Seagrass systems are characterized by high biodiversity. Their leaves offer substrata suitable for settlement and growth of a number of micro- and macro- colonists that form stratified assemblages characterized by a high diversity of species (Mazzella et al. 1989). The most abundant and diverse epiphytic organisms on seagrasses are algae. They range from unicellular diatoms and dinoflagellates to large macrophytes. Epiphytes contribute to the primary production of the seagrass, formation of sediment (Frankovich and Zieman 1994) and nutrient cycling (Borowitzka et al. 2006) and they are a direct as well as an indirect food source for many animal species (Gambi et al. 1992).

The sensitivity of Posidonia oceanica (L.) Delile to human impacts on ecosystems is an interesting biological indicator of these impacts in the coastal environment (Pergent et al. 1995). Montefalcone (2009) reviewed the use of P. oceanica as a bioindicator and found three levels of investigation: the ‘‘individual’’ level, in which the phenology of the plant (especially leaf biometry) provides information about its status and growth condition (Marbà et al. 2006); the ‘‘population’’ level, in which the structure (e.g. density and/or cover) and morphology of the meadow represent characteristic imprints of environmental conditions (Montefalcone et al. 2008); the ‘‘community’’ level, in which the associated flora and fauna (especially epiphytes) are similarly susceptible to environmental alterations (Cancemi et al. 2003). Seagrass epiphytes are known to be even more sensitive to environmental changes than the plant hosts (Nesti et al. 2009). Epiphytic microalgae composition, in particular, is influenced by abiotic factors such as temperature (Aligizaki and Nikolaidis 2006; Turki 2005), light, salinity, and nutrient availability (Armitage et al. 2006) and biotic factors such as grazing (Mazzella and Russo 1989), shoot length, density and morphology of the host plant (Sirota and Hovel 2006, Chung and Lee 2008, Mabrouk et al. 2011). Nutrient over-enrichment caused by anthropogenic activities has been associated with the shift of the structure of microepiphytic assemblages (Armitage et al. 2006) and the decline of seagrass (Green and Short 2003, Orth et al. 2006, Ben Brahim et al. 2010). For example, a reduction of leaf growth and a decrease in shoot density were observed around a sewage outfall (Pergent-Martini 1994) as a consequence of epiphyte overgrowth (Cambridge et al. 1986).

It is widely known that the distribution of epiphytic assemblages of P. oceanica both on leaves (Mazzella et al. 1989) and rhizomes (Piazzi et al. 2004; Balata et al. 2008) can change, but few studies analysing changes in epiphyte assemblages on leaves of P. oceanica in Tunisia has been published (Ben Brahim et al. 2010, Mabrouk et al. 2013), and to our knowledge no studies on microepiphytes have been conducted up to now. Many authors have bemoaned the lack of data in some Mediterranean regions, particularly North Africa (Ruiz et al. 2009). It is therefore of importance to undertake a study of the effects of urban pressure on P. oceanica dynamics under the impact of human activities vs. control conditions. We analysed, for the first time in the east of Tunisia (Mahdia), the structure and patterns of spatial variability of leaf epiphytic microalgae assemblages and tested their relationships with the degree of anthropogenic disturbance. We analysed the biometric parameters of Posidonia oceanica and its leaf microalgal epiphytes, looking at nutrient enrichment from sewage discharges as a cause for observed changes.

We attempted to answer the questions 1) how vulnerable are the biometric parameters of P. oceanica to these environmental pressures? and 2) which members of the epiphytic microalgae community should be considered as the most sensitive species? We thus compared plant biometric parameters and epiphytic microalgae assemblages of Posidonia oceanica between a station exposed to urban and industrial effluents, and two control stations, and examined their variability at spatial scales using a hierarchical sampling design (Underwood 1992).

The disturbed station exhibited quite higher concentrations of nitrite, nitrate, ammonium, phosphate and Si(OH)₄, TP and TN. Water transparency was higher at the control stations (Table 1). The lowest shoot density, adult and intermediate leaf length, leaf surface, leaf area index and leaf biomass were recorded in disturbed station D (Fig. 2). ANOVA analysis showed that most of biometric parameters (shoot density, leaf length, leaf surface and leaf area index) of P. oceanica differed between control and disturbed stations (Table 2).

