The floodplain woods of Tuscany: towards a phytosociological synthesis

The recent Landscape Plan of Tuscany Region pays particular attention towards the floodplain woods, for their conservation concern, landscape, and historical importance. The floodplain forest vegetation is defined as the natural and semi-natural wood vegetation located close to the rivers and submerged only during exceptional flooding. We gathered 180 both published and unpublished relevés coming from Tuscany and carried out in floodplain woods, where Alnus glutinosa , Fraxinus angustifolia subsp. oxycarpa , Quercus robur , Ulmus minor , Carpinus betulus , Populus nigra , P. alba and Salix alba were dominant, alone or in consociation with each other. On this dataset we performed a multivariate analysis, and the resulting groups were characterized by several points of view: floristic, ecological, chorological, by mean of diagnostic species, with the use of EIV. Moreover, their distribution in Tuscany was better defined. According to our results, six associations were found to be present in Tuscany. Thereby, one new association and three new subassociations were proposed. The syntaxonomic arrangement above the association level was discussed, with particular attention to the Italian Fraxinus angustifolia subsp. oxycarpa -rich communities. Finally, a comparison with the Annex I habitat types (sensu Directive 92/43/EEC) was carried out.


Introduction
Floodplain forests are considered one of the most widespread forest communities in Europe (Schnitler et al. 2007) and represent an important biodiversity hotspot (Ward et al. 1999;Geilen et al. 2004). Unfortunately, they are severely endangered ecosystems, threatened by land-use changes. In fact, the shrinking of alluvial forests throughout Europe began in remote times and then strongly intensified in the last two centuries, due to agriculture and urbanization; this phenomenon strongly af-fected most areas of the floodplains, leading to the process of "insularization" and fragmentation of the remaining forests. Finally, the growing number of invasive alien species such as Robinia pseudoacacia L., Ailanthus altissima (L.) Swingle, Acer negundo L. and Amorpha fruticosa L. represents an additional risk today (Schnitler et al. 2007;Lazzaro et al. 2020;Viciani et al. 2020).
For these reasons, the recent landscape plan of the Tuscany Administrative Region identifies the conservation and management of "alluvial soils", as one of the main concerns for conservation (Marson 2016), in order to respect the Code of Cultural Heritage and Landscape, which was approved as a regional law in the 2015 by the Tuscany Regional Council within the new Landscape Plan. The plan integrates the three main components of the landscape: aesthetic-perceptive (aesthetic values), ecological (environmental values of the landscape) and structural features (relationships between cultural and natural aspects structured over time).
According to many authors (e.g. Ellenberg1988; Bernetti 2005; Douda et al. 2016), alluvial forests include the vegetation established at a specific distance from the rivers and submerged only during exceptional floods. The soil of such forests is generally deep, with the water level frequently reaching roots of plants. The vegetation that occurs in these ecological conditions can be defined as azonal (Géhu 2006), as it is linked to particular geomorphological and edaphic conditions present in many climatic types, rather than only in a specific climate (Bernetti 2005;Blasi 2010). Alluvial forests are widespread across the continent, covering large areas in central and northern Europe, while representing only relic communities in southern Europe and the Mediterranean. They are dominated by deciduous broadleaves: the black alder (Alnus glutinosa (L.) Gartner) in the marshy areas and the English oak (Quercus robur L.) in the typical alluvial plains; in our region we also found the elm (Ulmus minor Mill.) and southern ash (Fraxinus angustifolia Vahl subsp. oxycarpa (Willd.) Franco & Rocha Afonso). In suburban areas, where the influence of river dynamics is severely limited (e.g. dead meanders), poplar coenoses (Populus nigra L., P. alba L. and P. canescens (Aiton) Sm.), with elm, and relict stands of white willow (Salix alba L.) can be found. According to Biondi and Blasi (2015), Douda et al. (2016) and also recent Italian works (e.g. Spampinato et al. 2019), the European alluvial plain forests can be as attributed to two classes: Alnetea glutinosae Br.-Bl. and Tx. ex Westhoff et al., 1946, for the swampy woods, and Salici purpureae-Populetea nigrae Rivas-Martínez & Cantó ex Rivas-Martínez, Báscones, TE Díaz, Fernández-González ex Loidi, 2001, for the periodically flooded riparian forests. On the contrary, Mucina et al. (2016) do not consider as an appropriate solution the use of the class Salici purpureae-Populetea nigrae and prefer to attribute part of the alluvial forests to Alno glutinosae-Populetea albae P.
