McKenna: Spatial structure and temporal continuity of South Georgian fish community 



479 



ment (Daniels, 1982) and feeds on invertebrate infauna, 

 primarily polychete worms (McKenna, 1991). 

 Nototheniops larseni was the second most abundant 

 species and is the most pelagically adapted of the com- 

 mon species (Targett, 1981). It is a relatively small 

 fish (<25cm) that feeds on krill and other pelagic in- 

 vertebrates (McKenna, 1991). Chaenocephalus aceratus 

 and P. georgianus are the other two icefish found in 

 the vacinity of South Georgia Island. Both species are 

 closely associated with the sea floor and feed heavily 

 on other fish and krill (McKenna, 1991). Notothenia 



squamifrons is a demersal rock cod 

 that prefers the deeper strata of the 

 region. It feeds on a wide variety of 

 benthic invertebrates but has a 

 preference for tunicates (McKenna, 

 1991). Nototheniops nudifrons is 

 one of the smallest species in 

 the region (10-5 cm). It lives in and 

 among the sedentary megainver- 

 tebrates growing on the bottom 

 (e.g., sponges). It feeds on benthic 

 epifauna, but most of its diet con- 

 sists of krill (McKenna, 1991). Other 

 species collected by the AMLR sur- 

 veys generally accounted for less 

 than 2% of the fish in the region. 

 For a more detailed description of 

 this community see Gabriel (1987), 

 McKenna and Saila (1989, 1991), 

 and McKenna (1991). 



Diversity values were moderate 

 throughout the region and relatively 

 consistent from year to year. The 

 overall diversities (H') for each sur- 

 vey (1986-87-1988-89) were 1.51. 

 1.94, and 1.83 (based on numerical 

 abundances) (Table 2). The value for 

 any given station ranged from to 

 2.097 (Fig. 3). Richness never ex- 

 ceeded 16 at a single station, but 

 was as low as 1 (which accounts for 

 the zero H' values) in a few cases. 

 Survey-wide richness was nearly 

 the same for 1986-87 and 1987-88 

 but declined in 1988-89 because of 

 the lack of rare species collected 

 during the survey. Overall evenness 

 for each survey ranged from 1.02 to 

 1.35 and was greater in the latter 

 two surveys. 



Significant associations were 

 found between species within each 

 survey. However, all were weakly 

 correlated (Table 3). Only three as- 

 sociations had r' values greater than 60 c r and none 

 explained more than 707r of the variability in their 

 distributions. None of the fifteen associations with r' 

 values greater than 50% were consistent from year to 

 year. However, three of these associations (C. aceratus- 

 P. georgianus, Artedidraco mirus-N. nudifrons. Mur- 

 aenolepis microps-N. nudifrons) persisted from 1986- 

 87 to 1987-88. The inverse (Q-mode) cluster analysis 

 appeared to classify the species into groups based on 

 their relative rareness, but there were exceptions to 

 such a classification in nearly every group. 



