Martinho et al.: Comparative feeding ecology of two elasmobranch species, Squalus blainville and Scyliorhinus canicu/a 
73 
where S' = the total number of species; and 
Pi =s the frequency of the ith species, 
and 2) Pielou’s index (J'), which indicates the evenness 
( i.e. , how evenly the individuals are distributed among 
the different species composing the diet of both species): 
J' = 
H' 
HL 
( 2 ) 
where H' is the number derived from the Shannon- 
Wiener index and H' max is the maximum value of H'. 
This last index varies between 0 and 1, when there is an 
equal distribution in numbers of all species. Adequacy 
of sample size was assessed by means of cumulative 
trophic diversity curves (as proposed by Ferry and Cail- 
liet, 1996), measured with the Shannon-Wiener index, 
by using all prey items, except unidentified ones. The 
cumulative numbers of 200 randomly chosen stomachs 
for S. canicula and 194 stomachs for S. blainville were 
plotted against the randomized cumulative trophic diver- 
sity. This analysis was performed in Estimates software 
(Colwell, 2009) with 50 randomizing runs. 
The relative importance of each prey item was as- 
sessed in three ways by using: 1) the numerical index 
(NI), i.e, the percentage of each prey item in relation to 
the total number of prey items (number of individuals 
of a prey category/total number of individuals among 
all prey categories )xl00; 2) the occurrence index (OI), 
the percentage of each prey item in all non-empty stom- 
achs, (number of stomachs containing a prey category/ 
total number of stomachs containing prey)xl00; and 
3) the gravimetric index (GI), the percentage of each 
prey item regarding the wet weight of all items (weight 
of individuals of a prey category/total weight of indi- 
viduals among all prey categories)xl00 (Hyslop, 1980). 
Feeding activity was evaluated by using the vacuity 
index (VI), the percentage of empty stomachs (number 
of stomachs with prey/total number of stomachs )x 100 
(Hyslop, 1980): low feeding activity is considered when 
high vacuity is observed. Diet overlap was evaluated by 
Schoener’s index (7 S ): 
7 S = 1-0.5 
which considers both the number (NI) and weight (GI) 
of ingested prey, and classifies them into dominant 
(QI>200), secondary (200>QI>20), or accidental (QI<20) 
categories; and 2) the index of relative importance (IRI) 
developed by Pinkas et al. (1971), defined as: 
IRI = (NI + WI) x OI, (5) 
which evaluates the relationships between the different 
prey items in each species’ diet, while considering the 
number (NI), weight (GI) and occurrence (OI) of each 
prey item. In order to classify the prey items accord- 
ing to the IRI, the method developed by Rosecchi and 
Nouaze (1987) was selected, in which the IRI values are 
ranked. The prey items that constitute 50% of the total 
sum are considered preferential, the prey items that 
constitute the next 25% are classified as secondary, and 
the remaining 25% are considered accessory prey items. 
The %IRI, defined as 
%IRI = 100 * IRI / ^ IRI, 
( 6 ) 
II 
PiA P iB I 
(3) 
was used to complement the information provided by 
the QI and IRI indices. For diet comparisons, prey taxa, 
excluding unidentified items, were grouped into the 
taxonomical categories referred to previously. 
Seasonal and maturity-stage-related variability 
in the diet of 5. canicula 
Seasonal and sex differences in the diet of S. canicula 
(the most abundant species) were analyzed by group- 
ing fish according to their maturity stage: adults or 
juveniles, and were presented according to the %IRI. 
Seasonal changes in diet composition were analyzed 
by partitioning the sampling period into four periods: 
winter (January to February), spring (March to May), 
summer (June to August), and autumn (September to 
November). Diet compositions of males and females 
were examined separately, in order to detect possible 
changes in habitat use. Differences in diet composi- 
tion and vacuity levels among life stages and seasons 
were assessed with analysis of variance (ANOVA, 
a=0.05). 
where p iA and p lB are the numerical frequencies of item 
i on the diet of species A and B, respectively (Linton et 
al., 1981). 
Diet overlap was measured by the I s ranges between 
0, when no food is shared, and 1, when there is the 
same proportional use of all food resources. Although 
there are no critical levels for this index, Wallace and 
Ramsey (1983) suggested that values higher than 0.6 
should be considered as biologically significant. 
To complement the information given by the tradi- 
tional indices, two mixed methods were also used: 1) 
the dietary coefficient (QI), adapted by Salgado et al. 
(2004), defined as: 
(4) 
Results 
Comparative feeding ecology of S. blainville 
and 5. canicula 
In total, 991 stomachs of S. canicula were analyzed. 
Total lengths ranged from 392 mm to 610 mm in males, 
and from 378 mm to 595 mm in females. For S. blainville, 
297 stomachs were analyzed, and total lengths ranged 
from 401 mm to 600 mm in males, and from 391 mm 
to 797 mm in females (Fig. 2). In total, the vacuity for 
S. canicula was 13.4%, and for S. blainville, 21.6%. The 
cumulative trophic diversity curves seemed to reach 
an asymptote, indicating that a sufficient number of 
QI = Nix GI, 
