Rogers et al.: Effectiveness of bycatch reduction devices in Louisiana inshore waters 
555 
net beginning 30 meshes from the tail. A3-mm chain 
was attached to the forward edge of this opening and 
a 10-mesh x 12-mesh long cover was attached to the 
rear edge of this opening. Half of a 3.8 x 7.6 cm float 
(“small float” in Fig. 2B) was attached under, and a 
4.5-mm rope was threaded along, the forward edge 
of this cover. A 13 x 28 cm conical float (“large float” 
in Fig. 2B) was attached 10 meshes behind the open- 
ing. The chain and floats created the escape opening. 
The Cameron shooter (Fig. 2C) was a 30-cm wide, 
45-cm long, 15-cm deep half cone of 12.7-mm alumi- 
num round stock. A 30-mesh opening was cut in the 
top of the trawl net, and the forward edge of the cone 
was inserted into the codend, 20 meshes back from 
the body of the trawl. The semicircular frame open- 
ing faced the codend and protruded inside the trawl 
net. 
The Eymard accelerator (Fig. 2D) had a polyethyl- 
ene-webbing accelerator funnel, 45 meshes in diam- 
eter and 24 meshes long surrounded by six 10 x 10 x 
10 mesh triangular openings. A 60-cm hoop of rub- 
ber coated cable was attached to pull the trawl net 
away from the funnel after initial dive tests indicated 
that the funnel blocked the escape openings. 
Personnel from the National Marine Fisheries 
Service (NMFS) Harvesting Systems Branch exam- 
ined the trawls several times by using scuba equip- 
ment, and adjustments were made to the trawl rig- 
ging and the BRD’s. Dye was injected into various 
parts of the nets around the devices to observe wa- 
ter flow, and the behavior of escaping fish was docu- 
mented. 
Sampling 
A6.7-m Boston whaler, powered by twin 115-hp out- 
board motors and equipped with a single-drum winch 
and double boom, was used to tow twin trawls off 
the stern (Harrington et al., 1988); one trawl con- 
tained a BRD, the other a bare control net. The nets 
were equipped with tickler chains, connected to an 
aluminum dummy door, and spread by two 0.6 x 
1.07 m pinewood trawl doors. Although twin trawls 
are not typically towed behind commercial vessels 
in Louisiana, the use of twin trawls to replace single 
large trawls off outriggers is increasing, particularly 
in the Gulf of Mexico (Watson et al., 1984). Smaller 
inshore vessels in Louisiana, without outriggers, typi- 
cally tow a single larger net behind the boat. Our 
twin-trawl rigging configuration was approved by the 
committee to ensure that the nets sampled an area 
as closely as possible, given the patchy nature of 
many species. Nets were towed for 20 minutes (time 
at towing speed); the speed over ground was main- 
tained between 2.0 and 2.5 knots (2.2 kn, average) 
by using a Global Positioning System, as recom- 
mended by the committee. Tows were made during 
daylight hours, near commercial shrimp boats when- 
ever possible. Average water depth and the salinity 
were recorded for each trawl tow. 
The four BRD’s were evaluated in each inshore 
shrimp zone (Fig. 1). Eighteen two-day sampling trips 
were made, three trips to each area during each sea- 
son. Each BRD net was towed 72 times over the year. 
Tows were divided among the areas and seasons. The 
towing order and trawl side for each BRD were ini- 
tially selected randomly, although the same nets were 
not used on consecutive tows owing to the time taken 
to empty the nets. The three control nets were num- 
bered and alternated to ensure equal pairing with a 
particular BRD net throughout the study. Sampling 
was conducted during the 1992 inshore shrimp sea- 
sons, although the short spring season necessitated 
sampling the week before the season opened in zone 1 
and a few days after the season in zone 3. This sched- 
ule was approved by the advisory committee. 
Samples were tagged and placed in mesh bags in 
ice and water. In the laboratory, organisms in each 
sample were identified, counted, and the biomass of 
each species in a sample was weighed to the nearest 
0.1 g. When numerous, individuals of a species within 
a sample were subsampled and the total number es- 
timated by weight. Standard lengths of most fishes, 
carapace widths of crabs, and total lengths of penaeid 
shrimp were measured. Organisms were measured 
in 5-mm length increments, designated by the lower 
end of the length range (e.g. 10-mm class=10.0 to 
14.9 mm). 
Statistical analysis 
Residuals were examined for univariate normality 
and homogeneity of variances prior to accepting the 
analysis of variance model. Normality was tested 
with the Wilk-Shapiro test, and a modified Levene 
test was used to test for homogeneity of variances. 
These tests indicated that the raw data were not dis- 
tributed normally and variances were not homoge- 
neous. The transformation ln(catch+l) was used to 
create a new variable that met the criteria of being 
approximately normally distributed with homoge- 
neous variances. This transformed variable was used 
in the analysis of variance (ANOVA). Statistical 
analysis was performed by using the Statistical 
Analysis System (SAS). 
Control nets The transformed catch ( both numbers 
and biomass of abundant species) of the three con- 
trol nets was used as the dependent variable in an 
ANOVA with season, area, and season-by-area terms 
