INTRODUCTION
One of the main goals of ecology has been to understand patterns in the composition and structure of the benthic assemblages. Recognizing the broad-scale processes and web of interactions among organisms, which operate in heterogeneous habitats, must include an understanding of functioning at smaller spatial scales (Underwood & Chapman, Reference Underwood and Chapman1998). In many cases, spatial patterns are considered in terms of patches, which are areas of a definite size with species composition differing from that in neighbouring ones (e.g. Wu & Levin, Reference Wu and Levin1994; Yakovis et al., Reference Yakovis, Artemieva, Fokin, Grishankov and Shunatova2005).
Biodiversity and faunal abundance are directly related to the complexity of marine and brackish habitats (e.g. Jones et al., Reference Jones, Lawton and Shachak1994; Attrill et al., 1996; Yakovis et al., Reference Yakovis, Artemieva and Fokin2004, Reference Yakovis, Artemieva, Shunatova and Varfolomeeva2008). For example Attrill et al. (1996, 1999) recorded the number of supralittoral species in the brackish environment of the Thames Estuary to be positively correlated with habitat complexity. Sandy–muddy sediments with a larger number of plant species, particles such as stones, shells, decayed wood and clay provided additional niches leading to much higher biodiversity than in locations lacking these complex features (Attrill et al., 1996, 1999). In the areas with relatively low diversity (e.g. estuaries and brackish seas) the heterogeneity of the substrate seems to be of prime importance for the sustainability of local ecosystems (Attrill et al., 1996, 1999). Therefore, documenting ecological patterns of such diversity hot spots is of special need in the light of proper ecosystem management and protection (Attrill et al., 1996, 1999).
Boulder fields assemblages have been the focus of many studies (e.g. Chapman & Underwood, Reference Chapman and Underwood1996; Chapman, Reference Chapman2002a,Reference Chapmanb; Cruz Motta et al., Reference Cruz Motta, Underwood, Chapman and Rossi2003; Le Hir & Hily, Reference Le Hir and Hily2005). With their spatial and temporal variations they constitute heterogeneous habitats conducive to high biodiversity (Chapman, Reference Chapman2002a). Smaller rocks are usually at constant state of recolonization following disturbance caused by overturning. Larger rocks being heavier are harder to overturn and thus offer much more stable substrate and usually support a more abundant biota (Osman, Reference Osman1977; Wilson, Reference Wilson1987). Together with their sessile inhabitants, boulders can create complex three-dimensional structures and provide additional niches for other organisms. The biotic part of such structures is often composed of animals (‘ecosystem engineers’) like corals, mussels or algae (e.g. Seed & O'Connor, Reference Seed and O'Connor1981; Seed, Reference Seed1996; Kuklinski & Bader, Reference Kuklinski and Bader2007; Yakovis et al., Reference Yakovis, Artemieva, Shunatova and Varfolomeeva2008). Such species influence the diversity and abundance of organisms living in these three-dimensional structures and while they may not be involved in direct trophic interactions they can modulate the availability of resources to other species by creating physical complexity (Jones et al., Reference Jones, Lawton and Shachak1994). Even when such three-dimensional structures are separated by only a few metres, they can harbour rather different faunal assemblages due to differences in their complexity (Yakovis et al., Reference Yakovis, Artemieva, Shunatova and Varfolomeeva2008).
