24 
Fishery Bulletin 11 6(1) 
block (i.e., the last point of contact between the vessel 
and the warp) and measured its coordinates relative to 
the heave sensor with a surveyor’s tape. In addition, 2 
measures of sea state, swell height and wave height, 
were estimated by the vessel captain during each tow. 
Recognizing that these are subjective estimates, we did 
not try to disentangle the exact meanings of the 2 val¬ 
ues, but instead followed the approach of Stewart et 
al. (2010) and simply added them together to create 
a composite value of sea state, henceforth referred to 
as wave height. In addition, the captain estimated the 
wave direction and determined the angular difference 
between wave direction and vessel heading. This direc¬ 
tional difference was recorded both as a numeric value 
(i.e., the absolute value of the cosine of the angle be¬ 
tween the 2 directions) and as a relative value (i.e., the 
vessel heading either toward or away from the wave 
direction). 
Analysis of wave height, vessel motion, and trawl motion 
data The data for each tow consisted of 1) wave height 
and the difference between wave direction and vessel 
heading, 2) remote heave at the trawl block, and 3) off- 
bottom distances at each of the 11 positions along the 
lower bridles and footrope. Mean and standard devia¬ 
tion of the heave and off-bottom measurements were 
computed for each of the 62 tows but the number of 
off-bottom values varied among the measurement posi¬ 
tions because of occasional malfunction of some BCS 
units. Henceforth, the standard deviation of the heave 
at the trawl block (SDH) will be used as a measure of 
vessel motion. 
We examined whether SDH is influenced by vari¬ 
ables other than wave height, in particular, the numer¬ 
ic and relative differences between vessel heading and 
the wave direction, by including both variables, along 
with wave height, in a linear model (using the function 
lm in R, vers. 2.0.0; R Development Core Team, 2004) 
of SDH. Analysis of variance was used to test for the' 
significance of the terms. Nonsignificant variables were 
removed and the model was re-fit sequentially until all 
remaining variables were significant. In addition, we 
used linear models to examine whether the mean and 
standard deviation of off-bottom measurements at each 
BCS position were influenced by SDH, using analysis 
of variance to test the significance of SDH. 
Because the driving variable for movement of the 
vessel, and potentially the trawl, was surface waves, 
the time series of vertical measurements from each 
trawl location can display periodic oscillations. As did 
O'Neill et al. (2003) and Politis et al. (2012), we as¬ 
sumed that coherence in the period of oscillations be¬ 
tween the vessel and the various measurement loca¬ 
tions on the trawl indicated that surface waves were 
the causative agent of trawl motion. Therefore, for 
some tows, measurements of heave and off-bottom dis¬ 
tance were subjected to spectral analysis (spectrum 
function in statistical software S+, vers. 6.2, TIBCO 
Software, Inc., Palo Alto, CA; Venables and Ripley, 
1994) to determine the major period of oscillation. The 
period was estimated as the inverse of the frequency 
at the spectral peak, multiplied by the sampling inter¬ 
val (the sampling interval differed among the various 
types of data). 
Wave height and herding of yellowfin sole 
Experimental design Flatfish are typically herded into 
the net path by direct contact with or close approach of 
the lower bridles. The proportion of the fish within the 
bridle contact area (i.e., area of the bottom contacted 
by the bridles) that are herded into the net path can be 
determined by using a herding experiment (Engas and 
God0,1989) in which the size of the bridle contact area 
is varied by repeatedly conducting trawl hauls with 
varying bridle lengths. In previous experiments with a 
survey trawl, this species was strongly herded (-60% 
of fish within the contact area were herded into the 
trawl path; Somerton and Munro, 2001), therefore yel¬ 
lowfin sole was considered a good candidate species for 
examining the influence of surface waves on herding. 
An experiment to examine the effects of surface 
waves on the herding of yellowfin sole, patterned af¬ 
ter the experiments described in Somerton and Munro 
(2001), was conducted aboard the 56.7 m stern trawler 
FV Cape Flattery during 11-16 August 2016 in the 
EBS near 60°33'N and 170°29'W. On each day of the 
experiment, 3 paired tows were completed according to 
all EBS survey operating protocols (Stauffer, 2004), ex¬ 
cept that for one tow of each pair, 27.4-m bridles were 
used, and for the other tow of the pair, 82.3-m bridles 
were used, instead of the standard 54.9-m bridles. The 
tow order of each bridle length was alternated in suc¬ 
cessive pairs to help reduce the potential for a tow 
order effect. Although the tow direction of each pair 
could vary with wind and wave direction, tow direction 
within each pair was always identical. During each 
tow, net and door spread was measured to the nearest 
0.1 m with an acoustic mensuration system (Marport 
Deep Sea Technologies Inc., Saint John’s, Canada) and 
tow length was measured with GPS from the first to 
last bottom contact of the footrope, determined by us¬ 
ing a BCS. Yellowfin sole were sorted from each catch 
and collectively weighed but not counted. Catch weight 
in each tow was later standardized by dividing by the 
swept area of the net (i.e., net width multiplied by tow 
length). Swell and wave heights were also estimated 
for each tow and, as in the previous experiment, were 
summed together and referred to as “wave height.” 
However, unlike the earlier sea-state experiment, we 
did not have a heave recorder and therefore wave 
height was used as the descriptor of sea state. Over the 
6 d of the experiment, 18 pairs of tows were completed. 
Analysis of the herding data An index of the strength 
of herding for each pair of tows was calculated as the 
catch ratio (i.e., the standardized catch from the long 
bridle tow divided by that of the short bridle tow in 
each tow pair). Because the 2 bridle lengths result in a 
strong contrast in bridle contact area, any factor that 
