O'NEIL and WEINSTEIN: FEEDING HABITATS OF SPOT 



For gut content analysis, fish from each of the 

 four trawl samples representing a given station 

 were pooled and then divided into several size 

 classes. Initially, 5 mm size increments were used 

 in order to corroborate the findings of others con- 

 cerning an ontongenetic shift in feeding habits of 

 spot. When mean standard lengths exceeded 20 

 mm, 10 mm size classes were adopted. 



Initially, up to 20 stomachs were removed from 

 randomly selected individuals in each size class. 

 Later, based on prey item diversity (Hurtubia 

 1973) comparisons for the June samples, 12 stom- 

 achs per size class was set as the upper limit 

 (O'Neil 1983). 



Stomach contents were pooled within size 

 classes and analyzed using the Carr and Adams 

 (1972) sieve fraction technique. After washing 

 stomach contents from each sieve (2, 0.85, 0.425, 

 0.25, 0.15, and 0.075 mm meshes) into a small 

 fingerbowl, a random subsample of approxi- 

 mately 5 mL was removed. The subsample was 

 placed in a labeled vial and the remainder was 

 filtered onto a preweighed 55 mm filter pad and 

 dried for 24 hours at 60°C. On the assumption 

 that food particles of roughly the same size have 

 approximately the same weight (Carr and Adams 

 1972), the total dry weight for each sieve fraction 

 was proportioned among the prey types identified 

 from its subsample. The Carr and Adams tech- 

 nique provided for rapid, accurate identification 

 of food items from a large number of stomachs and 

 has been used successfully by several investiga- 

 tors (Sheridan 1979; Stoner 1980; Livingston 

 1982; Lucas 1982). 



Statistical Analysis 



Dietary differences among various ontogenetic 

 groups, between creeks, between stations within 

 creeks, or for each month examined were com- 

 pared using "normal" classification methods 

 (Clifford and Stephenson 1975). Overlap of prey 

 utilization was then determined using the com- 

 plement of the Bray-Curtis dissimilarity mea- 

 sure: 



2 K-^2,i 



_J. 



n 

 2 ^^lj+^2j) 



and X2 are the values of the jth attribute for 

 any pair of entities (size, station, month). 



Separate matrices were constructed for each 

 comparison from untransformed, pooled monthly 

 data using COMPAH (Boesch 1977). The data in 

 each matrix were then clustered by the group- 

 average method (Lance and Williams 1967). Diet 

 information was based on dry weights of the 30 

 prey taxa categories, all of which were mutually 

 exclusive except for the unidentified (UID) and 

 miscellaneous (MISC) categories (Table 1). Prey 

 items contributing <0.1 mg of total dry weight 

 per size class were eliminated prior to the analy- 

 sis. The miscellaneous category contains the total 

 of all food items individually representing <2% of 

 the final dry weight. 



In addition to the clustering procedure, recipro- 

 cal averaging ordination (Guinochet 1973; Hill 

 1973) was used to provide independent verifica- 



Table 1 . — Prey categories used for tropic com- 

 parisons. All but unidentified (UID) and miscel- 

 laneous (t^ISC) are mutually exclusive feeding 

 categories. 



where n is the number of attributes (prey) and Xj 



787 