In this study we investigated rocks and their associated biota in the brackish environment of the Baltic Sea (Gulf of Gdansk). A sandy bottom dominates the coastal zone of the Gulf of Gdansk, however, in places large patches of stones and boulders exist. Several studies have indicated that these patch areas support relatively diverse assemblages of organisms (Haque et al., Reference Haque, Szymelfenig and Węsławski1997; Osowiecki, Reference Osowiecki2000; Osowiecki & Żmudziński, Reference Osowiecki and Żmudziński2000; Kruk-Dowgiallo & Szaniawska, Reference Kruk-Dowgiallo, Szaniawska and Schiewer2008; personal observations). HELCOM (1994) Helcom recommendation 15/5: System of coastal and marine Baltic Sea Protected Areas (BSPA) recommended that one of these sites (near Gdynia-Redlowo) should be included in the Baltic Sea Protected Areas. However, it is recognized that rocky areas on sediment enhance diversity and increase its conservation importance (e.g. Attrill et al., 1996); quantitative ecological studies of biota associated with rocks from the area are lacking. The main goal of the present investigation is to fill this gap in our knowledge concerning the rocky habitats of the Gulf of Gdansk where, given its low diversity, such habitats can be considered as local diversity hot spots. In addition to the faunistic aspect of this study, we also investigate some of the factors that influence rock-associated fauna.
MATERIALS AND METHODS
Sampling and laboratory analyses
The Gulf of Gdansk is located in the southern part of the Baltic Sea (Figure 1). It covers an area of 4940 km2, with a mean depth of 59 m (Majewski, Reference Majewski1990). Its water is brackish, with an average salinity of 5–8 psu. Furthermore, the fresh waters of the Vistula River influence coastal waters of the Gulf. Surface temperature varies between 0°C in winter to more than 22°C in summer (Kruk-Dowgiallo & Szaniawska, Reference Kruk-Dowgiallo, Szaniawska and Schiewer2008). The bottom of the Gulf is made up of soft sediments, largely coarse and medium sand with occasional silt deposits. Rocky patches of postglacial origin (pebbles and boulders) occur at several locations, usually near the shore (Majewski, Reference Majewski1990; Osowiecki & Żmudziński, Reference Osowiecki and Żmudziński2000). However, in the majority of cases, such patches are small and very shallow. The boulder field selected for this study is the largest in the bay and is located in the vicinity of Gdynia (Figure 1). It is approximately 3 km long and 0.2 km wide. Within it, boulder patches are separated by bare sand and/or seagrass. Environmental variables, e.g salinity, water temperature and anthropogenic influences are constant across the study area (Nowacki & Jarosz, Reference Nowacki and Jarosz1998).
Study area with sampling locations.
The study was conducted in spring 2007 in a shallow, coastal zone of the Gulf of Gdansk at a depth of ~3 m. Samples of rocks of various sizes (31 rocks from Station A and 40 rocks from Station B) were collected haphazardly from two randomly chosen locations by SCUBA diving (Figures 1 & 2). Each rock was placed in a separate zip-bag in situ by a diver in order to retain all the associated biota. After removing algae (and their associated fauna), each rock was dried and all encrusting fauna were determined to species level and counted. Stone surface area (as a proxy for size) and the percentage coverage of encrusting fauna were estimated using an inflexible net marked in a grid of square centimetres. Fauna indirectly associated with rocks were separated from algae and all macroorganisms (larger than 500 µm) were determined to the lowest taxonomic level, in most cases to species. All individuals were counted and their abundance expressed as individuals per 1 m2 of rock surface. In the case of the present study each colony of a bryozoan species was considered as one individual.
Rocks in situ with their associated biota at study locality.
We explored the influence of substrate stability and heterogeneity on species composition and abundance in rock-associated biota. We used the surface area of the rocks as a proxy for substrate stability. The size of the rock has previously been shown to be directly related to its stability (Osman, Reference Osman1977). The presence of any three-dimensional structure on a rock increases its complexity by adding extra niches (e.g. Attrill et al., 1996). Because the smoothness of the substrate was negligible, we used the number and dry mass of large engineering species (algae, bivalves and barnacle soft tissue with a shell) present on a given rock as a proxy for substrate heterogeneity. The dry mass of the samples was measured after 24 hours of drying at 60°C to constant weight.
