200 
Fishery Bulletin 116(2) 
Table 2 
The median and 90% simulation interval (SI) for the estimated number of years needed for simulated rockfish stocks to 
rebuild to the target biomass, the operating model number of years needed for the stocks to rebuild to target biomass, 
and the number of stocks that failed to rebuild to the target biomass determined by the estimation method (EM) and 
the operating model (OM) for each case and data scenario. 
Estimated number Operating model number Number of stocks 
of rebuilding years of rebuilding years that failed to rebuild 
Selectivity/data scenario 
Median 
90% SI 
Median 
90% SI 
EM 
OM 
Time-invariant 
Full data 
43 
(13-87) 
34 
(16-73) 
7 
4 
Reduced data 
31 
(19-61) 
34 
(14-83) 
1 
5 
Eliminated data 
25 
(14-72) 
37 
(14-87) 
35 
4 
Time-varying 
Full data 
31 
(13-91) 
35 
(13-85) 
13 
4 
Reduced data 
25 
(13-79) 
32 
(12-74) 
8 
2 
Eliminated data 
25 
(13-77) 
36 
(12-79) 
32 
5 
Table 3 
The median and 90% simulation intervals (SI) for the average catch of simulated rockfish stocks during rebuilding, the 
annual average variability of the catches (AAV) during rebuilding, and the AAV over all years for each case and data 
scenario. 
Average catch AAV during 
during rebuilding rebuilding AAV all years 
Selectivity and data scenario 
Median 
90% SI 
Median 
90% SI 
Median 
90% SI 
Time-invariant 
full data 
44.0 
• (15.3-78.9) 
6.0 
(3.7-11.5) 
3.2 
(2.1-4.7) 
reduced data 
28.1 
(14.6-57.9) 
7.7 
(4.0-14.5) 
3.5 
(2.3-5.3) 
eliminated data 
41.3 
(19.9-83.8) 
2.6 
(1.3-4.4) 
2.2 
(1.3-3.9) 
Time-varying 
full data 
31.7 
(11.0-75.4) 
7.3 
(4.4-17.5) 
4.2 
(2.7-5.9) 
reduced data 
25.1 
(15.6-68.0) 
8.9 
(4.5-20.7) 
4.5 
(2.6-9.8) 
eliminated data 
36.3 
(15.7-79.4) 
2.3 
(1.2-4.8) 
2.8 
(1.3-5.3) 
The reduced data scenario had the lowest median 
average catch during rebuilding (Table 3), and the me¬ 
dian rebuilding time was estimated to be shorter than 
the true time to recovery within the operating model 
(Table 2). The eliminated data scenario, which was en¬ 
tirely dependent upon historical data until the simu¬ 
lated stocks were projected to rebuild, essentially pro¬ 
jected the population forward with each assessment on 
the basis of the initial parameter estimates from the 
historical data and resulted in high median average 
catches during rebuilding and the lowest median AAV 
during rebuilding and across the entire management 
period (Table 3 ). 
The effect of time-varying parameters 
Time-varying annual deviations in natural mortality 
and fishery selectivity generally resulted in increased 
among-simulation variation in estimation errors than 
with the time-invariant case. The median error of es¬ 
timates of spawning biomass at the time of the first 
assessment exceeded the true values and were highly 
variable among simulations (Fig. 6, A-C ). The among- 
simulation variance in errors of estimates of spawning 
biomass decreased markedly for the full data scenar¬ 
io after the first assessment (Fig. 6A). However, this 
variability remained high for approximately the first 
25 years of the management period (assessments were 
performed every fourth year between years 50-74 ap¬ 
proximately) for both the reduced and eliminated data 
scenarios, until approximately 50% of the simulated 
stocks were estimated to be recovered and the fishery 
sample sizes increased to historical levels (Fig. 6, B 
and C). The full and reduced data scenarios resulted in 
median spawning biomass estimates that were gener¬ 
ally smaller than the operating model values (Fig. 6, A 
and B). However, the medians of the errors for relative 
spawning biomasses were variable over the manage- 
