A phytosociological survey of the Corynephorus canescens (L.) Beauv. communities of Italy

Abstract In Italy, Corynephorus communities are distributed along the medium course of the Ticino river and Sesia river and the internal sand dunes of Lomellina (through the Vercelli, Novara and Pavia provinces); these stations represent the southern limit of European distribution of this habitat. A phytosociological study was carried out to gain better knowledge of their composition; of their affinity or diversity against the central European communities; of their distribution and of the main threats to their conservation. Original and literature relevés (114) were elaborated producing a cluster analysis; correspondence analysis (CA), principal component analysis and Kruskal–Wallis test were carried on to characterize the clusters of relevés taking into consideration biological forms, chorological groups, Ellenberg indicator values and floristic groups. Italian Corynephorus communities can be attributed to the following syntaxa: Spergulo vernalis-Corynephoretum canescentis, Spergulo vernalis-Corynephoretum canescentis cladonietosum, Spergulo vernalis-Corynephoretum canescentis silenetosum nutantis and Spergulo vernalis-Corynephoretum canescentis artemisietosum campestris. Italian Corynephorus communities are included in the Habitat 2330 of the EU Habitat Directive. They are threatened by different factors (such as restricted areas of occurrence, alien plant invasion and natural dynamics) and they need to be managed if we want to conserve them.

Its native distribution occupies most of European central and western lowlands, from Ukraine and Belarus in the east, through all countries bordering the southern Baltic, southern North Sea and the Atlantic Ocean as far as south Portugal (Blunt 2006). The Carpathians and Alps act as a general boundary to the species range extension in the south, with the exception of an outlying population in the Hungarian plain between the Danube and Tisza rivers (Rychnovská 1963). The C. canescens also reaches the French Mediterranean coast of Languedoc, but is absent from most of Spain except for a spur in central Castille. Outside this principal range, there are scattered coastal populations in southern Norway and Sweden; the Eastern Baltic as far north as Riga and in Romania and northern Marocco. Blunt (2006) describes its distribution in Poland. Within the British Isles, C. canescens is chiefly a plant of coastal dunes and tidal beaches; it is very local on and near coasts of Jersey, E and W Suffolk, E and W Norfolk, S Lancs, rare inland in E Suffolk, probably introduced on coasts of Moray, Westerness and E Lothian, and naturalized in Staffs and Worcs (Stace 2010).
Another outlying population is in Italy in the north-western part of the Po River Plain; observations were made by Bertossi (1950), Corbetta (1968) and Assini (2007) on the typical inland fossil dunes called ''dossi'' in the area of the Lomellina (Lombardy Region, Pavia Province).
Recently, C. canescens has been subjected to the IUCN (International Union for Conservation of Nature) assessment procedure (Assini & Abeli, in press) and new sites were recorded (Varese et al. 2010) along the Sesia river and the Ticino river (Piedmont Region). According to this assessment, the grey hair-grass status is endangered in Italy. Where present C. canescens often dominates dry acidic grasslands, which are a declining European habitat designated for conservation under the EU Habitat Directive (Habitat 2330 -Inland dunes with open Corynephorus and Agrostis grasslands and Habitat 2340* -Pannonic inland dunes). Italian Corynephorus communities are included in the Habitat 2330. In Germany C. canescens grasslands are currently more endangered than bogs or calcareous grasslands (Jentsch & Beyschlag 2003). Corynephorus communities are very rare and endangered in Italy; thus, a phytosociological study was carried out with the aim to classify and describe the Italian communities, gaining better knowledge of: their vascular species composition; similarity or dissimilarity from some Corynephorus syntaxa quoted in the main vegetation prodromes of central Europe; and distribution and main threats to their conservation.

Study sites
The study area comprises almost all the sites where C. canescens is recorded in Italy; they are localized north of the Po river, in the west part of the Po Plain ( Figure 1). Twenty-nine sites were studied (Supplement 1).

