Feeding ecology of Bathyraja macloviana ( Rajiformes : Arhynchobatidae ) : a polychaete-feeding skate from the South-west Atlantic *

1 Laboratorio de Ictiología, Universidad Nacional de Mar del Plata, Funes 3350, B7602AYL, Mar del Plata, Argentina. E-mail: emabraga@mdp.edu.ar 2 Museo del Mar, Colón 1114, B7600FXR, Mar del Plata, Argentina. 3 Laboratorio de Bentos, Instituto Nacional de Investigación y Desarrollo Pesquero (INIDEP), V. Ocampo N° 1, B7602HSA, Mar del Plata, Argentina. 4 Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET).


INTRODUCTION
Skates are a significant element of the trophic food webs in benthic and demersal communities (McEachran and Musick, 1975;Ajayi, 1982;Ebert et al., 1991;Orlov, 1998).Within the skates, Bathyraja is the most diversified genus and comprises about 45 species (McEachran and Dunn, 1998;Compagno, 1999).These fishes have a worldwide distribution, being present in temperate and cold waters from the continental shelf to the upper slope (McEachran and Miyake, 1990).In spite of this, the knowledge of the feeding ecology of this genus is scarce (Ebert et al., 1991;Orlov, 1998).
The Patagonian skate B. macloviana is endemic to South America, distributed along the Atlantic and Pacific Oceans (Pequeño and Lamilla, 1993;Menni and Stehmann, 2000).In the ACS, it is distributed from 35°S to 55°S, between 80 and 250 m depth.It was also recorded in the Beagle Channel (Norman, 1937;Bellisio et al., 1979).
B. macloviana is the smallest species of the genus in the ACS, reaching 67 cm in total length (TL) (Cousseau et al., 2000), and it is one of the most important skates in terms of biomass in Patagonian waters (Sanchez and Mabragaña, 2002).In this area, B. macloviana matures at 53-58 cm of TL, suggesting a late maturity.Preliminary analysis of the diet suggested that this skate feeds on polychaetes (Sanchez and Mabragaña, 2002;Scenna, 2003).
Skates as a whole are becoming target species in South-west Atlantic fisheries (Agnew et al., 1998;Paesch and Meneses, 1999;Cousseau et al., 2000).The general low fecundity and late maturity typical of most rajid species would indicate that they are particularly sensitive to fishing pressure and overexploitation (Walker and Hislop, 1998).Most of the demersal fisheries in Argentina include by-catches of skates that were mostly discarded until the early 1990s.This situation changed and catches increased from 300 tons in 1991 to 14,856 tons in 1998 (Cousseau et al., 2000).Therefore, further studies on the biology and ecology of this group are needed for the proper management of the fishery.
The main goals of this work are to analyse the diet, feeding strategy and niche width of B. macloviana in the ACS.

Study area
The study area comprises a sector of the ACS, between 37ºS and 50ºS and 80 and 400 m depth in the southwest Atlantic (Fig. 1).The water masses of the shelf are of subantarctic origin.Subantarctic waters enter the shelf mainly between the Malvinas Islands and Tierra del Fuego.These water masses are modified substantially by the contribution of diluted waters from the Magellan Strait and freshwater from Negro River and La Plata River.Bottom temperature ranges from 4 to 15°C.The general pattern of circulation of marine currents follows the continental shelf with NNE direction (Guerrero and Piola, 1997).

Data collection
Bathyraja macloviana individuals were collected during the day from 43 sampling stations on two research cruises designed to assess demersal fish stocks in late summer (southern locations) and autumn (northern locations) 2001 by the R/V "Eduardo L. Holmberg" (INIDEP) (Fig. 1).Skates (n=147) were captured using a bottom trawl performed with an Engel mesh type.Total length (TL) and disc width (DW) of each specimen were measured to the nearest millimetre (mm), weight was measured to the nearest gram (g) and sex was recorded.Some stomach contents were analysed on board and prey items were identified and assigned to major taxonomic groups.In order to identify the prey groups to a lower specific level, 108 stomachs were frozen and analysed in the laboratory.
As part of a usual sampling regime carried out at each station, biological data (abundance, TL, sex and gut contents assigned to major taxonomic groups) of other rajids collected (n= 491) from the 43 sampling stations were recorded on board.

