Demersal and epibenthic assemblages of trawlable grounds in the northern Alboran Sea ( western Mediterranean )

The composition and abundance of megabenthic fauna caught by the commercial trawl fleet in the Alboran Sea were studied. A total of 28 hauls were carried out at depths ranging from 50 to 640 m. As a result of a hierarchical classification analysis four assemblages were detected: (1) the outer shelf group (50-150 m), characterised by Octopus vulgaris and Cepola macrophthalma; (2) the upper slope group (151-350 m), characterised by Micromesistius poutassou, with Plesionika heterocarpus and Parapenaeus longirostris as secondary species; (3) the middle slope group (351-640 m), characterised by M. poutassou, Nephrops norvegicus and Caelorhincus caelorhincus, and (4) the small seamount Seco de los Olivos (310-360 m), characterised by M. poutassou, Helicolenus dactylopterus and Gadiculus argenteus, together with Chlorophthalmus agassizi, Stichopus regalis and Palinurus mauritanicus. The results also revealed significantly higher abundances in the Seco de los Olivos seamount, probably related to a higher food availability caused by strong localised currents and upwellings that enhanced primary production. Although depth proved to be the main structuring factor, others such as sediment type and food availability also appeared to be important. Differences between shelf and slope assemblages could be in part related to a greater dependence on benthic resources in the former and a higher use of planktonic resources in the latter.


INTRODUCTION
The Alboran Sea is an interesting area due to its location on the border between three biogeographic marine regions, the Lusitanian, Mauritanian and Mediterranean regions, and the fact that it contains species typical of all three (Templado et al., 1986).This transition zone between the Atlantic and the western Mediterranean generates a highly peculiar and complex environment.Its hydrological characteristics are defined primarily by two main events: the Atlantic Jet, an input of surface waters through the Gibraltar strait which generates anticyclonic patterns in the current system and upwelling of organic-enriched deep-water (Heburn and LaViolette, 1990;Sarhan et al., 2000); and the Almería-Orán front, which forms at the meeting point of Atlantic and Mediterranean surface waters, at the eastern end of the Alborán Sea (Tintoré et al., 1988), representing a distribution barrier for several species and producing an enhancement of primary producers in its proximity (Fielding et al., 2001).
The Alboran Sea´s geomorphology is also a source of environmental variability.The continental shelf is characterised by two types of surface: narrow areas (10 km average) with low gradient slopes, and wide areas (21 km average) with steep slopes (Flanagan, 1972;Ercilla et al., 1992).The shelf break is located around 150 m depth and the foot of the slope at around 1200 m (Hernández- Molina, 1993).The Alboran Sea is a very active region inside the area of continental collision generated by the northward movement of the African plate relative to Europe (Maldonado et al., 1992).This tectonic situation between two major plates is characterised by a complex physiographic bottom (Ballesteros et al., in press).Volcanism is expressed in seamounts like the Seco de los Olivos and the Alboran Island (Maldonado and Comas, 1992).All these factors generate a high topographical complexity, a wide range of substrates, and different types of habitats.
This hydrological and geological diversity is bound to affect the biological richness of the region, which includes a wide range of species, many of which are of commercial interest.In the western Mediterranean trawling is the most important fishery, regarding both volume of catches and economic value (Sánchez et al, 2004).Several trawl fisheries belonging to Mediterranean fishing ports of Andalucía operate all year round in the area.The most important target species are Merluccius mer-luccius, Parapenaeus longirostris, Octopus vulgaris, Nephrops norvegicus and Lophius spp.due to their high economic value, and Micromesistius poutassou and Plesionika heterocarpus because of their abundant landings.
Integrated assessment of fisheries requires studies that focus on the whole ecosystem and not only on single species, and that consider fishing activities as key pressures affecting several ecosystem components (Gaertner et al., 2005;Massuti and Reñones, 2005).Therefore, it is highly necessary to develop studies like these, which identify the components, assemblage structure and functioning of ecosystems at a regional scale.
Furthermore, the presence of sensitive and vulnerable habitats, such as seamounts and volcanic crests, has resulted in a recent proposal for making the seamounts and volcanic cones of Alboran Sea a Marine Protected Area (WWF/Adena, 2006).The Texel-Faial criteria for the identification of species and habitats that need protection, defined by OSPAR (2003), requires a detailed knowledge of the quality, structure and functioning of the ecosystems and/or habitats proposed as MPAs.Despite the ecological and economical interest of the area, there is a lack of studies on Alboran demersal and epibenthic communities, with the exception of a few studies in the western Mediterranean focusing on demersal fish (Cartes et al., 2002;Gaertner et al., 2005), elasmobranch distribution (Bertrand et al, 2000), crustaceans (Abelló et al., 2002;Cartes et al., 2002) and cephalopod assemblages (González and Sánchez, 2002) that include the Alboran Sea as one of the regions studied.
The present study is the first attempt to describe the spatial distribution patterns of demersal and megaepibenthic fauna exploited by trawlers in the Alboran Sea.The main objectives were to characterise demersal and epibenthic assemblages caught by the commercial fleet operating in the area, and to establish the relationships between demersal and epibenthic fauna distribution patterns and environmental characteristics such as depth and bottom type.

