Poussard et al.: Discriminating between high- and low-quality field depletion experiments 279 
values of each of the 4 characteristics for a given field deple- 
tion experiment fell above or below the mean value for the 
characteristic from the simulation data set. This compari- 
son generated a 4-digit integer sequence (e.g., 1011) for any 
given field experiment with the numeral 1 assigned if the 
field experiment measure fell above the mean of the simu- 
lated experiment measures and a with the numeral 0 
assigned if the measure fell below the mean. 
The same set of integer sequences were calculated for 
each simulation and compared with the mean of the char- 
acteristics for all simulated experiments. Then the subset 
of simulations that have the same sequence as the field 
experiment was extracted from the data set. The means of 
the measures for this subset were again calculated and com- 
pared with those of the field experiment, generating a new 
4-digit sequence. This sequence, in turn, was used to extract 
a subset of simulated experiments of identical sequence. 
This process was repeated sequentially on each extracted 
subset, with the mean values for the simulated experi- 
ments being updated by using only the extracted subset, 
until none of the final subset of simulations had the same 
4-digit value as the chosen field experiment. This group of 
simulations was considered the most comparable to the field 
experiment in question out of all the simulations. This “most 
comparable” subset typically numbered 2-20 of the 9000 
simulations and was used to describe the average values 
for the simulated 4 characteristics and the average error in 
efficiency most appropriate for comparison with the known 
values (Tables 2 and 3). 
Each simulation in the extracted subset of simulations 
was run by using a specified dispersion of clams. The dis- 
tributions of clams were organized as follows: a relatively 
Table 2 
Parameters from 19 simulations that were most comparable with data from the depletion experiment 
OQ08-02. Type of clam distribution is denoted as uniform (NP), vertical bands (P), clams in half the 
area (HP), and diagonal across the area (T). Characteristics used to describe the experiments are the 
effective area swept (EAS), the coefficient of variation (CV) for the efficiency estimate (CV), and the CV 
of the k parameter (CV,), which is the negative binomial dispersion parameter. The estimated efficiency 
of depletion experiment OQ08-02 is 0.79507. The data set used in the simulations came from depletion 
experiments conducted during 1997-2011 for ocean quahogs (Arctica islandica) off the mid-Atlantic 
coast of the United States. See Figure 1 for the locations where the experiments were conducted. 
Simulation and OQ08-02 values 
Average values from simulations 
Values from experiment OQ08-02 
Mean absolute error in efficiency estimate: 
Parameter 
CVp CV_ EAS (m’) 
5.477 41.966 
9.361 35.067 
21,331.1 
24,604.8 
0.1003 Range: 0.0001—0.4000 
Parameter 
Error in 
efficiency 
estimate 
Density Clam 
(individuals/m”) 
0.6667 0.75 HP 
0.6667 1.50 
0.6667 3.00 
0.1517 0.75 
0.0633 0.75 
0.1500 0.75 
0.1050 1.50 
0.0633 1.50 
0.0883 1.50 
0.6667 0.75 
0.6667 0.75 
0.6667 1.50 
0.6667 3.00 
0.0905 0.75 
0.0572 0.75 
0.0359 0.75 
0.0778 0.75 
0.0017 0.75 
0.0483 3.00 
Ze ZZ, 
PUUARZgyseavvyyy Ae 
distribution efficiency 
True CV; CVx_ 
estimate estimate 
EAS (m7?) 
38.4412 
38.6189 
37.5421 
47.8368 
46.6667 
47.6762 
46.1634 
47.7372 
47.5397 
41.7476 
41.2644 
41.0180 
41.6268 
42.9225 
41.7308 
40.6921 
48.6502 
48.5524 
38.5597 
22,601.94 
22,601.94 
22,771.00 
22,551.50 
22,696.44 
23,019.59 
22,653.81 
22,696.44 
23,019.59 
22,601.94 
22,771.00 
22,771.00 
22,601.94 
20,596.54 
20,775.93 
20,679.71 
20,596.54 
20,807.33 
20,807.33 
