FISHERY BULLETIN: VOL. 81, NO. 3 



where S 



X,, 



= S, min(X„, X ik ) 



= similarity between entities; and k 

 = value of the ith attribute for entity; 

 = value of the ith attribute for entity 

 k. 



The Bray-Curtis similarity measure was chosen, 

 since it most accurately reflects similarity in these 

 types of analyses (Bloom 1981). 



Normal analysis compared similarities among sites 

 as indicated by the assemblages of fishes collected in 

 trawl tows (entities are sites; attributes are the 

 transformed species abundance scores). Inverse 

 analysis compared similarities among the distribu- 

 tion patterns of species [entities are species, and at- 

 tributes are the sites where they occur (Boesch 1977; 

 Clifford and Stephenson 1975)). The sorting strategy 

 was flexible with /3 = —0.25 (see Clifford and 

 Stephenson 1975 and Boesch 1977 for explanation). 



Nodal analysis was used to examine the cooccur- 

 rence of species and site groups based on patterns of 

 constancy and fidelity. Constancy, a measure of how 

 consistently the members of a particular species 

 group occur among the stations of a given site group, 

 was calculated by the expression: 



C t] = aj{n t n) (Boesch 1977) 



where C y = constancy of species group i in site 

 group; 



actual number of individuals of 

 species group i in site group; 

 number of entities in groups ; and;, 

 respectively. 



a„ — 



W, 



restricted to a particular site group was determined 

 by 



F tj = (a tJ Xnj/rij Za„) (Boesch 1977) 



where F = fidelity of species group i in site 



group; 

 a y = actual number of individuals of 



species group i in collection group; 

 n J — number of entities in group;. 



RESULTS 



Biomass 



Otter trawl catches from the sponge-coral habitat 

 were highly variable, ranging from 6 to 2,976 in- 

 dividuals (1.6-244.8 kg) in a standard tow (Table 1). 

 The means of the natural log-transformed values of 

 the number of individuals per haul from all reef sites 

 were not significantly different between day and 

 night (t = 1.135, df = 55), whereas mean weight of 

 fishes was significantly greater in day tows (£=2.145, 

 df = 55). Mean values and density estimates for all 

 trawl tows within the sponge-coral and open-shelf 

 habitats were 



Mean values/tow 



Mean density estimates'/ 

 ha 



No. Weight No. Weight 



Habitat individuals (kg) individuals (kg) 



Sponge-coral 363 29.3 384 31.0 



2 (268;491) 2 (24; 35.8) 2 (284; 520) 2 (25.4; 37.9) 



Open shelf 54 3.0 57 3.2 



2 (24; 121) 2 (2.2; 4.0) 2 (25; 128) 2 (2.3; 4.2) 



Fidelity, the degree to which a given species group is 



'Using 0.95 ha as the swept area of a standard tow. 

 2 Upper and lower 907c confidence limits. 



Table 1.— Mean catch/tow of demersal teleosts, 90' ■'<' lower and upper confidence limits (LCL and UCL), and ranges 

 for % Yankee trawl tows by site for the spring 1978 sponge-coral survey. South Atlantic Bight. 



540 



