Distribution and abundance of Citharus linguatula , Lepidorhombus boscii , and Solea vulgaris ( Osteichthyes : Pleuronectiformes ) in the Mediterranean Sea *

Pleuronectiformes constitutes a very distinctive taxon: adults of the species belonging to this group show a marked asymmetry of the body, a unique phenomenon among all Vertebrates. The body is highly compressed (“flatfish”), somewhat rounded on the eyed side and flat on the blind side. The specimens live and swim on the eyeless side. The upper side is pigmented and frequently highly mimetic, whereas the lower side is usually non-pigmented. Flatfish are typical marine benthic species, which live on soft, sandy or muddy bottoms. Most species are littoral or sublittoral, although deep-water species are also present. A total of 538 extant species of Pleuronectiformes are reported in the world, belonging to 117 genera and six families (Nelson, 1984). According SCI. MAR., 66 (Suppl. 2): 83-102 SCIENTIA MARINA 2002


INTRODUCTION
Pleuronectiformes constitutes a very distinctive taxon: adults of the species belonging to this group show a marked asymmetry of the body, a unique phenomenon among all Vertebrates.The body is highly compressed ("flatfish"), somewhat rounded on the eyed side and flat on the blind side.The spec-imens live and swim on the eyeless side.The upper side is pigmented and frequently highly mimetic, whereas the lower side is usually non-pigmented.Flatfish are typical marine benthic species, which live on soft, sandy or muddy bottoms.Most species are littoral or sublittoral, although deep-water species are also present.
A total of 538 extant species of Pleuronectiformes are reported in the world, belonging to 117 genera and six families (Nelson, 1984).According  SURVEY (1994SURVEY ( -1999)).P. ABELLÓ, J.A. BERTRAND, L. GIL DE SOLA, C. PAPACONSTANTINOU, G. RELINI and A. SOUPLET (eds.)Distribution and abundance of Citharus linguatula, Lepidorhombus boscii, and Solea vulgaris (Osteichthyes: Pleuronectiformes) in the Mediterranean Sea* to Fischer et al. (1987), 34 species belonging to six Families (Bothidae, Citharidae, Cynoglossidae, Pleuronectidae, Scophtalmidae and Soleidae) are reported for the Mediterranean basin.Many Pleuronectiformes species are valued as food sources and are intensely exploited in the world.Fishing activity involving flatfish is mainly carried out with towed gears (i.e.otter and beam trawls), although passive gears (i.e.set nets) are also commonly employed to catch these species.In the Mediterranean basin, flatfish provide an important contribution to commercial landings.Common sole, Solea vulgaris Quensel 1806 (common synonym Solea solea Linnaeus 1758), is undoubtedly the most important species of this group in the area, especially concerning its commercial value.According to FAO statistics (Stamatopoulos, 1993), annual landings of S. vulgaris in the Mediterranean increased from about 4,500 tons in 1972 to about 10,000 tons in 1992.Mediterranean trawl landings include Spotted Flounder Citharus linguatula (Linnaeus, 1758) and Four-spotted Megrim Lepidorhombus boscii (Risso, 1810), which are regularly present at domestic markets and represent important by-catches (Relini et al., 1999).On account of their ecological, morphological and behavioural peculiarities, Pleuronectiformes are currently the object of many studies.However, to date, there is still a lack of basic information for correct management of these resources, especially in the Mediterranean area.The "MEDITS" research project has made it possible for the first time to collect a large dataset in a wide area of the Mediterranean Sea by means of experimental trawl surveys (Bertrand et al., 2000(Bertrand et al., , 2002)), thereby improving knowledge on demersal species on a large geographical scale.

