Characterization and cartography of some Mediterranean soft-bottom benthic communities ( Ligurian Sea , Italy ) *

Soft-bottom benthic communities were studied along the Western coast of the Ligurian Sea with a new approach using both videocamera surveys and collected samples. The preliminary distribution of soft-bottoms and the definition of the limits and status of seagrass beds were carried out in September 1991, using an underwater vehicle provided with a videocamera and towed by a boat. Moreover, 90 benthic samples were collected at 5-40 m depth in order to characterize the macrobenthic soft-bottom communities. Six soft-bottom benthic assemblages and two sea grass biotopes (Cymodocea nodosa and Posidonia oceanica) were revealed by means of underwater images and multivariate analysis (TWINSPAN) on samples collected. The communities inhabiting the infralittoral sandy and coarse sediments corresponded to those previously described in the Mediterranean Sea, whereas a large complex transition between sandy and muddy communities was recognized on circalittoral soft-bottoms. Information obtained with this approach was used to draw a map of the investigated areas at 1:10,000 scale. The employment of the two techniques was cost effective for both time and research effort.


INTRODUCTION
During the last few years, cartography of benthic communities has been used succesfully in environmental management, such as planning parks and productive areas, and as a tool for detecting regression phenomena of interesting biotopes such as beds of Posidonia oceanica (Ardizzone, 1992;Ardizzone et al., 1993).
The charting process previously needs the careful definition of biological units that thereafter should be easily recognised.Afterwards, one of the problems in charting biological data is the definition of major discontinuities in an environmental continuum.
Mediterranean benthic communities were described from a qualitative and quantitative point of view by Peres and Picard (1964) and Picard (1965) respectively.Such assemblages are described by the presence of characteristic species and generally extend over wide areas, even though there is often local overlapping due to the presence of mixed sediments (Bourcier et al., 1979;Guille, 1970;Gambi and Fresi, 1981;Falciai et al., 1983;Fresi et al., 1983;Nodot et al., 1984;Gravina, 1986;Dounas and Koukouras, 1992).These transitional zones where species from both assemblages coexist have been defined as "ecotone", and sometimes an increase in the number of species and individuals has been observed (Bellan Santini, 1985).Moreover, the local abundance of a species in circalittoral biocoenoses has lead to the description of some "degradation facies" resulting from anthropogenic effects (Salen Picard, 1981;1985).In this view, identification of assemblages as proposed by Pérès and Picard (1964) is very difficult.
This paper is an attempt to characterize infralittoral and circalittoral soft-bottom communities with a new mixed approach: the "classical" bionomical study with the collection of benthic samples and the information obtained by underwater images recorded by a vehicle towed by a boat.

Study area
The study areas were two zones with a steep and rocky coastline located in the west Gulf of Genoa from 5 to 40 m depth (Fig. 1).The first zone (ZONE 1), about 15 km in length, is located between Albisola (Savona) and Arenzano (Genoa); the bot-  tom is steep down to 50 m depth, but more gentle down to 100 m depth.The second zone (ZONE 2), about 8km in length, is located further to the west, between Bergeggi (Savona) and Noli Ligure (Savona).There, the continental platform is so narrow that the bottom is very steep and the 100 m bathymetric very close to the coastline.

Underwater images and charting methodology
Bottom investigation was carried out using an underwater towed vehicle provided with a videocamera, and connected by a cable to a boat, where the images of the bottom were observed and recorded.The vehicle was located at a distance of 1-10 m on the bottom with a shot range of 1.5-15 m.Boat, cable and telecamera can be considered as a system unit, so that coordinates of recorded images can be compared to those detected by GPS on the boat if a little delay is considered.
Bottom images were recorded along 106 and 35 transects orthogonal to the coast in the zone 1 and zone 2 respectively.Observations along transects allow the best interpolation of discontinuous images of benthic communities located on a depth gradient.In order to produce a cartography map with a scale of at 1:10,000 transects were arranged at 150-200 m intervals, corresponding to 1/50-1/25 on the cartography scale, as suggested by previous experiences (Ardizzone, 1992).
Images recorded are interpreted before mapping and the information content is related to the restitution scale.For example, several characteristics of the Posidonia bed can be reported on maps at 1:500-1:5,000 scales (Meinesz et al., 1983), whereas only two main categories (Posidonia bed and Posidonia "dead matte") can be reported at lower resolution, as in the case of the present study.At this scale other seagrasses such as Cymodocea nodosa, were easily recognizable and mapped (Ardizzone, 1992).On the contrary, this method is not useful to identify softbottom communities which are mainly composed of infauna.
One problem arising in mapping biological data is the definition of spatial boundaries between communities that generally lie along a continuum.At some point, two ecological units can be inserted one in the other: this is the case of the upper limit of Posidonia meadows that often is composed of Posidonia patches on soft-bottom.In such cases if the restitution scale is 1:10,000, it may be appropiate to report only the biotope with a coverage more than 50%.Somewhere else, seagrass coverage could be continuous but with very low densities as reported in many cases for Cymodocea nodosa.In this case the Cymodocea meadow should be detected only if there are at least 50 shoots in a square meter.Furthermore, statistical analysis on faunal samples collected on bottoms partially colonized by the seagrass can support visual observations.

