SMITH ET AL.: ORGANIZATION OF NEKTON 



leadline hoop which helped fill contours along the 

 bottom. Samples were obtained by tossing the ring 

 from shore and then applying rotenone at a 30 ppb 

 concentration within the confines of the net 

 (Weinstein and Brooks 1983). Stricken fishes were 

 then captured with dip net or swept off the bottom. 



Fishes were sorted into 20 mm (or less) size 

 classes and up to 20 randomly selected individuals 

 from each class used for trophic analyses. Stomach 

 fullness was recorded as a relative fullness index 

 (RFI) value (Hyslop 1980). Stomach contents were 

 subsequently analyzed using a modified Carr and 

 Adams (1972) sieve fractionation technique. Total 

 dry weights for each sieve fraction were then ob- 

 tained and proportioned among the prey taxa 

 identified from a five drop subsample taken before 

 drying. On the assumption that particles of equal 

 size have approximately the same weight, this 

 method agglomerates food particles of roughly the 

 same size. 



The Carr and Adams technique provided rapid, 

 accurate identification of food items for a large 

 number of stomachs and has been used with suc- 

 cess by several investigators (Sheridan 1979; 

 Sheridan and Livingston 1979; Stoner 1980; 

 Livingston 1982). A useful modification employed 

 in this study was the application of a low pressure 

 stream of compressed air delivered through a Pas- 

 teur pipette which greatly aided in removing food 

 particles adhering to the finer screens of the 

 sieves. 



Numerical classification analysis used here is 

 similar to the procedures employed by Weinstein 

 (1979) and Weinstein and Brooks (1983). Briefly, 

 marsh creek communities and trophic ecology of 

 dominant species were compared by classification 

 methods using "normal" and "inverse" classifica- 

 tion (Clifford and Stephenson 1975). The former 

 method groups sites (or predators) by their species 

 (or prey taxon) attributes; while inverse classifica- 

 tion (used only for community analysis purposes 

 here) groups species according to their site of oc- 

 currence (i.e., the sites become the attributes of 

 the species). Similarity between sites (or pred- 

 ators) was calculated as the complement of the 

 Canberra metric index: 



N 



[l/n][!]\xij - x 2 j\Kxij + x 2 j) 



(1) 



where n = number of attributes, and x\j and x 2 j 

 are the values of the jth attribute for any pair of 

 entities. The merits of the Canberra metric index 



have been discussed by Clifford and Stephenson 

 (1975). 



Separate matrices were constructed for each 

 comparison from untransformed, pooled monthly 

 data and clustered by the unweighted pair, 

 group-average strategy (Clifford and Stephenson 

 1975). Species occurring at only one station (sin- 

 gletons) were eliminated prior to the community 

 analyses. Combined trawl and Wegener ring data 

 were used separately in these procedures. Den- 

 drograms for site and species dissimilarity (com- 

 munity analyses) were constructed and cross- 

 tabulated in a two-way coincidence table. 



All nekton were preserved in 10% buffered For- 

 malin 5 . Standard length (SL, carapace width for 

 blue crabs) was recorded for all taxa. Up to 30 

 individuals/species were measured from each col- 

 lection, subsampling for lengths was employed 

 when sorted collections contained more than 30 

 individuals of a given species. Prior to each collec- 

 tion, temperature and salinity were recorded with 

 an immersion thermometer and a temperature- 

 compensated refractometer. 



RESULTS 



Abundance and Seasonality 



Only two species — spot and the bay anchovy, 

 Anchoa mitchelli — comprised >90% of the total 

 number of individuals captured at Blevins Creek 

 and adjacent shoals. Using this same criterion, 

 upstream densities were more equitably distrib- 

 uted with four species in the creek and six on the 

 shoal sharing dominance (Table 1). Blue crabs; 

 white perch, Morone americana; and the hog- 

 choker, Trinectes maculatus, were in this group in 

 Goalders Creek, while in late summer and fall the 

 Atlantic croaker, Micropogonias undulatus, and 

 the weakfish, Cynoscion regalis, were also abun- 

 dant at the shoal station (Table 1). 



Species richness (S) was also greater in all 

 months in the Goalders Creek system compared 

 with the polyhaline Blevins Creek (Fig. 2) and was 

 significantly greater for the entire study period 

 (Wilcoxon sign-ranks test; P < 0.05). No apparent 

 trend, however, was evident in the number of indi- 

 viduals captured at each locality (Fig. 2), except 

 that peak catches of two dominant species, spot 

 and bay anchovy, were greater in Blevins Creek 

 and resulted in the large disparity in catches in 



5 Reference to trade names does not imply endorsement by the 

 National Marine Fisheries Service, NOAA. 



457 