MATERIAL AND METHODS
The "MEDITS" project (Bertrand et al., 2000(Bertrand et al., , 2002) ) was carried out in a wide area of the Mediterranean Sea, concerning mainly European waters, from the Alborán Sea to the South Aegean Sea.In the period 1994-1999, six late spring -early summer experimental trawl surveys were carried out, performing a total of 6,336 hauls within the depth range 10-800 m.An experimental trawl net (GOC 73) with 10 mm cod-end mesh size (knot to knot) was employed for the sampling.
The study area was subdivided into 40 geographical sectors and five depth strata: A: 10-50 m, B: 50-100 m, C: 100-200 m, D: 200-500 m, E: 500-800 m.Selection of sampling stations was based on a depth-stratified sampling scheme, taking into account the surface area of each stratum.A detailed map of the sampled area and further details of the sampling design, methodology and sampling gear are described in Bertrand et al. (2000Bertrand et al. ( , 2002)).
For each haul, the specimens caught of C. linguatula, L. boscii and S. vulgaris were counted and total weight of each species was recorded.Total length (TL, to the next lower 0.5 cm) was then recorded for each individual caught.Raw catch data (biomass and number of specimens collected) were converted to catch rates with a specifically developed software (Souplet, 1996) in order to compute abundance (number of specimens/km 2 ) and biomass indices (kg/km 2 ).Catch rates of the three species were analysed as mean value per year of sampling, geographic sector and bathymetric stratum.
Variation in biomass indices was investigated on a spatio-temporal basis.For this analysis, in order to obtain readable and effective information, the 40 geographical sectors were pooled in five geographic macro-areas, according to the "old" GFCM (General Fisheries Council for the Mediterranean) indications (Stamatopoulos, 1993;Relini et al., 1999): Morocco-Spain-France (ex GFCM 1.1 and 1.2), Sardinian-Tyrrhenian Sea (ex GFCM 1.3), Adriatic Sea (ex GFCM 2.1), Ionian Sea (ex GFCM 2.2), Aegean Sea (ex GFCM 3.1).Data concerning the two Moroccan sectors were not included in this analysis, since sampling in these areas only started in 1999.The variation of the biomass indices with macroarea, depth stratum and year, was studied using Generalised Linear Models (GLMs) (McCullagh and Nelder, 1989;Chambers and Hastie, 1992).The analyses were performed by applying the routines contained in the S-Plus programming environment (Becker et al., 1988).Following the approach of Stéfansson (1996) a gamma distribution was used in the analysis since the frequency distribution of the biomass indices was skewed and the variance proportional to nearly the square of the mean.The following generalised linear model was used: where: Biomass Index ijk : expected value of g/km 2 at the macro-area i, in the depth stratum j and in the year k; µ: overall mean; A i : effect of macro-area i; D j : effect of depth stratum j Y k: : effect of year k; interaction: any possible combination of interaction between two effects ε: error term assumed to be distributed normally.
Standardisation was performed on biomass indices expressed in g/km 2 .The constant 1 was added to the abundance values to account for zero values.Analysis of deviance to evaluate the significance of the main factors and interactions in the model was performed.All covariates are considered as fixed factors.
Subsequently, the demographic structure of the three species was studied considering the six years of sampling jointly, according to the five above mentioned macro-areas.

Citharus linguatula
The Spotted Flounder was frequently caught in the macro-areas 1.1-1.2 and 3.1 (with a percentage occurrence of 34% and 33%, respectively).In the other macro-areas, this species was collected in a percentage ranging from 10% to 18% of total hauls.It is worth noting that C. linguatula was never caught in some geographic sectors, such as north and south-east Sardinia, north Ionian Sea and southwest Adriatic Sea (Tables 1 and 2).This species was mainly found in the first three depth strata, its presence being occasionally noted down to 200 m.For this reason further analyses were confined to the first three depth strata.
Results from the analysis of deviance for the GLM model indicated that the differences in biomass indices observed among macro-areas as well as among depth strata and years were significant (Table 3).In addition, the variations explained by interactions between year and macro-area and between stratum and macro-area were also significant.The model accounted for 91% of the deviance, with most of the variation being due to differences between macro-areas.
Highest biomass indices were estimated in the Aegean Sea, followed by those of the Spain-France macro-area (Fig. 1; Table 4).The high values of this last macro-area were mostly due to the two sectors in the Gulf of Lions (Table 1).High biomass values, greater than 10 kg/km 2 , were also observed along the Moroccan and Sardinian coasts in 1999 (Table 1).Biomass indices were found to be significantly (p<0.0001)higher in the depth range 50-100 m (Table 3; Fig. 1), despite the fact that in some sectors the highest values were obtained in the other two strata.This finding supported the significance (p<0.001) of the interaction between stratum and macro-area.The differences (p<0.05)among years in all macro-areas studied were due to the notably lower biomass value obtained in the first year of study (Fig. 1 and Table 4).In general, the abundance indices expressed in number of individuals per km 2 provided a similar trend to that shown by the corresponding weight index (Table 2).
The size range of individuals caught during the six years of sampling was 4-34 cm TL, although most of the catch was comprised between 6 and 20 cm TL.Length frequency distributions obtained in each macro-area (Fig. 2) are characterised by two modes, the first corresponding to 7-9 cm TL and the second to 12-14 cm TL.
The catches obtained in the Sardinian-Tyrrhenian Sea macro-area showed the highest presence, albeit only occasional, of larger individuals (greater than 26 cm TL).Specimens caught in the Adriatic Sea were characterised by the smallest size range (from 5 to 24 cm TL).
Considering the entire "MEDITS" area, the number of specimens included in the first mode (lower  and year (1994-1999).Not sampled strata are indicated by '*'.(1994)(1995)(1996)(1997)(1998)(1999).Not sampled strata are indicated by '*'.Values higher than 500 individuals/km 2 are presented in bold.than 10 cm TL) was about 25% of the total catch.In the different macro-areas this value ranged from 18.7% in the Sardinian-Tyrrhenian Sea to 27.3% in the Morocco-Spain-France area.