Data analyses
In order to analyse whether samples were saturated with respect to the faunal species pool, species accumulation curves were plotted for both sites. Species accumulation curves describe the number of observed species (Sobs) as a function of the number of samples. The non-parametric estimator Chao2 (Chao, Reference Chao1987) was additionally selected to calculate the theoretical species richness from presence/absence in all the available samples. Richness estimators predict richness, including species not discovered in the sample, from the proportional abundances of species within the total sample (Soberón & Lorente, Reference Soberón and Lorente1993). The Chao2 estimator provides the least biased estimates of species richness for small numbers of samples and moderate species richness, as is the case for this study. Sobs and Chao2 were calculated using the PRIMER package (Clarke & Gorley, Reference Clarke and Gorley2001)
A series of multivariate techniques were used for statistical analyses of the faunal communities. All analyses were conducted on un-transformed, square- and four-root transformed data, in order to detect the effects of the dominant and rarer species.
To explore the influence of variables, regarded as proxies of substrate heterogeneity and stability (rock size, number of engineering species, percentage coverage, etc.) on faunal assemblages, a correlation-based principal component analysis (PCA) was used. Prior to the PCA a draftsman plot was produced to detect correlation between variables. Data for PCA analysis were fourth-root transformed and normalized.
To assess simultaneous response in the indirectly rock-associated faunal assemblages between sites and within sites of proxy variables (rock size, algal biomass and abundance of sessile fauna), permutational multivariate analysis of variance (PERMANOVA: Anderson, Reference Anderson2005) was used. PERMANOVA calculates P value using permutations, rather than relying on tabled P value (as traditional ANOVA), which assumes normality. The assumption of normality is difficult to achieve and cannot be properly tested according to ecological data. Analyses were conducted on four-root transformed data, to reduce the influence of highly abundant taxa, using Bray–Curtis dissimilarities. Data for rock-size were divided into three groups: small rocks—all of which were smaller than 100 cm2 (later indicated in the text as 100), medium rocks—between 100 and 200 cm2 (indicated as 200); and large—bigger than 200 cm2 (indicated as >200). A similar division was conducted for algal biomass (algal biomass <0.5 g, between 0.5–1 g, and >1 g) and sessile fauna abundance (density <5, 5–50 and >50). PERMANOVA analysis requires an equal number of observations per treatment. In our case 10 rocks were randomly included in each group. Finally, a second draftsman plot was performed (Pearson correlation coefficients: Clarke & Warwick, Reference Clarke and Warwick1994) to identify further correlations between mobile species densities, the number of recorded species and proxy variables. Using PRIMER software non-metric multidimensional scaling (nMDS) was performed on the similarity matrices of quantitative data to illustrate patterns in the spatial distribution of the samples in terms of faunal assemblages (associated fauna, sessile fauna and all faunal data) and analysis of similarities (ANOSIM) was performed to test for differences between sites according to different factors (e.g abundance of structural species and abundance of associated species).
RESULTS
A total of 26,609 faunal organisms from 25 animal taxa were found on 71 rocks totalling 12,450 cm2 surface area (Station A: 6422 cm2 and Station B: 6028 cm2).
The shapes of the species accumulation curves were very similar for both sites (Figure 3A). For site A the curve levelled before all the samples were included, while for site B the accumulation plot was levelling at the end of its curve. The total number of species observed was 23 for Station A and 24 for Station B. The Chao2 estimate of total richness gave mean 24 (SD±0.68) species for Station A and 26 (SD±0.97) species for Station B (Figure 3B). For Station A the points tended to form a curve of asymptotic shape even if a small number of samples were considered. For Station B the shape of the curve was slightly more variable.
Species accumulation curves for both investigated locations. Species richness estimated by the total number of all species (Sobs) (A) and Chao2 estimators (B).
Among the fauna there were three species that directly encrusted stones and 23 species indirectly associated with rocks (associated with algae growing on the rocks (Table 1). One individual of Caprella mutica Schurin was recorded for the first time in the Bay of Gdansk. The number of species per stone varied from 5 to 17, while their abundance varied from 19 to 926 individuals.
