Food habits of the broad nose skate, Bathyraja brachyurops (Chondrichthyes, Rajidae), in the south-west Atlantic

1 Laboratorio de Ictiología, FCEyN, Universidad Nacional de Mar del Plata (UNMdP), Funes 3350, Mar del Plata, B7602AYL, Argentina. E-mail: mbellegg@mdp.edu.ar, belleggia@inidep.edu.ar 2 Museo del Mar, Colón 1114, Mar del Plata, B7600FXR, Argentina. 3 Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Argentina. 4 Comisión de Investigaciones Científicas del Gobierno de la Provincia de Buenos Aires (CIC), Argentina. 5 Instituto Nacional de Investigación y Desarrollo Pesquero, Paseo Victoria Ocampo 1, Mar del Plata, B7602HSA, Argentina.

Skates, like other cartilaginous fishes, are characterized by slow growth, late attainment of sexual maturity and lower fecundity than teleost fishes (Stevens et al., 2000).Their life history characteristics make them particularly sensitive to overexploitation and once overfished, populations of skates need more time to recover than populations of teleost fishes (Stevens et al., 2000;Stehmann, 2002).Skates are increasingly targeted in fisheries conducted on the Argentinean continental shelf (Massa and Hozbor, 2003); catches increased from 300 t in 1991 to 14856 t in 1998 (Cousseau et al., 2000).In Argentinean harbours they are commonly landed with no species differentiation under the common name "skates".Therefore, further studies on the biology and ecology of skates are needed for proper management of these fisheries.
Despite the local diversity of Bathyraja in Argentinean waters, few studies have focused on the feeding habits and ecology of species of this genus.The diet of five species of Bathyraja have been briefly described on the Argentinean continental shelf (Sánchez and Mabragaña, 2002), and only the feeding habits of the Patagonian skate B. macloviana have been thoroughly studied (Mabragaña et al., 2005;Scenna et al., 2006).Brickle et al. (2003) provided some data on diet composition of B. albomaculata, B. brachyurops and B. griseocauda around the Malvinas Islands.The diet composition of skate species of other genera have been studied in the region (Lucifora et al., 2000;Koen Alonso et al., 2001;Sánchez and Mabragaña, 2002;Braccini and Perez, 2005;Mabragaña et al., 2005;Mabragaña and Gilberto, 2007;San Martín et al., 2007).Although understanding a predator's trophic interactions is crucial for developing sustainable management strategies (Robinson et al., 2007), trophic relationships among skate species in Argentinean waters remain unknown.
The objectives of this paper were to describe the diet composition and trophic level of B. brachy-urops on the Argentinean continental shelf, evaluate whether there is sexual, regional and ontogenetic variation in their feeding habits and determine their feeding strategy.This study provides the first detailed information on food habits of B. brachyurops in an extensive area of the south-west Atlantic in order to understand the role of the species in the food web of this region.

Sample collection
Specimens of B. brachyurops were collected from ten research cruises on the RVs "Dr.E. L. Holmberg" and "Capitán Oca Balda" of the Instituto Nacional de Investigación y Desarrollo Pesquero (INIDEP, Mar del Plata, Argentina), between 2003 and 2005 on the Argentinean continental shelf (36 o S-55 o S) (Fig. 1).Skates were captured using a bottom trawl of 200 m mesh in the wing and 103 mm in the codend.
Disc width (DW) of each specimen was measured to the nearest millimetre and sex was recorded.Stomachs were excised, fixed in 4% formaldehyde and analyzed in the laboratory.

