Patterns of spatial variability of mobile macro-invertebrate assemblages within a Posidonia oceanica meadow

The study evaluated patterns of spatial variability of mobile macro-invertebrate assemblages associated with Posidonia oceanica leaves within a meadow at three different depths and three different coastal wave-exposures. A total of 171 taxa was found, among them two Nemertea, 65 Mollusca, 91 Arthropoda, eight Annelida and five Echinodermata. The total number of taxa per sample was higher in intermediate and deep stands than in shallow ones, while no significant differences among coastal wave-exposures were observed. Multivariate analyses detected significant differences in the structure of assemblages in relation to depth and coastal wave-exposure. Mollusca were more abundant in shallow stands, while Crustacea and Echinodermata increased in intermediate and deep stands. Moreover, patterns of spatial variability changed with depth: shallow and intermediate assemblages showed high small-scale (tens of metres apart) variability, while deep assemblages showed high variability at intermediate scale (hundreds of metres apart).


Introduction
Natural systems may vary in space and time, following patterns of organismal distribution (Menge and Olson 1990;Schneider 1994). Moreover, the variability of natural assemblages is scale dependent (Underwood and Chapman 1996;Benedetti-Cecchi 2001;Terlizzi et al. 2007), making it crucial for ecologists to identify the main scales of variability and the processes generating these patterns (Levin 1992). Knowledge of spatial patterns of assemblages may allow discrimination between natural variability and human-induced effects in environmental monitoring surveys and impact evaluation studies (Hewitt et al. 2001;Bishop et al. 2002;Fraschetti et al. 2005) and sampling design pertinent to ecological questions (Underwood 1993;Benedetti-Cecchi et al. 2003).
Seagrasses and their associated assemblages represent one of the main coastal ecosystems and they are the object of a lot of studies because of their ecological relevance in coastal waters and their role as bioindicators (Kirkman 1996;Kuo et al. 1996). Seagrass meadows host complex macro-invertebrate assemblages both sessile, as epiphytes of leaves and rhizomes, and mobile (Mazzella et al. 1992). Mobile macro-invertebrate assemblages associated with seagrasses are particularly abundant and diverse thanks to the habitat complexity of meadows, and their role as refuges from predators and as a food supply (Heck and Orth 1980). In fact, the structure of seagrasses can create many microhabitats characterised by different environmental conditions, allowing the coexistence of species with different ecological requirements (Stoner 1980;Orth et al. 1984). Moreover, plants and their epiphytes represent an important food source for both grazers and detritivorous organisms Klumpp and Van der Valk 1984;Edgar 1999;Duffy et al. 2003). Although widely studied (Mattila et al. 1999;Atrill et al. 2000;Bostrom et al. 2006), patterns of spatial variability of macroinvertebrate assemblages of seagrass meadows and factors influencing these patterns are not completely known.
In the Mediterranean Sea, Posidonia oceanica (L.) Delile represents the most important seagrass in terms of meadow extent and their role in coastal systems (Mazzella et al. 1992). The structure of macro-invertebrate assemblages associated with P. oceanica has been widely studied (Russo et al. 1991;Scipione et al. 1996;Bedini et al. 1997;Zakhama-Sraieb et al. 2011;Albano and Sabelli 2012;Urra et al. 2013) and spatial variability of assemblages has been evaluated at a large scale (among meadows and habitats) and in relation to depth gradients (Russo et al. 1984;Scipione and Fresi 1984;Mazzella et al. 1989;Gambi et al. 1992Gambi et al. , 1995Borg and Schembri 2000;Borg et al. 2010;Bedini et al. 2011;Belgacem et al. 2011Belgacem et al. , 2013. However, little is known about spatial patterns of variability within meadows, and about mechanisms determining these patterns.
The present study aimed to evaluate patterns of spatial variability of mobile macro-invertebrate assemblages associated with the leaves within a Posidonia oceanica meadow, in relation to depth and coastal wave-exposure. For this purpose, a hierarchical sampling design was used to study a P. oceanica meadow at multiple spatial scales in relation to three different depths and three different coastal waveexposures.