Mean density values of species abundance (ind·m−2) and standard deviation (italic) for each rock size-class at investigated stations (B, Bivalvia; Br, Bryozoa; C, Crustacea; G; Gastropoda; H, Hydrozoa; I, Insecta; O, Oligochaeta, P, Polychaeta).
Three sessile animal species (Balanus improvisus, Mytilus trossulus and Electra crustulenta) were directly attached to the rocks and were present at both locations and on almost all rock size-classes, except for rocks below 100 cm2 in surface at site A, where M. trossulus was absent (Table 1). The community structure of these sessile species was homogeneous and no differences between sites were recorded (ANOSIM test: R = 0.056, P = 0.32; Figure 4C). Balanus improvisus dominated at each site, both numerically (Station A: 74% and Station B: 77%) and in terms of weight (Station A: 89% and Station B: 65%).
Non-parametric multidimensional scaling plots based on abundance and composition of all faunal data (A), only fauna indirectly associated with rocks (B) and fauna directly associated with rocks (C).
Five species of algae (Cladophora sp., Pilayella littoralis Linnaeus, Polysiphonia nigrescens Hudson, Ectocarpus confervoides (Roth) and Ulva sp.) were recorded on rocks at both sites. The mean total biomass of algae was not equal between locations. At Station A dry biomass reached 65.1 g·m−2 (SD±58.8) and at Station B 52.0 g·m−2 (SD±30.3). Based on the total data set for animals indirectly associated with rocks, oligochaetes were the most abundant group (45%) followed by crustaceans (27%) and polychaetes (26%). The most common species was the oligochaete Nais elinguis (32% at Station A and 22% at Station B), which was followed by the crustacean Heterotanais oerstedti (19%) and second species of oligochaete Paranais litoralis (12%) at Station A, and the polychaetes Pygospio elegans juv. (21%) and P. litoralis (13%) at Station B. The crustacean Idotea chelipes was recorded only at site B, while a clavid hydrozoan was found only at Station A. The mean total abundance of associated fauna was higher at Station A (28085 ind·m−2; SD±13651) compared to Station B (11916 ind·m−2; SD±5991). There was a clear difference in the compositon of the associated fauna between stations (ANOSIM test: R = 0.512, P = 0.001; Figure 4B).
In general, the abundance of species indirectly associated with rocks increased with the size of the rock and the algal biomass (r = 0.64, P < 0.001 and r = 0.78, P < 0.001 respectively; see also Figure 5). Not all variables for engineering species were correlated with faunal abundance and species number (e.g. M. trossulus abundance versus abundance of associated organisms r = 0.26, P > 0.05; M. trossulus mass versus number of associated species r = 0.36, P > 0.05). Only the number of B. improvisus individuals had significantly positive correlation with faunal abundance and species number (r = 0.78, P < 0.001; r = 0.82, P < 0.001 respectively).
Mean abundance of indirectly associated fauna at different size rock classes (A) and algal biomass classes (B).
The total variability explained by PCA was 73.3%. 58.4% variations were explained by the first principal component (PC1), with PC2 explaining 14.9%. The analysis indicated algal biomass, rock size and B. improvisus abundance as the primary PC1 variables explaining the variation in data. The remaining variables included in the analysis (percentage coverage, E. crustulenta and M. trossulus abundance) had minor influence on the observed data variation.
Significant differences in the indirectly rock-associated faunal assemblages between sites and within the site were found. Of the variables which were regarded as proxies for substrate stability and heterogeneity, algal biomass had significant effect on the assemblages within investigated stations (PERMANOVA: F = 4.91, P = 0.015) and between the sites (F = 23.86, P < 0.0001). These differences are apparent in the nMDS, however, the three algal biomass classes show some degree of overlap (Figure 6A). Consistent with the PERMANOVA results, rock size and B. improvisus abundance were also significant for faunal communities (F = 5.82, P = 0.031; F = 3.80, P = 0.023 respectively). Nonetheless, the differences between rock size-classes were smaller for Station B compared with Station A (Figure 6B). For Station A changes with increasing rock size can be seen, even if a PERMANOVA post-hoc pairwise test was not significant (P > 0.05) for all rock classes. A similar pattern is observed for number of B. improvisus individuals (Figure 6C).