Diet composition
Prey items were identified to the lowest possible taxonomic level using keys, field guides (Bastida and Torti, 1973;Menni et al., 1984;Boschi et al., 1992;Cousseau and Perrotta, 2000) and reference collections at the Ichthiological Laboratory of the University of Mar del Plata.Each prey item was counted and weighed to the nearest 0.01 g using a digital top loading balance.Diet composition was assessed according to the percentage frequency of occurrence (%F), the percentage of wet weight (%W), the percentage of number (%N) and the Index of Relative Importance (IRI = %F (%N + %W)) (Pinkas et al., 1971) expressed as a percentage (Cortés, 1997).
Prey were assigned to seven taxonomic categories (Pisces, Brachyura, Isopoda, Mollusca, Polychaeta, Anomura and Amphipoda) to analyze sexual, regional and ontogenetic variation in the diet.Regional dietary difference was assessed by comparing two study regions (Fig. 1): northern (35ºS-43ºS) and southern (43ºS-55ºS).The northern region is more influenced by the warm, salty waters of the Brazil Current which flows southward adjacent to the east-ern South American continental shelf.The southern region is characterized by the cold and relatively fresh subantarctic waters of the equatorward Malvinas Current (Boltovskoy, 1981;Olson et al., 1988).Ontogenetic variations were determined by analyzing the following size classes: I (<400 mm DW), II (400-500 mm DW) and III (>500 mm DW).
To assess sample size sufficiency (Ferry and Caillet, 1996;Cortés, 1997), the order of stomachs sampled were randomized 100 times, and the mean cumulative diversity of stomach contents (Shannon-Wiener diversity index) plotted as a function of stomach number.Cumulative curves were built separately for each combination of specimen groups considered in the comparative analyses.

Diet comparisons
Similarity in the composition of the diet (%W and %N of each taxonomic category) between regions and size classes was assessed using the multivariate statistical software PRIMER v5 (Clarke and Warwick, 2001).Similarity matrices were constructed using the Bray-Curtis similarity coefficient.Two-way nested analyses of similarities (ANOSIM) using regions and size classes as factors were used to determine whether the dietary composition of B. brachyurops was significantly influenced by location or ontogeny.One-way ANOSIMs were performed to identify any paired relationships.Similarity percentages (SIMPER) were used to identify which taxonomic categories characterized the dietary composition of each sample and which taxonomic categories made the greatest contributions to any dissimilarity.
Preliminary multivariate analysis indicated that there were no significant differences between the diets of females and males (ANOSIM %W: R statistic = 0.01, p = 0.069; ANOSIM %N: R statistic = 0.003, p = 0.254) and thus dietary data from both sexes were pooled for subsequent analyses.

Trophic level and feeding strategy
Trophic level (TL) was estimated to determine the position of B. brachyurops within the food web and was calculated following Cortés (1999) as: n TL = 1 + (Σ P j x TL j ) j=1 where TL j is the trophic level of each prey category j, P j is the proportion of each prey category j (using %IRI) in the diet of B. brachyurops, and n is the total number of prey categories.The taxonomic categories used to calculate the standardized trophic level of B. brachyurops were Teleosts, Chondrichthyes, Cephalopoda, Mollusca (excluding cephalopods), Brachyura, Isopoda, Polychaeta, Anomura and Amphipoda.The trophic level of each prey category was obtained from Ebert and Bizarro (2007).
In order to analyze the feeding strategy of each size class of B. brachyurops, the graphic method proposed by Amundsen et al. (1996) was used by plotting the prey-specific abundance (%Pi) of each prey category against %F.Pi was calculated as the weight of prey category i, divided by the total weight of prey in the stomachs that contained prey category i, expressed as a percentage.Prey points located at the upper right of the diagram are indicative of specialization of the predator population.In contrast, all prey points located along or below the diagonal from the upper left to the lower right reflect a generalized feeding strategy of the predator population.Furthermore, the distribution of points along the diagonal from the lower left to the upper right corner provides a measure of prey importance, with dominant prey at the upper and rare prey at the lower end.Levins' measure (B) was used for calculating niche breadth of each size class of B. brachyurops using the following equation (Krebs, 1989): where p i is the proportion of each prey category i in the diet and n is the total number of prey categories in the diet of B. brachyurops.The standardized Levins' index (B est = (B -1) / (n -1)) was used to express niche breadth on a scale from 0 (a narrow niche breadth) to 1 (a broad niche breadth).

