Distribution of the Mediterranean hake populations ( Merluccius merluccius smiridus Rafinesque , 1810 ) ( Osteichthyes : Gadiformes ) based on six years monitoring by trawl-surveys : some implications for management *

1 Dipartimento per lo studio del Territorio e delle sue Risorse (Dip.Te. Ris), Università di Genova Laboratorio di Biologia marina ed Ecologia animale, Via Balbi 5, 16126 Genova, Italy. E-mail: largepel@unige.it 2 National Centre for Marine Research (NCMR), Institute of Marine Biological Resources, Aghios Kosmas, 16604 Hellinikon, Athens, Greece. 3 Institut Za Oceanografiju i Ribarstvo, Setaliste Ivana Mestrovica 63, P.O. Box 500, 21000 Split, Croatia. 4 IFREMER, Station de Sète, Direction des Ressources Vivantes, Lab. Dinamique des Systemes Productifs, Rue Jean Vilar 1, BP 171, 34203 Sète Cedex, France. 5 Instituto Español de Oceanografía, Centro Oceanografico de Malaga, P.O. Box 285, 29640 Fuengirola, Spain. 6 Laboratorio Biologia Marina e Pesca, Università di Bologna, Viale Adriatico, 1/N, 61032 Fano (PS), Italy.


INTRODUCTION
Fish of the genus Merluccius constitute very important demersal resources both worldwide and in the Mediterranean.In the last thirty years the world catches have oscillated between 1 and 2 million tonnes, while Mediterranean catches have doubled to more than 44,500 tonnes (Fiorentini et al., 1997).The contribution of the three sectors, western, central and eastern Mediterranean (sensu FAO), as well as the catches since 1972 are shown in Figure 1.In recent years the central Mediterranean, which is formed by the Sicilian Channel and the Ionian and Adriatic seas, has provided the most relevant quantities.
The various species of hake are a difficult subject for fishery biology (Jones, 1974) owing to their biological characteristics.The morpho-physiological model of Merluccius is long-standing and successful, probably due to its flexibility with regard to environmental constraints.Fossil hake have been found dating back to the Eocene in the Pacific and the Oligocene in Europe (Alheit and Pitcher, 1995).Among the recognised reasons for their success some significant biological and ecological characteristics can be listed, such as: the broad extension of the vertical distribution which covers both shelf and slope, the prolonged period of reproduction (thus eggs and larvae can experience very different sets of environmental conditions) and the possibility of cannibalism (if density is too high, part of the population dies, thus giving support to its conspecifics) (Payne and Punt, 1995).
At present, of twelve known species of hake (Inada, 1981), only one is found along the eastern North Atlantic and the Mediterranean: Merluccius merluccius (L.), the "European hake" with two subspecies, one in the Atlantic and one in the Mediterranean (Cohen et al., 1990).Therefore, the target of the MEDITS project was Merluccius merluccius smiridus, the Mediterranean subspecies, possibly with a plurality of stocks, given the variety of the observed biological patterns regarding reproduction and growth.It should be remembered that, on a simple meristic basis (i.e.vertebral number), differences were detected between Atlantic and Mediterranean (Cadenat, 1950) as well as among western, central and eastern Mediterranean hakes (Maurin, 1965).Genetic studies confirmed the former distinction (Pla et al., 1991;Roldán et al., 1998) and first results in the central Mediterranean are confirming the latter (Lo Brutto et al., 1998).
In the Mediterranean subspecies, the characteristics which form the basis of the success of hake have been repeatedly recorded.In particular, regarding the long reproductive period, several different cases have been pointed out, which could represent a first basis for the identification of stocks: in the Adriatic, Zupanovic (1968) showed two peaks of recruitment, in spring and in autumn, probably linked to winter and summer spawning peaks; the same recruitment pattern was observed in the Ligurian Sea (Orsi Relini et al., 1986); in the adjacent Tyrrhenian Sea two spawning peaks were found, one in February-March and another in September (Biagi et al., 1995).Bouhlal (1973) observed three spawning peaks off Tunisia in summer, winter and spring; off the Balearic Islands, the spawning period is from November to May (Bruno et al., 1979); along the Catalan coast and Gulf of Lions, the spawning season lasts all year long with a peak in autumn (Recasens et al., 1998).In the Eastern Mediterranean different authors have presented a variety of results (Papaconstantinou and Stergiou, 1995): spawning periods are long, frequently lasting throughout the year, with autumnwinter and spring peaks.In the Greek seas, recruits appear in trawl catches in summer and through to early autumn, in well-defined areas such as Saronikos, Patroikos, Siggitikos and also in certain locations in the north and south Aegean Sea and the Ionian Sea (Papaconstantinou, 1999).Also, reproductive sizes and growth rates vary in different Mediterranean sectors.The biological parameters assumed in numerous stock assessments can be found in Fiorentino (2000).
In relation to fishing exploitation, such assessments generally agree with the opinion that hake populations are fully exploited or over-exploited (Fiorentino, 2000).However, given that in some areas yields have apparently been stable for some time (see Bertrand et al., 2000a, for the Gulf of Lions), the "refuge theory" has been proposed (Caddy, 1993).According to this theory, the fact that the spawners are caught in small numbers ensures the rich recruitment which sustains the fishery.Therefore, an assessment framework which puts the emphasis on maintaining spawning stock biomass and on the stock-recruitment relationship, rather than on maximising yield per recruit, has been suggested as a more appropriate basis for scientific management (Caddy, 1993).Until now, the stock-recruitment relationships (Alegría Hernández and Jukic, 1992) have not been exhaustively studied.
The MEDITS trawl survey represented an important occasion to work on homogeneous data from a series of surveys which were accurately standardised and covered the western, central and eastern Mediterranean.This paper deals with some general results.Biomass and abundance indices, recorded from the Alborán Sea to the Aegean Sea and in a depth range of 800 m during six consecutive years, are illustrated as well as their general trends.Size structures of the fished stocks are analysed in two main strata (continental shelf and slope) and total mortality coefficients are derived.The distribution of very young individuals (6-9 cm total length) is studied in order to show the most important summer nursery areas.As in few cases in the Mediterranean literature, the explored area is large enough to give evidence to regional differentiations; those concerning recruitment patterns represent an interesting contribution to the problem of stock identification.