The RELATE procedure in PRIMER software reveals significant correlations between physico-chemical and biometric parameters and abundance of microepiphytes (R=0.214, p=0.007).

For epiphytic microalgae on the leaves of P. oceanica, 58 species were counted, including 52 at the control stations and 44 at the disturbed station (Appendix 1). ANOSIM of epiphytic microalgae species abundances log(x+1)-transformed showed significant differences (R=0.591; p=0.01) between the control stations and the disturbed station. The same results were obtained (R=0.615; p=0.01) using presence-/absence-transformed data showing that the dissimilarities between stations are due both to species composition and their abundances.

Dinoflagellates, diatoms and cyanobacteria were common epiphytes on leaves (Fig. 3). Abundances of those groups were included in the univariate analyses of variance. Dinoflagellates differed significantly between stations (F_(2;12)=4.63, p=0.032) and between the contrast ‘disturbed vs. control stations’ (F_(1;25)=5.47, p=0.028) with high abundance at disturbed station (Tukey test). Significant differences were detected for Prorocentrales and Protoperidiniales when comparing at the level of the stations (F_(2;12)=5.96, p=0.016; F_(2;12)=3.19, p= 0.047, respectively) and between the contrast ‘disturbed vs. control stations’ (F_(1;25)=6.97, p=0.014; F_(1;25)=15.89, p<0.001), with high abundance at disturbed station (Tukey test). When the abundances of toxic dinoflagellates were grouped, significant differences were detected at the level of stations (F_(2;12)=5.57, p=0.019) and for the contrast ‘disturbed vs. control’ (F_(1;25)=3.68, p=0.043), with high abundance at the disturbed station. No significant differences for all levels were recorded for diatoms and for cyanobacteria. No significant differences were detected among control stations for all groups (Appendix 2).

SIMPER analyses showed that the average dissimilarity between the control and disturbed group was high (74.19%). This procedure also allowed us to determine the species that contribute to this dissimilarity: Prorocentrum concavum, P. rathymum, Pinnularia sp., Peridinium sp., cyst of Karenia selliformis, cyst of Polykrikos, Gymnodinium sp., Amphiprora constricta, Prorocentrum lima, Oscillatoria sp., Biddulphia, Poly-krikos sp., Pleurosigma sp., Coolia monatis and Amphidinium cartera (Fig. 4).

CCA analysis indicated that the axis I (eigenvalue λ1 =0.324) expressed 74.7% of the cumulative variance species-environmental variable. Eigenvalues (P-value) calculated with the CCA associated with the analysis in the Monte Carlo test were used to select two statistically significant environmental variables (p<0.05) that best explain variations (28.54%) of species composition: Si(OH)₄ and NH₄⁺ (14.27% and 11.86% of total variance, respectively). In the triplot diagram (Fig. 5) disturbed station samples were grouped in the left and are correlated with the axis I, which is defined by the environmental variables NH₄⁺ and Si(OH)₄ (intra-set correlation 0.639 and 0.439, respectively). Species placed together with the disturbed station were Protoperidinium sp. and Gymnodinium sp. The orthogonal projection of these species on environmental variables showed that they have important preferences for NH₄⁺ and Si(OH)₄. Most other species have a midway position between the three stations, which suggests that they have quite similar frequencies and abundances among the three stations.

Our study shows two main results: (1) A decline in meadow vitality at the disturbed station compared with control sites, and (2) The abundance and composition of epiphytic microalgae changes at the disturbed station through an increase of dinoflagellate abundance.

Our data showed a moderate nutrient enrichment at the disturbed station that did not reach an eutrophication stage because the concentrations were not high enough, but the transparency was very low due to suspended solids discharged from the outfall.