Sampling
The study area included the north Alboran continental shelf and slope between 100 m and 650 m depth, stratified in 5 depth strata according to tradi-tional fishing grounds: Terraira (50-150 m), Cantillo (151-275 m), Mediamar (276-350 m), Canto (351-460 m) and Fonela (461-640 m) (Fig. 1).Additional sampling was carried out on the soft bottoms of the summit of Seco de los Olivos (310-360 m), also known as Chella Bank, which is one of the small seamounts of the Alboran Sea and has an uneven topography of coarse sands and rocky bottoms.
In this area bottom trawling is performed 5 days a week for a maximum of 12 hours per day during daylight.Data were collected from a total of 28 hauls onboard eight commercial vessels representative of the trawlers constituting the trawl fleet of Motril, from August 2000 to June 2001.This study comes from a project in which data were obtained onboard commercial vessels that did not modify their usual behaviour, fishing when and where the skipper decided.The number of hauls in a fishing ground was determined by the frequency vessels went to that area in that period: 2 hauls in Terraira, 10 in Cantillo, 5 in Mediamar, 6 in Canto, 2 in Fonela and 3 in Seco de los Olivos (Fig. 1).A 44/60 otter trawl net was used, with slight inter-vessel variations (mean mesh size=42.3mm, SD=1.6).Average values of technical characteristics were: 40.1 TRB (SD=4.9),19.8 m in length (SD=1.2) and 219.3 CV of power (SD=35.6),again showing slight inter-vessel variations.Therefore, the vessels were very similar and did not show important differences in technical characteristics.Hauls lasted between 2 and 6 hours, so abundance data were standardised to kg•h -1 .Haul duration was determined by depth, since short hauls corresponded to shallower grounds and longer ones to deeper grounds.The geographical position of each haul was recorded using GPS (Global Position System).Total catch was estimated per haul.Wet weight of each commercial species was obtained from each haul.Discarded species composition was estimated on a sub-sample depending on the quantity of discarded catches and extrapolated to the total discards (Sánchez et al., 2004).This was necessary in order to identify all species caught by trawlers, because most of them are always discarded.It should be noted that the otter trawl collected only demersal and epibenthic species (Massutí and Reñones, 2005).No endobenthic species were used in the statistical analysis.
Sedimentary characteristics were obtained from Ballesteros et al. (in press), as a qualitative scale of mean particle diameter (Q 50 ) in coarse sands (>500 µm), fine sands (62-500 µm) and mud (<62 µm).Three types of sediment were defined in the area, ranging from those with a maximum percentage of mud to those with high gravel content.The continental shelf area fishing ground (Terraira) was characterised mainly by fine sands, the slope bottoms by mud with a low sand content and the trawlable areas of the Seco de los Olivos seamount by coarse sands (Ballesteros et al., in press).

Spatial distribution patterns of demersal and epibenthic fauna
Demersal and epibenthic species abundance data were fourth-root transformed and subjected to cluster analysis using the Bray-Curtis similarity index.Prior to the cluster analysis, species with pelagic behaviour (e.g.Scomber scombrus, Boops boops, Alloteuthis spp., myctophids and Pasiphaea sivado) as well as endobenthic species (e.g.Spatangus purpureus and Sternaspis scutata) were removed.The distance matrix was then processed using the UPGMA algorithm.SIMPER analyses were also run to identify the species responsible for the major similarities in the dendrogram.
Once cluster groups had been established, demersal and epibenthic fauna abundance (average wet weight h -1 ), species richness (average species number per haul) and Shannon-Wiener diversity (species wet weight per haul) for each cluster group were calculated.Differences in faunal abundance, species richness and species diversity among cluster groups were examined using a 1-way ANOVA test and Kruskal-Wallis 1-way ANOVA on ranks for homocedastic and heterocedastic data, respectively.When significant differences were detected, we ran pairwise "a posteriori" Student-Newman-Keuls and Dunn tests, respectively, to identify the groups responsible for such differences.