Non-parametric multidimensional scaling plots showing the effect of the biomass of algae (A), size of the rocks (B), and abundance of barnacles (C) on the faunal assemblages at Station A (black symbols) and at Station B (grey symbols). The stress value indicates that the two-dimensional plots give an adequate representation of the data.
DISCUSSION
Local biodiversity
Natural, undisturbed stony seafloors are colonized by an abundance of plants and animals (mostly invertebrates), so they might by regarded as ‘living rock’. Furthermore, the spatially complex structures developed by primary structural species provide food and shelter for a rich diversity of dependent species (Wenner et al., Reference Wenner, Hinde, Knott and Van Dolah1984) to the extent that even small or isolated rocky areas surrounded by soft sediment can be considered as oases of biodiversity and productivity at any latitudes.
For the Polish Baltic coastal zone, a lack of baseline information on the biological functioning of nearshore hard bottom dictates a need for research and monitoring. To our knowledge this study, is the first quantitative investigation of a stony habitat in the Gulf of Gdansk. The only other study that has focused on faunal assemblages on hard bottoms was performed by Andrulewicz et al. (Reference Andrulewicz, Kruk-Dowgiallo and Osowiecki2004) at the stony reef (8.5 to 18 m depth) of the Slupsk Bank (open Baltic water). In addition Kotwicki (Reference Kotwicki1996) investigated the fauna associated with algae growing on a shallow sandy bottom, in the inner basin of Puck Bay, part of the Gdansk basin; data from these sources are summarized in Table 2. Andrulewicz et al. (Reference Andrulewicz, Kruk-Dowgiallo and Osowiecki2004) recorded 27 faunal taxa, including Turbellaria and Nematoda, which are usually considered to be meiofauna (Table 2) as well as common representatives of cooler and deeper waters such as the isopod Saduria entomon and the polychaete Harmothoe sarsi (Hagerman & Szaniawska, Reference Hagerman and Szaniawska1990; Janas et al., Reference Janas, Wacial and Szaniawska2004). In total, there is only a 48% overlap in species composition between our study and that of Andrulewicz et al. (Reference Andrulewicz, Kruk-Dowgiallo and Osowiecki2004) (see Table 2). Kotwicki (Reference Kotwicki1996), who investigated shallower, inshore waters, observed 29 taxa (Table 2). There were several organisms found by Kotwicki (Reference Kotwicki1996) but not encountered in our study (see Table 2), e.g. the crustacean Idotea baltica or the gastropod Theodoxus fluviatilis. The most probable factors causing dissimilarities in our results are differences in habitat and sampling methodology. In addition, temporal fluctuations of faunal recruitment may also play an important role, even though sampling was performed in the same season of the year.
Checklist of associated macrofaunal species found on the Polish coast (stony and algal habitats): √, present; -, absent; *, sum of abundance of all four recorded species; **, recorded, but considered as meiofauna and not counted.
Our results, as well as a number of published studies, indicate that rocky habitats increase local biodiversity by attracting different sets of species to those found in nearby soft bottoms (e.g. Yakovis et al., Reference Yakovis, Artemieva and Fokin2004, Reference Yakovis, Artemieva, Shunatova and Varfolomeeva2008 and references therein). By introducing extra living space to the area, rocks promote both the abundance and biomass of local macrofauna communities. Although the number of species in our study (25) corresponds well with the number of macrofaunal species recorded for adjacent soft bottoms, there was very little overlap in species composition between these two habitats (e.g. Jażdżewski & Konopacka, Reference Jażdżewski and Konopacka1995; Żmudziński, Reference Żmudziński, Piesik, Piesik and Kurowska1996; Kotwicki, Reference Kotwicki1997).