Diet composition
All the cumulative diversity curves reached an asymptote, which indicates that the sample sizes were sufficient for describing and comparing the diets (Fig. 2).A total of 346 skates was examined, of which 265 (76.6%) were found with stomachs containing prey items at different stages of digestion.Males (n=125) ranged from 245 to 567 mm DW, while females (n=140) ranged from 268 to 692 mm DW.

Diet comparisons
Two-way nested analyses of similarities showed that there were no significant differences in overall diet between northern (n=118) and southern (n=147) populations (ANOSIM %W: R statistic = -0.037,p = 0.5; ANOSIM %N: R statistic = 0.037, p = 0.5).However, significant differences in the dietary composition of the size classes categorized according to percent of weight and percent of number were ob- served (ANOSIM %W: R statistic = 0.14, p = 0.001; ANOSIM %N: R statistic = 0.137, p = 0.01).
Pairwise comparisons within ANOSIM in terms of %W and %N showed significant differences between size class I (n= 92) and size class II (n= 96) and size class III (n= 77), but not between size class II and size class III.According to the ANOSIM test, larger differences occurred between size classes I and III (Table 2).SIMPER analyses showed that the average dissimilarity between size class I and size class II was high (76.45% by %W and 79.4% by %N).This is mainly due to differences in the importance of isopods and fishes in the diet of the two size classes, although crabs also contributed considerably to the dietary dissimilarity (Table 3).Average dissimilarity between size class I and size class III categorized by %W and %N were 79.33% and 82.13% respectively.SIMPER identified isopods, fishes and crabs as the prey categories responsible for these differences (Table 3).
The diet of size class I consisted largely of isopods.Individuals of size class II fed more frequently on fishes, followed by crabs.In size class III, fish was the most important prey category, followed by amphipods.Fish prey became increasingly important in the diet of larger specimens of B. brachyurops (Table 4).

Trophic level and feeding strategy
The trophic level of B. brachyurops was 4.16.The three size classes had similar values and were all tertiary consumers (TL size class I = 4.14, TL size class II = 4.12 and TL size class III = 4.19).However, %Pi-%F plots showed a pronounced and progressive change in diet of B. brachyurops with increased size, from a mixed dietary composition in the smallest skates to one dominated almost entirely by fishes in size classes II and III.Individuals of size classes II and III fed mainly on teleosts but small proportions of other prey categories were included in the diet of some specimens (Fig. 3, Table 4).The niche breadths in both classes were narrow (Fig. 3).Conversely, the graphical method proposed by Amundsen et al. (1996) demonstrated a more mixed feeding strategy for individuals of size class I, with a certain preference for isopods (Fig. 3, Table 4).Furthermore, the niche breadth value of small skates of B. brachyurops was wider (Fig. 3), which suggests a more generalized feeding strategy.

