510 



Fishery Bulletin 97(3), 1999 



most widespread encrusting organism was the reef- 

 building sabellariid worm Phragmatopoma lapidosa 

 (=P. caudata: Kirtley, 1994), often covering over 50"^ 

 of the hardbottom at all sites. 



A beach restoration project occurred at Carlin Park 

 in March and April of 1995. More than 350,000 m'^ of 

 beach-compatible sediments were excavated by a 

 cutter-head dredge from a site 0.8 km offshore and 

 hydraulically pumped along 1.8 km of shoreline. 

 Bulldozers extended the fill seaward to an estimated 

 width of 60 m. An estimated total of 4.9 to 5.7 ha 

 ( 12-14 acres) of nearshore hardbottom was buried.^ 

 Visual surveys of fishes were conducted for 12 months 

 before burial and 15 months after burial at both 

 Carlin Park (the impact site) and Coral Cove (the 

 control site). Little or no hardbottom was observed 

 in fish surveys at Carlin Park after the project. 



Survey protocol 



During each site visit, three to five transects within 

 two adjacent areas (=6-10 total transects/site) were 

 censused. These 2 x 15 m transects were randomly 

 located at depths ranging from one to four m. 

 Transects were deployed along random compass 

 headings at random distances between successive 

 transects. Random number tables were used prior 

 to site visits to determine the compass headings and 

 distances between transects (based on numbers of 

 fin kicks). All fishes observed within one m of each 

 side of the transect line were identified and tallied 

 by a snorkeler The survey zone extended from the 

 bottom to the surface and 2 m in front of the observer. 

 Ledges and sand-rock interfaces were examined for 

 fishes. Rocks were not overturned. The proportion of 

 hardbottom to sand was estimated within each 

 transect. An estimated 35"^ of the area within all 

 transects was sand. Surveys were conducted between 

 0900 and 1700 and avoided twilight periods. To com- 

 pare fish abundances at hardbottom and sand ar- 

 eas, identical transect methods were used at near- 

 shore sand plains greater than 50 m from any hard- 

 bottom structures. 



Monthly visual censusing occurred from April 1994 

 to June 1996 as permitted by nearshore visibility and 

 sea state. Discharges of turbid water from Jupiter 

 Inlet and wave resuspension of fine sediments some- 

 times resulted in turbidity levels that precluded sam- 

 pling. Samples were obtained for all months except 

 October through January when waves and turbidity 

 were typically prohibitive. 



' Davis, P. 1998. Palm Beach County Department of Environ- 

 mental Resources Management, 3.32.3 Belvedere Rd.. Bldg. .502, 

 W. Palm Beach, FL 33406. Personal commun. 



In addition to total abundances, early life stages 

 were also enumerated. Fork length was used for size 

 estimation. Following Lindeman (1986; 1997a), life 

 stages of grunts [Haemulon and Anisotremus) were 

 recorded as follows: newly settled (<2 cm), early ju- 

 venile (2-5 cm), juvenile (5-15 cm), and adult (>15 

 cm). For other families, the same newly settled size 

 range (<2 cm) was used. The early juvenile designa- 

 tion was used only for grunts because of the distinct 

 morphological features of the 2-5 cm size range 

 (Lindeman, 1986). Juvenile and adult stages were 

 based on size and pigment patterns reported in the 

 literature (e.g. Robins and Ray, 1986; Humann, 1994). 

 Species identifications of the newly settled or juve- 

 nile stages for certain taxa were limited by very simi- 

 lar morphological features (e.g. scarids, kyphosids, 

 gerreids, haemulids, clupeids >. Some early stage iden- 

 tifications were therefore recorded only at the genus 

 or family level. Collections of small schools of newly 

 settled grunts were made with hand nets at both 

 Jupiter sites in 1994 and 1995 to supplement field 

 identifications. All collections were deposited at the 

 Florida Museum of Natural History, University of 

 Florida. 



Data analyses 



To address the first objective of our study, two comple- 

 mentary multivariate methods were used to spatially 

 and temporally characterize the assemblages at the 

 three sites. The second objective was addressed by 

 univariate testing of the hypothesis that abundances 

 of different life stages would not differ significantly 

 within sites. The third objective was examined with 

 the hypothesis that numbers of individuals and spe- 

 cies would not differ significantly between an impact 

 site where allmost all the hardbottom was buried and 

 a control site that was unaffected by the burial. In 

 univariate analyses, data were standardized as the 

 mean number of individuals per transect and as the 

 mean number of species per transect. 



Samples were temporally unbalanced owing to the 

 inability to visually sample during portions of the 

 winter. Therefore, to examine the first objective, 

 multivariate ordination and classification (cluster 

 analysis) of a samples-by-taxa matrix for the entire, 

 unpooled data set were used. These analyses were 

 performed on a data set of 31 samples ( 16 from Coral 

 Cove, 12 from Carlin Park, and 3 from Ocean Ridge) 

 and 61 taxa. Each sample represented a site visit 

 where 6 to 10 transects were censused. Samples from 

 Carlin Park did not include postdredging site visits 

 because these samples contained few or no fishes. 

 The 61 taxa were those remaining from a total of 86 

 after eliminating taxa occurring only once across all 



