230 
Fishery Bulletin 1 14(2) 
Species 
— planehead filefish 
— lookdown 
— flying gurnard 
— brown burrfish 
— piquitinga anchovy 
— great barracuda 
— American harvestfish 
— gray triggerfish 
— largehead hairtail 
— Argentine anchovy 
— Brazilian sardinella 
— white snake mackerel 
— Atlantic chub mackerel 
— rough scad 
— bluefish 
— Atlantic thread herring 
— Castin leatherjacket 
— false herring 
— Atlantic bumper 
— Atlantic moonfish 
— American coastal pellona 
— dogtooth herring 
— ! 
100 
Figure 3 
Resulting cluster determined with the Ochiai similarity 
index matrix and used to assess the pattern of co-occur- 
rence of the main fish species (those species that totaled 
93% of the relative total frequency of occurrence) sampled 
in the spring-summer ECOSAR IV-VII cruises. The spe- 
cies groupings ( i.e. , groups A, B, and C) were arbitrarily 
defined at a cutoff level of 20% similarity. 
Table 7 
Summary of results from analysis, conducted with the 
distance-based, nonparametric linear modeling (available 
in DistLM in the PERMANOVA+ add-on for the software 
PRIMER 6) of the relationship between multivariate struc- 
ture of pelagic fish assemblages and oceanographic and 
spatial variables sampled during spring-summer ECOSAR 
IV-VII cruises in the southeastern Brazilian Bight. Data 
from these cruises (63 tows) were pooled because there was 
no interannual effect on the assemblage structure. The 
spatial variables were distance from shore (DS) and mean 
depth of the aggregation in the water column (Z a ). Models 
selected by the Bayesian information criterion and Akaike 
information criterion. Cumul. cumulative percentage of 
the variability of the fish assemblage structure explained 
by the model. SS=sum of squares. 
Variable SS Pseudo-F P-values Cumul. (%) 
Z a 7771.2 1.818 0.0190 3.0 
DS 9323.6 2.263 0.0036 7.0 
rough scad) and were principally responsible for the 
differentiation of the SACW assemblage structure 
from the CW+M assemblage structure (Table 6); 
group B, formed mostly (85% of species sampled) by 
species that were more abundant in the CW+M (e.g., 
the false pilchard, American coastal pellona, Atlantic 
thread herring, and castin leatherjacket [Oligoplites 
saliens]) and were relevant for distinguishing the 
assemblage structures of CW+M and SACW (Table 
5); and group C, a mixed grouping, composed of no- 
diagnostic species in SIMPER (Table 6) and of those 
species that were diagnostically abundant in either 
the CW+M (e.g., flying gurnard) or the SACW (e.g., 
the piquitinga anchovy) assemblage. 
The effect of environmental and spatial factors on fish 
assemblage structure 
Because there was no significant variation in fish 
assemblage structure among the ECOSAR cruises 
(Table 5), data from all cruises were pooled to pro- 
duce only one model for assessing the relationship 
between the variability in the assemblage structure 
and the oceanographic and spatial variables. The 
most parsimonious model was formed of 2 spatial 
variables: distance from shore and Z a (Table 7). 
These variables explained 7% of total variation in 
fish assemblage structure. The resulting pattern 
among samples that was revealed on the constrained 
dbRDA ordination for this model (Fig. 4) indicates 2 
clear gradients of shifts in the assemblage structure 
related to such variables. The vector of groupings of 
species (Fig. 3) on these gradients (Fig. 4) revealed 
2 principal tendencies: 1) an increase in abundance 
of SACW-associated species with increasing Z a and 
a decreasing distance from shore and 2) an increase 
in abundance of species associated with CW+M with 
decreasing Z a and an increasing distance from shore. 
Discussion 
Our findings reveal that the pelagic aggregations of 
fishes over the SBB shelf were dominated by highly 
abundant mid-trophic-level schooling species in as- 
sociation with predatory fishes (of which the Atlantic 
cutlassfish was the most representative). That result 
is consistent with the shelf pattern typically observed 
worldwide, including in the California Current (Litz 
et al., 2014), Colombia (Duarte and Garcia, 2004), 
Humboldt Current (Fablet et al., 2012), Southern 
Brazil (Soares et al., 2005), and Benguela (Pecquerie 
et al., 2004). In our study, the biomass of the aggrega- 
tions was dominated by Argentine anchoita, followed 
by rough scad, Atlantic chub mackerel, American 
coastal pellona, Brazilian sardinella, flying gurnard, 
and Atlantic cutlassfish, all of which may play a key 
role in the pelagic ecosystem of the SBB. 
Such patterns of dominance and aggregations 
were consistent with those previously found for the 