Data analysis
The diet was expressed as frequency of occurrence of major groups for all the skates collected on board (n= 638).On the other hand, prey items of B. macloviana analysed in the laboratory (n = 108) were identified to the lowest taxa possible, counted and weighed (wet weight) to the nearest 0.01 g.We assessed the prey importance by percentage frequency of occurrence (%F), numerical abundance (%A) and percentage of biomass (%B).An index of relative prey importance (IRI) was calculated for each prey category i as the product of %F i and %B i and expressed as a percentage, where: and n is the total number of food categories considered at a given taxonomic level (Cortes, 1997;Griffiths, 1997).Numerical percentage contribution was excluded in this analysis because it is biased towards small prey items (e.g.amphipods and isopods), and because it is difficult to estimate numbers in each category when there is mastication of food items (Hyslop, 1980;Griffiths, 1997).
The analysis was carried out at two levels: the lowest taxonomic groups possible and functional groups.
To evaluate the potential vulnerability to fish predation, prey items were assigned into functional groups from specific literature using aspects of life habits (mobility as the major criteria: motile, discretely motile and sessile) related to particular traits that influence invertebrate conspicuousness (Fauchald and Jumars, 1979;Boschi et al., 1992;Wolff, 1992).Stomachs with prey items that showed a high degree of digestion (e.g.only "polychaete remains" ) were removed to perform this analysis (n=21).
Similarity in the composition of the diet (% IRI of each prey item) between stations was determined by a non-parametric multivariate analysis (Field et al., 1982;Clarke, 1993;Clarke and Warwick, 2001) and the ANOSIM (Analysis of similarities) test (Clarke and Green, 1988;Clarke and Warwick, 2001).The ANOSIM test was used to search for differences in the diet between northern and southern sectors defined in the study area.This permutation test analyses differences between replicates within sites contrasted with differences between sites, computing an R statistic under the null hypothesis "no difference between sites".R falls between -1 and 1, so R is approximately 0 if the null hypothesis is true and R= 1 if all replicates within sites are more similar to each other than any replicates from different sites.Classification (group average sorting of the Bray-Curtis similarity measures based on 4th root transformed %IRI data) and ordination (multidimensional scaling (MDS) on the above similarity matrices) were performed using the PRIMER software (Clarke and Warwick, 2001).Niche width was analysed by the reciprocal of Simpson's diversity index (λ i ), the Levins index B (Levins, 1968) where p ij is the proportion of utilisation of resource j by species i.
Finally, the feeding strategy was analysed with the graphical method proposed by Amundsen et al. (1996), which incorporated the prey-specific biomass into Costello's (1990) analysis.This new parameter is defined as the percentage that a prey taxon comprises of all items in only those predators in which the actual prey occurs, or in mathematical terms: where Pi is the prey-specific biomass of prey i, Bi the stomach content (weight) comprised of prey i, and Bt i the total stomach content in only those predators with prey i in their stomach.Then this value is plotted against the frequency of occurrence on a two-dimensional graph and information about feeding strategies in terms of specialisation (narrow dietary niche width) and generalisation (broad dietary niche width) can be extracted.Additional information about the method is available in Costello (1990) and Amundsen et al. (1996).
Regarding functional groups, B. macloviana preyed upon infaunal and slow motile invertebrates, mainly discretely motile polychaetes and crustaceans (Table 2).Epibenthic crustaceans and motile polychaetes were also included, but in lower proportions (  Skates showed similar values in diet composition (% IRI) throughout the study area.The global test of the ANOSIM (R = 0.065, p= 79.3%) indicated no difference between sites.This similar diet was also evident in the classification (Fig. 2a) and ordination (Fig. 2b) analyses performed, in which all sampling sites were clustered together, from the coast of Buenos Aires (northern stations) to Patagonia (southern stations) (Figs. 1, 2).
A specialisation of B. macloviana was found, with most of the prey points positioned towards the upper right corner of the graph (Fig. 3).Typically, all individuals had been feeding on the dominant prey taxon, polychaetes, but small proportions of other prey types, like crustaceans, were included in the diet of some individuals.Echinoderms and molluscs were rarely found.At specific level, there was  a tendency to specialisation towards the discretely motile worm T. kerguelensis, while other prey types showed a low average contribution to the stomach contents (mainly amphipods Gammaridae and polychaetes Lumbrineridae and Nephtyidae) (Fig. 4).
The value of niche breadth (B= 3.37) for B. macloviana, calculated on the lowest specific level of preys, indicates a relative low variability in its diet.
Ten species of skates (n= 491) were collected together with B. macloviana.The most important prey items, except for Bathyraja albomaculata, were crustaceans, mainly burrowing isopods Cirolanidae, decapods and amphipods (Table 3).Demersal fishes, mainly Nototheniidae, were observed in high frequencies in two abundant and large species Dipturus chilensis, and Bathyraja brachyurops.On the other hand, Illex argentinus, the most abundant cephalopod in the ACS (Brunetti et al., 1998), was consumed by three large species B. griseocauda, B. brachyurops and D. chilensis.Finally, although reaching similar sizes and feeding primarily on crustaceans, two small skates, Psammobatis rudis and P. normani, showed differences in the other taxa consumed.In P. rudis fishes (juveniles of Nototheniidae) constituted the secondary item, and polychaetes and molluscs were of minor importance, whereas in P. normani polychaetes were consumed in high frequency, fishes were of minor importance and molluscs were absent from the diet.

