158 
Fishery Bulletin 119(2-3) 
Table 4 
Methods for bias adjustment of recruitment from 2 of the age-structured stock assessment 
models evaluated in this study: the Beaufort Assessment Model (BAM) and Stock Synthesis 
(SS). MSY=maximum sustainable yield; RO=unfished recruitment. 
Parameter BAM method 
Unfished recruitment 
Median-unbiased RO as input 
Median-unbiased RO and 
SS method 
Mean-unbiased RO as input 
and output 
mean-unbiased RO as output 
Unfished biomass 
Equilibrium recruitment 
Spawner-recruit parameters 
MSY-based reference points 
Median-unbiased h 
[o) 
oO 
oS 
aS 
2.0 
Based on mean-unbiased RO 
Mean-unbiased 
Median-unbiased 
Mean-unbiased 
Based on mean-unbiased RO 
Mean-unbiased 
Mean-unbiased 
Mean-unbiased 
=) 
©0 
| 
Median-unbiased h 
(@) 
fon) 
| 
So 
aN 
Figure 3 
(A) Relative difference (%) in unfished recruitment (RO) and (B) difference in steepness (h) over 
possible combinations of median-unbiased A and standard deviation of log recruitment (og), 
both indicated by the contour lines. The differences were determined by using the spawner- 
recruit parameter conversion function. Relative difference in unfished recruitment is defined as 
100(mean-unbiased RO-median-unbiased RO)/median-unbiased RO. Relative difference in steep- 
ness is defined as mean-unbiased h—median-unbiased h. 
The MAREs of RO and q were below 10% for all EMs 
(Table 5). The median RO over 100 iterations from the 
AMAK was generally lower compared with the estimates 
from the other EMs (cases 0-9), indicating that the AMAK 
initializes the population differently compared with the 
OM (Fig. 4). For case 12, in which the initial condition was 
simulated for an unfished equilibrium population, the 
estimated RO from the AMAK was similar to the estimates 
from the other EMs (Fig. 4). 
Under case 0, the estimated median selectivity at 
age over 100 iterations from all EMs has almost identi- 
cal patterns compared with the selectivity curves based 
on true values from fishery and survey sources (Suppl. 
Fig. 3) (online only). The MARKs of MSY, Fysy, and SSBysy 
were below 10% (Table 6), and the REs indicate similar 
variability among all EMs (Fig. 5). The AMAK produced 
relatively lower MSY and SSBygy, along with a lower 
estimate of RO, compared with the true values and esti- 
mates from the other stock assessment models. The REs 
for SSB, R, F, relative SSB, and relative F centered around 
zero over time and had similar variability patterns (Fig. 6, 
Suppl. Figs. 4-8 [online only]). The accuracy of stock status 
determination was 100% for overfished status determina- 
tion, and the accuracy of the overfishing status was 100% 
in most but not all years (Fig. 7). 
Recruitment variability level The 4 EMs accurately estimated 
model parameters, but the range of RE increased when op 
increased. The MARE in key parameters increased when op 
increased from the null case value of 0.2 (Tables 5 and 6). 