Vegetation data
The data for this study comprise original relevés and literature relevés. Thirty-four original phytosociological relevés (Pott 2011) were carried out during the spring-summer of 2009 using the Braun-Blanquet scale (1928). The size of the sampling plots was comprised between 5 and 15 m 2 , and it is consistent with the plot sizes discussed by Chytrý and Otýpková (2003) about low terrestrial vegetation and most types of herbaceous vegetation.
We only considered the cover-abundance of the lichen and moss strata in each relevés because no moss and lichen specialists were available while carrying out relevés. This is consistent with the fact that in our opinion and in literature the species of mosses and/or lichens do not play a basic role in the cluster analysis and in the syntaxonomical classification at association level. As stated by Bü ltmann (2005), together with bryophytes, lichens form syntaxonomically and often ecologically independent units, the microcoena. Furthermore, it is true that the lichen and moss composition is very important in the pioneer stages of the Spergulo-Corynephoretum and that such composition can change in relation to the habitat factors (according to Bü ltmann 2005, terricolous lichens and their microcommunities are sensitive indicators of microhabitat factors), but it also true that only species quantities change as response to   (Hasse & Daniëls 2006). According to Berg and Dengler (2005), only three moss species have diagnostic value at class and alliance levels in the Koelerio-Corynephoretea vegetation of Mecklenburg-Vorpommern, while many more lichen species have a diagnostic value for the same vegetation class. For this reason, lichen samples were carried out and classified in laboratory for a qualitative description of the lichen stratum of our Corynephorus communities.

Species taxonomy and nomenclature
Nomenclature of species which are present in Italy follows Conti et al. (2005); nomenclature of species which are present in Europe but not in Italy follows Flora Europaea (Tutin et al. 1964(Tutin et al. -1993. Nomenclature of syntaxa follows Mucina et al. (1993), Oberdorfer (1978) and Pott (1992). Nomenclature of lichens follows Nimis and Martellos (2008).

Data analysis
A table with a total number of 114 relevés was built.
Because we considered only the cover of lichen and moss strata, for each European relevé the highest cover-abundance value showed by one species of lichen was attributed to the lichen stratum; if two or three lichen species showed the highest value, than we increased the cover-abundance by one point. For example, if only one species of lichen showed the highest value of 2, we attributed this value to the lichen stratum; if two or three lichen species showed the highest value of 2, we assigned the value 3 to the lichen stratum. We did the same with moss species.
We did not include species with frequency of only 1-2 for analyses.
A matrix of 114 relevés6102 species was elaborated using the SYN-TAX package (Podani 1993).
A cluster analysis was performed using quantitative values (combined transformation of Van der Maarel 1979), similarity ratio and average link (UPGMA -Unweighted Pair Group Method with Arithmetic Mean).
Ellenberg indicator values (Ellenberg et al. 1992), readapted by Pignatti (2005), were calculated for all relevés taking into account the cover values (weighted mean); mean values were used as variables in a principal component analysis (PCA) based on a variance-covariance matrix, in order to show the ecological difference among the surveys. The PCA was performed using the PAST package (Hammer et al. 2001).
Biological and chorological spectra, based on species frequency, were performed to compare the described vegetation types. Life forms and chorological elements for the species present in Italy follow Pignatti (2005) and Poldini (1991); life forms and chorological elements for the species not present in Italy follow Oberdorfer (2001 Kruskal-Wallis test was performed to test the significance among the values of the biological forms, chorological types and ecological factors of the identified cluster vegetation. Correspondence analysis was performed on three matrixes: -a 5 6 8 matrix in which the eight vegetation clusters were described by the ratios between the total number of biological form occurrences in each cluster and the number of relevés in the cluster; -a 11 6 8 matrix in which the eight vegetation clusters were described by the ratios between the total number of chorological type occurrences in each cluster and the number of relevés in the cluster; Aliens were excluded from this elaboration because they does not really reflect the phytogeography of the communities, but they can be interpreted as an effect of disturbance and human pressure; -a 8 6 8 matrix in which the eight vegetation clusters were described by the ratios between the total number of species occurrences of each species group in each vegetation cluster and the number of relevés in each cluster (eight groups of species linked to the eight vegetation clusters have been identified taking into consideration both their syntaxonomical meaning in the previous literature and the differential role played in the analysed vegetation).
Kruskal-Wallis test and CA were performed by means of the PAST package (Hammer et al. 2001).