Fukarek ex Fabijanić, 1968, because the concept of Salici purpureae-Populetea nigrae (as presented in the original diagnosis) also includes the Salicetalia purpureae order, that is the type of the Salicetea purpureae, i.e. a shrubby riparian class. We prefer to use Alno glutinosae-Populetea albae, in accordance with some recent Italian literature (e.g. Poldini et al., 2020), in order to include only riparian forest vegetation.
The syntaxonomic scheme below the class rank, especially at alliance level, is very intricate and a plethora of classification schemes have been proposed, often based on very few and local surveys.
With this study, we aim to contribute to the phytosociological knowledge of the alluvial forests of Tuscany, including the true swampy forests dominated by Alnus glutinosa and/or Fraxinus angustifolia subsp. oxycarpa, the suburban forests with Populus sp.pl., Salix alba and/or Ulmus minor, the hygrophilous and meso-hygrophilous woods with Quercus robur and the meso-hygrophilous forests dominated by Carpinus betulus.

Matherials and Methods
In order to define and identify in the field floodplain woods, we carried out several literature and field investigations which also led to the publication of an updated map for Tuscany (Gennai et al. 2020). The dataset was extracted from published and unpublished relevés stored in the database of the Laboratory of Phytogeography (Dept. of Biology, University of Florence). We firstly selected the relevés in which the tree species Alnus glutinosa, Fraxinus angustifolia subsp. oxycarpa, Quercus robur, Ulmus minor, Populus alba, P. nigra, P. cancescens, Salix alba and Carpinus betulus, alone or in association with each other, determined at least the 75% of the canopy cover. In Appendix I we provided references of literature used for the selected relevés. In Appendix II we reported the correspondence between numbers of relevés in original reference and numbers attributed in our tables. We excluded the strictly linear riparian formations. To the obtained data-set, 25 unpublished relevés, surveyed following the classical approach of the Zürich-Montpellier school and further updates (Braun-Blanquet 1964;Dengler et al. 2005Dengler et al. , 2008Biondi 2011), were added. These relevés were carried out in sites at a maximum altitude of 200 m a.s.l. and with maximum inclination of 2°, where the larger parts of the Tuscan floodplain woods are distributed (see Gennai et al. 2020). The final data-set comprised 180 relevés with a total of 425 species. We considered as sporadic the species occurring in less than four relevés and with cover-abundance values ≤ 1; such species were excluded from the numerical analyses, reducing the total number of species to 152. The Braun-Blanquet cover-abundance scale was transformed according to the ordinal scale proposed by Van der Maarel (1979) and Noest et al. (1989): r = 1, + = 2, 1 = 3, 2 = 5, 3 = 7, 4 = 8, 5 = 9.
The matrix was numerically classified by mean of Cluster Analysis using Ward's method and by mean of NMDS analysis with the Bray-Curtis method as similarity measure; the analyses were performed using PAST software (Hammer et al. 2001).
The diagnostic species were statistically defined by the Phi coefficient of association (Chytrý et al. 2002) and its significance was calculated through a Fisher test. We considered a species as diagnostic of each group if phi > 0,20, with p < 0.01 (Douda et al. 2016).
A simplified high-level chorotype was attributed to each species, following Pignatti (2005).
The Ellenberg Indicators Values (Ellenberg et al. 1992) were attributed to species according to Pignatti (2005). We considered the following parameters: Light (L), Tem- perature (T), Continentality (C), Soil Moisture (U), Soil pH (R) and Nutrients (N). For each relevé, the total indicator value was calculated using the weighted averages of the presence/absence data of the species recorded in the relevé (except for sporadic species). To investigate possible different EIV characters of relevés and groups, the EIV vectors were passively projected onto NMDS ordination.
For the analysis of syntaxonomic data we referred to local and international literature separately cited in each paragraph. In particular, for the woods dominated by Fraxinus angustifolia subsp. oxycarpa we referred to the recent work by Poldini and Sburlino (2018), and we compared their outcomes with our results, through a NMDS analysis and a synoptic table.
Vascular plant species names follow the Portal to the Flora of Italy (2020) and Pignatti (2017Pignatti ( -2019. In the descriptions of syntaxa, we indicated as diagnostic the characteristic and differential species reported by the original authors and the species highlighted by the phi analysis. In proper terms, the species indicated by the original authors as characteristics must be mainly considered as differentials (Géhu 2006).
In the syntaxonomic tables, we also indicated the diagnostic species as reported by the original authors of the syntaxa at the association level, and the phi values of the species (when phi> 20) resulting from the Phi analysis. The attribution of species to Alnetea, Phragmito-Magnocaricetea and Alno-Populetea follows Poldini and Sburlino (2018) and Poldini et al. (2020).

