Poussard et al.: Discriminating between high- and low-quality field depletion experiments 285 
Efficiency 
Efficiency CV Efficiency CV 
k parameter 8.06) 0.20 k parameter 
k-parameter CV k-parameter CV 
Dredge width Dredge width 
No. of tows No. of tows 
om 0-0 9 6 
| nf  o an 
Figure 4 
Correlogram for depletion experiments that targeted ocean quahogs (Arctica islandica) off the 
mid-Atlantic coast of the United States between 1997 and 2011. Numerals in the squares are 
Pearson’s correlation coefficients. Circles denote significant correlations (P<0.05). White circles 
indicate positive correlations, and gray circles indicate negative correlations. Characteristics used 
to describe experiments include dredge efficiency and its coefficient of variation (CV); the k param- 
eter, which is the negative binomial dispersion parameter; CV of the k parameter; effective area 
swept (EAS); depth; dredge width; number of tows of the dredge; year; latitude; and clam density 
and its CV. 
The parameters describing clam distribution do not fall 
on any axis and are grouped in the middle of the corre- 
spondence analysis plots on all dimensions; therefore, they 
are not included in the figures to improve clarity. Although 
clam distribution clearly affects the outcome of individual 
experiments, as observed through simulation analysis 
(Poussard et al., 2021), this effect is distributed across the 
experimental spectrum, apportioning uncertainty in a rel- 
atively random way with respect to the field experimental 
data set. 
Discussion 
Estimation of efficiency 
The 4 error terms identify field depletion experiments 
that engender misgivings about their quality. Because 
the 4 characteristics used to generate 2 of the error 
estimates (Err1 and Err2), CVp, CVx, OS, and average 
EAS, are orthogonal to each other, identification of a sub- 
set of experiments flagged by Erri and Err2 indicates 
that these experiments are characterized by an unusual 
distribution of these 4 descriptive measures. It is import- 
ant to note that using a log transformation of EAS did not 
change which experiments were flagged by these 2 error 
terms. A close fit to the values of these 4 characteristics 
was not found among the 9000 simulations of Poussard 
et al. (2021) that covered a wide range of experimental 
protocols and field conditions of clam dispersion (Table 1). 
The absence of a close fit is reason to suspect that these 
experiments may be uninformative or may have produced 
inaccurate efficiency estimates. 
Error terms Err3 and Err4 relate to inferred inaccu- 
racies in the efficiency estimates, also gleaned from com- 
parison with the simulation data set of Poussard et al. 
(2021). Experiments flagged by these error terms are 
most comparable to simulations with high average error 