Material and methods
The study was carried out around the Island of Pianosa, in the Tuscan Archipelago National Park (northwestern Mediterranean Sea), in July 2010 ( Figure 1). The sampling period was chosen because it corresponds to the maximum vegetative development of Posidonia oceanica. The island has very low anthropogenic pressure and all activities are forbidden for about 2 km from the shore. The seafloor slopes gently, allowing the development, between 5 and 40 m depth, of a Posidonia oceanica meadow about 16 km 2 wide (Cinelli et al. 1995). The meadow is installed on matte (structures constituted of several layers of dead rhizomes and trapped sediment); rocky and sand bottoms are present above and below the meadow all around the island. The ecological quality of the site allows the elimination of variation due to anthropogenic disturbance; also, the structure of the meadow may be considered similar within a given depth range.
Three depth ranges were sampled: 8-10 m (shallow = S), 18-20 m (intermediate = I) and 30-35 m (deep = D). Three sides of the island corresponding to different coastal wave-exposures (east = e, south = s, west = w) were considered, and four areas 100s of m distant from each other were sampled at each depth and side. Western and southwestern winds are dominant in the area, making the western side the most exposed to water movements (Region of Tuscany 1993).
At each area, 10 replicate measures of density were made using a 40 × 40 cm quadrat and 10 shoots were collected to assess the meadow canopy (Pergent et al. 1995).
A hand-towed net (40 × 20 cm, 0.4-mm mesh size) was used to sample mobile organisms living in the leaf stratum. Two samples tens of metres distant from each other were collected at each area. Each sample consisted of a series of 60 strokes over the seagrass canopy (Russo et al. 1985;Buia et al. 2003). In the laboratory, taxa were identified and the abundance was expressed as number of individuals per sample.
Species composition and abundance, were analysed by permutational analysis of variance (PRIMER 6 + PERMANOVA, Anderson 2001). A three-way model was used with depth (S vs I vs D) and side (e vs s vs w) as fixed and crossed factors, area (four levels) as random factor nested in the interaction depth × side. Speciesabundance data were log(x + 1) transformed before calculation of Bray-Curtis index of dissimilarity. A two-dimensional non-metric multidimensional scaling (nMDS), based on centroids for replicated areas, was used for a graphical representation of the data. Pseudo-variance components in species composition and abundance were calculated for each spatial scale at each depth. The SIMPER (similarity percentages) routine was performed on untransformed data to establish which taxa contributed most to the dissimilarity.
Shoot density, leaf length, the number of taxa, the abundance of organisms per sample and the abundance of the main phyla were analysed by analyses of variance (ANOVA), with the same factors and levels used for multivariate analyses. Cochran's C-test was utilised before each analysis to check for homogeneity of variance, and data were transformed when necessary. The Student-Newman-Keuls (SNK) test was used for a posteriori multiple comparison of means (Underwood 1997).

Meadow structure
Shoot density ranged from 665.3 ± 33.5 to 181.4 ± 21.1, decreasing from shallow to deep stands (Figure 2b), while values of mean leaf length were higher in intermediate stands ( Figure 2b). ANOVA and SNK test detected significant differences among depths (even if non-homogeneity of data has to be considered for shoot density) but not among coast sides (Table 1).
The total number of taxa per sample was higher in intermediate and deep stands than in shallow ones, while no significant differences among sides were observed (Table 3; Figure 3a). Differences among depths and sides were not significant for the mean number of organisms per sample (Table 3) even if the abundance of the main taxa changed along the depth gradient: Mollusca were more abundant in shallow stands, while Crustacea and Echinodermata increased in intermediate and deep stands (Table 3; Figure 3b). Crustacea also differed between east and west sides (Table 3). PERMANOVA showed significant differences related to both depth and side for species composition and abundance of macro-invertebrate assemblages (Table 4). Pair-wise tests showed that depths were significantly different to each other. Differences were also significant between the east and west sides ( Table 4). The nMDS showed that shallow stands were well separated from intermediate and deep ones, and, within the shallow stands, the west side was separated from the others (Figure 4).    Pseudo-variance components increased from large (sides, kilometres apart) to small spatial scales (samples, tens of metres apart) at shallow and intermediate depths (Figure 5a and b), while deep stands had a higher variance at intermediate scale (areas, hundreds of metres apart, Figure 5c).
