Distribution and abundance of Octopus vulgaris Cuvier , 1797 ( Cephalopoda : Octopoda ) in the Mediterranean Sea *

The common octopus, Octopus vulgaris Cuvier, 1797, is a benthic cephalopod distributed on rocky, sandy and muddy bottoms from the coastline to the edge of the continental shelf (Mangold, 1983; Belcari and Sartor, 1999). This species has been long considered of cosmopolitan occurrence in temperate and tropical seas (Roper et al., 1984), but a possible occurrence of cryptic species among O. vulgarislike octopods is also reported (Mangold, 1998; Guerra et al., 1999). Thus, the distribution of O. vulgaris in a strict sense may be restricted to the Mediterranean Sea and eastern Atlantic Ocean (Mangold, 1998). The species is of great concern as an exploitable resource and therefore subject to intense fishing activity carried out mainly by various small-scale gears, such as traps, pots, fyke-nets and set-nets (Würtz and Repetto, 1983; Sánchez and Obarti, SCI. MAR., 66 (Suppl. 2): 157-166 SCIENTIA MARINA 2002

1993; Belluscio and Ardizzone, 1990), altough in the Mediterranean basin, octopus catches constitute an important fraction of the commercial landings of trawlers operating on the continental shelf (Tursi and D'Onghia, 1992;Belcari and Sartor, 1993;Quetglas et al., 1998).According to FAO statistics, the world annual landings of O. vulgaris were about 165,000 tons in 1996 (FAO, 1998).Despite this great commercial interest, studies defining distribution and abundance of stocks are still lacking in many areas.The species is often pooled together with Eledone cirrhosa and Eledone moschata in the commercial landings and in the Mediterranean fishery statistics (Sánchez and Martín, 1993;Belcari et al., 1998;Lefkaditou et al., 2000).This makes it impossible to assess individual catch levels as an index of stock biomass based on the assessment of landing data and statistics.To date there is still a lack of basic information for the correct management of this resource.
The present paper aims at contributing to the knowledge of the distribution, abundance and demographic structure of O. vulgaris collected by means of experimental trawl surveys carried out with a common methodology in a wide area of the Mediterranean basin.

Sampling
Six annual bottom trawl surveys that were mainly aimed at obtaining estimates of abundance indices for a series of demersal target species were carried out from late spring to mid summer in the Mediterranean Sea from 1994 to 1999 (Bertrand et al., 2000(Bertrand et al., , 2002)).The surveys covered 40 sub-areas belonging to 15 major areas (Table 1); Morocco was included in the project in 1999.A total of approximately 1,000 hauls was made during each survey in the depth range 10-800 m, by means of a standard trawl net GOC 73 having a cod-end mesh opening of 20 mm.Selection of sampling stations was based on a depth-stratified sampling scheme taking into account the surface area of each stratum; five depth zones were considered: 10-50, 50-100, 100-200, 200-500 and 500-800 m.Specimens were counted, weighed, measured (mantle length, ML, to the next lower 0.5 cm) sexed and assigned to a maturity stage by macroscopic analysis of the gonads.The same sampling protocol was used in all cases (see Bertrand et al., 2000Bertrand et al., , 2002)).

Data analysis
Catch data (number and weight) of specimens collected were analysed by means of specifically developed software (Souplet, 1996).In order to obtain estimates of abundance indices expressed in terms of both number of specimens and kg per km 2 , the surface of each sub-area and depth stratum was taken into account.The abundance estimations of the species were calculated as a mean value per year of sampling, geographic sector and bathymetric stratum.
Abundance and biomass indices were compared between strata and the interaction of year effect on the indices was evaluated, using two-way ANOVA and post hoc contrast analysis (Zar, 1984).In order to help normalize density indices (n/km 2 ) and biomass indices (kg/km 2 ) from heteroscedasticity, indices were log-transformed (log (x+1)) before ANOVA, thus eliminating the correlation between means and standard deviations, making the statistical analysis more robust (Zar, 1984).Pairwise comparisons, based on both indices, were made using the Scheffé test.The above estimates indicated that the abundance of O. vulgaris was negligible in waters deeper than 100 m.For this reason further analysis was confined to depth strata down to 100 m.
The mean density indices between areas were compared by one-way ANOVA in the depth strata where the species was more abundant, 10-50 m and 50-100 m.Using the Tukey (HSD) test, for unequal sample sizes, pairwise comparisons between areas 158 P. BELCARI et al.TABLE 1. -List of the major areas covered by the surveys.*Area 114 was not included in the analysis as only one-year data were available.A map of all the area sampled can be found in Bertrand et al., 2000Bertrand et al., , 2002)).year (1994-1999).Not sampled strata are indicated by '*'.The demographic structure of the species was firstly studied by computing the size frequency distributions for every major area (see Table 1).All the years were pooled, since the cruises were always performed at the same period of the year.Due to the narrow distribution and to the scarcity of catches, almost all the length frequency histograms considered here resulted from grouping together data coming from areas of the same country, when non-significant differences were found.Length frequency distributions in major areas were compared by means of the Kolmogorov-Smirnov test.Furthermore, samples that during the six years of study did not exceed 50 specimens were excluded (area 211).