DISCUSSION
Stomach content analysis of Bathyraja brachyurops showed that it is a piscivorous predator that feeds largely on benthic and demersal fishes on the Argentinean continental shelf.Other benthic invertebrates are present in the diet but in lower proportions.The dominance of fishes and cephalopods in the diet of large individuals of larger skate species has been reported by several authors (Ebert et al., 1991;Smale and Cowley, 1992;Orlov, 1998;Lucifora et al., 2000;Koen Alonso et al., 2001;Brickle et al., 2003).Koen Alonso et al. (2001) noted that smaller specimens of the long nose skate Dipturus chilensis in the Argentinean Sea preyed on benth-ic crustaceans, whereas larger individuals foraged mostly on fishes and molluscs.A general pattern for the feeding habits of skates was suggested by Braccini and Perez (2005), in which small individuals (either smaller species or young individuals of larger species) prey mainly on crustaceans and thus are secondary consumers (TL<4), while larger individuals feed on fishes and cephalopods and thus are tertiary consumers that occupy higher trophic levels (TL>4).Nevertheless, no differences were found between the trophic levels of the three size classes of B. brachyurops analyzed here, which ranged between 4.12 and 4.19.Ebert and Bizarro (2007) estimated the trophic level for 60 skate species, based on quantitative data obtained from different studies.Values ranged from 3.48 (Rajella caudaspinosa) to 4.22 (D. chilensis) and the greatest values for species of Bathyraja were 4.09 (B.griseocauda) and 4.08 (B.brachyurops).In our study the trophic level of B. brachyurops was 4.16 and thus conformed to the feeding pattern hypothesis as a tertiary consumer which occupies trophic positions similar to other upper trophic level predators e.g.marine mammals, seabirds (Wetherbee and Cortés, 2006), large teleosts and some shark species (Ebert and Bizzarro, 2007).One case of cannibalism was found in our study and this behaviour has been observed in other skates (Ellis et al., 1996).Ebert et al. (1991) analyzed the diet of several species of skates from the south eastern Atlantic, including one species of the genus Bathyraja (B.smithii), and found, similarly to the present study, that the most important prey were crustaceans and teleosts.In this study the most frequently consumed teleosts that contributed most to the diet in terms of weight were Patagonotothen spp., Argentine hake, and scorpaenids (Helicolenus dactylopterus).The Argentine hake is the most important demersal fishing resource on the Argentinean shelf (Bezzi et al., 2004).Furthermore, the Argentine hake is consumed by a lot of organisms, such as Illex argentinus (Santos and Haimovici, 1997), Squalus acanthias (García de la Rosa and Sánchez, 1997), Dipturus chilensis (Lucifora et al., 2000;Koen Alonso et al., 2001), dusky dolphins Lagenorynchus obscurus (Alonso et al., 1998) and 20 other fish species (García de la Rosa and Sánchez, 1997), which prey on different stages of ontogeny.
There were no sexual differences in the dietary composition of B. brachyurops, which indicates that males and females consume similar prey and occupy similar trophic roles.Similar findings have been reported in other studies (Braccini and Perez, 2005;Scenna et al., 2006;San Martín et al., 2007).Conversely, sexual differences in the diet have been found in other skate species of Bathyraja in the western Bering Sea (e.g. B. parmifera, B. aleutica, B. maculata, B. matsubarai, B. minispinosa) (Orlov, 2001).Sexual heterodonty has been investigated in B. brachyurops and it was found that females have wider, more rounded teeth on both the upper and lower jaws than males (Belleggia, 2007).Dental sexual dimorphism was originally proposed to be a result of feeding segregation among sexes (Du Buit, 1978).It is also well known that elasmobranchs exhibit complex reproductive behaviour in which the teeth are used by males for firmly gripping the female during copulation (Price, 1967;McEachran, 1977;Kajiura et al., 2000).Males and females of B. brachyurops have a similar diet, which suggests that differences in tooth morphology could be related more to reproductive behaviour than to feeding.Although all size classes of B. brachyurops occupy similar trophic roles and are tertiary consumers (TL>4), ontogenetic shifts in the diet composition were found in this work.Small specimens of B. brachyurops predominantly fed upon isopods, whereas fishes and crabs became important in the diet of larger skates.Small individuals of B. brachyurops that occurred around Malvinas Islands mostly foraged on benthic gammarid amphipods and isopods, whereas larger skates consumed fish and cephalopods (Brickle et al., 2003).Orlov (1998) found that larger species of Bathyraja in the northern Pacific (B. aleutica, B. maculata, B. parmifera, and B. matsubarai) preyed on large crustaceans, cephalopods and fishes, while smaller species (B.interrupta, B. minispinosa and B. violacea) fed on benthic invertebrates.Ontogenetic changes in the diet composition of skates have been reported for several other species (Pedersen, 1995;Skjaeraasen and Bergstad, 2000;Brickle et al., 2003), and may be attributed to morphological constraints (McEachran et al., 1976) or may simply reflect differences in the foraging ability of larger individuals.Dietary changes associated with the ontogeny of B. brachyurops may, in part, explain how food resources within a given area are partitioned to minimize competition between co-specifics at different life-history stages, as Ebert (2002) proposed for Notorynchus cepedianus.Ontogenetic change in feeding habits is an almost universal phenomenon in fishes and thus its occurrence in elasmobranchs is not surprising.Although many species of skates increase in size, there are also changes in habitat, movement patterns, swimming speed, size of jaws, teeth and other factors that result in variable exposure to prey or improved ability to capture different prey (Wetherbee and Cortés, 2006).