The SIMPER test showed the taxa that mostly contributed to differences between depths were the Mollusca Bittium latreillii, Chauvetia mamillata and Rissoa auriscalpium which were more abundant in shallower assemblages; and the Crustacea Hippolyte inermis, Cestopagurus timidus and Phtisica marina, the Mollusca Jujubinus exasperatus and Ocinebrina aciculata and the Echinodermata Asterina gibbosa which increased their abundance with depth (Table 5). Differences between the west and east sides of the island were mostly related to a higher abundance of Bittium latreillii, Cestopagurus timidus and Rissoa auriscalpium on the east side (Table 5).

Discussion
Results show that the structure of macro-invertebrate assemblages within the studied meadow varied in relation to depth and coastal wave-exposure. Moreover, patterns of spatial variability changed with depth. The structure of macro-invertebrate assemblages of Pianosa Island was similar to that described for other P. oceanica meadows, with a dominance of Mollusca and Crustacea (Russo et al. 1984(Russo et al. , 1991Scipione and Fresi 1984;Mazzella et al. 1989Mazzella et al. , 1992Gambi et al. 1992;Scipione et al. 1996;Borg and Schembri 2000;Bedini et al. 2011;Belgacem et al. 2011Belgacem et al. , 2013Zakhama-Sraieb et al. 2011;Urra et al. 2013). However, differences in patterns of diversity and abundance may be highlighted between the pristine Pianosa assemblages and those described in meadows of the same geographic area but subjected to a higher human pressure (Bedini et al. 1997(Bedini et al. , 2011. The total number of taxa was higher at Pianosa than in the other meadows of the continental coasts of Tuscany, where three meadows were sampled at three depths with a sampling effort (98 samples) higher than that of the present study (171 taxa at Pianosa Islands vs 136 taxa at  three other localities of Tuscany, Bedini et al. 1997Bedini et al. , 2011. The mean number of individuals and the mean number of species per sample were also higher at Pianosa (193.8 and 18.2,respectively) than in the other Tuscan meadows   1 and 14.8). The structure of the assemblages was strongly influenced by depth. Depth gradients in the structure of mobile epifauna of seagrasses are widely reported, especially for P. oceanica that can spread along a wide bathymetrical range Borg and Schembri 2000;Belgacem et al. 2011Belgacem et al. , 2013, even if this pattern does not seem to be consistent throughout the year (Bedini et al. 2011). Depth-related patterns of spatial variability have been interpreted as a consequence of the trophic distribution of invertebrates (Mazzella et al. 1992). In agreement with this model, the SIMPER test showed that herbivorous organisms (Bittium latreillii, Rissoa auriscalpium and Alvania lineata) were normally linked to shallower stands, where algal epiphytes are more abundant (Nesti et al. 2009), while carnivores-detritivores (Cestopagurus timidus, Hippolyte inermis, Phtisica marina, Asterina gibbosa, Siriella clausii) were mostly distributed in the deep parts of the meadow. However, depth-related patterns were also found within each trophic group, as shown by the higher abundance of Jujubinus exasperatus and other herbivores in the deeper stands. This finding highlights that other factors, such as hydrodynamism, temperature and dissolved oxygen (Russo et al. 1984;Scipione and Fresi 1984;Belgacem et al. 2013), may be important in determining changes in the structure of macro-invertebrate assemblages along a depth gradient. Changes in meadow density between shallow and deep stands may also influence mobile macro-invertebrates, both directly, modifying the habitat complexity, and indirectly, causing changes in epiphyte assemblages and in the effectiveness of protection from predators (Heck and Orth 1980;Orth et al. 1984;Gambi et al. 1998;Attrill et al. 2000;Bostrom et al. 2006;Belgacem et al. 2013).
