454 



Fishery Bulletin 103(2) 



with a stereoscopic microscope and dietary items were 

 identified to the lowest taxonomic level possible by using 

 specialized keys. Garth (1958), Rodriguez de la Cruz 

 (1987), Hendrickx and Salgado-Barragan (1991), and 

 Hendrickx (1996), were consulted for crustacean iden- 

 tification, whereas Jordan and Evermann (1896-1900), 

 Castro-Aguirre (1978), Allen and Robertson (1994), 

 Thomson et al. (2000), and FAO guides were used for 

 fish identification (Fischer et al., 1995). 



Both the number of individuals and weight of each 

 dietary category were quantified, and mean proportions 

 in terms of number (%7V V ) and biomass C7rWj ) were cal- 

 culated according to Tirasin and Jorgensen (1999): 



IX 



%X, = 



j=i 



■xlOO, 



where X = the number or weight of each taxa i in the 

 j th stomach; and 

 k = the number of dietary components found in 

 all stomachs analyzed, /;,. 



The percent frequency of occurrence of each component 

 was also obtained ( c 7cF). Finally, the index of relative 

 importance for each dietary category was calculated 

 (IRI, Pinkas et al., 1971; Rosecchi and Nouaze, 1987): 



IRI, =(%N i +%P i )x%F i . 



Relative importance index values were expressed as 

 a percentage of the total items analyzed (Cortes, 1997) 

 and results were graphically represented as a rectangle 

 of base %F and height 7c N + 7cW. 



Variance analysis was applied on transformed W'= 

 sin _1 (VW)l gravimetric proportions of the dietary 

 components (Zar, 1999) to evaluate both monthly and 

 ontogenetic variations in the feeding habits of C. nan- 

 nus. The number [q , =l+3.322(Log ]0 n)] and width of size 

 classes (w=R TL /q) were considered for analysis, where ;; 

 is the sample size and R TL =TL mRX -TL mm . 



For the analysis of trophic niche breadth, the nor- 

 malized version of the index proposed by Levins (1968) 

 was used. This index combines both the number of prey 

 resources used (k) (i.e., the trophic spectrum) and the 

 relative frequency with which each prey resource is 

 consumed (J). This represents the distribution function 

 of prey proportions in diet (Hespenheide 1975; Hurlbert, 

 1978): 



(n, \ 



Ba- 



k-1 



Because the ensemble of prey found in any given 

 stomach does not constitute independent samples (Hurl- 

 bert, 1984), p f was calculated as the proportion of indi- 



vidual fish (iV*) that consumed a certain food resource 

 in relation to the number of resources used by the total 

 number of fish: 



N 



YTn sothat I^ = 1 - 



Ba values range between and 1. Zero values indi- 

 cate that fish feed on only one prey type, representing 

 the minimum diet breadth and high feeding special- 

 ization. Unity values, on the other hand, indicate that 

 the species consumed all k food resources in the same 

 proportion (p ; = l/&), representing no selection among 

 prey types and the widest possible trophic niche (Gibson 

 and Ezzi, 1987; Labropoulou and Eleftheriou, 1997). Ba 

 values were calculated on the basis of matrix resources 

 (Colwell and Futuyma, 1971) both for each month and 

 for each size class. The percentage similarity measure 

 (/?) between size classes q' and q" (Renkonen, 1938; 

 Schoener, 1970; Hurlbert, 1978) was calculated as 



V-=l"2 



H\p Jq --p,A 



where p is the proportion of individual fish in each size 

 class that consumed a certain food resource, calculated 

 over the total number of stomachs per size class. 



Confidence intervals (CI 95% ) of Ba were obtained by 

 means of the bootstrap method (Mueller and Altenberg, 

 1985; Efron and Tibshirani, 1986) by considering two 

 thousand resamplings of the data (Hamilton, 1991). 



Results 



The 311 Cynoscion nannus examined ranged from 7.5 

 to 20.6 cm TL. Food was found in 287 (92%, rang- 

 ing from 85% to 98% among size classes) stomachs. 

 The trophic spectrum of C. nannus is composed of 29 

 dietary items (Table 1), which were classified into four 

 general categories: penaeid shrimp, fish, stomatopods. 

 and cephalopods. 



Penaeid shrimp constituted the principal dietary cat- 

 egory of C. nannus (A^ = 82.5%, W f = 35.4%; Fj=43.9%, 

 77?/j = 74.6%; Fig. 1), of which juvenile stages were the 

 most frequent (F f =23.4%). Fish were the second most 

 important category (7V f = 6.5%, W f = 36.5%, F v = 37.7%, 

 7ff/j = 14.5% ), followed by stomatopods of the Squilla 

 genus (iV—5.8%, W==8.6%, F I= 25.5%, 7/^ = 6.6%). The 

 cephalopod Loliopsis diomedae was the last category in 

 order of importance (JV==1.0%, W f = 12.4%, 7^ = 4.2%, 

 IRI; = 1.87c). 



Overall, significant differences in diet were found 

 between individuals of different size classes (F=1.03; 

 P<0.05). Values of the percentage similarity of diet (i?) 

 between size classes were, in general, <50% (Table 2). 

 R -values were relatively high only among size classes 2 