Sampling methodology
During the same period, a total of 90 soft-bottom samples were collected along 19 transects between 5-40 m depth for faunal analysis (Fig. 1).At shallow depth (down to 15 m) on compact fine sand, macrobenthos was collected using a 50 litre dredge with a mouth of 0.59 x 0.23 m, as proposed by Picard (1965).To get information on a single sampling point the dredge was externaly covered with a sackcloth to fill it immediately (Picard, 1965).The same instrument was used also on bottoms colonised by Cymodoca nodosa.Finest silty sediments located below seagrass beds were collected using the Van Veen grab covering an area of 0.1 m 2 (three replicates equivalent to about 50 litres of volume).Furthermore, in some representative sampling stations the same grab was used to collect sediment for grain size analysis (Fig. 1).
Macrobenthic fauna retained by a 0.5 mm mesh was fixed in 10% formol and sorted in the laboratory.Sediments were frozen as suggested by Holme and McIntyre (1984) and grain size analysis was performed by sieving through 2 mm and 0.063 mm meshes to sort gravel, sand and silt-clay fraction respectively (Wentworth, 1922).

Data analysis
Identification to species level was done for the majority of zoobenthic taxa to produce an abundance matrix.In order to get equal information from different systematic groups a selected matrix was prepared after elimination of species with less than 0.5% abundance inside each group, such as: molluscs, polychaetes, sipunculans, crustaceans, echinoderms, and cephalocords.TWINSPAN analysis (Hill, 1979) was performed on this matrix to point out the affinity between sampling stations.Mean and standard deviation of diversity values (H') (Shannon and Weaver, 1949), evenness (J) (Pielou, 1975) and redundance (R) (Patten, 1962) were then computed for each group of sampling stations resulting from multivariate analysis.
Finally, results of TWINSPAN were put into a GIS (ARC/INFO) to produce the cartography of the area.Limits of Posidonia meadows and softbottom were obtained from the discontinuities observed with videocamera using methods sug-gested in the previous chapter (at least 50% of coverage for Posidonia and at least 50 shoot/m2 for Cymodocea).With this procedure we obtained a map with polygons pertaining to two main classes; the first grouping all types of seagrass meadows and the second all the other soft-bottom benthic communities.This last set of polygons was subsequently classified according to the TWINSPAN results with the following "maximum likelihood" procedure: a Thiessen polygon (ESRI, 1991) was first generated around each sampling station; each polygon thus contained only one sampling station and was assigned to the class of the sampling station as derived from the TWINSPAN.By intersecting these polygons with those defined by the limits obtained with videocamera, they were then classified according to a "majority rule": if most of a "videocamera" polygon fell within a certain class of Thiessen polygon then it would take on that certain class.

Faunal composition
A total of 20,554 individuals belonging to 575 macrozoobenthic species was collected (Tab.1).Molluscs turned out to be the most abundant and diversified taxon at shallow depth (5-10m), whereas polychaetes were more important on deeper bottoms (11-40m depth).Although crustaceans, mainly amphipods and decapods, were less important as far as species and individuals are concerned, they were constantly present along the depth gradient.