Cluster analysis and syntaxonomy
Dendrogram resulted from cluster analysis (Supplement 2) showed eight clusters with nine relevés isolated and linked to them at low values of similarity; thus, they were considered atypical and excluded from the following syntaxonomical considerations.

Environmental conditions
Biological and chorological spectra Figure 3 shows the life forms (expressed as percentage data) in the different Corynephorus communities. Particularly, Spergulo-Corynephoretum is the richest in Therophytes (almost 50%), while Armerio-Festucetum Corynephorus subass. and Artemisio-Corynephoretum are the richest in Hemicryptophytes (more than 70%) and the poorest in Therophytes (less than 10%), with a discrete presence of Chamaephytes (comprised between 10% and 20%). The Therophyte percentage in the other communities is comprised between 20 and 30, while the Hemicryptophytes percentage is comprised between 50 and 70, with the exception of Thymo-Corynephoretum which shows a Hemicryptophyte percentage lower than 40, but the highest percentage of Geophytes (15). In this last community as in Spergulo-Corynephoretum artemisietosum, the Chamaephyte percentage is discrete (more than 15). In all Corynephorus communities Nanopha-nerophytes and Phanerophytes are scarce, with percentage generally lower than 5. If we consider only the two Italian communities, we can observe that Spergulo-Corynephoretum artemisietosum is very rich in Aliens and show a higher percentage of Circumboreal elements than Spergulo-Corynephoretum silenetosum. This last community is characterized by a high percentage of Orophytes (10).

Corynephorus communities in Italy 69
determining the variability between the cluster vegetation and confirm the differences between them. Only the Terophyte and Nanophanerophyte/ Phanerophyte biological forms and the Subatlantic chorological type are not significant. Figure 4 shows the CA results from biological forms and vegetation clusters matrix; it points out the linkage between biological forms and vegetation clusters, with: Thymo-Corynephoretum (6) particularly characterized by the Geophytes; Artemisio-Corynephoretum (7) and Armerio-Festucetum Corynephorus subass. (8) particularly characterized by the Chamaephytes; Spergulo-Corynephoretum artemisietosum (5) and Spergulo-Corynephoretum silenetosum (2) particularly characterized by the Hemicryptophytes; and Spergulo-Corynephoretum (3) particularly characterized by the Terophytes. Spergulo-Corynephoretum cladonietosum (1) and Spergulo-Corynephoretum festucetosum (4) are characterized by both the Terophytes and the Hemicryptophytes. Figure 4 also gives suggestions for the interpretation of the dynamic processes in which C. canescens communities are involved. The most pioneer community (Spergulo-Corynephoretum) is placed on the right of the plot; the most evolved communities, both from central Europe (Armerio-Festucetum Corynephorus subass. and Artemisio campestris-Corynephoretum canescentis) and from Italy (Spergulo-Corynephoretum artemisietosum and Spergulo-Corynephoretum silenetosum), are placed on the left. Spergulo-Corynephoretum festucetosum, Spergulo-Corynephoretum cladonietosum and Thymo-Corynephoretum are placed among the most pioneer community and the most evolved ones. Figure 5 shows the CA results from chorological types and vegetation clusters matrix; it points out the linkage between chorological types and vegetation clusters, with: Artemisio-Corynephoretum (7) and Armerio-Festucetum Corynephorus subass. (8) particularly characterized by Circumboreal, Eurosibiric and Pontic-Continental; Thymo-Corynephoretum (6) particularly characterized by Eurasiatic and European; Spergulo-Corynephoretum artemisietosum (5) particularly characterized by Paleotemperate and Eurimediterranean; Spergulo-Corynephoretum silenetosum (2) particularly characterized by Paleotemperate, Eurimediterranean and Orophytes; and Spergulo-Corynephoretum (3) particularly characterized by Eurimed-Subatlantic. Spergulo-Corynephoretum cladonietosum (1) and Spergulo-Corynephoretum festucetosum (4) result more heterogeneous without a particular chorological type characterizing them. Figure 6 shows the CA results from species groups and vegetation clusters matrix; it points out the linkage between species groups and vegetation clusters and also gives suggestions about the interpretation of the dynamic processes in which C. canescens communities are involved. The pioneer communities (Spergulo-Corynephoretum, Spergulo-Corynephoretum cladonietosum) and their differential species groups are placed on the right of the plot; the most evolved communities, both from Italy (Spergulo-Corynephoretum artemisietosum) and from central Europe (Armerio-Festucetum trachyphyllae, Artemisio campestris-Corynephoretum canescentis), are placed on the left together with their differential species. Spergulo-Corynephoretum silenetosum, Spergulo-Corynephoretum festucetosum and Thymo-Corynephoretum are placed among the pioneer and the most evolved communities. Table III shows the synoptic table of Corynephorus communities, indicating the percentage presence of the different species groups in the vegetation clusters.

