Parker et al.: Reef fish abundance, composition, and habitat use 



789 



Methods 



Research site selection 



Based on preliminary observations (1-2 May 1985), 

 30,000 m 2 of bottom in GRNMS were divided em- 

 pirically into sand, live-bottom, and ledge habitats 

 (Parker et al. 3 ). For detailed community analyses, 

 the live-bottom area was further divided into three 

 subunits. The habitat classifications and approxi- 

 mate proportional areas within GRNMS (calculated 

 from Hunt, 1974) were the following: 



Sand: sand or sand and shell bottom with bot- 

 tom relief (<20 cm) provided by sandy swales; 

 occasional (<1%) depressions or burrows (5-25 

 cm deep) in sand surrounded by algae, 

 macrofauna, and sometimes rocks; approxi- 

 mately 18% of GRNMS. 



Live-bottom: approximately 1 to 75% of bottom 

 composed of rock outcroppings covered by algal 

 and benthic macrofauna; little or no (<15 cm) 

 relief; sandy areas, 2 to 25 cm deep.underlaid 

 by rock; approximately 58% of GRNMS. 



a Sparse live-bottom: covers 1 to 25% of the 

 substrate. 



b Moderate live-bottom: covers 26 to 50% 

 of the substrate. 



c Dense live-bottom: covers greater than 

 50% of the substrate. 



Ledge: distinct rock outcroppings of 15 cm 

 to over 200 cm; associated rock bottoms 

 covered by algal and benthic macrofauna; 

 approximately 24% of GRNMS. 



Thirty-six potential sampling sites, 12 each 

 over sand, live-bottom, and ledge substrates, 

 were randomly selected from a pool of 76 lo- 

 cations of known habitat type defined by our 

 preliminary work and by existing Georgia 

 Department of Natural Resources data 

 (Nicholson 4 ; Hudson 5 ). Of the 36 sites, 14, 17, 

 9, and 12, respectively, were randomly se- 

 lected for sampling during four surveys: sum- 

 mer ( 15-22 August) and fall (13-18 Novem- 



Parker, R. 0., Jr.. R. S. Nelson, and A. J. Chester. 1988. 

 A quantitative approach to the estimation of reef fish 

 abundance and community composition in the Gray's 

 Reef National Marine Sanctuary using SCUBA divers. 

 Final Rep. Contract No. NA84DOC-C2004. U.S. Natl. 

 Ocean Serv., Washington, D.C., 71 p. 

 Nicholson, N. 1982. The Gray's Reef National Marine 

 Sanctuary Visual Reef Fish Censusing Workshop, fi- 

 nal report. Georgia Dep. Nat. Resour., Coastal Resour. 

 Div., Brunswick, 16 p. 

 ' Hudson, J. A. 1984. Summer Internship Report, 

 Valdosta State College, Georgia Dep. Nat. Resour., 

 Coastal Resour. Div., Brunswick, 33 p. 



ber) 1985, and spring ( 13-18 May) and summer (21- 

 27 August) 1986. Sampling was stratified by habitat 

 type. 



As a stratified random design, optimal allocation 

 of effort among habitats normally would be deter- 

 mined by the variance of the population size of a 

 particular species and the size of each habitat. For 

 this study we used species richness as a proxy for 

 variance because the focus was on a multispecies 

 assemblage. Although the relationship between spe- 

 cies richness and total variance is not clearly defined, 

 even approximate estimators of variance usually are 

 adequate for allocating sampling effort (Steel and 

 Torrie, 1960). 



Because the sample size of our preliminary work 

 on live-bottom habitat was roughly one-third the 

 number of intervals available for sand and ledge 

 habitats (Fig. 2), we extrapolated the live-bottom data 

 (10 species observed in eight sampling intervals) to 

 a hypothetical sample of 25 sampling intervals. On 

 both ledge and sand habitats approximately 71% of 

 the total species observed were encountered after 

 eight sampling intervals ( Fig. 2 ). Assuming the same 

 relationship for live-bottom data, 14 species would 

 have been encountered in 25 sampling intervals. 

 Prior experience by both diving investigators sug- 

 gests this is a reasonable approximation. Sampling 

 effort was allocated among the different reef habi- 

 tats in proportion to the product of the area of a given 



