28 
Fishery Bulletin 11 6(1) 
Table 2 
Slope, intercept, and P-value for linear regressions of 
vertical motion, expressed as the standard deviation of 
off-bottom distance (in centimeters) at various positions 
along the lower bridles and footrope, against vessel mo¬ 
tion, expressed as standard deviation of vessel heave at 
the trawl block from an experiment conducted in the 
eastern Bering Sea during September 2003. 
Position 
Intercept 
Slope 
P-value 
Center 
0.893 
0.0087 
0.005 
Corner 
1.016 
0.0071 
0.019 
Wing 
1.022 
0.0038 
0.103 
Bridle—25 m 
0.997 
0.0066 
0.0073 
Bridle—40 m 
0.803 
0.0274 
<0.001 
Bridle—50 m 
2.349 
0.0439 
<0.001 
tions at 5 levels of SDH (0, 20, 40, 60, 80 cm) and then 
plotted the predicted off-bottom distances as a function 
of SDH (Fig. 5). Comparing the off-bottom distances in 
this way emphasizes that, in terms of a mean response, 
the primary influence of SDH on off-bottom distance 
occurs in the region between the 25- and 40-m posi¬ 
tions, which is near the point where the bridle leaves 
the bottom in calm conditions (-31 m from the wing; 
Somerton and Munro, 2001). If a value of off-bottom 
distance is specified as a critical value for herding (e.g., 
in Fig. 5 a value of 2.5 cm is used), then as SDH in¬ 
creases, the bridle position at the critical value of off- 
bottom distance moves toward the wing tip, reducing 
the area exposed to herding. 
Although the change in the mean off-bottom distance 
at the center and corner positions on the footrope are 
not significant, the slopes of the regressions are nearly 
as large as those at the 40-m bridle positions (Table 
3), but there is more haul-to-haul variability. This re-' 
lationship indicates that the oscillations are also likely 
to impact escapement along the footrope. 
Influence of wave height on the herding of yellowfin sole 
In the yellowfin sole herding experiment, the catch ra¬ 
tio declined significantly with increasing wave height 
(,ff=1.96 - (O.Slxwave height), P=0.038, coefficient of 
determination [r 2 ]=0.24; Fig. 6). Because the catch ra¬ 
tio was based on pairs of tows taken near each other, 
in the same direction, and in quick succession, essen¬ 
tially all influences on catch were kept constant, except 
that of bridle length. Therefore, the decrease in catch 
ratio with increasing wave height indicates that herd¬ 
ing was decreased. 
Our hypothesized model expressing the influence of 
wave height on catch ratio is shown below with the 
estimated values of k l and k s : 
K w n +0.58(2sin(a)(L' bt -(36 + 2.29 H)) 
~ w n + 0.58(2sin(a)(L bt -(36 + 0.042#))' 
The estimated value of k l is considerably larger than k s , 
confirming the notion that the impact of wave height 
would be stronger for the long bridles. However, the fit 
of the model to the catch ratios was poorer (r 2 =0.12) 
than the fit of the wave height data itself to the catch 
ratios (r 2 =0.24). This result indicates that either the 
model was incorrectly specified, perhaps lacking other 
important factors influenced by wave height, or that 
the fixed parameter estimates (i.e., h and L 0 ^) obtained 
from previous studies were biased. 
Influence of wave height on annual estimates of yellowfin 
sole abundance 
The biomass of yellowfin sole estimated by the EBS 
bottom trawl survey, during the period 2005-2016, de¬ 
clined significantly with an increase in the mean wave 
height (biomass=3.37 - (0.67xwave height), P=0.028, 
r 2 =0.40; Fig. 7). The deviations between the survey 
biomass estimates and those produced by the yellow¬ 
fin sole stock assessment model, during the same time 
period, also declined significantly with an increase in 
the mean wave height (deviation=l.ll - (0.19 xwave 
height), P=0.044, r 2 =0.35; Fig. 7). This result indicates 
that wave height, perhaps through its apparent effect 
on herding, may produce a detectable bias in the an¬ 
nual estimates of yellowfin sole abundance in the EBS. 
Discussion 
Wave height and vessel motion 
It is clear that wave height influences vessel motion. 
This was shown by the strong correlation of SDH with 
estimated wave height. However, SDH was insensi¬ 
tive to vessel heading relative to wave direction, which 
is counterintuitive to most observers standing on the 
deck of a vessel. An alternate measure of vessel motion 
proposed by Politis et al. (2012) divides heave by time 
to create a measure of vertical speed, which is sensi¬ 
tive to vessel heading relative to the waves. We chose 
SDH as a measure of vessel motion because it is con¬ 
sistent with the subjective estimates of wave heights 
now provided by vessel captains and which are the ba¬ 
sis of our time series of sea states encountered during 
trawl surveys. 
An objective, repeatable, measure of wave height 
obtained with a vessel heave sensor is definitely pre¬ 
ferred to the subjective estimates now provided by ves¬ 
sel captains. Although heave sensors are nearly always 
included in the instrument suite on fishery research 
vessels, all NOAA bottom trawl surveys conducted by 
chartered fishing vessels along the U.S. West Coast do 
not use heave sensors (Stauffer, 2004); therefore the 
only available information on sea-state is the wave and 
swell heights estimated by the vessel captains. Such 
estimates are inadequate because they are subjective, 
likely to vary among vessel captains, and do not cap¬ 
ture variability in vessel motion introduced by factors 