Effect of environmental variables on demersal and epibenthic fauna distribution
To assess the amount of variation in demersal and epibenthic fauna abundance per haul related to 3 environmental variables (i.e.depth, geographical location and sedimentary characteristics), we used a canonical correspondence analysis (CCA).Depth was considered as a continuous variable recorded as depth of each haul.Geographical location was recorded as geographical longitude of each haul and sedimentary characteristics as a qualitative variable: mud (1), fine sand (2) and coarse sand (3).CCA calculations were based on log-transformed abundance of all species collected and down-weighting of rare species.The statistical significance of the first and all canonical axes together was tested with Monte-Carlo tests using 999 permutations under the reduced model.CCA results were presented graphically in a bi-dimensional ordination diagram generated by bi-plot scaling focusing on inter-species distances, in which species are represented by points and environmental variables by vectors.

Demersal and epibenthic fauna assemblages
The cluster analysis showed a depth-dependent grouping pattern (Fig. 2).Four groups appeared at a similarity level of 40%: (1) the Outer Shelf (OS), corresponding to the Terraira fishing ground (100-150 m); (2) the Upper Slope (US), corresponding to the Cantillo and Mediamar fishing grounds (151-350 m); (3) the Middle Slope (MS), corresponding to the deepest hauls located in the Canto and Fonela fishing grounds (351-640 m); and (4) the Seco de los Olivos seamount (SS) at 310-360 m depth.The first dichotomy (28%) showed group OS as the most different of the four groups, when shelf hauls were separated from slope hauls.The SIMPER analysis showed that the OS group was characterised by O. vulgaris, Cepola macrophthalma, Lophius spp.and Illex coindetti.These species, together with Pagellus acarne, were responsible for dissimilarities between the other three cluster groups (Table 2).The upper slope hauls (US) were characterised by M. poutassou, Plesionika heterocarpus, Parapenaeus longirostris and Gadiculus argenteus, whereas the middle slope hauls (MS) were typified by M. poutassou, Nephrops norvegicus, Caelorhincus caelorhincus and Phycis blennoides.The SS group was split in a second dichotomy out of the bathymetric pattern of clustering, showing its peculiarity with respect to DEMERSAL AND EPIBENTHIC ASSEMBLAGES OF ALBORAN SEA • 517 An ANOVA test and a Student-Newman-Keuls "a posteriori" test revealed that demersal and epibenthic fauna abundance was significantly higher in Seco de los Olivos (SS) than in the rest of the cluster groups (F = 11.60,p<0.001) (Fig. 3).However, there were no significant differences between cluster groups in either species richness or species diversity.

Effects of depth, geographical location and sedimentary characteristics on species distribution
The first three axes of CCA explained 30% of the faunal variation in the species per sample matrix, and 100% of the variation in the "speciesenvironment" matrix.Montecarlo tests indicated that both the first axis (p = 0.002) and all canonical axes together (p = 0.002) were significant.Therefore, the environmental variables under study are clearly responsible for at least some of the faunal variation.In the CCA, both depth and bottom type (Q 50 ) showed a comparable weight in building axes 1 and 2 (Fig. 4a).The Longitude factor was correlated with the Q 50 factor, although with less importance.Nevertheless, in the bi-plots (Fig. 4a, Fig. 4b) the high dispersion of samples and species followed a bathymetric diagonal axis.The relevance of the Q 50 factor and, to a lesser extent, of Longitude, was due to the singularity of the SS group, corresponding to the Seco de los Olivos fishing ground.Coarse and medium sand covered this ground, a bottom type different from the rest.Likewise, the Terraira fishing ground (OS group) had a different bottom type consisting of fine sand,   and was the shallowest ground in the study area.
The remainder of the hauls, corresponding to the slope groups (US and MS) and containing muddy bottoms, followed a bathymetric gradient.The same response was observed in the ordination of species (Fig. 4b), again following a diagonal bathymetric gradient.The greatest discrimination was due to species related to the OS and SS groups.The outer shelf assemblage (OS group) was characterised by fish like the spotted dragonet Callyonimus maculatus, the scaldfishes Arnoglossus imperialis and A. thori, gobids such as Deltentosteus quadrimaculatus and Gobius niger, John Dory Zeus faber, horse mackerel Trachurus trachurus, the triglid Lepidotrigla dieuzeidei, and invertebrates like the mollusc Octopus vulgaris, the urchin Echinus acutus and the anthozoan Pennatula rubra.Several species were found in this habitat but also in the deeper muddy bottoms of the upper slope (US group), such as the red bandfish Cepola macrophthalma, the squids Ilex coindetti and Todaropsis eblanae, and the cuttlefish Sepia orbignyana.
Finally, the top predators Lophius spp., Conger conger and Merluccius merluccius were located near the centroid of the analysis, since these three species were caught in all six fishing grounds (see Table 1), showing their ubiquitous distribution in the study area.seamount.Species codes are given in Table 1.