Structural analysis
Result of TWINSPAN on a selected set of 85 samples and 188 species revealed the presence of seven groups of sampling stations, primarily ordered according to bathymetry and subsequently to sediment texture (Fig. 2, 3).The first distinction was between samples located above and below 15 m depth.At the shallow depth, samples of coarse sediments with vegetal debris (I assemblage) were different from sandy samples (II and III assemblages).Similarly, at greater depths coarse sediment samples (IV and V assemblages) differed from finer ones (VI and VII assemblages).Finally, at the shallow depths the remaining two groups of stations reflected respectively the presence (II assemblage) and the absence (III assemblage) of Cymodocea beds.The IV assemblage was composed of coarse sediment samples mostly located  between 15 m and 25 m depth, whereas the V assemblage included deeper bottoms (30-40 m depth) with a higher fine fraction.Finally, the last two assemblages represented the deep samples, at 20-40 m depth (VI and VII assemblages) which differed in the silt-clay fraction.
Concerning the faunal composition of the seven assemblages, the first twenty species ranked in order of abundance and their ecological preferences are reported in Table 2 together with the mean values of some structural parameters for each assemblage.Although in assemblage I (gravel with vegetal debris) sandy-bottom and gravellicolous species were present, the most important ecological groups were rocky bottom and Posidonia species (Fig. 4).The diversity was high due to the presence of many rare species.Assemblage II (C.nodosa bed) was dominated by sandy bottom species which largely coexist with rocky bottom ones due to the presence of C. nodosa.The diversity was higher than for unvegetated sandy bottoms.Assemblage III (fine well sorted sand) fine sandy-bottom species were largely dominant and the lowest values of both biodiversity and species richness were encountered.In the Assemblage IV (coarse sand and gravel), gravellicolous species and those indicating instability were dominant.The fauna was very rich in species.Similarly, in Assemblage V (muddy biodetritic bottoms) species indicating instability and those having wide ecological distribution were found to be the most abundant and the diversity value was the highest.
Assemblages VI and VII corresponded to heterogeneous mixed sediments and were respectively characterized by widely distributed species (circalittoral muddy-sand bottoms) and species indicating instability (circalittoral sandy-mud bottoms).The diversity values were intermediate between sandy (II and III) and detritic bottoms (IV and V).

CARTOGRAPHY
Maps resulting from image data and sampling analysis are shown in Fig. 5.In ZONE 1, Cymodocea beds (Assemblage II) were extensively widespread from 5 m to about 20 m depth, where they sometimes bordered upon P. oceanica beds.Posidonia beds extended from 5 m to 25 m depth and were bordered by dead matte (an aggregation of roots and rhizomes persisting after the death of the leaves) on their lower limit.Infralittoral soft-bottoms not colonized by seagrass consisted of fine well sorted sand with Diogenes pugilator and S. subtruncata (Assemblage III), and coarse sand and gravel with S. taylori and S. glandulata (Assemblages IV) in the erosion channels inside the P. oceanica beds.In some restricted areas, an abundant local deposit of vegetal debris on coarse sediments was inhabited by a community characterized by E. pusillus and A. ephippium, with P. mucosa and S. polytrema as the dominant species (I assemblage).Below the seagrass beds, mixed CARTOGRAPHY OF SOFT-BOTTOM BENTHIC COMMUNITIES 33