Dynamics interpretation of the Italian Corynephorus communities
The successional trajectories for plant communities of acidic grasslands mainly depend on climatic conditions, soil pH, water and nutrient availability, disturbance regimes, and nearby seed pools (Jentsch & Beyschlag 2003). Stability and successional behaviour of plant communities on inland sand dunes are insufficiently known (Jentsch & Beyschlag    (Figure 7). Initially, the bare sand substrate is loosely covered by pioneer species such as C. canescens, Teesdalia nudicaulis, Rumex acetosella or Filago minima (Spergulo vernalis-Corynephoretum canescentis). These pioneers begin to fix the sand. As soon as the surface is consolidated, mosses and lichens, mainly of the genus Cladonia, may form a crust (Spergulo vernalis-Corynephoretum canescentis cladonietosum). In this stage, small soil disturbances (trampling, animals or others) provide open patches with bare substrate, which seem crucial for further seedling establishment of phanerogams (Jentsch & Beyschlag 2003  Corynephorus communities in Italy 9 72 S. Assini et al.

Threats to Corynephorus communities
As stated by Janišova et al. (2011), while in some cases dry grasslands are directly destroyed by building activities or mining, the more serious threats are those affecting vast areas, namely agricultural intensification, land abandonment and atmospheric nitrogen input. At local level, conservation and protection of Italian Corynephorus communities present different problems. Only places along the Ticino river and the Sesia river (Greggio) are included in protected areas (respectively, Ticino Park and Sesia Park). However, because of the insufficient knowledge, their management and monitoring have never been considered; thus, these communities are abandoned to succession.
In many areas, the principal factors that threaten their conservation are different. The surface area occupied by the community is often very limited (520 m 2 ). In many sites, the typical floristic composition is threatened by alien plant invasion. Different plants are involved in different sites: along the medium-low course of the Sesia, Eragrostis curvula (Schrad.) Nees is often abundant; in many In all these cases, the establishment of forests with which Corynephorus communities can be in contact and the natural dynamics of the Corynephorus communities threaten Corynephorus communities if there is no management action.
The presence of alloctonous fauna could also be a problem, and this needs to be better studied. In many Italian sites, Sylvilagus floridanus (Allen 1890) is abundant and replaces the native wild rabbit. According to Jentsch et al. (2002), wild rabbit represents an important factor in the persistence of Corynephorus communities because of the disturbance from digging galleries. The same is still not demonstrated for Sylvilagus which shows more reduced disturbance activity and population explosions producing faeces accumulations that could increase soil nutrient availability and thus accelerate natural vegetation dynamic.

