290 
Fishery Bulletin 119(4) 
Table 9 
Variables that fall on each of the 3 dimensions of correspondence analysis with loading factors <—0.5 or 20.5. Variables describe 
depletion experiments conducted during 1997-2011 to examine efficiency of hydraulic dredges for capturing clam species and to 
estimate stock density for populations of ocean quahogs (Arctica islandica) and Atlantic surfclams (Spisula solidissima) off the 
mid-Atlantic coast of the United States. Variables include dredge width; dredge efficiency and its coefficient of variation (CV); effec- 
tive area swept (EAS); CV of the k parameter, which is the negative binomial dispersion parameter; error terms Err2, Err3, and 
Err4; CV of clam density; region, Long Island (LI) in New York, New Jersey (NJ), or the Delmarva Peninsula (DMV) of Delaware, 
Maryland, and Virginia; overlap score (OS), which describes the overlap of tows of the dredge; species; and latitude. N/A indicates 
that the variable fell close to the origin and not strongly on the positive or negative side of the dimension. 
Dimension 1 
Negative Positive 
Variable (<-0.5) (20.5) Variable 
Dredge width (m) 3.81 
Efficiency High Low OS 
2.54, 3.30 Dredge width 
Efficiency CV Low High Species 
EAS High Low EAS 
k-parameter CV N/A Low Region 
Err2, Err3, and Err4 N/A High Depth 
Density CV Low High Latitude 
Region N/A DMV Density CV 
used to create the distribution of clams. The k parame- 
ter is indirectly related to clam distribution and tow dis- 
tance (Hennen et al., 2012). Results from the simulations 
of Poussard et al. (2021) indicate that the estimates of k 
are higher with a uniform distribution of clams and lower 
with a more irregular distribution. This parameter is 
influenced by the same conditions of the experiment that 
influence efficiency, but correspondence analysis does not 
cleanly separate the k parameter or its CV from other vari- 
ables, such as the efficiency estimate, density estimate, 
depth, region, dredge width, and the values of CV, and 
CVp>. (Figs. 7 and 8). The results of correspondence analy- 
sis identify a tendency for low uncertainty in the k param- 
eter (and in CV) to be associated with low efficiency and 
high CV; in comparison with those of other experiments. 
Experiments with these characteristics also were flagged 
by the error terms. 
Poussard et al. (2021) reported that the dispersion of 
clams on the bottom clearly can cause a decrease in per- 
formance in the depletion experiment. This outcome is 
exacerbated by low tow number and low amount of tow 
overlap. In practice, even an ideal experiment, with many 
dredge tows and a high degree of overlap in the tow paths, 
would appear to be susceptible to producing an unreli- 
able efficiency estimate if the distribution of clams in the 
benthos is irregular. Clam dispersion is a random effect 
for the field experiments, despite its documented impor- 
tance in determining outcomes. This notion is consistent 
with the fact that the locations for the experiments were 
chosen without any a priori knowledge of the characteris- 
tics of clam dispersion at the site. As Hennen et al. (2012) 
concluded, the ability to have divers determine or to use 
remote optical methods to determine the size, location, 
Dimension 2 
Negative 
(<-0.5) (20.5) 
Dimension 3 
Negative Positive 
(<-0.5) (20.5) 
Positive 
Variable 
3.30 Dredge width 3.30 N/A 
N/A Efficiency N/A High 
Atlantic EAS N/A High 
surfclam 
N/A k-parameter CV High N/A 
NJ 
Low 
N/A 
N/A 
and approximate density of clam aggregations would be 
useful in choosing the site and tow pattern in a depletion 
experiment. 
Factors affecting field outcomes 
The size of the dredge is related to the efficiency esti- 
mated, with larger dredges being used with experiments 
with higher efficiency estimates. Small dredge sizes 
(2.54 m [8.33 ft] and 3.05 m [10.00 ft]) were used in many 
experiments, and they contributed disproportionately to 
the subset identified by estimates of Err2, Err3, and Err4 
(Fig. 6). It may be that smaller dredges are harder to con- 
trol precisely, leading to greater uncertainty in the exact 
position of the dredge, uncertainty that can lead to error in 
the estimation of efficiency (Hennen et al., 2012; Willberg 
et al., 2013). The majority of flagged experiments identi- 
fied by estimates of the 4 error terms were conducted in 
1997, 1999, and 2005, and among these experiments are 
those categorized as having lower efficiency estimates 
with more uncertainty in the estimate, relative to those of 
the other experiments. Although speculative, 2 possibili- 
ties may explain this trend. A wider dredge may be inher- 
ently more efficient because loss in efficiency is likely 
associated with the encounter of clams near the lateral 
edges of the dredge knife blade, and these clams are a 
lower fraction of the potential catch with the larger dredge. 
In addition, the narrow dimension of the depletion rectan- 
gle was generally set at 10 dredge widths; therefore, the 
largest dredge (3.81 m [12.50 ft]) was used to deplete 
larger regions, possibly reducing the influence of small- 
scale variations in clam dispersion within the depletion 
rectangle. It is noteworthy that experiments with the 