MATERIAL AND METHODS
Within the framework of the "MEDITS" project six annual bottom trawl surveys were performed from 1994 to 1999 in the Mediterranean Sea, covering 40 sub-areas belonging to 15 geographical sectors (Bertrand et al., 2000b(Bertrand et al., , 2002) ) (Fig. 2).At the beginning (1994), the project included four Euro-    (1994)(1995)(1996)(1997)(1998)(1999).Not sampled strata are indicated by *.Values higher than 100 kg km -2 are presented in bold.Hauls were performed during daylight hours and their duration was fixed to half an hour on the shelf trawling grounds and to one hour on the slope.Collected data included number, weight, gonad maturation stage and total length measurements for the target species (a total of 30, including M. merluccius).
The same gear and sampling protocol were used in all cases (for more details see Anonimous, 1998, Bertrand et al., 2000b, 2002).

Data analysis
From the collected data, abundance indices by sub-area, year and depth stratum were computed and expressed in terms of both number of individuals (N) and weight (kg) per square kilometre.
Generalized   cance level of 0.05, and using a stepwise backward method.
In addition, on the basis of the total length (cm) measurements taken from a representative number of individuals, the summary statistics together with the yearly percentage length-frequency distributions were estimated for each sub-area and major depth stratum (10-200, 200-800 m), corresponding to shelf and slope areas.Estimates were made using specifically developed software that takes into account the surface of each sub-area and depth stratum (Cochran, 1977;adapted by A. Souplet).
Mean lengths, standard deviation and coefficient of variation (CV) were derived from length-frequency distributions.Using LCCC (Length Converted Catch Curves, Gayanilo et al., 1994), the coefficient Z of instantaneous total mortality was calculated for the 15 studied geographical sectors: this was a rough estimate in which sex distinction was not made and the growth function was common for all areas: parameters L inf = 73 cm and K= 0.15 were estimated as average values taking into account Fiorentino's (2000) compilation in the context of the FAO-GFCM Stock Assessment Committee.
Length-frequency distributions were also used to point up recruitment events occurring shortly before the period during which the MEDITS surveys were carried out (May-July).The MEDITS recruitment patterns represent only part of the general recruitment processes in the Mediterranean hake populations, given the temporal limits of the surveys and the already mentioned variety of the reproductive seasons (see Introduction).Another limit of MED-ITS data for recruitment studies is due to the fact DISTRIBUTION AND ABUNDANCE OF MEDITERRANEAN HAKES 27 TABLE 3 (Cont.).-Merluccius merluccius: Mean abundance (ind.km -2 ) estimated from the MEDITS trawl surveys by depth stratum, geographical sector and year (1994)(1995)(1996)(1997)(1998)(1999).Not sampled strata are indicated by *.Values higher than 10000 ind.km -2 are presented in bold.that trawl surveys which are aimed at the youngest hake require a specific timetable of sampling in respect of the factor light (Orsi Relini et al., 1997).So the following quantitative aspects must be considered only provisionally, in the sense that they are probably underestimated.On the other hand, given that MEDITS procedures were applied in the same way over a very large area and during several years, these provisional results assume a great relevance when they show patterns of regional differences.
In each geographical sector the Gaussian components of fish whose modal length was in the range 6-9 cm were isolated using Bhattacharya's method.The extracted numbers of juveniles were transformed in mean numbers per square kilometres for comparison among sectors.The same groups were also figured by geographical sector, year and main stratum (shelf and slope), after a standardization based upon a factor proportional to the area of each sector.

