Mediterranean and Atlantic deep-sea fish assemblages : differences in biomass composition and size-related structure *

1 IEO Centre Oceanogràfic de les Balears, P.O. Box 291, 07080 Palma de Mallorca, Spain. E-mail: enric.massuti@ba.ieo.es 2 Scottish Association for Marine Science, Dunstaffnage Marine Laboratory, Oban, Scotland, PA37 1QA, United Kingdom. 3 Institut Mediterrani d’Estudis Avançats (CSIC/UIB), Miquel Marquès 21, 07190 Esporles, Spain. 4 Museu de Granollers Ciències Naturals, Francesc Macià 51, 08400 Granollers, Spain. 5 Muttlebury’s Mead, Chard Street, Thorncombe, Chard TA20 4NB, United Kingdom.


INTRODUCTION
Estimates of biomass and abundance give an indication of the role and relative importance of populations in food webs, including the amount of energy needed for maintenance (e.g.Platt, 1985;Bergmann, 1987).A knowledge of total abundance and biomass of deep-sea fauna and how these are partitioned among different functional and size categories can be a potentially useful approach for understanding trophic relationships and energy flow into and through these ecosystems (Thurston et al., 1994, Haedrich, 1996).
Furthermore, size distribution can provide information on the evolutionary responses of species and on community metabolism, and can also reflect the influence of fishing activity.However, there have been few studies of the species size and biomass spectra for deep-sea demersal fish assemblages.The only available information is from the Porcupine Seabight and the Madeira Abyssal Plain in the eastern North Atlantic (Haedrich and Merrett, 1992) and from the upper slope off southern Namibia in the southeastern North Atlantic (Macpherson and Gordoa, 1996).
In recent years, the deep-sea ichthyofauna of the Mediterranean Sea and the Atlantic Ocean has been extensively studied.In the western Mediterranean, both community structure (e.g.Stefanescu et al., 1993;Moranta et al., 1998) and distribution and biology of the main species have been studied (e.g.Massutí et al., 1995Massutí et al., , 1996;;Morales-Nin et al., 1996;Carrasón and Matallanas, 1998, 2001, 2002).In the eastern North Atlantic, there are also many publications, both on species composition and structure of assemblages (e.g.Merrett et al., 1991;Koslow, 1993) and the ecology of some dominant species (e.g.Mauchline and Gordon, 1984;Gordon and Duncan, 1985;Coggan et al., 1998;Allain, 2001).
From the results of the papers cited above, clear differences can be inferred between Mediterranean and Atlantic deep-sea fish assemblages.These differences are due not only to distinct species composition (e.g.Haedrich and Merrett, 1988;Stefanescu et al., 1993), but also to the maximum size of individual species (Tortonese, 1960;Stefanescu et al., 1992) .It is important to note, however, that misleading conclusions in comparative studies can arise from the use of different methodologies.Thus, in recent years, the importance of sampling gear when one is comparing fish assemblages has been demonstrated (e.g.Merrett et al., 1991;Gordon et al., 1996).
For this reason, our aim is to make a standardised comparison between biomass composition and sizerelated structure for both the whole fish assemblage and the most common species found in the deep sea of the western Mediterranean (Algerian and Balearic basins) and eastern North Atlantic (Rockall Trough and Porcupine Seabight).These areas are characterised by distinctly different and well-documented oceanographic conditions, surface productivity and seasonality, biogeographical affinities, biodiversity and fishing exploitation patterns.

