Gibbons et al : Habitat use by demersal nekton on the continental shelf in the Benguela ecosystem 



479 



data and contrasted species found over soft bio- 

 active substrata (sole, Austroglossus microlepis, 

 and to a lesser extent hakes) with those over 

 more mixed and hard substrata (false jacopever, 

 Sebastes capensis). The other species were located 

 midway along this axis (Fig. 2). 



PC2 accounted for almost 25% of the variance 

 in the data set (Table 4). The significant load- 

 ings on the environmental variables essentially 

 contrasted broken expanses of hard substrata 

 (cobbles, and the biota associated with more 

 open areas) with unbroken expanses of "lifeless" 

 bedrock (Table 4). Those species that were previ- 

 ously contrasted along the PCI axis were grouped 

 together (false jacopever and sole) and set against 

 those that were positioned midway between them 

 (kingklip, Genypterus capensis, and goby) (Fig. 2). 



PCS was responsible for almost 10% of the 

 variance in the data set (Table 4). It can be 

 interpreted in terms of biological richness and 

 contrasted gobies and soles with kingklip and 

 hake (Table 4). 



South A total of 10 species were seen in these 

 samples, and their mean environments are given 

 in Table 5. PCA of the correlation matrix gener- 

 ated from the dominant species' means produced 

 four PCs with eigenvalues greater than one 

 (Table 6). The first two PCs explained almost 

 75% of the variance in the data, although PCI 

 extracted only -60%. A clear trend that differen- 

 tiated the species was determined by PCI, which 

 contrasted dragonets and soles with kingklip 

 and false jacopever (Fig. 2). The loadings on the 

 variables of PCI (Table 6) opposed hard substrata 

 and their associated epifauna at one extreme with 

 sand and its associated organisms and structures 

 at the other. PC2 contrasted gobies, kingklip, and 

 hake with false jacopever and, to a lesser extent, 

 jacopever, Helicolenus dactylopterus (Fig. 2). In- 

 terpretation of this PC is difficult because few 



1 

 0.5 



^A 



0-- 



§ -0.5 + 



CO 

 LL 



-1 



-1.5 -I- 



-2 



Sole 



False Jacopever 



Cuttlefish 



Hake 



Kingklip 



Goby 



-" 1 1 1 1 1 1 (— 



•1.5 -1 -0.5 0.5 



Sand Factor 1 



B 



H 1 1 1 1 



1 1.5 2 



Rocks 



1.5 -r 



1 - - False Jacopever 



0.5 -- 



-- 



-0.5 -- 



-1 -- 



-1.5 



Jacopever 



Grenadier 



Dragonet 



Sole 



Kingklip 



Goby 



Cuttlefish 



Hake 



-+- 



■+- 



■+- 



■+- 



-t 



-2 -1.5 



Rocks 



-1 



-0.5 



Factor 1 



0.5 



1 1.5 



Sand 



Figure 2 



Distribution of species scores along the first and second axes repre- 

 senting habitat use by nekton species in the sites to the north (A) and 

 south (Bl of the Orange River mouth. PCI (north and south) repre- 

 sented habitat use according to hard and sofl substrata. PC2 (north) 

 corresponded to the broken or continuous nature of the substratum, 

 and PC2 (south) represented biological richness (see text). 