Some studies on the feeding habits of skates have described them as generalist predators (McEachran et al., 1976;Orlov, 1998), although some species have been regarded as specialist predators (Ebert et al., 1991;Braccini and Perez, 2005;Scenna et al., 2006;San Martín et al., 2007).A gen-eralist predator has a broad dietary niche, in contrast to the small variety of prey found in the stomach contents of a specialist predator.In our study, the Levins' standardized index indicated that the niche breadth was widest for small skates, whereas the niche breadth of the larger skates was relatively narrow.The analysis of the prey-specific abundance in relation to frequency of occurrence showed that small individuals of B. brachyurops have a generalist strategy with isopods as the main prey.However, larger skates, demonstrated a marked specialization in teleosts.
As in this study, Brickle et al. (2003) found that adult specimens of Bathyraja brachyurops were active predators and preyed mainly on fishes and cephalopods.However, the Argentine short fin squid Illex argentinus appeared to be the most important prey among cephalopods in the diet of B. brachyurops on the Argentinean continental shelf (the present study) and the loliginid squid Loligo gahi was the most common (in terms of numbers and frequency of occurrence) in the diet around the Malvinas Islands (Brickle et al., 2003).The difference would be attributed to different distribution patterns observed in these two species of cephalopods.Loligo gahi is the coldest water dwelling loliginid species, and reaches its highest abundance in waters associated with the Malvinas Current which derives from the Antarctic Circumpolar Current (Hatfield and Des Clers, 1998).In contrast, I. argentinus presents latitudinal migrations and its concentrations during different times of the year are governed by feeding, sexual maturation and egg-laying (Brunetti et al., 1998).It is a species associated mainly with waters of the Patagonian shelf (Haimovici et al., 1998).
Bathyraja brachyurops occurs sympatrically with B. macloviana and B. albomaculata along the Argentinean continental shelf (Cousseau et al., 2000), which have been reported to be polychaete consumers (Sánchez and Mabragaña, 2002;Mabragaña et al., 2005;Scenna et al., 2006).Species that share the same habitat may rely on differences in the utilization of food resources to avoid competition (Schoener, 1974).In our study we found that polychaetes made a low contribution to the diet of B. brachyurops (%IRI =0.44).The differences in feeding habits between these two polychaete consumers and B. brachyurops would indicate low feeding competition among these species.
Prior to this diet study, no published information was available on the food habits of B. brachyurops on the Argentinea continental shelf.To date our understanding of the Argentinean benthic ecosystem is scant.Although the results of this research will contribute to increasing our knowledge about benthic communities, further studies are warranted to fully understand the food webs of benthic and demersal communities on the Argentinean continental shelf.

Fig. 1 .
Fig. 1. -Study area showing the location of sampling sites where Bathyraja brachyurops were collected on the Argentinean continental shelf.The rectangle in the inset represents the study area.

Fig. 2 .
Fig. 2. -Cumulative prey diversity curves for each trophic group of Bathyraja brachyurops on the Argentinean continental shelf.The dark line indicates the mean diversity estimated by the Shannon-Wiener diversity index as a function of sample size, and the light lines indicate the standard deviations.

Table 1 .
-Percentage of occurrence (%F), number (%N), wet weight (%W) and percentage of the Index of Relative Importance (%IRI) for each prey present in the diet of Bathyraja brachyurops.

Table 2 .
-One-way ANOSIM results for Bathyraja brachyurops between diets of different size classes analyzed according to percent of weight and percent of number.

Table 3 .
-The contribution of prey categories to observed dietary differences among B. brachyurops size classes determined by SIMPER analyses.Prey categories are listed in descending order of percentage contribution and only categories contributing >2% to the observed differences are shown.

Table 4 .
-Percentage of occurrence (%F), number (%N), wet weight (%W) and percentage of the Index of Relative Importance (%IRI) for each taxonomic category present in the diet of different Bathyraja brachyurops size classes.