Study areas
The western Mediterranean is a subtropical, semi-enclosed area separated from the Atlantic by a sill in the Strait of Gibraltar, with a high degree of environmental stability for both temperature (12.8-13ºC) and salinity (38-38.6‰)below a depth of 200 m (Hopkins, 1985).The two sampled basins of the western Mediterranean (Fig. 1), the Balearic (north of the Balearic Islands) and the Algerian (south of the Balearic Islands) basins are characterised by different oceanographic and geomorphological conditions (e.g.Beckers et al., 1997;Pinot et al., 2002).
The Algerian basin acts as a reservoir for water of Atlantic origin.By contrast, the Balearic basin has a large cyclonic circulation, with two permanent fronts linked to the Northern Mediterranean Current and the Balearic Current (Millot, 1999).These frontal boundary regions, reinforced by the formation of a winter water mass with minimum temperature in the water column, are particularly relevant in the general oligotrophic context of the Mediterranean Sea as biologically active locations (Estrada, 1996).Moreover, the presence of numerous submarine canyons in the Balearic basin can also influence the environmental conditions in this area.These geomorphological structures play an important role in the transport and concentration of sediment to greater depths and have been proved to be areas of high productivity (Monaco et al., 1990).Available estimations give the annual primary production in the northwestern Mediterranean as between 77 and 100 g C m -2 year -1 (Minas et al., 1988), although only about 1-5 g C m -2 may reach the benthic ecosystems at 800 to 1000 m depth (Miquel et al., 1994).
The Rockall Trough and the Porcupine Seabight (Fig. 1) are temperate regions characterised by marked seasonal changes in primary productivity related to increased summer temperatures (11-14°C) in the upper water column and a seasonal thermocline.Below this, the environmental conditions are virtually constant throughout the year, with a gradual decrease in temperature with increasing depth, from about 10°C at 600 m to about 3.5ºC at 2000 m (Ellett et al., 1986;Rice et al., 1991).There is some evidence that the two areas could constitute a marginal oceanic region in the eastern Atlantic, where greater than average production takes place compared with adjacent regions (Mauchline, 1990).Although direct measurements of annual primary production are not available, the general rate for these areas appears to be within the range 60-100 g C m -2 year -1 , with a vertical flux to the deeper layers of about 12-20 g C m -2 (Berger et al., 1989).

Sampling and data sets
Data were derived from a series of bottom trawl surveys carried out between 1978 and 1998 by different groups of researchers, studying the distribution and ecology of deep-sea benthic and benthopelagic communities.Catch data from a total of 214 hauls were used in the analysis (Fig. 2 western Mediterranean; and 39 trawls between 530 and 1955 m depth (Rockall Trough) and 77 trawls between 407 and 1993 m depth (Porcupine Seabight) in the eastern North Atlantic.
As long-term routine monitoring of fish assemblages tends to be confined to the economically more important shelf species, there are relatively few long time-series of data on deep-sea fish assemblages using standardised sampling methods.For this reason, catches from different years and seasons have been pooled in our analysis.
It has also been recognised that there are considerable differences between catches, in terms of species composition, relative abundance and length frequency, from otter bottom trawls towed on single or paired warps (e.g.Merrett et al., 1991;Gordon et al., 1996).Therefore, only samples obtained by a single warp trawl were analysed.In the Rockall Trough, Porcupine Seabight and the Balearic basin, the gear used was a standard OTSB-14 (e.g.Rucabado et al., 1991), whereas in the Algerian basin the modified semi-balloon otter trawl OTMS-27.5 (Sardà et al., 1998) was used.Both gears were towed at 2.5 knots and had similar vertical openings (1.5-2.0 m), with their effective horizontal openings considered to be around 6.7 and 14 m respectively and a codend mesh size of 12 mm.
Values were obtained for the number and weight of individuals, as well as the length frequencies of demersal fish species at each station.Depending on the morphology of the species examined, total length, head length, pre-anal length, standard length or gnathoproctal length was measured for the total catch or for random sub-samples.Biomass and abundance were standardised to a common sampled area (1000 m 2 ), according to the methodology most commonly employed in studies of deep-sea fish assemblages.This calculation takes into account the greater path width of the OTMS -27.5 trawl.This trawl was specifically designed to improve the catch rates of decapod crustaceans, but it also yielded some significantly different catch rates of some fish species (Sardà et al., 1998).Weight was calculated using available length-weight relationships obtained during these samplings.