Biomass and abundance
The biomass and abundance of hake presented, respectively, as kg per km 2 and number of individuals (N) per km 2 in each sub-area, caught in the six MEDITS surveys per stratum and per year, are shown in Tables 2 and 3 (tables are presented along a west to east geographical gradient).Catch variability was high.The highest values of biomass (Table 2) were recorded in Sardinia, with peaks of more than 540 kg per km 2 and in the eastern sector, especially in Argosaronikos, with peaks of more than 260 kg per km 2 .Significant quantities also appeared in the Tyrrhenian Sea, the Gulf of Lions, and in Croatian, Apulian and Ionian waters.Throughout the period studied, no figure higher than 100 kg/km 2 was registered in Spanish, Moroccan, Sicilian or northwestern and central Adriatic waters.Average biomass in the main geographical sectors showed increases and decreases (Fig. 3), but the research of statistically significant trends on the basis of the Pearson's correlation coefficient was generally negative; the only exception was found in sector 132 (Ligurian and north central Tyrrhenian Sea) (Table 4) where biomass slowly increased.
Maximum values of abundance in N per km 2 (> 20,000) were recorded in Ligurian-Tyrrhenian waters (Table 3).In general all values higher than 28 L. ORSI RELINI et al.  4 gives the result of the research of trends by the Pearson's correlation coefficient: only one, in bold, is significant (r = 0.89; p = 0.05).Pooling the catches of the 15 main sectors to obtain absolute numbers, a total of 521,687 individuals were collected during the samplings, of which 302,848 on the continental shelf and 218,839 on the slope (Table 5).It is remarkable that 44.3% of these fishes were found in sector 132 (Ligurian and north central Tyrrhenian Sea), which represents, respectively, 8% and 10% of the overall shelf and slope of the MEDITS sea beds (Table 1).Second in terms of abundance (15.2% of the total catches) was area 133 (Sardinia), adjacent to the former, with, respectively, 6% and 5% of the overall shelf and slope MEDITS trawlable surface (Table 1).On the other hand, samples from one or both levels, shelf and slope, were very scanty in some areas.