Abundance indices
Octopus vulgaris was distributed along the whole western Mediterranean basin, even if in a discontinuous way; in the central basin the species was fairly sparse, appearing only as a few individuals per square km, or not at all, in the Adriatic and northern and north western Ionian sea, and slightly increasing again in the eastern Mediterranean (Tables 2 and 3).The species showed a narrow depth distribution, mostly restricted within 100 m; below 200 m, catches were scarce and at further depths, occasional.
The species abundance and bathymetric distribution were evaluated by comparing the variation of density and biomass indices between the five depth zones in the years 1994-1999, using two-way ANOVA design, 5 (depth zones) x 6 (years), without taking into account the effect of different sectors (Tables 4 and 5).According to these tables, there was no interaction effect between the variables, that is, the effect of depth on the abundance and biomass indices was not affected by the effect of year (p>0.05),while significant differences (p<0.001)among the depth zones were estimated.Figure 1 shows that the shallower the bathymetric zone the more abundant the species.Pairwise comparisons indicated that the mean abundance indices were different among all the depth zones (p<0.001),except those between the two deeper zones, 200-500 m and 500-800 m (p>0.05),where the lowest abundance indices were computed (Fig. 1).Mean density indices were compared among major areas in the depth strata 10-50 m and 50-100 m.ANOVA detected significant differences between areas (p<0.001) for both depth strata.The post hoc procedure was used to explain any pattern of means and to test their statistical significance.Pairwise comparisons between areas indicated the occurrence of three homogeneous groups of areas in the zone 10-50 m and 50-100 m (Table 6).According to the table, in the depth stratum 10-50 m there were no clear distinct groups of areas, except that of the northern and central Adriatic Sea (area 211), which showed very low abundance indices in comparison to the other ones.The group 2 consisted of areas with generally lower mean abundance values in comparison to the group 3. Areas not belonging contemporarily to groups 2 and 3 stand out from the others.As a matter of fact, the Alborán Sea, Alicante sector, Gulf of Lions and Sardinian waters (areas 111, 112, 121, 133), showed constantly higher densities, while the area 221, south Adriatic Sea and western Ionian Sea, showed the lowest values in the group.For the stratum 50-100 m, the first group, consisting of 11 areas, and the third one, consisting of 9 areas, presented the lowest and the highest mean abundance values, respectively.Alborán Sea and Sardinian waters remained the areas where the greatest abundance of the species was found.

Size structure
Length composition analysis of O. vulgaris in the 10 areas considered, showed, in most cases, sizes comprised between 3 and 20 cm ML (Fig. 3).
Small individuals were present in the entire study area.The smallest specimens (1 cm ML) were found in the Catalan Sea (area 113), while the maximum