Conclusions
The Italian stations represent the extreme southern border of European distribution of Corynephorus communities, and this is the reason for which Italian communities lack typical central European species (such as Spergula morisonii, Mibora minima, Carex arenaria and H. arenarium) and can be differentiated by unique species combinations.
Italian Corynephorus communities can be attributed to the following syntaxa: Spergulo vernalis-Corynephoretum canescentis, Spergulo vernalis-Coryne-phoretum canescentis cladonietosum, Spergulo vernalis-Corynephoretum canescentis silenetosum nutantis and Spergulo vernalis-Corynephoretum canescentis artemisietosum campestris. The first two communities represent pioneer stages, while the last two ones represent more mature stages. Particularly Spergulo-Corynephoretum silenetosum and Spergulo-Corynephoretum artemisietosum seem to be exclusive of Northern Italy and are more rich of thermophilous elements than the most part of European communities; as seen before they are well distinguishable chorologically and ecologically.
Between the analysed European communities, the interpretation of Thymo angustifolii-Corynephoretum is not clear and should be studied in depth: Mucina et al. (1993) consider the association a geographical vicarious of the Spergulo-Corynephoretum, while Chytrý (2007) considers it as synonymous of the Spergulo-Corynephoretum. Our results show this association to be different from the Spergulo-Corynephoretum on the basis of the biological forms and chorological elements; it also results well distinguishable ecologically.
Italian Corynephorus communities, as the other dry acidic grasslands of Europe, need to be managed if we want to conserve them.
Literature references dedicated to the management and conservation of the Festuco-Brometea dry grasslands are numerous (just recently Hegedü šová & Senko 2011; Janišova et al. 2011;Vassilev et al. 2011), while those dedicated to the management of the Koelerio-Corynephoretea grasslands are scarce.
Anyway, suggested management actions for Italian Corynephorus communities should include: a small-scale mechanical ground disturbance, the enhancement of Corynephorus populations and the invasive plant control.
As stated by Jentsch (2004), mechanical soil disturbances may cause mobility and redistribution of soil nutrients and induce germination of seeds stored in the soil seed bank. Small-scale soil disturbances also create bare sandy substrates and provide safe sites for novel seedling establishment in otherwise closed vegetation with high proportions of cryptogams; they especially maintain a niche for short-lived species and permit continual, low-density recruitment of perennials. For these reasons, we are experimenting a mechanical ground disturbance consisting in a ploughing of the upper soil layer (first 5 cm) with removal of the cryptogam crust in five plots at Cergnago (PV), comparing them with five plots not disturbed. The experiment begun in 2009, but a long-term monitoring (almost for 6-10 years) will be necessary before having interesting results.
In 2005, we have also introduced C. canescens in a Nature 2000 site (IT2080008, Boschetto di Scaldasole, Lombardy region, Pavia province) which is a remnant of an inland sand dune where the grey hairgrass was not present. We are monitoring the plant, but also in this case a long-term monitoring will be necessary to test the efficiency of our introduction.
About the invasive plant control, we have not experiments/works in progress, but we think that the control of almost the invasive woody individuals (particularly of the following species: Robinia pseudacacia, P. serotina, A. altissima) should be possible by means of punctual treatments, also chemical, in Corynephorus communities where invasive trees are still sporadic and scattered. Chemical treatments are of course not suggested where the density of invasive plants is high.
Also for the control of invasive herbaceous species, we have no works in progress, but we think that studies on the ecology and functional traits (Ordonez et al. 2010) of invasive herb plants should be realized to better understand their invasion success.
Furthermore, the knowledge level of lichens and mosses (dynamism, ecology and conservation value) should also be increased for better suggestions about the conservation and restoration of the Italian Corynephorus communities.