506 
Fishery Bulletin 96(3), 1998 
■C 
Gulf of Alaska (610) n=95 
r 2 =0.90 
10 - 
0 -I 1 1 1 1 i 1 1 1 1 1 1 1 
01 23456789 10 11 12 
Central Aleutians (542) n=163 
F=0.89 
50 
40 
30 - 
20 
10 
0 
0 1 23456789 10 11 12 
Eastern Aleutians (541 ) n=94 
r 2 =0.75 
50 t 
40 
30 
20 
10 
0 
0 1 
Western Aleutians (543) ri= 1 58 
F=0.79 
r 4**++-*- H 
2 3 4 5 6 7 8 9 10 11 12 
Age (years) 
Figure 2 
Atka mackerel age-length dat a by fishery management areas, collected from resource assessment surveys of the 
Gulf of Alaska and Aleutian Islands in 1993 and 1994, respectively. The fitted von Bertalanffy growth curves are 
plotted. Sample sizes (n ) and the coefficients of determination (r 2 ) are given. 
ber of alleles per locus. Samples were tested indi- 
vidually with chi-square analyses for conformance 
to expected Hardy- Weinberg proportions and as one 
pooled sample. The latter method tests the assump- 
tion that the samples are not different (i.e. repre- 
sent one interbreeding population). Differences 
among samples were examined with Wright’s fixa- 
tion index F gT and Nei’s unbiased genetic distance 
statistic (Nei, 1978) and with standard chi-square 
contingency table analysis for the P 095 loci. Values 
of F st range from zero, indicating no population dif- 
ferentiation, to one, indicating fixation for alternate 
alleles. Genetic analyses were conducted using the 
BIOSYS-1 computer program (Swofford and Selan- 
der, 1981). 
Results 
Age, weight, and iengtSi data 
Sample sizes were relatively small when stratified 
by area, ranging from 95 to 163. Sampled fish lengths 
ranged from 21 to 49 cm, ages from 2 to 10 years, 
and weights from 0.084 to 1.9 kg (Figs. 2 and 3). The 
relatively high coefficients of determination ( r 2 ) in- 
dicated good fits to the von Bertalanffy and length- 
weight models (Figs. 2 and 3). An examination of the 
residuals also indicated no obvious patterns, and the 
variances of the residuals were homogeneous by age 
and area for the von Bertalanffy model, and by length 
and area for the length-weight model. The fitted 
growth curves for each management are shown to- 
gether in Figures 4 and 5; Table 3 gives the estimated 
parameters and standard errors. The F - tests revealed 
that differences in the growth curves among areas 
were significant (P<0.0005) for both the age-length 
and length-weight data. Because of the low sample 
sizes for the Gulf of Alaska and eastern Aleutian Is- 
lands data, and low numbers of fish in the Gulf of 
Alaska greater than five years, these data were com- 
bined and fit to the von Bertalanffy model. An 
F-test to test for differences between these two spe- 
cific data sets was highly significant (P<0.0005). 
The results of the two-way factorial analysis of age- 
length and length-weight data were highly signifi- 
cant (P of F<0.0001, Tables 4 and 5). To account for 
unequal sample sizes among factors, the AN OVA re- 
sults in Tables 4 and 5 should be interpreted sequen- 
tially. The terms in the tables were added in the or- 
der presented, and the F-tests were given for each 
term in sequence. In addition, an AN OVA model was 
evaluated where the area factor was added first. The 
result of that F-test was also highly significant (P of 
F<0.0001). An examination of the residuals from the 
age-length and length- weight AN OVA models showed 
