Table 7 
Results of Wilcoxon rank sums tests for depletion experiment 
variables classified by error terms. Experiments were flagged 
because estimates of variables were at or above the 80th 
percentile for one or more error terms. The tests were used 
to evaluate relationships between data from the field exper- 
iments flagged by one or more error terms with the rest of 
the data set from experiments conducted during 1997-2011 
for ocean quahogs (Arctica islandica) and Atlantic surfclams 
(Spisula solidissima) off the mid-Atlantic coast of the United 
States. A dash indicates that the relationship between an 
Poussard et al.: Discriminating between high- and low-quality field depletion experiments 287 
in efficiency estimates and potentially have high average 
error in efficiency themselves. Nine out of 10 experiments 
flagged by Err4 were also flagged by Err3, as these 2 error 
terms are very similar. These experiments may be unin- 
formative or may have produced inaccurate efficiency 
estimates. 
Ultimately, because of the nature of the error terms 
and the inability to evaluate all possible experimental 
conditions (e.g., all possible tow numbers and clam dis- 
tributions, or all possible true efficiencies), the inference 
that the flagged experiments produced uninformative or 
inaccurate efficiency estimates cannot be affirmed. In 
aggregate, however, the analysis of the field experiments 
points to a subset of field experiments of lower quality 
than the remainder. 
Interestingly, the experiments flagged by Err1, which 
might identify suspect experiments, exert less influence on 
P>|Z| the final objective of determining the efficiency of hydrau- 
lic dredges. The distribution of these experiments is unbi- 
ased relative to the remaining experiments, regardless of 
the characteristic used for comparison (Table 5). The same 
cannot be said for Err2, Err3, and Err4. The series of 16 
depletion experiments with estimates of Err2, Err3, and 
Err4 that fall at or above the 80th percentile is clearly 
biased relative to the remaining experiments, on the basis 
of the results of Wilcoxon rank sums tests (Table 7), and 
the existence of this bias is reinforced by correspondence 
analysis (Figs. 6-8). In addition, the direction of bias is 
noteworthy. Experiments identified by Err2, Err3, and 
Err4 are characterized by lower efficiency estimates on 
error term and a variable are not significant (P>0.05). Vari- 
ables that describe the experiments include dredge efficiency; 
clam density; the k parameter, which is the negative binomial 
dispersion parameter; effective area swept (EAS); and num- 
ber of tows of the dredge. CV=coefficient of variation. 
Error term 
Variable Err1 Err2 Err3 
Efficiency 0.0454 0.0004 
Efficiency CV 0.0180 
Density ~ 
Density CV 0.0250 
k parameter = 
k-parameter CV 0.0340 
EAS 0.0001 
No. of tows - 
Table 8 
Comparison of mean and median values of model parameters between depletion experiments with parameter estimates at or above 
the 80th percentile for error terms Err2, Err3, and Err4 and experiments with estimates below the 80th percentile for the same error 
terms. The parameters are dredge efficiency and its coefficient of variation (CV); the CV of the k parameter, which is the negative 
binomial dispersion parameter; clam density and its CV; overlap score (OS), which describes the overlap of tows of the dredge; and 
effective area swept (EAS). Data used in the model are from depletion experiments conducted during 1997—2011 for populations of 
ocean quahogs (Arctica islandica) and Atlantic surfclams (Spisula solidissima) off the mid-Atlantic coast of the United States. 
Error term 
Err3 
Type of <80th >80th <80th >80th <80th >80th 
Parameter value percentile percentile percentile percentile percentile percentile 
Efficiency Mean 0.654 0.464 0.667 0.384 0.666 0.419 
Median 0.645 0.551 0.652 0.381 0.652 0.435 
Efficiency CV Mean 19.496 134.744 44.402 34.089 45.202 31.920 
Median 19.232 25.299 16.789 32.075 17.325 27.392 
k-parameter CV Mean 41.785 59.215 42.368 58.497 42.644 55.783 
Median 32.924 31.953 33.139 29.869 33.257 29.913 
Density (individuals/m”) Mean 1.389 1.356 1.345 1.507 1.367 1.421 
Median 0.743 0.887 0.743 0.969 0.743 0.915 
Density CV Mean 162.998 3933.081 1054.153 292.294 1077.931 273.350 
Median 135.765 189.542 132.375 238.345 132.924 205.655 
OS Mean 0.948 0.997 0.967 0.917 0.969 0.915 
Median 0.911 1.103 0.926 1.020 0.940 0.961 
EAS (m?) Mean 18,048.6 99,251.5 40,370.7 6584.3 40,827.9 8134.3 
Median 15,489.3 10,666.3 16,066.2 6302.6 16,050.7 6400.2 