Lepidorhombus boscii
The Four-spotted Megrim was caught in all the macro-areas, more frequently in the western and eastern regions (percentage of occurrence: 31, 34 and 36% in Spain-France, Sardinian-Tyrrhenian and Aegean macro-areas, respectively) as compared to the Central regions (13% in the Adriatic macro-area and 19% in the Ionian macro-area).The species showed a wide depth distribution, being caught in all the depth strata.However, occurrence between 10 to 50 m was virtually restricted to the northern Aegean Sea (Tables 5 and 6), and this stratum was therefore not considered for GLM analysis.
The variation in biomass indices observed with varying macro-areas and depth was statistically significant (Table 7), while variation explained by the different years was non-significant (p>0.6).Moreover, differences in biomass indices were influenced by the interaction between stratum and macro-area.The model reduced null deviance from 318.8 to 45.9, which is similar to obtaining an r 2 of 0.86 in a normal regression.Results for the macroareas showed that biomass indices were higher in Spain-France and Aegean GFCM regions, while in Adriatic and Ionian macro-areas significantly lower values were obtained (Fig. 3; Table 8).Also, the high values obtained in the Spain-France macro-area were due to those of the Gulf of Lions (Table 5).The depth factor accounted for most of the variation explained (37.6%).Biomass indices of L. boscii were significantly higher in strata C and D, for a depth from 100 to 500 m (Fig. 3; Tables 7 and 8).
Fairly similar results were obtained for indices expressed in number of individuals (Table 6).In this case, highest values were also obtained in the Gulf of Lions, but high abundance indices were also observed along the Sardinian coasts.This could be due to differences in the size composition of the catch in the macro-areas studied.(1994)(1995)(1996)(1997)(1998)(1999).Not sampled strata are indicated by '*'.Values higher than 10 kg/km 2 are presented in bold.Catches of L. boscii were characterised by a very wide size range, from 2 to 48 cm TL, although presence of specimens smaller than 5 cm TL and larger than 36 cm TL was only occasional.Some differences were found in the population demography at macroarea level (Fig. 4).Catches obtained in the Aegean Sea showed the widest size range (from 2 to 48 cm TL).However, it was difficult to single out a predominant size from the histogram, as specimens were equally shared among several size classes.In the other four macro-areas the size range proved to be more restricted, especially in the Adriatic Sea (from 7 to 34 cm TL).The histograms showed a fairly similar shape, with the first modal class ranging from 8 to 11 cm TL.The number of specimens smaller than 12 cm TL was higher in the western areas (about 35.5% of the total catch) than in the Adriatic Sea and the eastern macro-areas (from 13.6 to 26.0%).

