 
98 
1989 
29 
35 
39 
23 
105 
87 
1990 
49 
58 
48 
36 
74 
92 
1991 
82 
64 
26 
0 
65 
97 
1992 
32 
0 
25 
73 
79 
39 
1993 
31 
50 
18 
47 
0 
46 
1994 
24 
64 
21 
72 
13 
37 
Total 
492 
745 
423 
533 
907 
791 
Females 
1982 
15 
110 
49 
65 
97 
83 
1983 
0 
80 
129 
57 
102 
41 
1984 
0 
143 
71 
33 
137 
83 
1985 
41 
154 
40 
43 
57 
129 
1986 
0 
113 
97 
48 
57 
103 
1987 
152 
67 
24 
36 
112 
51 
1988 
58 
38 
46 
47 
31 
100 
1989 
0 
82 
113 
39 
92 
96 
1990 
47 
81 
86 
54 
38 
129 
1991 
85 
73 
35 
0 
69 
146 
1992 
35 
0 
91 
86 
79 
67 
1993 
28 
67 
100 
61 
0 
101 
1994 
29 
55 
49 
61 
20 
76 
Total 
490 
1,063 
930 
630 
891 
1,205 
probability that an individual of a given length (cm) 
was mature or immature. The following equation was 
fitted independently to 1992, 1993, and 1994 female 
yellowfin sole length-maturity data by using the Sta- 
tistical Analysis System maximum likelihood proce- 
dure LOGISTIC (SAS Institute, 1989): 
_j_ g-(n+(5 L+a area+8 D+A L D) 5 (l) 
where MAT = mature proportion of female yellowfin 
sole given its length (L), area, and bottom depth (D) 
of capture. 
Area was treated as a factor indicating either 
northwest or southeast areas. Depth was treated as 
a continuous variable. Length, area, and depth coef- 
ficients were represented by a, and S, respectively, 
and /j denoted the intercept (on the logit scale). A 
length x depth interaction term ( L-D ; Eq. 1) with 
coefficient A was also tested for significance. Non- 
significant (P>0.05) highest order terms were re- 
moved from the model. 
Length at 50% maturity (L 50 ) was calculated by 
substituting 0.5 for MAT in Equation 1 and solving 
for L as follows: 
•^50 
// ■ '/ area + 5 D + A • L D 
( 2 ) 
Due to sample-size inequalities (Table 2), length at 
maturity comparisons between northwest and south- 
east areas were limited to females sampled in 1993 
and 1994. 
Table 2 
Summary of female yellowfin sole length-maturity collec- 
tions during the 1992-94 Alaska Fisheries Science Center 
eastern Bering Sea groundfish bottom trawl surveys. A = 
Age, length, and maturity data collected by sex-cm inter- 
val; L = Length and maturity (random measurements); 
O = Ovaries, lengths, and maturity data collected by size 
category, >25 cm TL. Ages were determined for 53 of these 
specimens. 
Number of samples 
Year 
Sample 
type 
Northwest 
Southeast 
1992 
A 
107 
218 
L 
0 
1,260 
1993 
A 
98 
162 
O 
256 
512 
1994 
A 
133 
158 