The abundance of rock-associated macrofauna is above the range reported for the Gulf of Gdansk sandy habitats. The total mean density revealed by this study varied between 10,000 and 23,500 ind·m−2 of rocks surface area (Table 1), while for earlier studies of sandy habitats it ranged from 0 to 14,000–15,000 ind·m−2 of sea bottom surface (Haque et al., Reference Haque, Szymelfenig and Węsławski1997; Osowiecki & Żmudziński, Reference Osowiecki and Żmudziński2000). A comparison of these data indicates that rocks, and the engineering species colonizing them, provide much more living space per unit surface area than does a sediment habitat; their fractal dimensions (Gee & Warwick, Reference Gee and Warwick1994) are considerably greater. Boulders often offer sheltered living space (Le Hire & Hily, 2005). They influence the benthic boundary layer and reduce environmental disturbance including currents created by wave action, temperature or salinity variation as well as protection from predation (Kraufvelin & Salovius, Reference Kraufvelin and Salovius2004; Le Hire & Hily, 2005; Norling & Kautsky, Reference Norling and Kautsky2007). Moreover, they physically trap detritus and faeces from associated fauna, which may also facilitate abundance, via nutrients provision, of these associations (Norling & Kautsky, Reference Norling and Kautsky2007).
Factors affecting assemblage structure
A detailed examination of the data indicates that structure of faunal assemblages within the patches is driven mostly by differences in the dominance of the associated fauna rather than by species composition, which was similar among the sites, or by structural species (Table 1; Figure 4). Some of the assemblages’ patterns could have been driven by organismal biology. All three animal species that were attached to rocks (M. trossulus, B. improvisus and E. crustulenta) produce a large number of planktotrophic larvae (e.g Bayne, Reference Bayne1964; Brusca & Brusca, Reference Brusca and Brusca2002). While such larvae give these organisms the potential to colonize over large distances, they are also responsible for highly variable interannual variation in recruitment. It might be suggested that such a life history would lead to homogeneous colonization of the area, which was evident in our data when only these sets of organisms were analysed (Figure 4C). On the other hand, among indirectly rock-associated fauna there are a number of brooders (e.g. H. oerstedi, G. salinus and C. multisetosum), which have limited larval dispersion, often limited to the immediate vicinity of the parent (e.g. Waage-Nielsen et al., Reference Waage-Nielsen, Christie and Rinde2003; Väinölä et al., Reference Väinölä, Witt, Grabowski, Bradbury, Jażdżewski and Sket2008). Therefore life strategies can lead to either spatial patchiness (e.g. indirectly rock associated fauna) or homogeneity (e.g. sessile fauna) (Figure 4).
Faunal associations can also be affected by different physical and chemical properties of algae (Parker et al., Reference Parker, Emmett Duffy and Orth2001; Goodsell & Connell, Reference Goodsell and Connell2008). It was found that quantitative descriptors of algae, such as the ratio of surface area to biomass, can play a role in epifaunal assemblage structure (Parker et al., Reference Parker, Emmett Duffy and Orth2001). Aquatic plants with higher ratios of surface area to biomass are more structurally complex and therefore are capable of supporting more abundant and diverse epifaunal assemblages (Parker et al., Reference Parker, Emmett Duffy and Orth2001). However, other studies have shown that macrobenthic assemblage structure can be related to particular attributes of the habitat, for instance the presence of a given algae species (e.g. Edgar, Reference Edgar1983; Parker et al., Reference Parker, Emmett Duffy and Orth2001; Kraufvelin & Salovius, Reference Kraufvelin and Salovius2004). Associated faunal assemblages often differ to a large extent among algae species (Edgar, Reference Edgar1983; Parker et al., Reference Parker, Emmett Duffy and Orth2001). In our study, there were five species of filamentous algae associated with rocks. Differences in their relative occurrence, and therefore the level of complexity among locations could lead to numerical differences in the associated fauna. There are a number of studies indicating that an increase in epifaunal abundance, biomass and diversity is directly related to plant quantity (e.g. Attrill et al., 1996; Parker et al., Reference Parker, Emmett Duffy and Orth2001; Christie et al., Reference Christie, Jørgensen, Norderhaug and Waage-Nielsen2003; Norderhaug et al., Reference Norderhaug, Christie and Fredriksen2007).