Data analysis
For analysis, the data were grouped into the following depth intervals: 400-800 m, 800-1400 m and 1400-2000 m.These depth strata approximately correspond to the upper, middle and lower slopes pro-posed by Haedrich and Merrett (1988) in the North Atlantic, and to the different fish groups identified along the continental slope of the western Mediterranean (Stefanescu et al., 1993;Moranta et al., 1998).
Standardised catch data were pooled for these depth intervals, and the ecological parameters of species richness, biomass (kg 1000 m -2 ) and abundance (n.1000m -2 ) were determined.For comparative purposes, the geometric mean [ln (x+1)] was preferred to the arithmetic mean for biomass and abundance, in order to minimise the negative effect of extreme values.The normality of these parameters was examined for standard skewness and kurtosis, and the homogeneity of variance was investigated using Cochran's test.The examination of the data, with unbalanced numbers of hauls per region and depth strata, indicated that the assumptions of normality and homogeneity of variances were violated.However, according to Underwood (1997), the analysis of variance may be considered robust and not greatly affected by violations of the assumptions where samples are relatively large (n > 6) and a more conservative significance level is adopted (in this case p < 0.01).Two-factor orthogonal analysis of variance (ANOVA) was used to test differences between the mean values of the ecological parameters for both studied areas and depth strata.After ANOVA, a Bonferroni multiple range test was applied to determine which means were significantly different.
Cluster analyses using the Bray-Curtis index and the Unweighted Pair Group Method with Arithmetic Mean (UPGMA) as the aggregation algorithm was applied to calculate and visualise similarities between the taxonomic composition, both in terms of biomass and abundance of main groups, by depth strata and area.For the analysis, only the groups representing more than 0.1% of the total catch in the assemblage were considered.The Bray-Curtis index was also applied to calculate the similarity percentage between areas in terms of species composition.
Biomass by size class for each assemblage was estimated following standard methodologies, using log 2 groupings of fish weight (g) on the x-axis (e.g.Macpherson and Gordoa, 1996).Correspondence analysis was used to examine the patterns of variation of assemblages in relation to different size classes.This technique has the advantage of providing a simultaneous representation of variables (assemblages) and descriptors (size classes) in a common system of orthogonal axes.

Biomass and abundance
The samples analysed yielded a total biomass of 11153 kg belonging to 140 species of 46 families (Appendix).Despite the different sampling effort in each area, large differences between the Mediterranean and the Atlantic assemblages were observed in the total catch.The 1.4 and 1.6 km 2 sampled in the Algerian basin and the Rockall Trough respectively yielded a total catch of 719 and 6089 kg respectively.Similarly, in the Balearic basin (4.7 km 2 sampled) and the Porcupine Seabight (3.2 km 2 sampled), the total catch was 976 and 3369 kg respectively.
No interactions were identified between the variables area and depth strata, and standardised biomass and abundance values were found to be signif-icantly different for both of these variables.The highest values were from the Rockall Trough and the lowest values were from the two Mediterranean areas.There was no significant difference between biomass and abundance values for the two Mediterranean areas (Table 1).There were significant differences in biomass between depth strata due to the high biomass values of middle slope assemblages, whereas for abundance the only significant difference was between the upper and lower slope.