DISCUSSION
The present study on the feeding ecology of Bathyraja macloviana showed its narrow food niche and a marked tendency to specialisation on polychaetes of low motility in the southwest Atlantic.The most important prey items observed in 9 other species of skates in ACS (see Table 3) were crustaceans, squids and fishes; only Bathyraja albomaculata also consumed polychaetes in high frequency.
A central aspect of niche theory and feeding strategies is the generalist-specialist dichotomy (Pianka, 1988).A generalist predator has a broad dietary niche, whereas the food niche of a specialist is narrow.To our knowledge, no record of specialisation on worms has been reported for skates.However, specialisation on crustaceans and fishes is known for several skates from the eastern north Atlantic (Ajayi, 1982;Ellis et. al., 1996), north Pacific (Orlov, 1998), southeast Atlantic (Ebert et al., 1991) and southwest Atlantic (Lucifora et al., 2000;García de la Rosa, 1998).Polychaetes were found in some generalist feeders but mostly constituted secondary items in the diet (McEachran et al., 1976;Orlov, 1998).In Raja radiata, for example, worms were found in the diet, together with other items, and varied among areas of the east coast of North America (McEachran et al., 1976).Ebert et al. (1991) described the diet of several species of skates from southwest Africa, including other species of the genus Bathyraja (B.smithii), and observed that the most important prey items were crustaceans and fishes, whereas polychaetes were not important food items.In the same way, on the other side of the southern Atlantic (the ACS), crustaceans and fishes constituted the most important prey items observed in the majority of the skates.However, in contrast to southwest African skates, two worm-consumers were reported in the ACS, the Patagonian skate B. macloviana and the white spots skate B. albomaculata (Sanchez and Mabragaña, 2002;Scenna, 2003;Ruocco, 2004 and present study).Similar values of %IRI and a relative low niche width were estimated from spring B. macloviana samples taken in the northern ACS (38-44°S) (Scenna, 2003).
Records of specialisation on worms in the ACS are not common.Two coastal fishes, the hawkfish Cheilodactylus bergi (Bruno et al., 2000) and the apron ray Discopyge tschudii (Garcia, 1984) displayed analogous values of polychaete preys to B. macloviana, feeding almost exclusively on worms.
Previous studies on the diet of skates (not only Bathyraja) from outer ACS revealed that the largest species usually preyed on fishes and molluscs (squids), while the smallest ones included crustaceans and polychaetes in the diet (García de la Rosa, 1998;Lucifora et al., 2000;Paesch et al., 2000;Kohen Alonso et al., 2001;Sanchez and Mabragaña, 2002 and present study).Studies from South Patagonian waters (48-55°S ) indicated that of 11 skate species, only B. albomaculata (n= 10) and B. macloviana (n= 45) showed occurrence frequency of polychaetes > 65%, and Psammobatis normani (n= 60) contained worms as secondary items (Sanchez and Mabragaña, 2002).These species displayed a similar feeding pattern in the present study (Table 3).
