Brown et al.: Simple gear modifications for bycatch reduction in a shrimp trawl fishery 
383 
used to ensure consensus in the selection of gears to be 
tested. This process engaged industry participants and 
created ownership in the selections made, contributing to 
the project’s success. 
Some elements of the study design mandated by the 
NCMFC differed from typical collaborative research meth¬ 
ods and BRD testing. For example, paired tow testing is 
best conducted during normal fishing conditions so that the 
true effects of experimental gear, relative to the standard 
gears used, can be determined (Eayrs, 2012). A condition of 
the award that sponsored the testing in 2015 required that 
industry partners contribute a 50% cost match. Requiring 
large industry matching funds can change the perceptions 
and expectations of the industry sector relative to those of 
scientific collaborators (Harte, 2001). Specifically, projects 
that deviate from a collaborative approach (e.g., request¬ 
ing that fishermen fish out of season or, more importantly, 
fish by using experimental gear during normal conditions) 
operate outside of established parameters (Conway and 
Pomeroy, 2006). Collaborative research is difficult at best, 
and expectations of scientists and industry participants 
often do not align when only one of the partners—in this 
case, the shrimp industry—is asked to make concessions 
(Stephenson et al., 2016). 
Our methods followed NOAA BRD testing protocols 
with a few exceptions. First, the establishment of a goal of 
40% bycatch reduction over that achieved by the current 
industry trawl with TED and BRD installed was above 
and beyond normal testing design, requiring BRD combi¬ 
nations to be evaluated. Federal certification of a BRD pro¬ 
totype requires the use of the BRD to result in a consistent 
reduction of total finfish bycatch of at least 30%, by weight, 
when compared with the use of a naked (TED, but no 
BRD) control net (NMFS 11 ). Assuming, for instance, that 
by using our control net with a BRD we achieved a 30% 
reduction of finfish bycatch and by using our experimen¬ 
tal net we achieved our goal of a 40% reduction of finfish 
bycatch by weight, we theoretically would have achieved 
the equivalent of 58% reduction over a naked net or nearly 
twice the federal requirement. Second, the protocol for reg¬ 
ular switching of experimental gear to each side of the ves¬ 
sel was not always followed for a variety of reasons: lack of 
performance for a given gear, weather, or vessel captains 
not following agreed upon protocols. Finally, NOAA test¬ 
ing protocols require that testing achieve a minimum of 
30 successful paired tows. The sponsor of our field sea¬ 
son in 2015 requested that 60 successful paired tows be 
attempted but offered no scientific basis for deviation from 
the NOAA testing protocol of 30 paired tows (NMFS 11 ). 
Although the level of decrease in shrimp catch is not 
part of the NOAA testing protocol, it was discussed during 
workgroup meetings. The results of previous collaborative 
testing with industry in the Gulf of Mexico indicate that 
a 10% loss in shrimp catch may be acceptable, as long as 
bycatch is greatly reduced by the gear (Crowley, 2014). 
During discussions in our study, members of the work¬ 
group from the fishing industry indicated that a 3-5% 
decrease in shrimp catch would likely be acceptable but 
that acceptable loss would ultimately depend upon the 
level of bycatch reduction achieved. Because discussions 
on this topic happened largely before preliminary results 
were obtained, it will be worthwhile to reexamine the issue 
of shrimp loss in future collaborations now that significant 
finfish bycatch reductions have been observed. 
Gear evaluation 
The use of 2 statistical procedures to evaluate the paired 
tows resulted in similar outcomes in most cases. Unlike the 
paired Gtest, the randomization test did not assume that 
the data were normally distributed. Another advan¬ 
tage of the randomization procedure was that it did not 
require tows to be dropped from the analysis. Finally, the 
P-values computed in the randomization procedure are 
exact and not asymptotic. 
Experimental gear costs 
Fabrication and installation costs of the 14 experimental 
gear configurations varied considerably. However, cost 
was not a good predictor of performance. The estimates 
of gear costs included in Table 1 were provided by a local 
net maker and workgroup member; lower costs may be 
expected if fishermen build and install their own shrimp 
trawls. Further, we do not have costs for building a new 
experimental gear (costs are only for modifying control 
gear). Shrimp trawl components, particularly codends, 
have a finite life span and must be replaced with heavy 
use. Additional cost savings are likely if new gear is to be 
built rather than existing gear is to be modified. 
Recommendations 
Although our initial findings are promising, continued 
testing is necessary for repeatability and to ensure con¬ 
sistent and reliable performance by gears under variable 
conditions prior to widespread voluntary adoption (Murray 
et al., 1992). Shrimp catch and bycatch results varied con¬ 
siderably by vessel and between 2015 and 2016, although 
fishing was conducted in the same general area during the 
relatively short shrimp season. 
Further, we recommend continuing the industry work¬ 
group and evaluation of experimental gears and gear com¬ 
binations on both smaller class vessels (<12 m) and vessels 
that operate in the nearshore Atlantic Ocean (<4.83 km 
from shore) in both the summer brown shrimp and 
autumn white shrimp fisheries. Smaller vessels accounted 
for approximately 66% of the fleet and for 64% of the effort 
by number of trips over the 5-year period 2013-2017 in 
North Carolina (Bianchi 18 ). Although the majority of 
shrimp trawling effort in North Carolina typically occurs 
in estuarine waters, approximately 25% occurs in the 
nearshore Atlantic Ocean primarily <4.83 km from shore 
(i.e., in state managed waters). 
18 Bianchi, A. 2017. Personal commun. Div. Mar. Fish., North 
Carolina Dep. Environ. Qual., 3441 Arendell St., Morehead 
City, NC 28557. 