GLM analysis of biomass, abundances, European Union countries, depth and years of sampling
Results of the GLM analysis are given in Tables 6  and 7.The significance level was set to 0.05.The model was considered to be strong, explaining more than 82% (in the case of biomass) and 90% (in the case of abundance) of the observed variation in the data (adjusted R squared statistic = 0.824 and 0.920, respectively).From the variables examined and their first order interactions, the main effect of year as well as its interactive effect with depth zone were not found to be statistically significant.Depth zone, country, country*depth zone interaction and country*year interaction were all statistically highly significant.active effect of country*year on hake biomass.Results demonstrate that the range of values of the estimated marginal means of the country*year effect in Italy (0.476) and Greece (0.544) were less dispersed than those in Spain (0.657) and France (0.610).This supports the idea that smaller interactive effects might be observed within, rather than between, these two groups of countries (i.e., IT-GR and ES-FR).
The interactive effect of country*depth zone is presented in Figure 5. Results show that the range of values of the estimated marginal means of the latter effect in Italy (0.776), Greece (0.641), and France (0.623) were less dispersed than those in Spain (1.236).In all three countries (Spain, France and Italy) the 500-800 m depth zone was found to carry the lowest hake biomass; in Greek waters this was 30 L. ORSI RELINI et al.  observed in the 0-50 m depth zone.With the exception of its shallower waters (<50 m), Spain was found to have the lowest hake biomass compared to other countries.On the other hand, Greek waters with depths greater than 50 m carried the highest hake biomass.The 0-50 m depth zone had the least dispersed variation in the estimated marginal means of hake biomass.Figure 6 gives the estimated marginal mean of the interactive effect of country * year on hake abundance, indicating significant interactions between years.With the exception of Italy, all other countries showed significant interannual variability in total hake abundance.Figure 7 shows that the 500-800 m depth zone had the lowest level of hake abundance in all countries.Thus, eliminating the factor z800 (depth zone 500-800 m), the range of values of the estimated marginal means between the four countries presents two groups of countries.The first is formed by France (0.606), Greece (0.865) and Italy (0.94), and the second is Spain (1.392).On the other hand, the 100-200 m depth zone carried the highest numbers of hake specimens.In general, the 50-200 m depth zones in all countries were characterised by the highest levels of hake abundance.These depth zones (i.e., 50-100 and 100-200 m) were also characterised by the lowest variability in hake abundance between countries.

Size analysis
Length analysis showed that in all surveyed areas, with the exception of Corsica, Sardinia, the northern Adriatic and southern Aegean Seas, average lengths on the slope were higher than on the shelf (Table 5).In the western Mediterranean, with the exception of Corsica and Sardinia, average lengths were lower than in the Adriatic, Ionian and Aegean Seas.Definite trends in average length were not detected, except in area 111, Alborán (a=-0.6681;b=15.833;R 2 = 0.8789; Fig. 8).
Length-frequency distributions (Fig. 9) showed a general prevalence of small sizes, confirming the fact that the first age classes form the bulk of the fished stock.A considerable portion of larger sizes appeared in Corsica and Sardinia, and in the Ionian and Aegean Seas.On the other hand, a deficit of slope population seemed to characterize some segments of the Spanish coast.

Mortalities
The values of Z obtained for all MEDITS sectors are given in Table 8.They were, for all the investigated areas, fairly high.In some areas Z values exceeded 1, mostly in the case of western Mediterranean sectors: Alborán (1995;1999), Alicante (1997;;1998), Catalan (1994-1999)), Corsica (1994), Ligurian (1998;1999), Sardinia (1998), northern Adriatic (1995;1998), east Ionian (1995;;1997) andArgosaronikos (1995;1996;1998).However, some of these values were probably influenced by the massive presence of juveniles; the absolute maximum value of 2.48 in the Ligurian and northern-central Tyrrhenian Seas in 1998 was clearly linked to the large number of recruits found in the area (Table 9 and Figs. 10 and 11).In the case of the Morocco sector, where assessment of Z was made on only one set of annual data, the obtained value (Z=0.91) also describes significant fishing pressure in bottom trawl fisheries.

Recruitment patterns
In the MEDITS length frequency distributions, groups of fish identified by their small modal length were recorded from 6 cm total length onward.Findings of recruits varied considerably from year to year (Table 9) and only on the shelf of two sectors, Alicante and Ligurian-Tyrrhenian, were recruits recorded every year.However, this variability was offset by taking six years into account and in every sector did recruits appear at least one time.In total, 50 distinct groups of recruits were found on shelf fishing grounds and 29 on those of the slope (Figs. 10 and 11).The groups on the shelf were formed by Table 9. -Numbers of recruits (groups of fish identified by a modal length of 6-9 cm in length frequency distributions) derived from each geographical sector and year on shelf and slope fishing grounds.Estimated mean n/km 2 are also indicated.smaller fish and their absolute numbers were generally higher than on the slope (Table 9), indicating the fact that recruitment occurs in shelf waters and the fish, after some growth has taken place, move to the slope.Analysis of the geographical distribution of recruits on the shelf in the western Mediterranean (Table 9 and Fig. 10) showed numbers increasing from Alborán (note the scales of the individual graphs in Fig. 10) to a maximum in the northern sectors (Catalan, Gulf of Lions and Ligurian-Tyrrhenian Sea); thereafter they decreased in the southern Tyrrhenian Sea and northern Sicily.In the central Mediterranean, only the northeastern Adriatic presented high numbers; those in the eastern sectors were poor.Slope fishing grounds adjacent to the most crowded shelf areas presented important numbers of recruits.On the slope (Fig. 11) of area 132 (Ligurian-Tyrrhenian) in 1998, recruits of a modal length of 8 cm reached considerable densities, including the absolute maximum.

