482 



Fishery Bulletin 91(3), 1993 



peaks are real boundaries be- 

 tween communities (Fig. 11, page 

 488). Figure 12, page 488 sum- 

 marizes the most pronounced 

 boundaries from each of the gra- 

 dients examined. 



poral trends also existed. However, all of the trends 

 revealed by skewer analysis were weak; the primary 

 mode fell into either the smallest or second smallest 

 class interval of Tau (Fig. 10, page 487). 



Gradient analysis, using GRADSECT.BAS, measures 

 the rate of change of the species assemblage along the 

 gradient of interest. However, it offers no objective de- 

 cision about the value that the rate must reach in 

 order to be considered significant. Thus, it will iden- 

 tify ecotones along any gradient and it is a matter of 

 subjective judgment to determine which (if any) of the 



Discussion 



The diversity values for the South 

 Georgian fish community were 

 generally greater than those re- 

 ported by Targett (1981) for the 

 same region. They were within 

 the range of those reported for de- 

 mersal fish in tropical estuarine 

 areas (Yanez-Arancibia et al., 

 1980). These values place the 

 South Georgia fish community on 

 the relative, global diversity con- 

 tinuum, but the wide range of 

 values demonstrates that diver- 

 sity alone is a poor model of this 

 community. 



The richness component of di- 

 versity was low and similar to 

 that reported by Targett (1981), 

 but the evenness was consider- 

 ably larger than had been re- 

 ported for this region. This obser- 

 vation is consistent with the 

 finding that a significant shift in 

 the species assemblage from 

 1986-87 to 1987-88 was associ- 

 ated with an increase in diver- 

 sity because of an increase in the 

 evenness component of that index 

 (McKenna and Saila, 1991). The 

 increasing number of stations 

 that poorly fit the logarithmic se- 

 ries distribution with time also 

 supports the hypothesis that the 

 community was changing toward one which was less 

 dominated by a small number of species. This appar- 

 ent change in the community was compounded in 1988- 

 89 by the use of different sampling techniques and 

 gear. The sampling during that season was insuffi- 

 cient to accurately represent the abundance of the rar- 

 est species. A bias was also introduced by the small 

 gear used for that survey. In all previous AMLR sur- 

 veys the mackerel icefish (C. gunnari) was the most 

 abundant species. The smaller individuals of this spe- 

 cies tends to be more pelagic than many of the others 