Species richness and taxonomic composition
The number of species captured was 43 in the Balearic basin, 47 in the Algerian basin, 80 in the Rockall Trough and 104 in the Porcupine Seabight (Table 2).The percentage similarities within both the Mediterranean and Atlantic areas were much MEDITERRANEAN VS ATLANTIC DEEP-SEA FISH ASSEMBLAGES 105 higher than between the Mediterranean and Atlantic assemblages.A similar trend was evident when the number of species in common between areas was examined (Table 2).
In the Mediterranean, there was no obviously predominant family over the whole depth range (Table 3).The dominant families of the upper slope assemblages were Gadidae and Scyliorhinidae in terms of biomass and Gadidae and Macrouridae in terms of abundance.On the middle slope, Moridae and Alepocephalidae dominated in terms of biomass, while Moridae, Alepocephalidae and Macrouridae dominated in terms of abundance.On the lower slope, Alepocephalidae was the most important family in terms of biomass, whereas Chlorophthalmidae and Macrouridae dominated in terms of abundance.
By contrast, in both the Atlantic areas, the family Macrouridae was the most important in terms of biomass throughout the whole depth range surveyed (Table 3), although other important families included Gadidae and Chimaeridae on the upper slope, Moridae and Alepocephalidae on the middle slope and Synaphobranchidae on the lower slope.Synaphobranchidae and Macrouridae were the most important families in terms of abundance at all depth strata.106 E. MASSUTÍ et al.The dendrograms derived from the standardised values of biomass and abundance for taxonomic groups showed clear differentiation, in the first cluster, between Mediterranean and Atlantic assemblages (Fig. 3).The second cluster in the Mediterranean assemblage separated the upper from the middle and lower slope, both in terms of biomass and abundance, while in the Atlantic this bathymetric segregation was not so clear and the second dichotomy separated the Rockall Trough and the Porcupine Seabight assemblages in terms of biomass, but not abundance.

Size-related structure
The biomass spectra for the whole depth range surveyed in each area (Fig. 4) showed a clear gradient from the Balearic basin, in which 88% of the biomass was found between size classes 3 and 9 with a mode at 7, to the Rockall Trough, in which 80% of the biomass was situated between size classes 8 and 13, with a mode at 10 Similar trends were observed for the selected depth strata (Fig. 5), with greater differences on the upper slope, where a higher percentage of the biomass was comprised of size classes ≤ 6 in the Mediterranean assemblages and of size classes ≥ 5 in the Atlantic assemblages.The only exception was found on the lower slope, where the biomass spectrum from the Algerian basin was similar to those found in the Atlantic assemblages.The correspondence analysis applied to biomass spectra by depth strata showed that data were arranged in a parabolic form (Fig. 6), which signified a clear gradient.The first axis (F1) explained 58% of the variability and discriminated between the Mediterranean assemblages, in which biomass was mainly distributed in small and middle size classes, and the Atlantic assemblages, which were more related to larger size classes (10-13).Clear size differences were also observed, at a species level, by comparing the length of ten species captured in significant numbers within the four areas (Table 4).In all cases, both maximum (with the only exception of Helicolenus dactylopterus) and mean lengths were higher in the Atlantic than in the Mediterranean.By contrast, minimum lengths were more similar, with the only exceptions of Centroscymnus coelolepis and Helicolenus dactylopterus, where the minimum size in the Atlantic was at least double that found in the Mediterranean.

DISCUSSION
The standardised comparisons of this study confirm the clear differences between the deep-sea demersal fish fauna of the Mediterranean and the Atlantic, which were already suggested four decades ago by Tortonese (1960) and more recently shown, at a species level, by Stefanescu et al. (1992) andD'Onghia et al. (1998).