DISCUSSION
Octopus vulgaris is undoubtedly one of the best known cephalopod species.It has been the object of a number of studies, related mainly to its biology, ecology, behaviour and physiology (Voss, 1977;Wells, 1978;Mangold 1983;Villanueva, 1995).Despite this fact, field studies are still limited to few areas ( Belcari and Sartor, 1999).The MEDITS project allowed the collection of field data in a wide area of the Mediterranean, thus providing useful information on distribution, demographic structure and abundance estimates from areas not yet widely investigated.As a matter of fact, it is well known that the species is exploited mainly by small scale gears, targeted to catch large individuals.However, octopus catches are consistent in trawl landings as well, and have undergone a considerable increase as a consequence of the improvement of fishing techniques, like the introduction of the so-called "French net", a bottom trawl net with a large vertical opening (Belcari et al., 1998).When using otter trawls, captures are possible for that part of the population living on soft, sandy or muddy, bottoms.
The MEDITS trawl surveys confirm the wide geographical distribution of the species: O. vulgaris was collected in the all major areas investigated, although notable differences were evidenced among the 40 geographic sectors.Depth distribution, mostly restricted within 200 m, but with greatest densities in the shallowest 100 m, confirmed data already reported in the literature for the Mediterranean (see Belcari and Sartor, 1999, for a review).The spatial distribution followed a patchy pattern, with noticeable fluctuations among areas.The highest abundances were detected around the Sardinian coasts, the Alborán Sea, the Alicante region and Morocco, whereas the lowest abundances were found in the Adriatic Sea.Arguably, low or high densities of the species could be correlated with scarcity or abundance of rocky bottoms, essential for the development of all the phases of the life cycle of the species.
Size frequency distributions of O. vulgaris showed a wide size range.The maximum length registered was 27 cm ML.The same size is reported as the longest one in other areas of the Mediterranean (Sánchez and Obarti, 1993).However, large individuals were present in a very low proportion.As a matter of fact, it is known that trawling captures all sizes, but especially the small ones (Sánchez and Obarti, 1993).Considering that mature males are reported from 6.5 ML, but that the majority of them become mature at larger sizes, and that mature females are reported from 13 cm ML (Ruby and Knudsen;1972;Belcari and Sartor, 1993;Sánchez and Obarti, 1993), it appears that most of the population sampled was constituted by immature specimens.This is probably a consequence of various factors: (a) samplings refer to only one season of the year, when largest octopuses progressively disappear from trawling grounds, mainly as a consequence of migration to rocky bottoms and coastal habitats for spawning (Mangold-Wirz, 1963;Mangold, 1983); (b) unsuitability of the trawl net to capture large, mature individuals.
Investigations carried out in the Thracian Sea during May-July on sandy, muddy bottoms and Posidonia oceanica meadows within 35 m depth, using fyke nets and pots (Kallianiotis et al., 2001) showed that the majority of the trapped individuals (99.45%) were fully mature, with a modal mantle length of 14-15 cm.
The MEDITS surveys highlighted the relevant presence of small individuals in almost all analysed areas; the dominant mode of 5-6 cm ML in the histograms of a great extent of the Italian coasts and, above all, in Morocco and Corsica waters, correspond to the cohort of trawl net recruits.Taking into account that MEDITS samplings were mostly carried out from May-June onwards, this result confirms previous data reporting the presence of juveniles in the summer period in the Ligurian and Tyrrhenian seas (Relini and Orsi Relini, 1984;Belluscio and Ardizzone, 1990).The different patterns showed in the other analysed areas, where higher modes were evidenced or where it was not possible to single out a predominant size, could be variously interpreted.In the Mediterranean, the reproductive period of O. vulgaris seems to extend almost throughout the year, from January-March to July-October, with one or more spawning peaks (Mangold-Wirz, 1963;Mangold and Boletzky, 1973;Guerra, 1975;Sánchez and Obarti, 1993).As a consequence of a prolonged spawning season and long brooding period, a wide size range can be detected and support the existence of cohorts with different growth rates depending on the hatching period, as proposed for other cephalopod species (Boyle, 1983;Mangold, 1983;Belcari, 1996).Furthermore, postponements of the maturity peaks could lead to different compositions of the demographic structure of the sampled population.Finally, a not perfectly synchronous sampling in the different MEDITS areas of a species characterized by a short life span and high growth rate and having a reproductive period which is hard to be defined (Nixon, 1969;Guerra, 1992 ), could be partly responsible for some of the differences shown in the present study.
size (27 cm ML) was recorded in the Ionian Sea (eastern Sicily).Recruits with modal length of 5-6 cm ML constituted the dominant mode in the Ligurian and Tyrrhenian seas, Sicilian Channel, south Adriatic and Ionian Sea(areas 132, 134, 221 and  222).In Morocco and Corsican waters (areas 114, 131) recruits represented up to 50 % of the entire sampled population.The pairwise comparisons of length frequency distributions made by the Kolmogorov-Smirnov test revealed significant differences (p< 0.05) among all samples analysed.

TABLE 2 .
-Octopus vulgaris: Mean density (n/km 2 ) estimated from the MEDITS trawl surveys by depth stratum, geographical sector and
FIG. 1. -Density and biomass indices of Octopus vulgaris in the different sampled bathymetric strata.Mean; Box: Mean ± Standard Error; Whisker: Mean ± Standard Deviation.