The multivariate analysis illustrates a decrease in seagrass leaf biometric parameters (shoot density, leaf length, leaf surface and leaf area index) at disturbed vs. control stations, thus indicating the inability of disturbed meadows to withstand increasing urban interferences. The number of leaves per shoot was not different between stations, a finding which is consistent with previous studies (Capiomont et al. 2000, Mabrouk et al. 2013). A decline in shoot density, leaf area index, leaf surface and leaf length at disturbed stations compared with control sites has been found by several authors (Leriche et al. 2004, Ben Brahim et al. 2010). These results cannot be assigned only to nutrient enrichment; several factors may also explain this decline: first, the decrease in water transparency at the disturbed station (Table 1) involves the reduction of light intensity necessary for photosynthesis. Second, the decrease in salinity following the discharge of sewage could affect plant biometric parameters (Ben Alaya 1972). In addition, pollutants affect the vitality and physiology of Posidonia (Augier and Maudinas 1979, Pergent-Martini and Pergent 2000). The National Sanitation Utility in Tunisia (ONAS 2008) claimed that discharge into the sea in northern Tunisia caused high turbidity and sediment pollution (high content of total organic carbon, fairly high levels of nitrates and metals such as Fe, Mn, Zn, Pb, Ni, Co, Pb and Cd).

The diversity of epiphytic microorganisms on the leaves of Posidonia oceanica was high along the Mahdia coast. confirming previous findings indicating that vegetated ecosystems are ideal habitats for benthic diatoms and other epiphytes (Cummins et al. 2004). This is because seagrass leaves and algal thalli may represent a surface area one order of magnitude greater for the colonization and growth of diatoms (Zieman 1989). Our results show that the abundance and composition of microepiphytes changes at the disturbed station, with an increase in abundance of dinoflagellates, especially from the orders Prorocentrales and Protoperidiniales. These results are similar to those found by Ben Brahim et al. (2013) in southern Tunisia (Gulf of Gabes). Some fertilization experiments have shown shifts in relative dominance among algal species, with a shift towards cyanobacteria (Armitage et al. 2006) and diatoms (Frankovich et al. 2009) under nutrient enrichment.

There is an evident difference of epiphytic composition between control and disturbed stations, but those changes can be attributed to the moderate nutrient enrichment or to the low transparency detected at sites around the outfall. It is known that nutrient enrichment stimulates the growth of epiphytic algae (Moore and Wetzel 2000) and causes changes in their species composition (Wear et al. 1999). However, each microalgal group displays a unique spatial pattern in response to N and P enrichment (Armitage et al. 2006). Our data show that there is an increase in dinoflagellates at the disturbed station compared with control sites but no significant differences are found for diatoms and cyanobacteria. Armitage et al. (2006) also found that diatoms did not response to nutrient enrichment. It seems that dinoflagellates tolerate better those environmental variations. Baker et al. (2007) have suggested that some taxa of this latter group may have several significant ecophysiological differences when compared with diatoms, dealing with variations in their chemical and physical environment (e.g. nutrients, light and temperature), such as a lower affinity for nutrients, considerable nutritional diversity, and motility. In addition, Prézelin and Haxo (1976) have shown that in dinoflagellates grown under low light intensity, a greater proportion of the chlorophyll a is localized in PCP complexes (peridinin chlorophyll a proteins), providing an increased light-gathering capacity. Light is a factor that influences the composition of epiphytes at the polluted station. Indeed, turbidity at this station reduces the light intensity that reaches the meadow and hence the dominance of some tolerant species such as members of the order Prorocentrales. These species have an optimum light intensity lower than 10% of sunlight (Morton et al. 1992).

Phytoplankton species respond in different ways to nutrient enrichment, probably based on their life cycle characteristics, such as growth rate and their absorptive capacity (Philippart et al. 2000). According to Lepoint et al. (2007), it is difficult to predict which group would be favored by the moderate intake of nutrients. Lin et al. (1996) found that nutrient enrichment (NO₃^–, NH₄⁺, PO₄^–, either alone or in combination) did not result in an increase in the abundance of epiphytes on Zostera marina. These communities respond in complex ways to the addition of nutrients. The seasonal influence on the response of algae epiphytic nutrient enrichment seems important (Neckles 1993). For example, the increased epiphytic load on Z. marina because of the high nitrogen loading occurred only in summer (Hauxwell et al. 2003). Therefore, the dominance of a group is either transient or steady but highly seasonal, coinciding with seasonal events of nutrient inputs (Ierodiaconou and Laurenson 2002). Harrison et al. (1991) found that high organic matter and nutrient enrichment caused the increase in centric diatoms (such as Thalassionema, Rhizosolenia, Striatella, Skletonema). Shen (2001) found that the limitation of silicon did not cause a change in species abundance of diatoms but had an effect on cell size, which was smaller than those at control stations.