Biomass and abundance
The biomass and abundance indices were lower for the Mediterranean assemblages than for the Atlantic.The factors that cause these differences are likely to be complex.Although the influence of surface productivity on demersal megafauna is poorly understood, pelagic production and its arrival, as phytodetritus, on the sea floor has been reported as one of the most important sources of organic matter for the northwestern Mediterranean (Miquel et al., 1994) and eastern North Atlantic (Rice et al., 1986) deep-sea ecosystems.Although similar ranges of surface productivity have been estimated in these areas, the availability of organic matter at the seabed in the Atlantic, which has been estimated as an order of magnitude higher than in the Mediterranean (around 20% and 2-5%, respectively; see Material and Methods), could contribute to the differences in density of fish fauna between the two areas.The higher temperature below 200 m in the Mediterranean could contribute to a faster decompo-MEDITERRANEAN VS ATLANTIC DEEP-SEA FISH ASSEMBLAGES 109 sition of the organic matter during its transport through the water column before its arrival to the sea floor as food source.
Studies on the diets of deep-water fishes in the Rockall Trough (Mauchline and Gordon, 1991 and references cited therein) have shown the dominance of meso, bathy and benthopelagic prey.Therefore, the enhanced demersal biomass down to about 1000-1500 m depth most probably results from the efficient transfer of surface production, via overlapping food chains and vertical migration of prey.Conversely, in the western Mediterranean, it has been suggested that this energy transfer mechanism only reaches about 800-1000 m depth (Cartes, 1998).
The higher values of fish abundance obtained from the upper slope of the Balearic basin could be related to the presence of submarine canyons, which were absent in the Algerian basin.It has been shown that these geomorphological structures are a means of transporting sediments rich in organic matter from the continental shelf, and that they also create a focus for diel migrations of plankton from upper levels (Macquart-Moulin and Patriti, 1993).These canyons can act as recruiting grounds for both fish (Stefanescu et al., 1994) and crustaceans (Cartes et al., 1994).Similarly, the Rockall Trough has a gentle slope, with no canyons, and except at around 400-500 m, is composed of soft mud with an increasing biogenic content with depth (Mitchell et al., 1997), while the Porcupine Seabight is bisected by canyons (Rice et al., 1991) that separate a northern and a southern trawlable area.The northern area is comprised of very soft sediments, especially at mid-slope depths, and dense aggregations of the sponge, Pheronema carpenteri (Rice and Thurston, 1990).Multivariate analysis separated the catches of the larger paired warp trawl between these two areas, but the OTSB catches were not separable (Gordon et al., 1996).
The biomass indices showed some similar bathymetric trends between all four areas, with a maximum on the middle slope, while abundance indices were more variable.The peak of biomass in both the Balearic and Algerian basins occurs at between 1000 and 1200 m depth (Stefanescu et al., 1993;Moranta et al., 1998 respectively), which results both from some medium and large-sized species reaching their greatest abundance at these depths (e.g.Alepocephalus rostratus, Morales-Nin et al., 1996) and from other species reaching their greatest biomass due to a bigger deeper trend (e.g.Trachyrincus scabrus and Phycis blennoides; Massutí et al., 1995 and1996 respectively).The isobath of 1200 m is considered to be about the lower limit of the mesopelagic fauna and the reduction of this trophic resource probably accounts for the decline in biomass on the Mediterranean lower slope (Stefanescu et al., 1993).In the Rockall Trough, biomass had maximum values at mid-slope depths.This peak, which was evident using several different trawls (Gordon, 1986), is most probably related to the depth range of greatest potential vertical and horizontal impingement of epi-and mesopelagic fauna on the slope (Mauchline and Gordon, 1991).Similar results were obtained using several trawl types in the Porcupine Seabight (Merrett et al., 1991).

Species richness and taxonomic composition
Atlantic assemblages are composed of a larger number of species than those of the Mediterranean, which has been widely described as being less diverse than the Atlantic for some groups (Bouchet and Taviani, 1992).The more constant physical environment and the recent origin of the deep-sea fauna of the Mediterranean (Pérès, 1985) probably accounts for the fewer number of species.In the Atlantic, the Porcupine Seabight has a greater number of species than the Rockall Trough.The Rockall Trough is a semi-enclosed area, which is only open to the deeper water in the south.The relatively shallow sills to the west and north create a physical barrier to the movement of deep-water fish species.The markedly different hydrographic differences between the Rockall Trough and the Norwegian Sea have resulted in almost no similarity in the deep-water fish fauna between these areas (Bergstad et al., 1999).On the other hand, there are many species in common between the Rockall Trough and the Porcupine Seabight (Gordon et al., 1996).The Porcupine Seabight also has affinities with the northwest African slope, where there is a high diversity in some families (e.g.Alepocephalidae; Merrett and Domanski, 1985), and the presence of some species has been attributed to the presence of a Lusitanian (Mediterranean) influence in the area (e.g.Hoplostethus mediterraneus; Ratz, 1984).Mediterranean water is clearly identified in the Porcupine Seabight (Rice et al., 1991), but previous reports of traces of Mediterranean water in the Rockall Trough are now being questioned (New and Smythe-Wright, 2001).In addition to these biogeographic factors, the proximity of the slope of the Porcupine Seabight to abyssal depths compared to that of the Rockall Trough could also increase the species number in this area, through the capture of juveniles of abyssal-living species (e.g.