DISCUSSION
Andalusian fishermen defined the Alboran fishing grounds as bathymetric strata with boundaries at 150, 275, 350 and 460 m.Only the small "Secos" seamounts differed from this bathymetric pattern.Our results were consistent with this definition, and confirmed depth as the first structuring factor, revealing three main assemblages, with boundaries at 150 and 350 m.This faunal zonation has also been found in other Mediterranean Sea areas (Abelló et al., 1988;Ungaro et al, 1999;Gaertner et al., 2005).The first assemblage corresponded to the outer shelf, since the shelf break is located at 150 m depth (Hernández- Molina, 1993).The other two assemblages were located on the slope and have been identified as upper and middle slope.Depth also represents the main structuring factor in other areas of the Mediterranean Sea (Abelló et al., 1988;Demestre et al., 2000;Kallianiotis et al., 2000;D´Onghia et al., 2004;Gaertner et al., 2005;Massutí and Reñones, 2005), although it has often been argued that marine organisms may respond to a combination of depthrelated factors such as food availability, light, temperature and pressure (Cartes et al., 2004a).
In addition to being the shelf break depth, 150 m is also a hydrographical boundary, since the limit of the surface Atlantic Jet and the deeper Mediterranean waters is located approximately at this depth (Fielding et al., 2001).Furthermore, a high thermal and saline stability of the water mass below 150 m has been described for the whole Mediterranean Sea (Hopkins, 1985), determining the eurybathic distribution of most deep-Mediterranean species (Cartes et al., 2004b).Regarding the second boundary (at 350 m depth), between upper and middle slope, a faunal renewal belt around 300 m for megafaunal assemblages has been described (Cartes et al., 2004b).
The fourth assemblage corresponded to the small seamount Seco de los Olivos.The environment of this fishing ground was clearly different since rocky bottoms, inhabited by a great variety of corals and gorgonians, were interpolated with (trawlable) sandy bottoms.Therefore, the areas available to trawlers, characterised by coarse sands, were very limited and influenced by the near rocky bottoms.Similar geomorphologic formations in the study area, which are commonly known as 'Secos', have traditionally been fishing grounds.All of them have volcanic origin and can be considered as seamounts and special habitats (Maldonado and Comas, 1992;Ballesteros et al., in press).Recently, the scientific community has paid particular attention to these habitats (OSPAR, 2003) and proposed them as protected areas, although many studies are still needed to demonstrate their special characteristics and vulnerability.The Seco de los Olivos seamount showed significantly higher faunal abundances than the rest of the fishing grounds.This may indicate high food availability at this site due to strong localised currents and upwellings, which may enhance primary production and plankton biomass, resulting in alteration of the whole trophic web (Rogers, 1994;Koslow, 1997).Indeed, the planktophagous M. poutassou, as well as other low trophic level fish such as Gadiculus argenteus, Macrorhamphosus scolopax and Trachurus picturatus, displayed a higher abundance on the Seco de los Olivos seamount.
In addition to the bathymetric factor, the ordination analysis identified sediment type as another key factor structuring faunal assemblages in the study area.Nevertheless, all slope fishing grounds had the same bottom type, muddy sands, and were grouped into two different assemblages, upper and middle slope.The faunal renewal boundary of 350 m (Cartes et al., 2004b) generates a degree of variability higher than sediment type.The importance of sediment diameter in the canonical correspondence analysis was because the two most different assemblages also corresponded to different sediment types, i.e. fine sands on the shelf and coarse sands on the Seco de los Olivos seamount.Other authors have mentioned that particle size distributions alone may not always be the best guide for predicting epibenthic and demersal community types and that other factors, such as seabed morphology, appear to have a greater influence (Duineveld et al., 1991).However, to clarify the role of the sediment in structuring assemblages, a more intensive sampling would be required.
