152 
Fishery Bulletin 119(2-3) 
Table 1 
Description and values for index variables, structural parameters, state variables, derived variables, and stochastic deviation 
used in the operating model to generate simulated data that served as inputs for the 4 age-structured stock assessment models 
evaluated in this study. Models were tested under different cases, including the null case (case 0), which has one fishing fleet and 
one survey, with fully selected fishing mortality rate (F) increasing linearly with time. Also provided are the value or expression 
in case 0 and indication of whether a parameter is estimated in the estimation model. The parameter natural mortality rate at 
age is assumed to be constant across years and age classes. 
Symbol Description 
Index variables 
y Years 
a Ages 
Structural parameters 
Asymptotic average length (in millimeters) 
Growth coefficient (per year) 
Age at mean length of zero (in years) 
Length—weight coefficient 
Length—weight exponent 
Slope of maturity ogive 
Age at 50% maturity (in years) 
Natural mortality rate at age (per year) 
Proportion of females at each age 
Median-unbiased unfished recruitment (in number) 
Median-unbiased steepness 
Fishery selectivity slope 
Case 0 value or 
expression Estimated 
{1,2,...,Y} and Y=30 
{1,2,...,A} and A=12+ 
800 
0.18 
-1.36 
2.50 x 10°° 
3 
3 
2.25 
0.2 
0.5 
1 million 
0.75 
1 
Fishery selectivity age at 50% selection (in years) 2; 
Survey selectivity slope 
2 
Survey selectivity age at 50% selection (in years) 1.5 
Shape of the fully selected F in year y 
Linear increase with 
f,=0.01 and f,=0.40 
Annual sample size for age composition of fishery landings 200 
Annual sample size for age composition of survey 200 
State variables 
R Annual recruitment in year y (in number) 
y 
SSB, Spawning biomass in year y (in metric tons of female biomass) 
Abundance at age a in year y (in number) 
Ay Abundance in year y (in number) 
B, Biomass in year y (in metric tons) 
ee Landings at age a in year y (in number) 
selectivity patterns, number of surveys, and bias adjust- 
ment of recruitment on performance of the EMs (Table 3). 
Because our goal was to demonstrate that the 4 EMs are 
similar at their core, we started with simple cases to com- 
pare the performance of the EMs before adding in additional 
complexity. It would have been harder to interpret the dif- 
ferences found in the results if more complicated cases had 
been included in the simulation. We did not implement 
these cases to evaluate how model misspecifications intro- 
duced in the EMs could affect the model estimates; there- 
fore, we made the correct assumptions about parameters 
(e.g., fixing them at correct values) and governing processes 
(e.g., stock—recruit relationship and selectivity pattern). 
Recruitment variability level In case 0, the standard devi- 
ation of annual recruitment variability (og) in log space 
(Continued on next page) 
was 0.2. Two additional levels were explored to examine if 
higher recruitment variability with op of 0.4 (case 1) and 
0.6 (case 2) affected the performance of the EMs. 
Process error in fishing mortality A comparison between 
case 3 and case 0 addressed the question of whether addi- 
tional process error in F affected the performance of the 
EMs. In all cases, the standard deviation of the log fully 
selected F was used for generating annual deviations in 
fully selected F. In case 0, the same set of annual devia- 
tions was used for each iteration. In case 3, we randomly 
generated stochastic sets of annual deviations per itera- 
tion. Case 0 did not start with stochastic sets of annual 
deviations per iteration because it can be difficult to inter- 
pret differences in results if the process error in F is con- 
founded with other settings or assumptions. 