It seems that the response of the epiphytic community depends on the nature and degree of nutrient enrichment. For example, Silicon (Si) played the most important role in the growth and development of diatoms, while dinoflagellates were mostly controlled by phosphorus (P) availability (Chikhaoui et al. 2008). When phosphorus loading increased, a shift from diatoms to dinoflagellates was observed (Hodgkiss 2001). In the same vein, in Tunisian aquaculture lagoons, blooms of toxic dinoflagellates have been shown to develop when the N:P ratio drops in autumn (Romdhane et al. 1998). Increases in the N:Si ratio were proportional to the increase in flagellates (Smayda et al. 2004) and the decrease in diatoms abundance (Anderson et al. 2002).

The epiphyte-grazer interaction also plays an important role in controlling the abundance and diversity of epiphytes. Indeed, the epiphytes of marine macrophytes are a food source for a range of grazers and predators and, in turn, they influence the diversity and abundance of epiphytes by removing the substrate and biomass of the host plant (Jernakoff et al. 1996, Borowitzka et al. 2006). Grazers can be highly selective (Cattaneo 1983), making the potential effects of nutrient enrichment undetectable (Neckles et al. 1994, Heck and Valentine 2007), and they may thus have a strong effect on the spatial pattern of the periphytic community (Sarnelle et al. 1993). Some grazers (scrapers) feed preferentially on tightly attached diatoms (Mazzella and Russo 1989), whereas others (surfers) favour stalked and filamentous diatoms (Tall et al. 2006). Neckles et al. (1994) showed that numbers of diatoms decreased in the presence of grazers and showed little response to nutrient enrichment. This may be a possible explanation for the absence of variation of diatoms at our studied sites.

Another hypothesis that may explain changes in epiphytic microalgae at the polluted station is that the abundance of epiphytes is correlated with leaf phenological parameters of the host plant (RELATE procedure). The decrease in biometric parameters of Posidonia at the disturbed station (Fig. 2, Table 2) induces changes in the abundance of microepiphytes. This result has been also found by previous studies (Johnson et al. 2005, Mabrouk et al. 2011)

Our results also show an increase at the disturbed station in the abundance of potentially toxic dinoflagellates such as Alexandrium minitum, Amphidinium carterae, Karenia selliformis, Coolia monatis, Karlodinium veneficum, Ostreopsis siamensis, Prorocentrum concavum, P. minimum, P. rathymum and P. lima. Some of them are potential toxin producers (Nakajima et al. 1981). Epiphytic Prorocentrum species are mainly associated with okadaic acid and the production of analogues (Faust 1991, Nascimento et al. 2005). These findings support those of Romdhane et al. (1998) and Armi et al. (2010) in Tunisia and those of Aligizaki and Nikolaidis (2006), Mangialajo et al. (2008) and Totti et al. (2010) in the northern Mediterranean Sea. All those surveys demonstrated that the increase in dinoflagellates, including the toxic species, is related to the increase in nutrient enrichments which is species-specific (Anderson et al. 2002). These results are particularly useful in this area, with the recent establishment of several fish farms, since epiphytic dinoflagellates are easily resuspended in the water column (Mabrouk et al. 2012). Moreover, Marr et al. (1992) concluded that the underestimation of toxic dinoflagellates associated with a toxic event might be due, in part, to the lack of sampling of the benthic and epiphytic communities.

Finally, it is necessary to improve a management programme to protect P. oceanica and its associated epiphytes in Ben Ghayadha, respecting the guideline of UNEP/WHO (1996), which recommended that discharges in the direct vicinity of Posidonia beds should be avoided whenever possible.

The present study underlines the use of epiphytic microalgae assemblages in seagrass ecosystems as general indicators of anthropogenic disturbance to seagrass meadows. It was clear that the effect of nutrient loading and turbidity could be depicted in the community structure and diversities of seagrass-associated epiphytes (Balata et al. 2008, Ben Brahim et al. 2010, Mabrouk et al. 2013, this study). During the period of study (August), nutrient enrichment (mainly Si(OH)₄ and NH₄⁺) and turbidity were associated with an increase in abundance of dinoflagellates , especially Protoperidiniales and Prorocentrales. The results of this survey can be used to compare with other regions in the Mediterranean subjected to same conditions where we can observe the deterioration in vitality of Posidonia oceanica meadow and increases in dinoflagellates.