de Silva and Condrey: Patterns in patchy data discerned from Brevoortia patronus bycatch 
203 
Table 5 
Comparison of reduced logit models with the full model. Diff 
full model and reduced model. 
= difference between 
Likelihood 
ratio statistics 
Model 
comparisons 
df 
G 2 
P>G 2 
Diff 
(df) 
Diff 
(G 2 ) P- Value 
Full model 
Season= Apr-Jun, Jul-Aug, Sep-Oct 
Area = 11-12, 13-14, 15-16, 17-18 
12 
17.82 
0.1213 
Reduced model I 
Season=spring, summer, fall 
Area = 11-12, 13-18 
6 
9.76 
0.1352 
6 
8.06 0.237 
Reduced model II 
Season=Apr-Aug, Sep-Oct 
Area = 11-12, 13-14, 15-16, 17-18 
8 
9.99 
0.2659 
4 
7.83 0.098 
Reduced model III 
Season= Apr-Aug, Sep-Oct 
Area = 11-12, 13-18 
4 
7.54 
0.1100 
8 
10.28 0.246 
2 Test for seasonal differences 
in the odds of observing a set 
with high bycatch given the 
set was sampled west of the 
Mississippi River; 
3 Test for area differences in 
the odds of observing a set 
with high bycatch given the 
set was sampled in the early 
season; and 
4 Test for area differences in 
the odds of observing a set 
with high bycatch given the 
set was sampled in the late 
season. 
The contrasts were written 
similarly to those for the “full 
model.” For the four contrasts, 
we used an a level of 0.025 as 
significant. 
Of the four contrasts, two 
were significant. The first indi- 
cated that in areas east of the 
river, the odds of observing a set 
with high bycatch was significantly different between 
sets sampled in the early season and sets sampled 
in the late season (Wald % 2 =11.41, df=l, P>X 2 - 
0.0007 ). The odds of observing a set with high bycatch 
east of the river in the early season was 11 times 
greater than the odds of observing a set with high 
bycatch in the late season. The second significant 
contrast indicated that for sets sampled in the early 
season, the odds of observing a set with high bycatch 
east of the river was significantly different from the 
odds of observing a set with high bycatch west of the 
river (Wald x 2 =7.99, df=l, P>y 2 =0.QQ47). In the early 
season, the odds of observing a set with high by catch 
east of the river was 2.7 times greater than observ- 
ing a set with high bycatch west of the river during 
the same period. 
Bycatch species associations 
Of the 62 species groups observed, 20 occurred in two 
or fewer sets. The most frequently occurring species 
were Atlantic cutlassfish, Trichiurus lepturus (44% 
of sets), Atlantic croaker, Micropogonias undulatus 
(38% of sets), Spanish mackerel, Scomberomorus 
maculatus (36% of sets), sand seatrout, Cynoscion 
arenarius (35% of sets), and gafftopsail catfish, Bagre 
marinus (34% of sets). In terms of total abundance 
(Table 6), Atlantic croaker, sand seatrout, and Atlan- 
tic bumper, Chloroscombrus chrysurus, accounted for 
71% of the total releasable bycatch. 
Species included in the main table were Atlantic 
croaker, sand seatrout, crevalle jack, Caranx hippos, 
gafftopsail catfish, Spanish mackerel, and Atlantic 
cutlassfish (Table 6 ). Species included as supplemen- 
tary variables were striped mullet Mugil cephalus, 
unidentified requiem sharks, gulf butterfish, Peprilus 
burti, cownose ray, Rhinoptera bonasus, spotted 
seatrout, Cynoscion nebulosus, Atlantic bumper, 
blacktip shark, Carcharhinus limbatus, red drum, 
Sciaenops ocellatus, unidentified penaeid shrimp, 
hardhead catfish, Arius felis, brown shrimp, Panaeus 
aztecus, cabbage head jellyfish, Stomolophus 
meleagris , bull shark, Carcharhinus leucas, and uni- 
dentified tonguefish (Soleidae) (Table 6). 
The fate of releasable bycatch Correspondence analy- 
sis on the fate-by-species table for the entire fishing 
season indicated that the first two axes explained 97% 
of the total inertia (conceptually similar to variance) 
and offered a good representation of the fate-species 
associations. From the two-dimensional plot (Fig. 6) we 
discerned three major and one minor groupings. 
Species primarily associated with being released 
dead or disoriented were unidentified requiem 
sharks, red drum, crevalle jack, and bull sharks. Spe- 
cies secondarily associated with being released dead 
or disoriented were cownose rays and blacktip 
sharks. These last two species were primarily asso- 
ciated with being released healthy and appeared to 
form their own minor grouping. 