The size of rocks in the Gulf of Gdansk caused spatial varability in the composition of the associated macrofaunal assemblages (Table 1; Figure 6B); the larger the rock size the higher the number of both species and individuals (Figure 5). These results confirm the findings of a number of earlier studies (e.g. Osman, Reference Osman1977; Sousa, Reference Sousa1979; Barnes & Kuklinski, Reference Barnes and Kuklinski2003) which have shown that the size of rocks is of prime importance for colonization patterns. Most studies revealed a similar pattern to the one recorded here. As rock size increases there is both an increase in availability of surface for colonization and substrate decrease in the probability that it will be overturned by wave action (Wilson, Reference Wilson1987). Increased surface area increases the probability of an organism settling as well as enabling a larger number of biota to coexist. With increased stability of the substrate there is a greater chance that organisms will achieve reproductive age before the next disturbance occurs (e.g. turnover of the rock caused by storm) (Sousa, Reference Sousa1979). There is also the possibility that rocks might be overturned during the period of larval settlement or immediate post-settlement leading to individuals failing to establish. Moreover, with increased rock size we recorded an increased number of engineering organisms (e.g. barnacles—r = 0.61, P < 0.001; algae—r = 0.71, P < 0.001), which may have caused the observed increase in other macrofaunal individuals. The number of B. improvisus individuals and the biomass of algae had significant effect on the associated faunal assemblages (Figure 6). Barnacles and algae enhance the complexity of smooth surfaces of stones (personal observation), which creates additional living space (Yakovis et al., Reference Yakovis, Artemieva, Fokin, Grishankov and Shunatova2005; Norling & Kautsky, Reference Norling and Kautsky2007).
The results suggest that rock associated macrofauna is influenced by combined effect of rock size and surface complexity created by structural organisms. It appears that factors influencing rock-assemblages at the two locations studied follow similar trends, however, patterns at Station A are slightly easier to detect (Figure 6).
CONCLUSIONS
Boulder fields are important components of coastal ecosystems, particularly in low diversified habitats dominated by sandy bottom as in the brackish waters of the southern Baltic Sea. As indicated above they can be regarded as local diversity hot spots. This study also showed that hard-bottom benthic assemblages are affected by a complex of the factors. Substrate stability and heterogeneity are of prime importance for faunal assemblages associated with rocks. There were statistical differences in assemblages' structure between sites within the rocky patch. Yet the observed differences were mostly due to differences in numerical abundance rather than species composition. Further investigations of boulder patches together with the surrounding sandy habitat are needed to describe interaction between these two habitats more precisely and provide information allowing the determination of the specific functional importance of existing rocky habitats.
ACKNOWLEDGEMENTS
The authors would like to thank Jakub Korthals for diving assistance, Lech Kotwicki and Marek Zajaczkowski for logistic support. Additionally we are grateful to Lech Kotwicki for providing us with his unpublished data. Dr Timm Tarmo (Institute of Zoology and Botany, Estonia) is acknowledged for his help with identification of oligochaetes. We also would like to thank Mike Kendall, Suzanne Williams and an anonymous referee for comments leading to an improved manuscript. The study has been completed thanks to support of DYNARC EU project and a grant from the Polish Ministry of Science and Higher Education (No. N N304 270434).