An interesting finding of the study is the differences in patterns of spatial variability among depths. Pseudo-variance components showed an increasing trend from larger to smaller spatial scales in shallow and intermediate assemblages, following a pattern widely reported for benthic assemblages, including seagrass epiphytes (Piazzi et al. 2004;Balata et al. 2007). A high small-scale variability characterises seagrass systems, as a consequence of a patch distribution of organisms related to biotic interactions or habitat heterogeneity (Vanderklift and Lavery 2000;Lavery and Vanderklift 2002). In contrast, the studied deep assemblages showed high variability at an intermediate scale, among areas hundreds of metres apart. Similar changes in spatial variability between shallow and deep meadows have been observed for epiphytic assemblages (Nesti et al. 2009). Differences in meadow structure may determine changes in the associated faunal assemblages at spatial scales similar to that observed in the present study (Parker et al. 2001). However, the studied meadow had a quite uniform structure at the same depths, excluding a similar model. High variability of mobile macro-invertebrate assemblages at small scales in shallow P. oceanica stands has already been described, and is considered related to the more stressful characteristics of a shallow environment as compared to deeper ones . Recruitment may also have an important role, as the dispersion of larvae is influenced by water movements and stratification (Witman and Dayton 2001). The decreasing flow velocity with depth in temperate seas (Denny and Wethey 2001) may reduce the connectivity between sites below the thermocline, concurring to increase the variability at this spatial scale (Witman and Dayton 2001).
Sand bottom and rocky reefs were present below the deep edge of the meadow, and they could influence P. oceanica faunal assemblages. In fact, both modifications in the structure of assemblages near the deep limit of the meadow and changes in patterns of spatial variability could be related to a possible enrichment by organisms typical of the surrounding habitats (Mazzella et al. 1989;Urra et al. 2013). Some species typical of mobile bottoms (the Crustacea Leucothoe serraticarpa, Eurydice pulchra and Pagurus alatus) and rocky reefs (the Crustacea Leucothoe procera, Dynamene bidentata, the Mollusca Coralliophila meyendorffii and the Polychaeta Syllis gracilis) were found in the deep part of the meadow. The Mollusca Bolma rugosa, typical of bottoms characterised by detritus, was found only in juvenile forms, suggesting a nursery role of meadows for this organism. Moreover, proximity of rocky reefs may influence the intensity of predation and the recruitment of invertebrates, determining important changes in the structure of assemblages (Tuya et al. 2010;Urra et al. 2013). The role of surrounding habitats for P. oceanica invertebrate assemblages was not an aim of this study but may represent a goal for further investigations.
Differences in seagrass-associated fauna between edges and the internal part of seagrass meadows has been already described and considered relative to larval settlement patterns, post-settlement processes and accumulation of organisms that are seeking refuge from predation (Virnstein and Curran 1986;Bologna and Heck 2002). These 'edge effects' could have a key role in determining the structure of meadow-associated assemblages, and represent an important aspect to be considered in interpreting patterns of variability in studies aiming to distinguish, at a small scale, between natural changes in ecological conditions and those induced by human pressure.
Differences in assemblages related to coastal wave-exposure are not easily interpreted. Water movement represents one of the main factors structuring species composition of coastal benthic assemblages. In the study area, the west side is the most exposed to waves. Both wave intensity and the direction of dominant currents can determine changes in the structure of invertebrate assemblages through disturbance, larval dispersion or food supply (Commito et al. 1995;Harriague andAlbertelli 2007: Van Colen et al. 2010). Water movements probably determine the main environmental differences among sides of the island, but other abiotic factors, such as light intensity or sedimentation rates, may be related to coastal waveexposure.
The study was limited only to one period of the year; thus, it did not take into account seasonal changes of faunal distribution along the bathymetrical range Bedini et al. 2011). Despite this limit, the results highlighted important patterns of variability within a P. oceanica meadow, suggesting that different causes, such as shoot density, epiphyte load and distance to other habitats, can determine small-scale changes in the structure of invertebrate assemblages. Results obtained in a very pristine meadow, such as that of the present study, may represent a useful tool to compare macro-invertebrate assemblages associated with meadows subjected to different ecological conditions. In this context, patterns of variability highlighted by the present study may give useful information to be considered in planning sampling designs suitable for separating natural variability within meadows from effects of ecological alterations.
Journal of Natural History 2577