DISCUSSION
On a basis of a total of 4,800 positive hauls, the MEDITS surveys have made it possible, for the first time, to study hake in an area corresponding approximately to three quarters of both the latitudinal and the longitudinal range of the Mediterranean.The study of such rich material has developed knowledge of the distribution and abundance patterns, thus highlighting general management implications which are in line with the MEDITS purposes.The study also has shown aspects revealing the biological diversity of the Mediterranean hake populations.

GLM results
The bathymetric distribution of hake in the Mediterranean Sea displays a wide range stretching from 25 to 1000 m depth.The species is mainly abundant, however, at depths ranging from 100 to 400 m.Hake abundance increases with depth, depending on size and latitude, while average body size of hake increases, in general, as depth increases.
Different authors have studied the bathymetric distribution of hake in the Mediterranean Sea, such as Massutí and Daroca (1976) off the Spanish coast, Tsimenidis et al. (1978) in the Aegean Sea, Vas-silopoulou and Papaconstantinou (1987) in the Ionian, Zupanovic and Jardas (1986) in the Adriatic, Orsi- Relini et al. (1989) in the Ligurian Sea, Sartor et al. (1990) in the Tyrrhenian Sea, and Campillo et al. (1991) in the Gulf of Lions.Studies in the Ligurian Sea demonstrated that the nursery areas are situated from 50 to 250 m depth (with the main concentration located around 100 m), while the eggs and larvae appear in the pelagic environment at depths ranging from 50 to 150-200 m (Orsi- Relini et al. 1989).In the Gulf of Lions, distribution and abundance of age class 0, I and II have been studied using catch data from the trawler fleet (Campillo et al., 1991).Main concentrations of group 0 were found from 100 to 150 m depth, whereas groups I and II appeared to be distributed along the continental shelf from 30 to 150 m.Regarding the distribution of young hake (age 0) in the Greek Seas, it was found that they appear in trawl catches in summer through early autumn, in well defined areas such as the Saronikos, Patraikos, Siggitikos and also in certain locations of the northern and southern Aegean and the Ionian (Vassilopoulou and Papaconstantinou, 1987;Papaconstantinou and Stergiou, 1995;Papaconstantinou, 1999), in depths varying between 110 and 250 m.
The results of the present study provided a quantitative basis for the division of the bathymetric distribution of hake in the Mediterranean Sea, since the depth zones, as well as the Country * Depth Zone interaction was found to be statistically significant.From Figures 5 and 6 it is obvious that both the abundance and the biomass exhibit minimum values in the 500-800 m depth zone, which is in accordance with the findings of other studies, indicating that the species displays a rather limited abundance in the Mediterranean in waters deeper than 500 m (Papaconstantinou and Stergiou, 1995;Oliver and Massutí, 1995).Highest abundance values coincide with the 100-200 m depth zone in which most nursery grounds in the Mediterranean are located, according to the results of the various studies, as well as of the findings from the present MEDITS programme.In relation to the distribution of the biomass and abundance in other depth zones, the greatest difference is found between biomass values in Spain and in the rest of countries, while in relation to abundance, minor differences appear.In general, it could be suggested that higher biomass/abundance values are found in Greece and Italy and then France, while lower values are shown in Spain.
The main determining feature associated with the structure of hake abundance is depth since it reflects the changes from the continental shelf to the continental slope.However, other biotic and abiotic characteristics do play a role in the depth distribution of hake between the four areas/countries.The main factors which possibly contribute to such a geographical differentiation is the fishery status in each area/country, correlated directly to the applied management scheme and the extent and bottom type of the continental shelf and slope in these countries.Nevertheless, the most important quantitative boundary for all areas was located at around 200 m, which separated the hake distribution between continental shelf and upper slope.At this depth, a remarkable increase with respect to hake biomass has been noted in Spain, Greece and Italy, while concerning abundance, the highest values were found in Spain, Greece and France.On the other hand, hake found in deep waters tend to have a much broader length range than those inhabiting the shallow depths.
Taking into consideration the smaller interactive effects which have been observed for the country*year effect on biomass and country*depth zone on abundance between Italy and Greece, it is suggested that the status of the hake stocks in the two countries is more similar in relation to other Mediterranean countries.This similarity could be possibly connected with certain abiotic and biotic factors prevailing in the eastern Mediterranean, as well as to the fishery management systems of the two countries.In fact, the management systems of Greece and Italy are quite close, exhibiting considerable differences, as compared to those established in Spain and France (ACC, 1994).
Nevertheless, in general, management measures (i.e.closed seasons and areas, limited issue of licenses, minimum legal landing sizes and mesh size regulation) seem to be inadequate.In fact, despite the enforcement of such measures, these resources appear currently overfished (Papaconstantinou and Stergiou, 1995;Oliver and Massutí, 1995).This must be attributed mainly to the multi-species, multi-gear nature of the Mediterranean fisheries, which pose certain difficulties in designing and implementing uniform protective measures.Based on the results of the MEDITS programme since 1994, the derived debate suggested a network of marine protected areas, established on nursery and spawning grounds of hake in the Mediterranean Sea.This should ensure that both juvenile and adult hake are protected against exploitation, as long as those areas amount to a "significant" proportion of the total habitat.Management options in those areas would include the creation of "no-takes" zones, limited access zones, control of fishing gears, introduction and management of fishing permissions etc.