Solea vulgaris
The Common Sole was found in all macro-areas, but with a much lower frequency of occurrence with respect to the other two species, always lower than 12%.S. vulgaris was never caught in south-east Corsica, the north-west Ionian Sea and many sectors of the south-west Adriatic Sea (Tables 9 and 10).The distribution range of this species was mostly restricted to the first two depth strata.Only in a few cases notable abundance was also observed between 100 and 200 m (West Morocco, West Gulf of Lions and East Ionian Sea).
Results of analysis of deviance from the GLM model, performed for the first three depth strata, are shown in Table 11.The macro-area and depth stratum proved to be significant (p<0.0001)factors in explaining the differences in biomass indices (Table 12), as well as the interaction between stratum and macro-area (p<0.01) and between year and stratum (p<0.05).This model accounted for 78.6% of total deviance, with most of the variation being due to differences between the three depth strata (32.9%).Biomass indices were significantly higher in strata 10-50 m and 50-100 m (Table 12).Results for the macro-areas showed that differences were particularly evident for the Adriatic macro-area, although   (1994)(1995)(1996)(1997)(1998)(1999).Not sampled strata are indicated by '*'.Values higher than 10 kg/km 2 are presented in bold.and year (1994-1999).Not sampled strata are indicated by '*'.Values higher than 10 kg/km 2 are presented in bold.(1994)(1995)(1996)(1997)(1998)(1999).Not sampled strata are indicated by '*'.Values higher than 100 individuals/km 2 are presented in bold.(1994)(1995)(1996)(1997)(1998)(1999).Not sampled strata are indicated by '*'.Values higher than 100 individuals/km 2 are presented in bold.the lowest observed biomass index was found in the Ionian macro-area (Fig. 5; Table 12).Sizes of the specimens collected ranged between 6 and 49 cm TL.Given the low number of specimens in the catches, difficulty was encountered in examining the demographic characteristics of this species.A similar size frequency distribution characterised by two distinct groups was observed in the macro-areas of the Western Mediterranean: the histograms were characterised by specimens smaller than 14 cm TL and by larger individuals with modal class varying from 26 to 32 cm TL, representing the majority of the catches (Fig. 6).In the other three macro-areas the catches were mostly characterised by specimens greater than 15 cm TL, the group of small individuals being absent.

DISCUSSION
The species forming the object of this study, Citharus linguatula, Lepidorhombus boscii and Solea vulgaris, constitute the most important fishing resources among Mediterranean flatfish.The "MEDITS" project has represented an important tool in order to provide large scale data on their distribution and abundance patterns and demographic structure, which constitutes useful information for assessment purposes.
The three species showed a wide geographical distribution, since they were collected in all the macro-areas investigated, but with some differences in occurrence patterns among the 40 geographic sectors studied.Thus, while L. boscii was found practically everywhere, C. linguatula and S. vulgaris showed some discontinuities, being absent in some areas of the Corsican, Sardinian, Ionian and Adriatic Seas.The scarcity of C. linguatula and S. vulgaris in some sectors could be due to the reduced extension of the continental shelf and/or of the soft bottoms.
The biomass of the species studied showed highest values in some regions such as the Gulf of Lions and the Greek seas.S. vulgaris was also abundant along the Moroccan and Sardinian coasts.This distribution pattern could be due to differences both in environmental features (shelf/slope ratio, availability of resources, presence of nursery areas) and in the exploitation pattern due to the local fishing fleets.
Particularly for L. boscii and C. linguatula, an important fraction of the catch was represented by juveniles.In these species, the first mode of the size distributions of the catch can be identified as belonging to specimens in the first year of life, according to the estimations of Planas and Vives (1956), Vassilopoulou and Papaconstantinou (1994) and García-Rodríguez and Esteban (2000) for C. linguatula, and of Bello and Rizzi (1987), Mannini et al. (1990) and Ungaro and Martino (1998) for L. boscii.In addition, for the latter species, most of the population sampled was constituted by immature specimens.Assuming that the female's first maturity size is 23.6 cm TL (Ungaro and Martino, 1998), the percentage of catch under this size was greater than 80%.
With regard to S. vulgaris, most of the catch was constituted by adults.About 60% of the specimens collected were larger than the size at first maturity (25 cm TL, Fischer et al., 1987, corresponding to two-three years of life, Froglia and Giannetti, 1986).Specimens belonging to the first size group, when present in the catch, were juveniles, assuming that size at one year is 15-20 cm TL (Ramos, 1982b;Piccinetti and Giovanardi, 1984;Froglia and Giannetti, 1985;Vianet and Quignard, 1986;Pagotto and Piccinetti, 1988;Paci et al., 1989).The low catches of juveniles obtained in some areas could be attributable to the spatio-temporal pattern of the recruitment of these species.Froglia (1993) reported that the juveniles of S. vulgaris in the Adriatic Sea are concentrated in lagoons and in coastal areas of the northern part of this basin from spring to late summer.This could probably be the main reason for the absence of recruits observed with the "MEDITS" surveys in the Adriatic Sea.
However, the high percentage of small size specimens in the catch is not necessarily an index of overexploitation, because the picture obtained through this study only refers to one season of the year, which may correspond to the recruitment period.For example, along the coast of Tuscany, Abella et al. (1997) reported the smallest specimens (mean TL of 8 cm) of C. linguatula in May, a period coinciding with that of "MEDITS" surveys.
For these species, very few studies focusing on the exploitation state are available at present.In the Gulf of Lions, the observation on commercial landings of S. vulgaris showed a declining trend from 1972 to 1996 (Farrugio and Marin, 1999), suggesting an overexploitation state of this species.A first approach based on analytical models was undertaken for L. boscii in the South Adriatic Sea (Ungaro and Marano, 1995), with results showing an exploitation level close to equilibrium.
However, it is important to take into account that some behavioural characteristics of flatfish, such as circadian activity rhythms (Walsh and Hickey, 1993;Burrows, 2001), could naturally contribute to protecting populations against high fishing intensity, reducing the catchability coefficient.The results of this study highlighted the need to increase the amount of information collected.Such information should concern not only biological and ecological aspects, but also data on landings, catch and effort.At the moment, given the lack of information on the exploitation status of these species, a precautionary approach is recommended in the management of these resources.
RESOURCES: THE MEDITS INTERNATIONAL TRAWL FIG. 1. -Results of the main effects of the model fitted to Citharus linguatula biomass indices, incorporating logarithmic link and gamma variance functions.Each plot represents the contribution of the corresponding variable to the fitted linear predictor: macro-area (above), depth stratum (centre) and year (below).The fitted values are adjusted to average zero; broken vertical bars indicate standard errors.The width of the solid bars at the base of the plots is proportional to the number of observations at each level of the factors.