Regarding faunal composition, top predators like Lophius spp., M. merluccius and C. conger were dominant in all assemblages.The outer shelf group was characterised by typical shelf species, such as Octopus vulgaris, Arnoglossus spp., Callyonimus maculatus, Pagellus acarne and Cepola macrophthalma (Ungaro et al., 1999;Gaertner et al., 2002;González and Sánchez, 2002;Kallianiotis et al. 2004), most of them of benthophagous feeding habits.Cartes et al. (2002) described shelf fishes and crustaceans as more dependent on benthic resources, while plankton resources are utilised more by slope species.This pattern is a consequence of the decrease in benthic biomass with depth and the higher density of zooplankton offshore at the shelf break (Cartes et al., 2002).Slope groups, including the Seco de los Olivos seamount, were dominated by M. poutassou, which is a key trophic species that represents a link between nekton and demersal communities.In the upper slope group, the dominant species were the fish Caelorhincus caelorhincus and G. argenteus, and the crustaceans Plesionika heterocarpus and Parapenaeus longirostris (Abelló et al., 1988;Gaertner et al., 2005).This pandalid dominance again indicates a planktonic origin of the trophic resources (Massutí and Reñones, 2005).Another planktophagous fish, Gadiculus argenteus, was abundant on the upper slope, but especially on the Seco de los Olivos seamount, showing a bathymetric affinity for the 150-350 m depth stratum (Gaertner et al., 2002).Species that dominated the middle slope assemblage were typical deep-sea species, such as Galeus melastomus, C. caelorhincus and Phycis blennoides (Gaertner et al., 2002;Massutí and Moranta, 2003;Cartes et al., 2004b;D'Onghia et al., 2004), and species with mud affinity like Nephrops norvegicus (Cartes et al., 2004b).
None of the species caught on the Seco de los Olivos seamount were seamount-aggregating fish (Morato et al., 2006), although most of them have already been described as seamount fauna: e.g.Chlorophthalmus aggassizi, Pagellus bogaraveo, Molva macrophthalma, Synchiropus phaeton and Trachurus picturatus (Shcherbachev et al., 1985;Fock et al., 2002).Some of them have been regarded as hard-bottom fauna; for example, Kallianiotis et al. (2004) grouped C. aggassizi and P. bogaraveo in an upper slope assemblage inhabiting coralliferous stones, and Palinurus mauritanicus needs the proximity of rocky habitats for refuge (Goñi and Latrouite, 2005).Other species were characteristic of this bathymetric range (H. dactylopterus, P. blennoides, G. argenteus, and C. caelorhincus).The absence of seamount-aggregating species (p.e.orange roughy, alfonsinos) could be related to the deeper-dwelling character of these species.Furthermore, sampling on the Seco de los Olivos seamount was carried out on sandy bottoms, whereas rocky habitats, seamount slopes and less-trawled areas are more appropriate for these species (Koslow, 1997).Fauna inhabiting seamounts has been placed by several authors at the extreme end of the vulnerability spectrum (e.g.Fock et al., 2002;Morato et al., 2006).Therefore, fisheries in such areas need to be carefully controlled, since they are characterised by a boom and bust sequence, showing signs of overexploitation within a short period after the beginning of the fishery (Koslow et al., 2000;Morato et al., 2006).
The results of the present work should be taken into account for future assessment and management of trawl fisheries in the Alboran Sea, which should be based on the whole ecosystem.An integrated approach is especially necessary in vulnerable habitats such as seamounts, which are highly sensitive to anthropogenic impacts (Fock et al., 2002;Morato et al., 2006).Therefore, all the information available on habitats like the Seco de los Olivos seamount will be highly appreciated in the assessment of fisheries in the Alboran Sea.
FIG. 1. -Map of the study area (north Alboran Sea, western Mediterranean Sea) showing the location of the 28 hauls in the 6 fishing grounds.

FIG. 3
FIG. 3. -(a) Megabenthic fauna abundance (kg•h -1 ), (b) species richness, and (c) diversity (H') per cluster group.Bars represent mean (± SD) values.Uppercase letters (A to D) refer to mean values arranged in decreasing order.Groups of underlined letters indicate non-significant differences between pairs of means according to "a posteriori" Dunn's tests.
FIG.2.-Dendrogram of hauls based on Bray-Curtis similarity of megafauna abundance.For information on species responsible for major nodes in tree topology see Table2.Acronyms of assemblages: OS, Outer Shelf; US, Upper Slope; MS, Middle Slope; SS, Seco de los Olivos seamount.

TABLE 2 .
-Species contributing most to dissimilarity between the four groups resulting from the cluster analysis.ABU= average abundance in the cluster group; % CUM, cumulative percentage of species contributions; OS, outer shelf; US, upper slope; MS, middle slope; SS, Seco seamount.