Introduction

Mussel bed (Mytilus galloprovincialis LMK) assemblages can develop in clean and moderately polluted as well as polluted waters (e.g. Bellan-Santini et al. 1994; Damianidis and Chintiroglou 2000). Therefore, biomonitoring of these assemblages can provide valuable information about the impact of pollution on hard bottom communities (e.g. Wenner 1988).

There is adequate information about the structure of M. galloprovincialis assemblages from various regions in the Mediterranean Sea, especially from the western Mediterranean coasts (Bellan 1969, 1980; Bellan-Santini 1969, 1981; Desrosiers et al. 1982, 1986; Hong 1983; Tursi et al. 1984; Tsuchiya and Bellan-Santini 1989), while information on the Aegean Sea is relatively limited (Kocatas 1978; Topaloglou and Kihara 1993; Lantzouni et al. 1998; Damianidis and Chintiroglou 2000). Many authors have put emphasis on the importance of these assemblages in biomonitoring studies (see Thiel and Ullrich 2002). Damianidis and Chintiroglou (2000) reported that the abiotic factors at all sampling sites within Thermaikos Gulf do not fluctuate in time. Therefore, any variation in the composition of M. galloprovincialis assemblages has to be attributed to biotic factors.

This study aims at the investigation of the spatial and temporal structure of the fauna associated with M. galloprovincialis assemblages on the eastern coast of Thermaikos gulf.

Methods

Study area

The sampling sites were selected on the basis of their historical background, as well as the exposure of the assemblages and the depth of their occurrence. Of all locations with artificial hard substrate along the east coast of Thessaloniki Bay, two appeared to be very similar in bathymetric distribution and exposure of M. galloprovincialis populations: the piers of Agia Triada (ST1) and Perea (ST2) (Fig. 1). These piers were constructed 25 years ago and are supported by concrete pillars, which comprise the substrate for the mussel beds. At these sampling sites, the M. galloprovincialis populations develop a uniform physiognomic aspect, with large numbers of mussels covering an area that extends from the lower infralittoral zone down to a depth of 2.5 m.

Fig. 1
figure 1

Map showing Thermaikos Gulf and the two sampling sites (ST1, ST2)

Full size image

Physico-chemical factors

During this study, physico-chemical factors such as salinity (S‰), conductivity (μS/cm), water clarity (m), dissolved oxygen (mg/l), temperature (°C) and total hydrodynamics were measured. All measurements were made using WTW (Wissenschaftlich-Technische Werkstaetten, Weilheim, Germany) and Lovibond Checkit (Dortmund, Germany) micro-electronic equipment and water clarity was examined using the Secchi disc. These measurements were conducted monthly at each sampling site and the results have already been reported by Damianidis and Chintiroglou (2000).

Sampling methods

Sampling was carried out while scuba diving. Samples were taken as described by Chintiroglou and Koukouras (1992). The area covered by the quadrat sampler was 400 cm2 (Stirn 1981; Bakus 1990). Three replicates were taken each time. The samples, 24 in total, were collected during winter and summer of 1994 and 1995. After sampling, the specimens were preserved in 10% formalin and were transferred to the laboratory for further treatment. All samples were collected by the same scuba diver.

Data analysis

Common biocoenotic methods were employed to analyze the faunal composition of the M. galloprovincialis assemblages (Bellan-Santini 1981; Damianidis and Chintiroglou 2000). Hence, the numerical abundance (N) on a scale of 1 m2, the mean dominance (D) and the frequency (f) were estimated. Also Shannon-Weaver’s (H′), and Margalef (d) and Pielou’s Evenness (J′) were calculated on a log2 basis (Daget 1979).

Seasonal differences in the mean numbers of individuals were tested using one-way ANOVA and multiple comparisons. All data were then converted to logarithms. The Spearman rank correlation coefficient (r s) was employed to determine the relation of the number of mussels (Ab M) with faunal abundance (mAb) and richness (R).

The numerical abundance data, obtained per sampling station, were analyzed using cluster and multidimensional scaling (mds) techniques, based on the Bray-Curtis similarity, using the PRIMER package (see Clarke and Green 1988; Clarke and Warwick 1994). The square root transformation was applied in order to increase the contribution of the rare species (Clarke and Warwick 1994). The significance of the multivariate results was assessed using the ANOSIM test. SIMPER analysis was performed in order to identify the percentage contribution of each species to the overall similarity within a site and the dissimilarity among sites (Clarke 1993). The above were carried out to examine the similarity degree of samples, in both space and time.

Results

Composition of the assemblage

A total of 100 species were found associated with M. galloprovincialis assemblages in Thermaikos gulf (Table 1). The distribution of these species in major taxonomic groups is given in Table 2. The dominant groups are polychaetes (37.5%) and crustaceans (30.9%). As shown in Table 1, 17 species were distinguished as “very common” (f≥50%), 26 as “rare” (f<10%), and 57 as “common” (10%<f<50%).