Coryphaenoides leptolepis and Coryphaenoides carapinus).
There are some very clear differences in the families that dominate the biomass between the Mediterranean and the Atlantic.In the Mediterranean, a different family dominates each of the bathymetric zones: Gadidae on the upper slope, Moridae on the middle slope and Alepocephalidae on the lower slope.By contrast, in the Atlantic the family Macrouridae was dominant in each bathymetric zone.The different food availability and the partitioning of the main trophic resources within these ecosystems (among mostly fish in the Atlantic and between decapods and fish in the Mediterranean) may explain some of these differences.The dominant Macrouridae in the Atlantic (Coryphaenoides rupestris, Nezumia aequalis and Coryphaenoides guentheri) feed on small macroplankton (Mauchline and Gordon, 1984).On the upper and middle slope of the Mediterranean, this trophic resource may also be used by benthopelagic decapods, which in turn could provide food for the dominant fish and/or for their prey [Gadidae, Scyliorhinidae, Macrouridae and Moridae (Macpherson, 1978(Macpherson, , 1980;;Carrasón et al., 1997, Carrasón andMatallanas, 2002)].The Alepocephalidae are dominant on the Mediterranean lower slope.They have a conservative feeding strategy (float and wait), typical of oligotrophic conditions (Haedrich, 1996), which seems to confer an advantage over the more active foragers (Macrouridae and Synaphobranchidae) that predominate on the Atlantic lower slope.

