Barber et at: Demersal fish assemblages of the northeastern Chukchi Sea 
203 
170 ° 155 ° 
Figure 3 
Relative richness (number of species), species diversity (Shan- 
non index), and evenness of demersal fishes at 48 and 16 sta- 
tions sampled during 1990 and 1991, respectively, in the north- 
eastern Chukchi Sea, Alaska. SW = Shannon Wiener. 
Peninsula. A second association ( II ) was formed 
near a station cluster that bisected the north- 
ern offshore association (VI) but was more 
closely related to association I. The northern 
offshore association (VI) consisted of two rela- 
tively distant clusters, whereas the northern 
inshore association (III) consisted of two closely 
related clusters, one made up of two stations. The 
central offshore association (IV) was formed by 
two clusters. Finally, there was the central on- 
shore association (V) in Ledyard Bay, which con- 
sisted of four closely related and two distantly 
related stations. The cluster analysis yielded simi- 
lar results when B. saida was not included in the 
analysis. In all associations, B. saida made up 
over 90% of the abundance (Table 3). 
The most distinctive assemblage was VI, 
which had the fewest species, lowest abundance, 
and least diversity and evenness (Table 3). In 
comparison, associations I and V had much 
greater values for all these measures. Associa- 
tion I had the greatest number of species; the 
top five species in order of abundance were B. 
saida , Myoxocephalus sp., H. robustus, G. 
tricuspis, and Lrycodes palearis. Association II 
had the second most abundant species; the top 
five species in order of abundance were B. saida, 
L. raridens, M. verrucosus, G. tricuspis, and 
Clupea harengus pallasi. 
Bottom salinity and percent gravel were iden- 
tified through discriminant analysis as key fac- 
tors separating assemblage groups. The first 
axis accounted for 72%, the second axis for 28% 
of the variation (Table 4). Bottom salinity 
showed the strongest association with axis 1, 
whereas percent gravel was strongest in axis 
2. The lines superimposed on Figure 5 enclose 
stations of similar environmental conditions. 
There is relatively little overlap of groups III 
and V; the former is characterized by low bot- 
tom salinity and high gravel, whereas the lat- 
ter is intermediate in salinity and gravel (Fig. 
5). Stations 14 and 15 were classified together, 
with lowest salinity and percent gravel. There 
is overlap at the boundaries of groups I and VI, 
which suggests that there is a gradation in en- 
vironmental conditions. Group VI is associated 
with more saline water but includes a wide 
range of percent gravel. 
A principal component analysis, which in- 
cluded all environmental data, supports the 
discriminant analysis but suggests that other 
variables are also important determinants of 
fish associations (Table 5). This analysis indi- 
cated that bottom salinity, water depth, and 