Table 1 A list of species found in the Mytilus galloprovincialis assemblage at two stations in Thermaikos Gulf, in summer and winter of 1994 and 1995. W Winter, S summer, f frequency, mAb mean abundance, D p partial mean dominance, d diversity index, H′ Shannon Index, J′ Equitability Index
Full size table
Table 2 Distribution of the species found in the assemblage to the level of major taxa
Full size table

Among the very common species, the polyclad Stylochus sp. and the decapods Pilumnus hirtelus and Pisidia longicornis have been described as the main predators of mussels (see Galleni et al. 1977; Damianidis and Chintiroglou 2000). The cirripeds Balanus perforatus and B. trigonus, the polychaete Serpula vermicularis and Bryozoa sp1, are well-known as organisms that often settle on mussel shells (Bussani 1983; Damianidis and Chintiroglou 2000). The rest of the very common species are free motile organisms which employ various modes of feeding. The tube-building peracarids are detritivores (Barnard 1958, 1963; Isaac et al. 1994), Ophiothrix sp. (Pérès 1976) is a suspensivore, and the polychaete Staurocephalus rudolphii is a carnivore (Fauchald and Jumars 1979).

Relationships between fauna and structure of mussel beds

According to Tsuchiya and Nishihira (1986), the morphology and relative age of the mussels in an assemblage can play a significant role in the composition of the associated fauna. Therefore, the correlations of mussel abundance with species richness and faunal abundance, respectively, were examined.

For both summer and winter samples, faunal abundance (mAb) and richness (R) were not correlated (P>0.5) with the mussel abundance (Ab M).

Diversity

A total of 17,090 individuals, representing 100 faunal species, were examined. As shown in Table 1, the number of species and the diversity indices (H′, d and J′), were determined for each sampling station and season (winter/summer). The number of species ranged from 37 to 49 in winter, and from 45 to 60 in summer. The diversity indices ranged proportionately as they appeared to be higher in the summer samples (Table 1).

Abundance

The comparison of the faunal abundance in time (within and between years) and in space (sampling sites) was based on the examination of the null hypothesis that the abundance of the fauna does not differ significantly. One-way ANOVA was used to detect the exact differences and showed an equal distribution of the abundance of the fauna in winter as well as in summer samples (F=0.58, df=1,3, P=0.64 in winter samples, and F=2.47, df=1,3, P=0.14 in summer samples).

Affinity of sampling in space and time

The affinity of all sampling sites is given in Fig. 2. Both analyses (cluster and non-metric mds) indicate a separation of samples into three main groups at about 50% similarity level. The winter samples from 1995 form group A, the winter samples from 1994 group B and the summer samples from both years (1994 and 1995) group C. Summer and winter samples are separated, indicating seasonality of the benthic assemblages. The two stations were not separated. Samples from successive years were only separated in winter. The stress value for the two-dimensional mds configuration is 0.01, indicating an excellent ordination of samples (Clarke and Warwick 1994). The performance of a one-way ANOSIM test gave global R=1 at a significance level of P<0.005, so the separation of the three groups (A, B and C) was confirmed. Group A showed an average similarity of 62%. As identified by SIMPER analysis, five species (M. galloprovincialis, B. perforatus, Elasmopus rapax, Pisidia longicornis and B. trigonus) were responsible for 60% of the average similarity, and 15 species for 90%. Group B reached an average similarity of 57%, with seven species (M. galloprovincialis, Pisidia longicornis, Ophiothrix fragilis, B. perforatus, B. trigonus, Pilumnus hirtelus and Athanas nitescens) covering 60% of this similarity, and 22 species 90%. Finally, group C reached an average similarity of 67%, with eight species (Corophium sp., M. galloprovincialis, B. trigonus, O. fragilis, Pisidia longicornis, Prionospio malmgreni, S. vermicularis and Pilumnus hirtelus) being responsible for 60% of the similarity, and 26 species for 90%. As regards the divergence between groups, we found that group A had an average dissimilarity percentage of 50% with group B (21 species contributed 60% and 55 species 90% of this value) and 57% with group C (16 species contributed 60% and 54 species 90%), while the dissimilarity percentage between groups B and C was 50% (22 species contributed 60%, and 58 species 90%). It is quite clear that only a few species are important for characterizing the groups, while most of them are important for differentiating the groups.

