596 



Fishery Bulletin 101 (3) 



100 



80 



60 



40 



20 



Group I 



Group I 



^fr^ 



to in oi 



OD CD 00 in CD CD 



y- f- y- Oi 



2. -— X C >-- 



LO CD 3- ^ — - 



9? 5 5 5 





s. --~ — 



< 



LU 



Z 



!ii <r> ^ 



r r 5 



< 



05 



Group I 



V ^ 



^;^ I 



Figure 4 



Dendrogram of the cluster analysis (Euclidean distance, average linkage method) for geographic 

 patterns of feeding habits of Raja clavata. In parentheses is given the authorship of the studies: 1 = 

 Ellis et al .( 1996 ); 2 = Gibson and Ezzi (1987 ); 3 = Ajayi ( 1982 ); 4 = Quiniou and Andriamirado ( 1979 ); 

 5 = Olaso and Rodriguez-Mann ( 1995); 6 = Cunha et al. (1986); 7 = the present study; 8 = Ebert et 

 al. (1991); 9 = Smale and Cowley (1992). 



polychaetes, cephalopods, penaeids, mysids, seabreams (Pa- 

 gellus sp.), and myctophids were consumed more by rays 

 caught in deeper waters (Table 2 ). Schoener's overlap index 

 for individuals captured at different depth intervals (Table 

 3) indicated low overlap (=0.50), supporting the results of 

 the top-down concordance coefficient analysis. 



Finally, the diet of rays caught in coastal areas and 

 offshore banks were not significantly correlated (C.p=0.44, 

 P>0.05), indicating that thomback rays feed upon differ- 

 ent prey depending on the environment. The Diogenidae, 

 Liocarcinufs spp., "other reptants," and "other Pisces" were 

 more important prey for rays in coastal areas, whereas 

 polychaetes, penaeids, cephalopods, mysids, seabreams 

 iPagellus sp.), and myctophids were more important for 

 rays caught at offshore banks (Table 2). However, Schoen- 

 er's index showed a high level of overlap (0.69) between the 

 diets of rays caught in the different locations — most likely 

 due to the high dominance of two benthopelagic fishes in 

 their diets (75.4% and 69.1% for coastal areas and offshore 

 banks, respectively I. 



Published information on the diet of Ihornback rays is 

 summarized in Table 4. Estimations of mean trophic levels 

 vary from 3.1, for the smallest size class ( South Wales: <25 cm 

 TL), to 4.2 for the Azorean thomback ray (this study; size 



Table 3 



Schoener's diet overlap index for thornback rays iRaja cla- 

 vata) size classes and for different depth strata. 



Depth (m) 



101-200 201-350 



Total length (cm) 



61-70 71-80 81-93 



0-100 0.40 0.29 49-60 0.83 0.66 0.76 



201-350 0.50 61-70 0.77 0.77 



71-80 0.62 



classes 49-60 and 81-93 cm TL). The arbitrarily chosen 

 cutoff in the cluster analysis was set at 60% dissimilar- 

 ity, which divided the dendrogram into three groups with 

 similar feeding patterns (Fig. 4). Cluster group I grouped 

 the Azorean poi)ulations (all size classes) and had an esti- 

 mated trophic level of'4. 14 ( ±0.09 SD). Cluster group II con- 

 tained all other medium and large size classes (i.e. >40 cm 

 TL), with the exception of small rays from the Canta- 

 brian Sea, North Spain (17-49 cm TL), and one small- to 