Results and Discussion
Both the cluster analysis dendrogram (Suppl. material 1: Figure S1) and NMDS results (Fig. 1) confirmed a clear separation between the relevés respectively dominated or co-dominated by Quercus robur and Carpinus betulus on one side, and Fraxinus angustifolia subsp. oxycarpa, Alnus glutinosa, Ulmus minor and Populus spp./Salix alba on the other side. The Cluster Analysis dendrogram (Suppl. material 1: Figure S1) shows that the first separation divides the mesohygrophylous woods dominated by Quercus robur from the other ones. As regards the NMDS ordination, he swampy forests with Alnus glutinosa (A) can be found in the first quadrant, while the linear woods of the interdune wet areas (locally named "lame") with Fraxinus  angustifolia subsp. oxycarpa (F) are placed in the second quadrant. The mesohygrophylous forests with Quercus robur (Q) and those dominated by Carpinus betulus (C) are in the third quadrant, while the thickets with Ulmus minor (U) and the woods with poplars and/or white willow (P) can be mainly found in the fourth quadrant.
The resulting EIV vectors projected onto NMDS ordination ( Fig. 1) show that the woods dominated by Fraxinus angustifolia subsp. oxycarpa and Alnus glutinosa are positively influenced by soil moisture (U), light (L), continentality (C), and negatively influenced (but to a little extent), by temperature (T), while the woods dominated by Quercus robur and Carpinus betulus show an opposite trend. Nutrients (N) and soil pH (R) seem to affect the ordination to a lesser extent.
Statistical and floristic analyses allowed to recognize six different groups, interpreted as different vegetation types and analyzed through Phi analysis. Diagnostic species of these six groups, together with their frequencies, were reported in Table 1. Diagnostic species were indicated in light grey (phi > 20) and dark grey shading (phi > 30).
The results of the chorological analysis of the vegetation types are reported in Fig. 2.
The distribution in Tuscany of the six types of floodplain forests is shown in Fig. 3.
According to the results of our analysis and on the basis of the literature taken into account, we propose the following syntaxonomic scheme. In general terms, we confirmed the associations already reported for Tuscany, with the exception of some minor syntaxonomic formal changes, and the increase in the diagnostic species due to the increase in the number of relevés considered.
Distribution in Tuscany (Fig. 3): These woods were found in the north-western Tuscany, especially in the plains of Serchio and Arno rivers: Selva di San Rossore (Gellini et al.1986), Macchia lucchese , Cerbaie (Arrigoni 1997), Tenuta di San Rossore , Tenuta di Migliarino (Sani et al. 2011), Lago di Massaciuccoli Viciani et al. 2017), Lago di Porta (Lastrucci et al. 2016). Syntaxonomy: the communities dominated by Alnus glutinosa can be referred to the association described by  for the Selva di San Rossore (PI), and Ecology and Chorology: Fraxinus angustifolia subsp. oxycarpa can tolerate short periods of moderate water deficit and low concentrations of salt in soil waters. In general terms, this species is more Mediterranean, relatively more thermophilous than Alnus glutinosa. From a chorological viewpoint, these woods are firstly characterized by the European-Mediterranean and Eurasian chorotypes (Fig. 2). According to the EIV analysis ( Fig.  1), these woods are positively related to Soil Moisture (U) and Light (L).
Syntaxonomy:  referred this type of woods to the Carici remotae-Fraxinetum oxycarpae Pedrotti corr. Pedrotti 1993, a riparian forest association. The woods dominated by Fraxinus angustifolia subsp. oxycarpa here presented, instead, are clearly characterized by a higher presence of swamp species with respect to riverine Carici remotae-Fraxinetum oxycarpae coenoses. For example, species like Stachys sylvatica, Symphytum tuberosum, Ranunculus lanuginosus are not present ), while several Carex species were frequently found.  suggested the establishment of a new subassociation, Carici-Fraxinetum oxycarpae alnetosum glutinosae, but the name is invalid on the basis of Art. 5 of the ICPN (Theurillat et al. 2021). Furthermore, the two relevés attributed to this subassociation were included in the Hydrocotylo-Alnetum table, therefore, we decided to not validate the subassociation. However, it must be noted that the relevés 33-47 of Table 3, where Alnus glutinosa has high cover values, show a transition to the swampy woods of Hydrocotylo-Alnetum.
Fraxinus angustifolia subsp. oxycarpa, in Tuscany, has broad ecological needs; according to Mondino and Bernetti (1998) it can be found in swamp areas with Alnus glutinosa and Quercus robur, riparian habitats with Ulmus minor and, finally, in hilly meso-hygrophilous conditions with Quercus cerris (Scoppola and Filesi 1995;Terzi et al. 2020). Marshy woods are often character-

Table 2.