Fig. 2
figure 2

Affinity of the sampling sites according to cluster and non-metric multidimensional sampling analyses

Full size image

Discussion

Photophilic soft algae communities probably have the highest faunal and flora diversity of all benthic biocoenoses in the Mediterranean (Pérès 1982; Bellan-Santini et al. 1994). The assemblage of mussel beds of M. galloprovincialis in Thermaikos Gulf belongs to this type of biocoenosis, showing high diversity with respect to species abundance. One hundred animal species were recorded during this study, 17 of which were characterized as very common (f≥50%), and 57 as common (10%<f<50%). The recorded species have been reported by many authors as members of the assemblages of sublittoral photophilic algae (Bellan-Santini 1969; Saldanha 1974; Kocatas 1978; Bellan 1980; Hong 1983; Desrosiers et al. 1986; Marinopoulos 1988; Chintiroglou and Koukouras 1992; Topaloglou and Kihara 1993; Bellan-Santini et al. 1994; Lantzouni et al. 1998; Baxevanis and Chintiroglou 2000; Damianidis and Chintiroglou 2000).

The biodiversity of M. galloprovincialis assemblages in Thermaikos Gulf differs from those reported for other Mediterranean and northern European sites (Table 3). Most authors have used similar methods for the study of hard substrates (based mostly on work by Bellan-Santini 1969) (Table 3), yet data from different geographical areas are difficult to compare. The highest species abundance (R=131) was found in the infralittoral zone of the Portuguese coast. The clean waters of Marseilles, Ismir and Thermaikos Gulf also show high species abundance, while species abundance was relatively low in the Bosporus, on Danish coasts, in the midlittoral zone of the Portuguese coast and in the polluted waters of Marseilles. The respective values in Japan and Chile ranged around 65, while 89 species were counted in Italy and 56 in Aberffraw (N Wales).

Table 3 Literature data on M. galloprovincialis assemblages. Q Quadrat, C core, INFR infralittoral, MID midlittoral, P polluted, NP non-polluted
Full size table

Table 3 also shows that polychaetes and crustaceans are the most important taxonomic groups of the assemblage, contributing almost 50% of the total faunal species abundance. The abundance of these groups, however, varies from one area to another and also seems to be dependent on the specific features of each study area (polluted/non-polluted; midlittoral/supralittoral), a fact that was also reported by Thiel and Ullrich (2002). Saldanha (1974), for instance, recorded fewer polychaetes and more crustacean species on the coasts of Portugal, while D’Anna et al. (1985) recorded exactly the opposite for Sicily. Kocatas (1978) found 35 polychaete and 32 crustacean species in the Izmir Bay, whereas Topaloglou and Kihara (1993) reported 10 polychaete and 22 crustacean species for the Bosporus. Seed and Suchanek (1992) drew the same conclusions, even though the Mediterranean mussel assemblages were not the focus of their work. The total diversity of mussel assemblages in the studied areas shows no significant variations. Furthermore, significant similarities exist at the taxonomic level. It should also be noted that, according to Damianidis and Chintiroglou (2000), there is no significant difference in the composition of the dominant species: this implies that the structure of the polychaete fauna of M. galloprovincialis assemblages in the Mediterranean Sea is largely homogeneous.

An important part of the study of mussel assemblages is the study of interactions between the structure of mussel populations and the associated fauna (Tsuchiya and Nishihira 1986; Lintas and Seed 1994). These studies have produced contradictory results. According to Tsuchiya and Nishihira (1986), the structure of mussel beds in the Pacific Ocean (Japan) has a direct effect on the diversity of the assemblages. Lintas and Seed (1994) suggested that the fauna associated with M. edulis appear to be related largely to mussel density. Damianidis and Chintiroglou (2000) reported similar results on the polychaete fauna of mussel beds in Thermaikos Gulf. In the present study, however, no relationship was found between mussel density and the abundance and diversity of the assemblage. As regards the structure (= distribution of size classes of the populations) of the M. galloprovinicialis populations, there is certain information in Le Breton and Chintiroglou (1998) indicating an uneven distribution in space and time. Although information is still limited, it appears that biotic interactions have a strong effect on the composition of the assemblage, which seems to decrease when the assemblage is studied as a whole. In this case, the composition of the assemblages is largely even in space and time. Nevertheless, attention should be paid to any variations in order to understand the biotic interactions in hard substrate assemblages (see Damianidis and Chintiroglou 2000).

The evenness of the studied assemblage showed some variation in space, and particularly in time, while the total homogeny of the faunal composition remained unaffected. The separate analyses conducted by Lantzouni et al. (1998) and Damianidis and Chintiroglou (2000) produced similar results.

In conclusion, there is now adequate information about M. galloprovincialis assemblages, although there are still open questions. For biomonitoring of marine benthic assemblages, knowledge of the structure and function of M. galloprovincialis assemblages can play an important role. The main advantage of studying such assemblages, on a smaller or wider scale, is their similar physiognomic appearance (Reish 1971; Wenner 1988). As these assemblages can be found in clean as well as in polluted waters, they are of great interest in biomonitoring studies (see Wenner 1988).