WENNER: SPONGE-CORAL HABITAT FISHES 



CABLE 



VESSEL 



LIGHT 



Figure 3. 



WEIGHT 



- System for deployment of underwater television equip- 

 ment used in habitat documentation. 



Fishes caught in nets that were not damaged during 

 trawling operations were identified, measured, and 

 weighed. 



Initial calculations showed that the variance of the 

 number of individuals per tow and the weight per tow 

 far exceeded the mean and approximated a negative 

 binomial distribution. Therefore, data were trans- 

 formed (In \x + 1]) before analysis to standardize the 

 variance and approximate the normal distribution 

 (Taylor 1953; Elliott 1977). The number of species 

 per tow was normally distributed and thus was not 

 transformed. The Bliss (1967) approximation was 

 used in retransforming the data from logarithmic to 

 original units. 



Since occasional catches of large elasmobranchs 

 and large catches of pelagic fishes contributed 

 significantly to the variance, biomass estimates were 

 made only for demersal teleosts. Density estimates 

 were calculated by the swept area method (Rohr and 

 Gutherz 1977), with the sweep of the net being 8.748 

 m (Azarovitz 4 ) and 1.080 km distances covered dur- 

 ing a 10-min tow. These density estimates should be 

 viewed as minimum, since the effectiveness of the % 

 Yankee trawl in sampling reef fish populations is 

 unknown. 



Species diversity (//') and its components, evenness 

 (</') and richness, were calculated for elasmobranchs 

 and demersal teleosts in each trawl tow using the 

 following formulae: 



H 



where H' 



s 



(p,)(log 2 p,) (Pieloul969) 



index of species diversity expressed 



in bits/individual 

 S = number of species 

 Pi = proportion of total sample belonging 



to ith species; 



J = HIH\ 



(Pielou 1969) 



where J' = 

 H = 



equitability or evenness 

 observed species diversity 

 log,S; 



Species richness = 5 — 1/lniV (Margalef 1968) 



where 5 = number of species 



N = number of individuals. 



Normal and inverse cluster analysis (Clifford and 

 Stephenson 1975) were used to analyze trawl data. 

 Prior to the analysis, data were edited to eliminate 

 species occurring in only one trawl tow. These 

 species have no discernible distribution pattern and, 

 therefore, contribute no information to the analysis 

 (Boesch 1977). Species abundance scores were then 

 log 10 transformed, thus reducing the dominance of 

 species having high abundance. The Bray-Curtis 

 similarity coefficient (Clifford and Stephenson 

 1975) (= the Czekanowski Quantitative Index of 

 Bloom 198 1) was used on the modified data set and is 

 expressed by 



4 T. Azarovitz, Northeast Fisheries Center, National Marine 

 Fisheries Service, NOAA, Woods Hole, MA 02543, pers. com- 

 mun., 1977. 



539 