FIG. 3
FIG. 3. -Results of main effects of the model fitted to Lepidorhombus boscii biomass indices, incorporating logarithmic link and gamma variance functions.Each plot represents the contribution of the corresponding variable to the fitted linear predictor: macro-area (above), depth stratum (centre) and year (below).The fitted values are adjusted to average zero; broken vertical bars indicate standard errors.The width of the solid bars at the base of the plots is proportional to the number of observations at each level of the factors.

FIG. 5
FIG. 5. -Results of main effects of the model fitted to Solea vulgaris biomass indices, incorporating logarithmic link and gamma variance functions.Each plot represents the contribution of the corresponding variable to the fitted linear predictor: macro-area (above), depth stratum (centre) and year (below).The fitted values are adjusted to average zero; broken vertical bars indicate standard errors.The width of the solid bars at the base of the plots is proportional to the number of observations at each level of the factors.

TABLE 1 .
-Citharus linguatula: Mean biomass (kg/km 2 ) estimated from the MEDITS trawl surveys by depth stratum, geographical sector Values higher than 10 kg/km 2 are presented in bold.

TABLE 2
. -Citharus linguatula: Mean abundance (in number of individuals /km 2 ) estimated from the MEDITS trawl surveys by depth stratum, geographical sector and year

TABLE 4
. -Citharus linguatula: mean biomass indices, expressed in kg/km 2 (with standard error, in brackets), estimated from the MEDITS trawl surveys by macro-area, depth stratum and year.

TABLE 5 .
-Lepidorhombus boscii: Mean biomass (kg/km 2 ) estimated from the MEDITS trawl surveys by depth stratum, geographical sector and year

TABLE 8 .
-Lepidorhombus boscii: mean biomass indices, expressed in kg/km 2 (with standard error, in brackets), estimated from the MEDITS trawl surveys by macro-area, depth stratum and year.

TABLE 9 .
-Solea vulgaris: Mean biomass (kg/km 2 ) estimated from the MEDITS trawl surveys by depth stratum, geographical sector and year

TABLE 10 .
-Solea vulgaris: Mean abundance (in number of individuals/km 2 ) estimated from the MEDITS trawl surveys by depth stratum, geographical sector and year

TABLE 10 (
Cont.).-Solea vulgaris: Mean abundance (in number of individuals/km 2 ) estimated from the MEDITS trawl surveys by depth stratum, geographical sector and year

TABLE 11 .
-Solea vulgaris: analysis of deviance table for generalised linear models fitted to MEDITS biomass indices (g/km 2 ) obtained in the years 1994-1999.

TABLE 12
. -Solea vulgaris: mean biomass indices, expressed in kg/km 2 (with standard error, in brackets), estimated from the MEDITS trawl surveys by macro-area, depth stratum and year.