198 
Hshery Bulletin 95(2), 1997 
the 2 trawl hauls. Mean abundance and biomass of 
each species at each station were determined by aver- 
aging the results from the 2 trawl hauls, except at the 
few stations where only 1 sample could be collected. 
To investigate diversity, we used the number of 
species for richness (S) and calculated Shannon’s 
Index ( H ) (Pielou, 1977). Abundance and total unique 
species of both samples were combined for each sta- 
tion. Shannon’s index was calculated as 
k 
n log n - ^ log f t 
H = ^ 
n 
where n = total number of fish; 
f t - number of individuals in species i ; and 
k = the number of species (Zar, 1984). 
By using Shannon’s Index (H), “evenness” was esti- 
mated with the equation 
, H 
v = > 
Ins 
where v' = measure of evenness; and 
s = the number of species present. 
Fish assemblages were identified and their rela- 
tionship to physical oceanographic conditions deter- 
mined in a two-stage process. The first stage used 
cluster analysis of species abundance by station, fol- 
lowed by discriminant function and principal coordi- 
nate analyses of environmental data. Cluster analy- 
sis based on species abundance at each station was 
used to determine fish assemblages. Following the 
recommendation of Clifford and Stevenson (1975), 
the most commonly occurring species (21 species, 
each of which made up >0.1% of abundance) were 
chosen on the basis of a preliminary examination of 
abundance data. These species made up 99.6% of the 
total abundance, 98% of the biomass. Prior to calcu- 
lating similarity indices, abundance (X) was trans- 
formed (In [X+l]) to normalize the data (Clifford and 
Stevenson, 1975). Similarity indices were calculated 
as 1 - D, where D is the Bray-Curtis dissimilarity 
index (Clifford and Stevenson, 1975) adapted from 
Lance and Williams (1967). 
The algorithm for D is 
ZK-M 
D = , 
n 
^ X lj +X 2j) 
i = 1 
where n = number of individuals in species i; and 
j - number of stations. 
Similarity index values range from 0 to 1; a value of 
1 indicates identical species composition between 2 
stations and a value of 0 indicates no common spe- 
cies between stations. Following Clifford and 
Stevenson (1975), a range of similarity indices was 
used to determine major groupings. From prelimi- 
nary inspection of the data, it appeared that group- 
ings could be distinguished with indices of 0.5-0. 6. 
These indices were used as our reference for exam- 
ining the resulting dendrograms. 
Relationships between environmental conditions 
and fish assemblages were evaluated by using the 
following data subsets: 1) environmental (water 
depth, bottom temperature, and bottom salinity); 2) 
sediment type (arcsine-transformed percent of mud, 
sand, and gravel); and 3) abundance of infaunal and 
epifaunal mollusks. Sediment type and mollusk data 
values for those stations nearest ours were taken 
from Feder et al. (Footnote 1, sediment type) and 
Feder et al. (1994, mollusks). 
Multiple discriminant function analysis (DFA) and 
principal coordinate analysis (PCA) were used to 
evaluate the relationship between fish assemblages 
and environmental parameters. Mud, bottom tem- 
perature, epifaunal biomass, and invertebrate infau- 
nal biomass were not included in the analyses be- 
cause they were highly correlated with gravel, bot- 
tom salinity, epifaunal abundance, and infaunal in- 
vertebrate abundance, respectively. PCA was used 
to validate the results of the DFA and to determine 
whether other variables were influencing assem- 
blages. To control for multicollinearity, we discarded 
one of any pair of variables with -0.8 < r > 0.8. 
Abundance, commonality in species occurrence, 
ranks, and diversity were used to determine whether 
there was congruity between years at stations 
sampled in 1990 and 1991. Species ranks were com- 
pared by using the Wilcoxon signed-ranks test (Siegel 
and Castellan, 1988). 
Results 
Abundance and biomass 
A combined total of 66 species of 14 families were 
collected in 1990 and 1991 (Table 1). In 1990, two 
species of gadids, Boreogadus saida and Eleginus 
gracilis, made up 82% of the abundance and 69% of 
the biomass. Cottids, pleuronectids, and zoarcids 
made up an additional 15% of total abundance and 
24% of total biomass in 1990. On the basis of percent 