Size-related structure
Biomass spectra show the distribution of overall biomass within the assemblages and can indicate the relative importance of different fish sizes with respect to energy flow (Haedrich and Merrett, 1992).According to these authors, flat spectra correspond to areas where food supply is rather uniform over the entire year, whereas spiky spectra characterise areas where production is pulsed.The observed patterns in the Mediterranean and the Atlantic both show a predominantly modal biomass spectrum, suggesting that their most important energy source might be seasonal.
The biomass spectra show clear differences between Mediterranean and Atlantic assemblages where, respectively, small-middle and middle-large size classes predominate.Where the same species occurs in both the Mediterranean and the Atlantic, those in the Mediterranean tend to attain a smaller adult size.As a consequence, these fish will have smaller mouths and will therefore utilise a different component of the available food resource.Competition for this food resource by the invertebrate fauna, notably the decapod crustaceans, may be an important factor in determining the overall biomass spectra (Cartes et al., 2001).
The differences in biomass spectra between Mediterranean and Atlantic assemblages depend on the depth strata.Differences were greatest on the upper slope and these are most likely related to the fishing exploitation of this bathymetric range in both Mediterranean areas, where important bottom trawl fisheries targeting decapod crustaceans have been carried out since the 1960s (Bas et al., 1985).Fishing exploitation has been demonstrated as an important factor affecting distribution of biomass within benthic fish assemblages (Macpherson and Gordoa, 1996).On the unexploited lower slope, biomass spectra were similar in the Atlantic assemblages.However, some differences could be observed in the Mediterranean, where the distribution was skewed to small-sized classes compared to the Atlantic.In the Balearic basin, there were two evident modes at small and middle-size classes, while in the Algerian basin there was a main mode at the middle-size class.Differences between the deep-sea fauna of these two western Mediterranean basins have been reported previously, not only in fish fauna (Moranta et al., 1998), but also in bathyal decapod crustaceans (Maynou and Cartes, 2000).These differences have been attributed to the fact that trophic webs in the Algerian basin depend more directly on food of planktonic origin, while in the Balearic basin, where there are important submarine canyons, benthic prey enhances food availability and allows a broader trophic spectrum.Thus, the greater percentage of biomass on the lower slope of the Algerian basin corresponds to the middle-sized species Alepocephalus rostratus (almost 75%), whose diet is based on gelatinous macroplankton (Carrasón and Matallanas, 1998), while in the Balearic basin its importance decreases (representing 36%) and other small-sized species such as Bathypterois mediterraneus and Lepidion lepidion, which prey upon benthopelagic crustaceans (Carrasón and Matallanas, 2001) and benthic decapods (Carrasón et al., 1997) respectively, also represent an important fraction of the biomass (23 and 21% respectively).The greatest similarities between Mediterranean and Atlantic deep-sea fish assemblages were found on the middle slope.This might be due to the similar trophic webs present in this bathymetric range, in which a great proportion of the biomass (> 60%) corresponds to the macronekton foragers and/or epibenthos predators (Macrouridae and Moridae) and to the macroplanktonivores (Alepocephalidae).
The differences in the biomass structure between Mediterranean and Atlantic assemblages were also evident at a species level, when both mean and maximum lengths of species caught in all surveyed areas were compared.The small size of Mediterranean deep-sea fishes has already been discussed by Tortonese (1960) and more recently by Stefanescu et al. (1992).According to these authors, it is evident that this phenomenon is very complicated and is undoubtedly the result of a combination of factors of disparate origin (e.g.latitudinal gradients on fish size, ecological factors affecting body size).More limited resources and a high temperature in the Mediterranean (~ 13ºC compared to 10 to 4ºC in the eastern Atlantic areas) could also be assumed to play an important part.
In summary, the differences found in demersal deep-sea fish assemblages between the Mediterranean and the Atlantic reinforces not only the general view of the Mediterranean deep-sea as an oligotrophic ecosystem (e.g.Bouchet and Taviani, 1992), but also the view that there are distinct differences in the size-related structure of the fish populations.The primary cause of the differences in size structure is probably a result of adaptations at both the species and ecosystem level to different trophic relationships between the two areas.The historical evolution of the Mediterranean basin, its geomorphology and markedly different environmental conditions and fishing pattern must also contribute to the differences.
FIG. 1. -Map showing the four study areas: Balearic (BB) and Algerian (AB) basins in the western Mediterranean; and Rockall Trough (RT) and Porcupine Seabight (PS) in the eastern North Atlantic.The 500, 1000 and 2000 m isobaths are shown.
FIG. 5. -Percentage of total biomass by size-class and depth-strata in the four studied areas: Balearic (BB) and Algerian (AB) basins, in the western Mediterranean; and Rockall Trough (RT) and Porcupine Seabight (PS), in the eastern North Atlantic.

TABLE 2 .
-Total number of species (in bold) in each studied area and common species and percentage similarity (from Bray-Curtis similarity index) between the areas: Balearic basin (BB); Algerian basin (AB); Rockall Trough (RT); Porcupine Seabight (PS).

TABLE 3 .
-Percentage biomass and abundance (between brackets) composition for the most important families obtained by depth-strata in each area.
APPENDIX.-Fish species captured in the four studied areas (BB, Balearic basin; AB, Algerian basin; RT, Rockall Trough; PS, Porcupine Seabight), with their bathymetric range (m) and frequency of occurrence (%) within this depth interval.Cont.).-Fish species captured in the four studied areas (BB, Balearic basin; AB, Algerian basin; RT, Rockall Trough; PS, Porcupine Seabight), with their bathymetric range (m) and frequency of occurrence (%) within this depth interval.