Biomass and size structures
Considering the distribution of biomass and the observed size structures, where there is a dominance of young fish, as well as the calculated (albeit rough) estimates of Z, a general condition of growth overfishing is apparent.Fishing grounds which presumably are correctly fished, or at least less overfished and thus show rich biomass values, appear limited to some subareas of the central and eastern Mediterranean.On account of their high biomass indices, some western Sardinian fishing grounds, where trawling has developed more recently than in other Italian seas, would seem to represent the potential yield of hake when not overfished.
However, when trends of biomass and average size have been analysed -albeit in the brief time range of six years -no negative trend has been found (with the exception of the Alborán Sea).Consequently, the refuge theory (Caddy, 1993) is supported and also can be extended to large scale patterns.In fact, the slope around Sardinia could host a consistent number of spawners and the derived larval flow, given that main currents move toward the north, could enrich the Liguro-Tyrrhenian sector with unparalleled densities of juveniles.But in the eastern Mediterranean, the rich biomass of Argosaronikos is apparently not related, as shown later, to significant concentrations of young individuals in adjacent areas.Different reproductive seasons must be taken into account.

Recruitment patterns
Recruitment patterns have been analyzed on a detailed basis of areas and subareas, taking into account the fact that quantitative aspects registered in the MEDITS surveys are only facets of a more complex reality.Nevertheless, the analysis of recruitment processes has pointed out important regional differences in the surveyed area.Since the study carried out by Maurin (1965), very few papers have considered Mediterranean hake in relation to large areas of their distribution.In their review of world hake, Alheit and Pitcher (1995) considered two units: a western Mediterranean hake (Oliver and Massutí, 1995) and an eastern Mediterranean hake (Papaconstantinou and Stergiou, 1995), probably on the basis of the differences in growth rates which were available at that time in the literature (Oliver et al., 1992;Tsimenidis et al., 1978).In this approach, the eastern Mediterranean hake was supposed to include the Ionian and the Adriatic Seas in its distribution area.MEDITS surveys, which take place in late spring-early summer, have exhaustively sampled recruits in the northwestern Mediterranean and in the northern Adriatic but not in the eastern Mediterranean and the southernmost explored areas of the central sector.This feature is in accordance with the different timing of recruitment already reported.In the Ligurian Sea, two recruitments per year, in spring and in autumn, were observed, with the former generally stronger than the latter.During spring, the 100-150 m depth interval hosted the most substantial densities of juveniles and during summer, the spring recruits began their descent along the slope (Orsi Relini et al., 1989 a and b).Also, recruitment in the Adriatic takes mainly place in spring, while in the eastern sector, at the time of the MED-ITS survey, recruitment was only at its start and the fish were probably still dispersed and too small to provide significant samples.
So, MEDITS surveys have (1) highlighted the main nurseries of the western hake population.Positioned above 42°latitude N, in the Gulf of Lions and the Ligurian-Tyrrhenian Seas, these nursery areas correspond exactly to the core of the population subunit identified on the basis of the highest vertebral number in the Mediterranean (higher than 52, according to Maurin, 1965).