DISCUSSION AND CONCLUSIONS
The employment of two techniques such as the classical bionomical approach and the analysis of underwater images allowed us to draw the maps of two areas of the Western Ligurian Sea up to 40 m depth.
Classical sampling strategy such as the collection of samples by means of dredges and grabs on softbottoms and of scuba divers on hard bottoms are always valid systems to get information for drawing bionomical maps at a scales up to 1:500.For scales between 1:500 and 1:50,000, a mixed approach, based on information obtained both by benthic sampling methods and underwater video images, revealed itself to be mostly effective in terms of both time and expense (Ardizzone, 1992).In fact, on soft-bottoms the video images can supply some information (ripple-marks, organic debries, large benthic invertebrates), but they are inadequate to characterize the benthic communities.In such case, the analysis of benthic samples remain the most exhaustive supply, but a preliminary investigation of soft-bottoms with an underwater vehicle and videocamera is useful to settle the sampling design.
In the present study, information on the Posidonia beds (extension, dead matte) obtained by video images was reported on a 1:10,000 map together with results of statistical analysis on samples collected.With this last procedure shallow infralittoral bottoms were found to be inhabited by fine well sorted sand community (III) which borders upon Posidonia and Cymodocea beds (II).Intermatte channels inside the Posidonia meadows were characterized by the community of coarse sand and gravel (IV), even if the community structure changes when a significant deposit of vegetal debris occurs (I).Circalittoral bottoms were inhabited by three main communities corresponding to muddy-biodetritic sediments (V), muddy-sand (VI), and sandy-mud (VII).
In general, some considerations about the definition and classification of infralittoral and circalittoral soft-bottom communities arise from this study.On infralittoral bottoms, a fine well-sorted sand community largely corresponds to that reported by Pérès and Picard (1964) and many authors reporting on the Mediterranean coasts (Picard, 1965;Massé, 1972;Falciai et al., 1983).When C. nodosa densely colonizes fine sand, the increase of spatial heterogeneity and therefore of microhabitat availability leads to an increase in diversity and abundance causing a differentiation in community structure (True Shlenz, 1965;Giangrande and Gambi, 1986;Lanera and Gambi, 1993).Consequently, in the case studied, C. nodosa can be considered a key-stone species of the infralittoral zone (sensu Paine, 1966).A similar consideration has already been reported for Posidonia oceanica (Sarda, 1991;Somaschini et al., 1994).
Bottoms of coarse sand and gravel, characterized by Sphaerosyllis taylori, Sphaerosyllis glandulata, Parapionosyllis labronica, and Calyptraea chinensis, are more abundant between 10 m and 20 m depths in the intermatte channels inside Posidonia beds.This assemblage is very similar to that of coarse sand and gravel exposed to bottom currents (SGCF) (Pérès and Picard, 1964;Picard, 1965;Dauvin, 1988).The community composition changes if a large amount of vegetal debris is deposited on gravels, with Phyllodoce mucosa, Echinociamus pusillus, Anomia ephippium, Spirobranchus polytrema and many amphipods increasing in importance.
A more complex situation was observed in the circalittoral zone, down to the lower limit of the Posidonia beds.Pérès and Picard (1964) and Picard (1965) reported the presence of different circalittoral assemblages, such as coastal detritic (DC), muddy detritic bottoms (DE), terrigenous muds (VTC), and shelf-edge detritus (DL) (Peres, 1967).The three different assemblages characterising the upper part of the circalittoral zone of the investigated areas do not overlap with those previously described by Peres and Picard.
A difficulty in defining soft-bottom circalittoral communities has been reported already by other authors: large overlapping of soft-bottom circalittoral communities was observed along the French coast (Bourcier et al., 1979;Nodot et al., 1984); Salen Picard (1981;1982) described many degradation facies of coastal detritic (DC) and muddy detritic bottoms (DE); and Guille (1970) observed many transition zones among these assemblages so that he recognized only one circalittoral soft-bottom community with many sub-communities or "facies".A large transition zone between sandy and muddy bottoms, mainly characterized by widely distributed species has been observed by Febvre Chevalier, 1969;Gambi and Fresi, 1981;Falciai et al., 1983;Fresi et al., 1983;Zavodnik et al., 1985;Gravina, 1986.Effects of alteration in benthic communities also became evident through the analysis of the Posidonia meadows.In both zones, the inferior limit of the Posidonia beds is composed of dead matte.Its presence is evidence of a progressive change in community distribution and can be the result of alterations in the sedimentological equilibrium.A comparison with previously collected data from ZONE 1 shows that only 20 years ago living Posidonia was present where the dead "matte" is today (Drago and Albertelli, 1976).
FIG. 4. -Percentage composition (number of individuals) of ecological groups as detected by the autoecology of species collected in the seven assemblages.

TABLE 1 .
-The systematic groups collected during the study.
Indicator species identified by the analysis are reported.The most important features of biotopes are indicated below.
CARTOGRAPHY OF SOFT-BOTTOM BENTHIC COMMUNITIES 31 FIG. 2. -Cluster (Q mode) resulting from TWINSPAN.FIG. 3. -Percentage composition by weight and standard deviation (vertical bars) of sediment texture in the seven benthic assemblages.

TABLE 2
WE wide ecological distribution.Mean values and standard deviation of some structural parameters are given for each assemblages.
. -The twenty most important species ranked in order of abundance (nº of individual collected) for the assemblages (I-VII) identified by TWINSPAN (see text).For each species the ecological group is reported: FS: fine sandy bottom; GR: gravellicolous; IN: indicator of instability; MB muddy bottom; MI: mixticolous; PB: Posidonia bed; RB: rocky bottom; UK: Unknown;