Contente and Rossi-Wongtschowski: Fish assemblages in the southeastern Brazilian Bight 
227 
26 A 
24 
22 
20 
18 
16 
14 
M 
32 
33 34 
35 
36 
BC C 
M M 
S 
o 
<D 
Q. 
E 
,0) 
.33 
34 35 36 37 
M M 
S Q 
o S S S 
s s s 
35.2 35.4 35.6 35.8 
Cf 
M 
M 
M 
26 
24 
22 
20 
18 
16 
14 
24 ' v 
22 
20 
18 
16 
14 | 
12 
35.0 
23 D 
22 
21 
20 
19 
18 
17 
33 
28 p 
26 
24 
22 
20 
18 
16 
14 
33 35 37 
Salinity 
Figure 2 
Plots of mean temperature versus 
mean salinity used for identification 
of the water masses at the mean depth 
where the tows were conducted during 
the spring-summer ECOSAR cruises: 
(A) III, (B) IV, (C) V, (D) VI, and (E) 
VII. The water masses are Coastal Wa- 
ter (C), South Atlantic Central Water 
(S), and Mixed Water (M) 
M 
M 
S 
34 
C 
35 
36 
ing indicates a wide distribution and persistency over 
the SBB and over the years of this study (1995-2010). 
The Atlantic cutlassfish and Brazilian sardinella, also 
abundant, displayed high frequency in the aggregations 
over the years (FO T =70% and 51%, respectively), also 
indicating wide distribution and persistency (Table 4). 
Large schools of the flying gurnard were relatively com- 
mon in aggregations from 4 cruises, IV-VII (i.e. , from 
2008 through 2010), having been captured principally 
(68% of 25 tows) at midwater depths (>3 m above the 
bottom) and secondarily (32%) on the bottom (<3 m). 
Large schools of the Atlantic chub mackerel (~3 metric 
tons [t], cruise IV), gray triggerfish (~0.7 t, cruise III), 
American coastal pellona (-0.6 t, cruise VI), and blue- 
fish ( Pomatomus saltatrix; - 0.2 t, cruise V) were domi- 
nant in aggregations during only one cruise (Table 4). 
The other species showed both low abundance and low 
frequency (Table 4), reflecting either association with 
dominant species or scattered, low-biomass groupings. 
The effect of water masses on fish assemblage structure 
We found significant differences in the structure of fish 
assemblages among the water masses, but not among 
the ECOSAR cruises (Table 5). This result was not af- 
fected by changes in the order of insertion of terms 
of the PERMANOVA model (Table 5). Moreover, this 
significant result was due to the location (which indi- 
cated difference) of the fish assemblage structure of 
each water mass in multivariate space rather than 
to differences in the dispersion within assemblages of 
each cruise (PERMDISP, F 1 76 =4.16, P>0.05). The post- 
hoc permutational t-test revealed that the assemblage 
structure of SACW differed significantly from those of 
CW and M (SACWxCW: t=14.12, P=0.013; SACWxM: 
t= 14.79, P=0.003), but the assemblage structures of the 
remaining 2 water masses did not differ significantly 
(CWxM: t=0.94, P>0.05). Consequently, the data of CW 
and M were grouped together before SIMPER analysis 
was begun. 
The main species responsible for the difference in 
assemblage structure between the SACW and the com- 
bination of CW and M (CW+M) — the species that cumu- 
latively accounted for -75% of the difference — were the 
Argentine anchoita, Atlantic cutlassfish, rough scad, 
piquitinga anchovy, and white snake mackerel (Thyr- 
sitops lepidopoid.es), all of which were most abundant 
in SACW, and the Brazilian sardinella, flying gurnard, 
false pilchard ( Harengula clupeola), American coastal 
pellona, and Atlantic thread herring ( Ophistonema 
oglinum), all of which were most abundant in CW+M 
(Table 6). 
Patterns of species association 
The resulting cluster for identifying species associa- 
tions (Fig. 3) indicates the existence of 3 groups of spe- 
cies: group A, formed principally (70% of species sam- 
pled) by species that were most abundant in the SACW 
(e.g., the Argentine anchoita, Atlantic cutlassfish, and 
