Hulson et al.: Distribution of sampling effort for age composition of multiple species 
339 
0.4 0.6 
SD in log recruitment 
0.1 0.2 0.3 0.4 0.5 
CV in survey index 
> 
o 400 
- c 
□ Roundfish 
□ Flatfish 
■ Rockfish 
o E0 
□ El 
A £2 
V E3 
o 200 - 
p E0 = 0.75 
'□ 
0 2 4 6 8 
SD in log recruitment / CV in survey index 
Figure 7 
Correlations of (A) standard deviation (SD) in log-scale recruitment, (B) the coefficient 
of variation (CV) of the survey index, and (C) SD in log-scale recruitment divided by 
the CV of the survey index with the resulting percent change in the CV in biomass 
during the final year from the estimation models across survey age-composition sam- 
ple sizes for each survey index uncertainty case (E0-E3) evaluated (correlation in text 
shown for case E0). p=Pearson’s correlation coefficient for all survey index uncertainty 
cases; pEo=Pearson’s correlation coefficient for survey index uncertainty case E0 only. 
tribution. of sample size. We will discuss 3 of these: 1) 
cost, 2) intrahaul correlation, and 3) commercial value 
of the species. 
Cost, in terms of collecting age samples, would be 
defined as the cost in time (which is proportional to 
labor costs) required to both collect and read any given 
otolith. For example, when otoliths are collected, it is 
somewhat more difficult to obtain an otolith from a 
rockfish than from a roundfish or flatfish. There could 
also be differences in the amount of time it takes to 
obtain lengths of certain species. In terms of reading 
otoliths, more time is required to read a rockfish otolith 
than a flatfish or roundfish otolith, if for no other rea- 
son than that rockfish are longer-lived and have more 
annuli to count than flatfish or roundfish. Additionally, 
we found that aging error had a relatively larger influ- 
ence on rockfish species than on roundfish or flatfish. 
This finding would increase the cost in time because 
more otoliths would need to be read to obtain the same 
amount of information in the age-composition data. 
Cost could also be a function of sampling method, with 
the highest cost associated with fixed-allocation 2-stage 
sampling and lower costs associated with proportional 
allocation 2-stage sampling or SRS (in terms of the 
sample size necessary to achieve the same amount of 
uncertainty). Methods have been developed to include 
