700 
Fishery Bulletin 95(4), 1997 
the larger (faster growing) fish in a given age class. 
Given Goodyear’s (1995) findings and our nonran- 
dom sampling design, it may be that our growth 
models overestimated length-at-age to some degree 
for all but the youngest age classes, but probably not 
as much as Goodyear found in his study. Although 
we had sampling quotas for each year, region, sex, 
and 10-cm size interval combination, for many dif- 
ferent reasons we invariably exceeded those quotas 
for all but the rarest size classes, often greatly for 
the most common length intervals; Table 2 and Fig- 
ure 1 provide clear evidence of this. Because of this 
oversampling, our actual design fell somewhere be- 
tween simple random sampling and length-stratified 
sampling, and thus should have reduced the bias to 
some extent. Given this rather small potential bias 
and the fact that our sample sizes and spatial and 
temporal coverages greatly exceeded all previous 
1 986-1 992 
1977-1978 
700 
600 
500 
400 
30 0 
200 
1 00 
0 
700 
600 
>. 
O 500 
CD 
O' 400 
CD 
300 
200 
1 00 
0 
400 
300 
200 
1 00 
Age (yr) 
Figure 2 
Age distributions by sex and region of king mackerel included in the analysis and collected during 1977-78 and 1986-92. 
