Walsh et al Juvenile fish assemblages on the southeast United States continental shelf 



259 



1998; Munroe, 1998). Identification to species was not 

 possible for all fish. Taxa were separated into larval, 

 juvenile, and adult age classes according to life-history 

 characteristics from the literature and standard length 

 distributions (SL, mm) from the beam trawl collec- 

 tions. Fish were classified as larvae if fin development 

 was not complete (and defined as "larvae at settlement 

 stage" when fin formation was complete and pigmenta- 

 tion incomplete; fish were classified as juveniles when 

 pigmentation was complete, and were classified as adults 

 if they exceeded the reported minimum SL for sexual 

 maturity. Because the mesh of the beam trawl is large 

 enough to extrude larvae and less efficient at catch- 

 ing large (>150 mm SL) fish (Kuipers, 1975), analyses 

 in this study were conducted only on settlement and 

 juvenile stages, which were combined and referred to 

 as juveniles. Taxa were also classified as not reef-associ- 

 ated, strongly reef-associated or weakly reef-associated 

 based on habitat-use information (Chester et al., 1984; 

 Humann, 1994; Parker et al., 1994; Parker and Mays, 

 1998; Lindeman et al., 2000). Taxa found predominantly 

 on rocky-reefs (e.g., Cetitropiistis striata ) were classified 

 as strongly reef-associated. Those that use rocky-reefs 

 and other habitats (e.g., Diplectrum formosum) were 

 classified as weakly reef-associated. 



Standard catch per unit of effort (CPUE. fish/5-min 

 bottom tow) was calculated for each taxon. CPUE for 

 each taxon at each station was the average of all rep- 

 licate tows. Three juvenile stage data sets (i.e., cross- 

 shelf, inshore, and offshore) were compiled by using data 

 from the station groups (Table 1, Fig. 1): the cross-shelf 

 data set consisted of data from stations one through 

 seven (Fig. lA); the inshore data set consisted of data 

 from stations predominately inside the 20-m isobath 

 (Fig. IB), and the offshore data set consisted of date 

 from stations >30 m in depth (Fig. IC). 



For statistical analyses, the three data sets were subdi- 

 vided into two groups so that rare taxa would not great- 

 ly influence the classification of assemblages. The first 

 group consisted only of abundant taxa, and the second 

 group included abundant and rare taxa. Cross-shelf and 

 offshore data sets were subdivided into taxa that made 

 up at least lQ'7c (abundant) and 1% (abundant and rare) 

 of the collections at any one station. The inshore data set 

 was subdivided at 5% and 1% levels. The data sets were 

 further condensed by eliminating all taxa not identified 

 to species level with the exception of Stenotomus sp. and 

 Rypticus sp. which were probably single species. Several 

 abundant taxa were excluded from the analysis because 

 they were only identified to genus and have multiple spe- 

 cies common in the area, and each species may differ in 

 distribution: i.e., Etropus spp. (4 species), Prionotus spp. 

 (14 species), Sphoeroides spp. (11 species), Microgobius 

 spp. (3 species), and Bothus spp. (3 species). As a result, 

 the numbers of taxa in the rare and abundant data sets 

 used in analyses were as follows: cross-shelf 82 (rare) 

 and 28 (abundant) (Table 2); inshore 28 and 13 (Table 3), 

 and offshore 51 and 17 (Table 4). 



Preparation of environmental data 



Season and five environmental (or hydrographic) vari- 

 ables were chosen in an attempt to explain variation in 

 the juvenile fish data. Cruises were assigned to one of 

 four seasons (Table 1) based on wind (Atkinson et al., 

 1985) and temperature regimes (Marancik et al., 2005). 

 Environmental variables were calculated as in Marancik 

 et al. (2005). Briefly, CTD measurements of temperature 

 (°C) and salinity were used to derive density (a,, Kg/m'^). 

 Bottom temperature, salinity, and density were used 

 in the multivariate analyses with juvenile fish because 

 benthic habitats were sampled and bottom values were 



