102 
Fishery Bulletin 106(1 ) 
A B 
Figure 5 
Canonical correspondence Analysis (CCA) ordinations of the beam trawl survey data (1991-95) 
displaying the station assemblages for midsummer only (A), late summer only (B), and the species 
assemblages for midsummer only (C) and late summer only (D). Solid lines within each figure box 
(A-D) enclose the boundaries of the identified assemblages. In A and B, each sampling station is 
identified by a different symbol and the arrows depict the gradient of each environmental variable. 
In C and D, species that occupy the same area of the graph are grouped by short lines and arrows 
denote their true locations. 
near-ridge, and offshore). Within the Beach Haven Ridge 
beam trawl samples, 55% of the fish used in the analysis 
were found in both species assemblages, leaving nearly 
half of the species captured to be found in only one 
assemblage or the other. This is in stark contrast to the 
2001-06 Beach Haven Ridge otter-trawl data subset, 
where 82% of the species were found in at least two of 
the three assemblages and only 18% of the species (silver 
perch, Atlantic silverside [Menidia menidia ], northern 
kingfish [ Menticirrhus saxatilis], and Gulf stream floun- 
der t Citharichthys arctifrons ]) were present in only one 
assemblage. Thus it appears that there is a definite 
gradient along the transect, represented by changes in 
species present in the beam trawl and by the relative 
abundances of shared species in the otter trawl. 
Although cross-shelf gradients in demersal fish as- 
semblages have been identified along the northeast 
United States (Sullivan et ah, 2000), northwest United 
States (Mueter and Norcross, 1999), southwest United 
States (Johnson et ah, 2001), and worldwide (Gray and 
