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



477 



Table 1 



Distribution of number of samples (n) and area (m^) in the different areas, among the major substratum types (by nature). Sand = 

 soft sand, of variable texture and topography. With or without biogenic holes. Depth variable. Bioturbation maybe evident. Cobbles = 

 fist-size rocks of claystone, sandstone, or quartzite. Of smooth and rounded appearance. Scattered or aggregated (sometimes 

 densely) on sediment surface, and occasionally projecting through. With or without sediment layer or epifauna. Rock debris = 

 smaller than boulders but larger than cobbles or gravel. Generally of irregular shape and not always smooth. Composed of sand- 

 stone or claystone. May project through, or sit on, sediment surface. Scattered or aggregated densely. With or without .sediment on 

 upper face. Epifauna maybe present. Boulders = massive, freestanding outcrops of rock, of high vertical relief, generally project- 

 ing through the sediment surface. Composed of either sandstone or claystone. Usually with some sediment on upper face. With or 

 without encrusting fauna. Flat bedrock = slab-like sandstone or claystone rock of low vertical relief (<0. 5 m), on sediment surface, 

 usually with a thin sediment layer over upper surface. Epifauna on edges. 



North 



South 



Major substratum 



Substratum nature 



Boulders 



Cobbles 



Cobbles 



Rock debris 



Flat bedrock 



Sand 



Sand 



hard 



hard 



mixed 



hard 



hard 



mixed 



soft 



that samples could be collected or to zoom in on particular 

 features, and at other times it moved away from the bot- 

 tom. Videotape recorded during these times, and when 

 visibility was impaired due to silt disturbance, was not 

 examined. The camera was rarely operated for the full du- 

 ration of any one dive, and its use depended largely upon 

 the interest of the geological observer. The video camera 

 was equipped with a clock that was synchronized to that 

 aboard the support vessel, so that it was possible to relate 

 sections of the videotape to the actual position of the sub- 

 mersible on the dive-track. Illumination of the sea floor 

 was provided by five 150-W halogen projectors arranged 

 along a bar above the dome-window. A pair of red lasers 

 (positioned 50 cm apart) were projected onto the sea floor 

 to provide a moving scale. 



Although the entire videotape from each dive was 

 inspected, only samples of videotape were examined in 

 detail and analyzed. The samples from each tape met the 

 following criteria: 



1 they were recorded along linear sections of the dive- 

 track, as determined from the track-chart (to avoid 

 looping and re-examination of previously sampled 

 areas); 



2 the submersible was moving at a slow (0.25 knots) but 

 constant speed; and 



3 there was no suspended sediment to impair visibility. 



Samples were in multiples of selections that were 3 1 seconds 

 in length. The bottom cover in each sample was first catego- 

 rized by major (>50%) and subsidiary (>20%) substratum 

 type (see descriptions in legend to Table 1), and then all 

 sedentary biological features were recorded (as present or 

 absent). A total of 381 samples was analyzed, most of which 



were in the northern area (Table 1). All nekton were identi- 

 fied and counted. Counts were subsequently converted to 

 densities (m^) with a knowledge of both the horizontal dis- 

 tance traversed by the Jago (calculated from the dive track 

 with Autocad (Autodesk Inc., 1988]), and the width of the 

 video- frame (calculated by using the distance between the 

 two lasers, approx. 1-1.5 m). 



Statistical analysis 



Associations with environmental features The associa- 

 tion between nekton and features of the physical and bio- 

 logical environment was examined by following Felley et 

 al. ( 1989) and Felley and Vecchione (1995). In each sample, 

 the primary substratum was given a score of 0.75, and the 

 subsidiary substratum was allocated a score of 0.25: sed- 

 entary biological features were scored as either 1 (present) 

 or (absent). The mean environment was then calculated 

 for each species, after weighting values of each of the envi- 

 ronmental variables in each sample by density of individu- 

 als in that sample. Samples where no individuals were 

 seen, made no contribution (since they had a weight of 0), 

 whereas samples where the species was abundant made 

 a marked contribution to the species mean. The mean for 

 a variable reflected the relative state of that variable in 

 samples where the species was most likely to be found, 

 and can be thought of as its preference for that variable. 

 A correlation matrix was then generated from all means 

 for the measured environmental variables. Patterns of 

 habitat use by species were reflected in patterns of inter- 

 relations among variables. Principal components analysis 

 (PCA) was then performed on the correlation matrix. PCA 

 resolves patterns of interrelationships among variables 

 into a smaller set of composite variables (PCs) to which 