Moreover, MEDITS data suggest that (2) the population of the central Mediterranean is not homogeneous and the Adriatic is to be considered a northwestern "enclave".The Adriatic hake, with its peaks of recruitment in spring (recorded by MED-ITS) and autumn and a vertebral number of 51.99 (Piccinetti and Piccinetti Manfrin, 1971) is very close to the northwestern hake.Both recruitment patterns (hence reproduction) and vertebral numbers could be linked to latitudinal gradients, i.e. to physical factors such as temperature and light.In the case of hake, which generally lives under the thermocline, light is probably the most important factor.If these differences are phenotypic or genotypic is a matter for future research, but a recent study of four polymorphic loci in hake sampled in the Adriatic, Tyrrhenian, southern Sardinia and Sicily Channel (Lo Brutto et al., 1998) is apparently in line with the second hypothesis: in fact no differences were found, except in the case of one sample collected near Malta (which was interpreted as a casual effect, but could be referred to the Ionian hake, i.e. a transitional form to the eastern hake (vertebral number 51.87) according to the racial groups proposed by Maurin (1965).If such genetic results are confirmed, the Adriatic hake, and more in general the Italian hake populations, could be older than the present distribution of Italian seas i.e. than the last steps of the orogenesis of the Apennine Chain which isolated the Adriatic Sea.
In any case, the observed differences in the biology of hake populations have important management implications, especially if measures specifically addressed to juveniles are to be envisaged.
FIG. 2. -MEDITS geographical sectors (111 to 225) with indications of partner countries (in parenthesis) and their shelf and slope surfaces.

FIG. 3
FIG. 3. -Average biomass recorded in the total depth range (10-800 m) in different sectors.Table4gives the result of the research of trends by the Pearson's correlation coefficient: only one, in bold, is significant (r = 0.89; p = 0.05).
Figures 4-7 present the profiles of the estimated marginal means of the generalized linear model.Figure 4 shows the estimated marginal mean of the inter-DISTRIBUTION AND ABUNDANCE OF MEDITERRANEAN HAKES 29 FIG. 4. -Results of the GLM analysis.Effect: Country * Year.Estimated marginal means of log 10 (kg/km 2 +1) (ES=Spain; FR=France; GR=Greece; IT=Italy) (Lines are used to join scatter plots of each country only for easier interpretation).

FIG. 10
FIG. 10. -Gaussian components by year of the groups of recruits identified in length frequency distributions recorded on shelf fishing grounds.N are relative numbers, obtained by a factor proportional to the area of each sector.

TABLE 2
. -Merluccius merluccius: Mean biomass (in kg km -2 ) estimated from the MEDITS trawl surveys by depth stratum, geographical sector and year
lagh and Nelder, 1989)was used to analyse trends of hake biomass and abundance (number of specimens) in relation to country, year and depth stratum area.Given the complexity of the data (in terms of strata and years), only the four countries forming the initial MEDITS team were studied: Spain, France, Italy and Greece.A GLM provides a way of estimating a function of the mean response (called the link function) as a linear function of the values of some set of predictors.The model used was: glm [Log 10 (hake biomass)] = β + Country + Year + Depth zone + Country*Year + Country * Depth zone + Year * Depth zone with β being an intercept term.In GLMs the standard linearity assumption is extended to include any underlying probability distribution from the exponential family (including the Poisson, gamma, normal and binomial distributions).structure left in the residuals appeared only as noise.Backward stepwise elimination was used to select a set of significant covariates.In each model, only significant parameters were included, at the signifi-26 L. ORSI RELINI et al.

TABLE 4 .
-Pearson's correlation coefficient between sampling year and hake biomass for the different geographical sectors studied.The only significant trend is presented in bold.

TABLE 5 .
-Number and size characteristics of hake sampled on the continental shelf and slope of each sector(1994)(1995)(1996)(1997)(1998)(1999), with main size characteristics (minimum and maximum size recorded, average length with standard deviation and variability coefficient).