The results of the MDS (which presented a low resolution plot with no segregation of clear sites groups) and the ANOSIM test (R value close to zero implying little or no segregation of groups) indicated that the diet of B. macloviana was similar throughout the study area, consisting mainly of the Opheliidae T. kerguelensis.Opheliidae are found worldwide, usually in sandy sediments and muds.These worms are burrowers, feeding on detritus in sediment (Fauchald and Jumars, 1979).The dominance of T. kerguelensis in the diet suggested that B. macloviana fed deep in the substrate.On the other hand, Mianzan et al. (1996) found that this species may also feed on ctenophores, a feeding behaviour that can be explained by the later finding of nearbottom aggregations of these planktonic organisms (Costello and Mianzan, 2003;Colombo et al., 2003).
Species sharing the same habitat may rely on differences in the utilisation of food resources to avoid competition (Schoener, 1974).Several species of skates occur sympatrically with B. macloviana along the ACS (Cousseau et al., 2000).Most of them are potential food competitors, being typically benthophagous and reaching similar sizes during the ontogeny.Bathyraja macloviana feeds particularly on discretely motile polychaetes, excluding other macrobenthic invertebrates present in higher abundances on the bottom, while most of the sympatric skates of similar sizes prey mainly on crustaceans.B. macloviana preyed on polychaetes, a faunistic group characterised by low relative abundance in the ACS.The feeding pattern displayed by B. macloviana might reflect morphological differences in the structures involved in detection, capture and digestion of preys as well as different alimentary behaviours, allowing a low dietary overlap with other sympatric skates in the ACS.
FIG. 1. -Location of sampling sites of stomach contents of Bathyraja macloviana, corresponding to southern (+) and northern (•) regions of the study area.

FIG. 2
FIG. 2. -Feeding patterns of Bathyraja macloviana in relation to the region of the ACS analysed.(a) Classification of stations using %IRI of prey taxa, (b) MDS ordination of stations using %IRI of prey taxa (Stress: 0.21).S = southern stations, N = northern stations.

FIG. 4
FIG. 4. -Feeding strategy plot for Bathyraja macloviana.Contribution of prey taxa expressed as percentage of prey-specific biomass.Points represent different prey types.

TABLE 1 .
-Stomach contents of Bathyraja macloviana (n = 108) from the study area, presented as percentage of occurrence (%F), abundance (%A), wet biomass (%B), and percentage of the index of relative importance IRI (%F x %B) for each prey taxa (%IRI) (see text for details).

TABLE 2 .
-Stomach contents of Bathyraja macloviana (n = 87) from the study area presented as percentage of occurrence (%F), wet biomass (%B) and percentage of the index of relative importance IRI (%F x %B) for each functional groups (%IRI) (see text for details).

TABLE 3 .
-Stomach contents presented as percentage of occurrence (%F) of major taxa and range TL (cm) of rajids (n = 491) collected simultaneously with the patagonian skate B. macloviana in the study area.