Lindeman and Snyder: Nearshore hardbottom fishes of southeast Florida 



51 



samples. Within each sample, counts for individual 

 taxa were averaged over all transects to provide val- 

 ues for the matrix. These values were log-trans- 

 formed [logjQ(7i-t-l)] to prevent abundant taxa from 

 dominating the ordination or classification results. 



The transformed matrix was analyzed by corre- 

 spondence analysis (CA), a method that employs a 

 two-way weighted averaging algorithm to produce 

 simultaneous ordination of sites and taxa (Gauch, 

 1982; Jongman et al., 1995). These analyses were 

 performed with the program CANOCO (ter Braak, 

 1988). From the same log-transformed data matrix, 

 normal (samples) and inverse (taxa) resemblance 

 matrices were generated by using the Bray-Curtis 

 dissimilarity index (Bray and Curtis, 1957). Normal 

 and inverse resemblance matrices were clustered 

 separately by the unweighted paired-group method 

 of averaging (UPGMA) (Sneath and Sokal, 1973). All 

 dissimilarity and cluster analyses were computed 

 with NTSYS-pc software (Rohlf, 1997). 



To address the second project objective, numbers 

 of life stages per transect were compared within each 

 site. Data were analyzed by using a parametric one- 

 way ANOVA when variances were homogeneous 

 (Bartlett's test). A posteriori comparisons of differ- 

 ences among means employed Tukey's HSD test. 

 Variances of numbers of life stages of grunts per 

 transect at the two Jupiter sites remained heteroge- 

 neous after logji/n-t-l ) transformation and a Kruskall- 

 Wallis non parametric, single classification ANOVA 

 was used. Probability was calculated using the x~ 

 approximation. Two-sample ^tests for unequal vari- 

 ances were used to compare numbers of individuals 

 at hardbottom and natural sand sites. Only hard- 

 bottom samples from months when natural sand sites 

 were sampled (March and April 1995) were used for 

 these tests. In all statistical tests, differences were 

 considered significant at P<0.05. 



The third objective, examining dredging effects at 

 the impact site (Carlin Park) and the control site 

 (Coral Cove), employed a BACIPS (before after con- 

 trol impact paired series) design (Stewart-Oaten et 

 al., 1986; Osenberg and Schmitt, 1996). This ap- 

 proach compares differences in variables between 

 sites over time before and after the impact. The dif- 

 ferences in the paired series were examined by two- 

 sample Ntests by using the mean number of both 

 individuals and species as the variables (Stewart- 

 Oaten, 1996). 



Results 



A total of 352 transects was sampled at the two Ju- 

 piter sites: 204 at Carlin Park and 148 at Coral Cove. 



a 60 



40 



^ 20 



5 - 



10 



15 



Abundances 

 among sites 



Coral 

 Cove 



Carlin 

 Park 



Ocean 

 Ridge 



Figure 2 



Mean number of individuals and species per transect (with 

 95% confidence intervals i for all nearshore hardbottom 

 sites. Only predredging data were used for Carlin Park 

 site. Sample sizes: Coral Cove: 148 transects; Carlin Park: 

 112; Ocean Ridge: 36. 



At Carlin Park, 112 transects were sampled before 

 burial and 92 after. At Coral Cove, 58 transects were 

 sampled before the burial of the Carlin Park reef and 

 90 after. Eight transects were sampled over natural 

 sand habitats at Carlin Park and six at Coral Cove. 

 An additional 36 hardbottom and 6 sand transects 

 were sampled at Ocean Ridge. 



Family and species abundances 



Thirty-six families of fishes were censused among 

 the three hardbottom sites. The most speciose fam- 

 ily was the grunts and margates (Haemulidae) with 

 11 species of Haemulon and Anisotremus. The 

 wrasses, parrotfishes, and damselfishes (Labridae, 

 Scaridae, and Pomacentridae) had seven, six, and 

 five species, respectively. Four species each of jacks 

 (Carangidae), snappers (Lutjanidae), and clinids 

 (Labrisomidae) were recorded. These seven families 

 contained 50^7^ of the total species censused. Eighty- 

 six taxa (77 identified to species) and 10,491 indi- 

 viduals were censused at all sites (Appendix). At 

 Coral Cove, 64 species and 5093 individuals were 

 recorded. At Carlin Park, 53 species and 4438 indi- 

 viduals were recorded. At Ocean Ridge, 48 species 

 and 960 individuals were recorded. Mean numbers 

 of both species and individuals per transect were 

 similar among all sites (Fig. 2). 



