He et al Interactions of age-dependent mortality and selectivity functions in age-based stock assessment models 
203 
Table 3 
Key assessment model estimates on biological, fishery, and management parameters and their comparisons with their correct 
values. For virgin spawning output, depletion, and overfishing limit (OFL, in metric tons [t] ), the statistics are expressed as the 
percent difference of medians between simulations and assessments with 2.5% and 95% percentiles seen in parentheses. Values 
equal to zero indicate no difference between the simulation and assessment models. For steepness and catchability, the estimated 
values and their ranges are used. The true value of steepness (h) is 0.6 and the true values of two survey catchabilities (q 2 and 
q 2 ) are 1.0. Selectivity fits are indicated by three categories: good, fair, or bad. 
Run no. 
Virgin spawning 
output (B 0 ) 
Steepness 
Ui) 
Adult survey 
catchability (q x ) 
Juvenile survey 
catchability ( q 2 ) 
Selectivity 
fit 
Depletion 
OFL (t) 
1 
0.5 
0.59 
0.98 
0.99 
Good 
-2.0 
-2.2 
(-19.9, 26.8) 
(0.35, 0.75) 
(0.84, 1.14) 
(0.84, 1.19) 
(-27.0, 31.1) 
(-22.3, 18.2) 
2 
-21.3 
0.76 
1.35 
3.62 
Fair 
-6.3 
-32.4 
(-41.9, 2.5) 
(0.61, 0.87) 
(1.16, 1.56) 
(3.03, 4.28) 
(-32.9, 37.0) 
(-48.3, 11.7) 
3 
1.0 
0.59 
0.97 
0.99 
Good 
-2.1 
-1.7 
(-20.3, 32.3) 
(0.35, 0.74) 
(0.82, 1.17) 
(0.83, 1.18) 
(-27.7, 27.2) 
(-21.4, 19.5) 
4 
7.5 
0.60 
0.95 
2.80 
Bad 
-4.3 
-9.7 
(-21.1, 42.0) 
(0.36, 0.74) 
(0.81, 1.21) 
(2.36, 3.39) 
(-30.7, 31.8) 
(-26.2, 9.4) 
5 
5.3 
0.59 
0.88 
0.96 
Fair 
-2.7 
-0.1 
(-17.7, 32.6) 
(0.36, 0.73) 
(0.77, 1.04) 
(0.81, 1.14) 
(-28.3, 29.6) 
(-20.9, 20.6) 
6 
9.2 
0.61 
0.87 
2.72 
Bad 
-3.9 
-10.1 
(-20.9, 47.4) 
(0.35, 0.73) 
(0.73, 1.02) 
(2.32, 3.16) 
(-30.6, 31.7) 
(-24.9, 9.3) 
7 
-11.6 
0.63 
1.13 
1.11 
Bad 
0.8 
-6.6 
(-30.0, 13.2) 
(0.40, 0.78) 
(1.00, 1.32) 
(0.95, 1.32) 
(-25.8, 31.3) 
(-27.9, 11.6) 
8 
-29.6 
0.78 
1.55 
3.96 
Bad 
-5.1 
-35.8 
(-47.4, -6.4) 
(0.64, 0.90) 
(1.36, 1.78) 
(3.36, 4.76) 
(-32.9, 33.0) 
(-51.6,-19.1) 
9 
1.6 
0.57 
0.99 
1.00 
Good 
-2.0 
-3.4 
(-18.7, 25.0) 
(0.31, 1.00) 
(0.84, 1.13) 
(0.84, 1.17) 
(-26.1, 30.4) 
(-20.5, 15.9) 
10 
-24.1 
1.00 
1.33 
3.50 
Fair 
2.1 
-29.6 
(-43.2, -0.5) 
(0.63, 1.00) 
(1.13, 1.52) 
(2.93, 4.22) 
(-29.8,46.6) 
(-46.5, -9.7) 
11 
1.8 
0.58 
0.97 
0.99 
Good 
-1.8 
-2.3 
(-19.8, 33.0) 
(0.32, 1.00) 
(0.83, 1.14) 
(0.84, 1.17) 
(-27.5, 31.8) 
(-20.5, 16.6) 
12 
9.5 
0.60 
0.95 
2.80 
Fair 
-6.9 
-10.1 
(-18.9, 45.8) 
(0.34, 1.00) 
(0.81, 1.17) 
(2.37,3.32) 
(-31.1, 24.2) 
(-25.8, 6.1) 
13 
0.3 
0.62 
0.98 
1.00 
Good 
-1.5 
-2.4 
(—26.1, 37.8) 
(0.32, 1.00) 
(0.82, 1.15) 
(0.77, 1.30) 
(-29.0, 33.8) 
(-21.4, 22.7) 
14 
-0.4 
0.58 
0.98 
1.19 
Good 
-0.9 
-3.7 
(-24.1,-32.6) 
(0.33, 1.00) 
(0.82, 1.16) 
(0.79, 1.79) 
(-27.5, 36.9) 
(-24.0, 21.0) 
15 
-1.1 
0.61 
0.98 
1.03 
Good 
0.0 
-4.1 
(-24.4, 35.6) 
(0.33, 1.00) 
(0.83, 1.16) 
(0.78, 1.29) 
(-28.0, 31.7) 
(-22.31, 21.1) 
16 
22.9 
0.55 
0.96 
1.67 
Fair 
-6.0 
-19.7 
(-1.9, 110.4) 
(0.21, 1.00) 
(0.80, 1.13) 
(0.68, 4.33) 
(-33.2, 34.0) 
(-52.2, 20.3) 
small (Tables 3 and 4), and their distributions were 
centered near zero and symmetrical (Fig. 4). Other 
key assessment outputs, such as steepness (h) and the 
two catchability coefficients (q 1 and q 2 ), in compari- 
sons between the simulation and assessment models 
also matched very well (Table 3). Selectivity was also 
estimated well in the stock assessment models (Fig. 5). 
In this setting, with no model specification error, the 
estimation model performed very well; 100% of runs 
finished and 100% MGC values were smaller than speci- 
fied critical values (Table 5). 
Effects of misspecified M on assessment results 
If selectivity functions were asymptotic and correctly 
specified in the assessment models (runs 1 and 2), 
population depletion was generally well estimated, even 
when natural mortality was misspecified in assess- 
ment models (Table 3). However, the OFL estimates 
were lower by more than 32% than the true values if 
young and old fish were characterized by increasing 
natural mortality, but M was assumed to be constant 
in the assessment (Table 3, run 2). The estimated 
