592 



Fishery Bulletin 101(3) 



E t/5 

 o 



3.2 



2.8 



2.4 



2.0 



1.6 



1.2 



0.8 



0.4 



0.0 



[IRI={%N + %W) X %0] (Pinkas 

 et al., 1971) and the 9(IRI (as 

 %77?7, = 100 X IRIJ1IRI,) were 

 calculated for each prey category 

 and used in diet comparisons. 

 Prey taxa occurring in less than 

 five stomachs were grouped into 

 higher taxonomic categories. 

 Ontogenetic differences in the 

 diet of thornback rays were 

 examined by grouping fish into 

 four size classes (49-60, 61-70, 

 71-80, and 81-93 cm TL). The 

 diet of thornback rays was also 

 analyzed by sex, depth (0-100, 

 101-200, 201-350 m), and area 

 of capture (coastal areas and off- 

 shore banks). No further analy- 

 ses were performed for tope 

 shark because their diet was 

 dominated by only one prey cat- 

 egory (see "Results" section). To 

 determine if the most important 

 preys were similar for different 

 groups of rays, weighted corre- 

 lation and concordance analyses 



were used (Zar, 1999). These methods were preferred to 

 conventional rank correlation methods (e.g. Spearman) 

 because they emphasize the high ranking given to the 

 most important prey categories. Differences in the rank- 

 ings of IRI values for prey categories between three or more 

 groups (e.g. three size classes) were tested for significance 

 with the top-down concordance method iC\= top-down 

 concordance coefficient) (Zar, 1999). For paired groups (e.g. 

 males and females) the top-down correlation method (r,p= 

 top-down correlation coefficient) was used (Quade and 

 Salama, 1992; Zar, 1999). Schoener's dietary overlap index 

 (Schoener, 1970) (as C„= 1-0.5 1 1 P^, - P^, | , where P^- was 

 the proportion (based on %IRI ) of food category / in the diet 

 ofx; and P^,, was the proportion of food category i in the diet 

 ofy) was used to measure the diet overlap between sex, size 

 classes, depth strata, and area of capture. 



Cluster analysis was used to describe geographic simi- 

 larities in the feeding habits of thornback rays. A preda- 

 tor-prey matrix was built from published data. When more 

 than one index was available, the following criteria were 

 used to choose between indexes: IRI or '/< IRI, ^O, %N, %W, 

 %Volume. The number of prey categories included was 

 based on the quality of the description found in the pub- 

 lished sources. Eleven different categories were obtained. 

 A distance matrix was then calculated by using Euclidean 

 distance, and the hierarchical form of analysis was applied 

 (Clarke and Warwick, 1994 ). The grouping of predators was 

 based on the "average linkage method," and a dendrogram 

 was used as a graphic form of representation. Finally, tro- 

 phic levels (TLd^) were estimated for each of the samples {k) 

 by using the method proposed by Cortes ( 1999) [as TLVf^=l+ 

 'I^,* X TLu^t, where TLi', is the trophic level of each prey 

 category as estimated by the author, P,^, is the proportion of 

 prey category i in sample k]. Mean trophic levels were also 



Raja clavala 



Galeorhimts gateus 



10 20 30 40 30 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 

 Number of stomachs witti contents 



Figure 2 



Randomized cumulative trophic diversity curves for thornback ray (Raja clavata) and 

 tope shark (Galeorhinus galeus) . 



estimated for groups resulting from the cluster analysis, 

 and differences between them were tested by using one- 

 way ANOVA (Zar, 1999). 



Results 



Thornback rays were caught at depths ranging from 10 to 

 350 m, but primarily (95%) shallower than 250 m. Out of 

 237 stomachs examined, the contents of four appeared to 

 have been regurgitated (1.7%), seven contained bait only 

 (2.9%), 88 were empty (37.1%), and 138 contained prey 

 (58.2%). Rays with stomachs containing food measured 

 from 49.0 to 93.0 cm TL. All tope sharks were caught 

 between 10 and 150 m depth, except for one individual 

 taken at 300 m. Out of 365 stomachs examined, 174 (47.7%) 

 were empty, seven ( 1.9% ) contained fish hooked on the long- 

 line and 184 stomachs (50.4% ) contained prey. Sharks with 

 stomachs containing food ranged from 58.0 to 153.0 cm 

 TL. The cumulative trophic diversity curves of both spe- 

 cies appeared to reach an asymptote, suggesting that a 

 sufficient number of stomachs were analyzed for both the 

 thornback ray and tope shark (Fig. 2). 



Thornback ray 



The main diet components of thornback rays were fish 

 (%IRI=81 .6) and crustaceans reptants (%IRI=17.4) (Fig. 3). 

 Fish occurred in 84.1% of stomachs that contained food, 

 and represented 78.0% of total prey weight and 50.2% of 

 total prey number (Table 1). Two benthopelagic species, 

 the snipefish {Macroramphosus scolopax (%IRI=34.01) and 

 the boarfish (Capros aper (%IRI=26.8]), were by far the 

 predominant fish prey items. However, some pelagic fish 