Continuation.  5 5 5 5 5 5 5 5 4 4 4 3 5 5 5 4 5 5 5 5 5 5 5 5 5 5 5 4 4 5 4 5 3 3 3 5 5 5 4 4 4 4 4 3 3 2 2 24 Periploca graeca L. ized by a vegetation mosaic of forest stands with Juncetalia maritimi, Magnocaricion elatae and Phragmition. A similar situation is reported for the Bosco della Mesola site, in North-Eastern Italian Adriatic coast (Piccoli et al. 1983;Piccoli and Gerdol 1984;Gerdol et al. 2018). In in a similar landscape context, for the Tuscan north-western Thyrrenian coast,  proposed the new association Carici elatae-Fraxinetum oxycarpae, not validly published according to the Art. 5 of the ICPN (Theurillat et al. 2021). Moreover, it was based on only two relevés that, in our opinion, are not representative of a wood, being the cover value of the tree layer very low. For these reasons, we decided not to validate this name and we here propose to attribute these coenoses to a new association named Limnirido pseudacori-Fraxinetum oxycarpae, which is characterized by the presence of a high contingent of Phragmiti-Magnocariceta species compared to other similar associations, i.e. Fraxino-Quercetum roboris observed for the low interdune swamp areas of Migliarino-San Rossore by , Carici remotae-Fraxinetum oxycarpae reported by Pedrotti ( , 1993, , Mondino and Bernetti (1998), Arrigoni (1998), Mercadal and Vilar (2013) and, finally, Alno-Fraxinetum oxycarpae (Arrigoni 1998). The floristic differences between the other associations dominated by Fraxinus angustifolia subsp. oxycarpa described for Italy (Tables 1 and 2 in Poldini and Sburlino 2018) and the similar coenoses detected in the Tuscan floodplain woods are displayed in a synoptic table (Table 4) and in the dedicated NMDS ordination (Fig. 4). Table 4 shows that columns on the left side (columns 1-12) can be assigned to the swampy alliance Frangulo alni-Fraxinion, while those at the right side (columns 13-20) can be attributed to the riparian alliance Carici remotae-Fraxinion (Poldini and Sburlino 2018). According to this scheme, we refer the association Limnirido-Fraxinetum oxycarpae to the alliance Frangulo-Fraxinion. This interpretation was also confirmed by the cluster analysis (Suppl. material 1: Figure S1) through which the riparian woods are clustered together and separated from the swampy woods. Even if these coenoses also host a high rate of Alno-Populetea and Populetalia species (Table 3), the attribution of this new association to the Alnetea and Alnetalia is also justified for its ecological position, as it is typically located in the interdune swamps. Also, the Fraxinus angustifolia subsp. oxycarpa communities recently indicated for Corsica (Gauberville et al. 2018) as Carici remotae-Fraxinetum can be probably attributed to Limnirido-Fraxinetum, considering their floristic composition and ecology.
In the end, we note that in Tuscany the distinction between the alliances Alnion glutinosae and Frangulo-Fraxinion is not so clear, as many diagnostic species are in common, and a comprehensive work devoted to clarify this problem is needed.
FRAXINO OXYCARPAE-QUERCETUM ROBORIS Gellini, Pedrotti ex Venanzoni, 1986. Holotypus: Table 3, rel. 14 in Gellini, Pedrotti ex Venanzoni, 1986 (corresponding to rel. n. 40 in Suppl. material 2: Ecology and Chorology: Meso-hygrophilous, Mediterranean woods developing on fertile and deep soils. The chorotypes spectrum is very similar to that of the Fraxinus angustifolia subsp. oxycarpa woods, apart for a slight increase in European species (Figure 2). These communities were found in the more elevated areas of the swamps, in contact with the climatic woods, which act as suppliers of many euryoecious species, tolerating seasonal variations of hygromorphic conditions. According to the EIV analysis (Fig. 1), these woods are positively related to Temperature (T), and negatively to Soil Moisture (U), Continentality (C) and Light (L).
Syntaxonomy: English oak forests in Tuscany were originally described by  for the Selva di San Rossore and referred to the association Fraxino oxycarpae-Quercetum roboris . The presence of the association was then confirmed by Mondino and Bernetti (1998) and Arrigoni (1998) for the same areas. The association can be divided into four subassociations (Suppl. material 2: Table S1): 1. subass. typicum Rels 1-45 in Suppl. material 2: Table  S1; it represents the typical aspects of the association (holotypus -the same of the association - Table 3, rel. 14 in    ; 2: Tuscany, Tenuta di San Rossore ; 3: Tuscany, Selva di San Rossore ; 4: Tuscany, Tenuta di Migliarino (Sani et al., 2011); 5: Tuscany, Bosco dell'Ulivo ; 6: Tuscany, Tenuta del Tombolo di Pisa (AAVV, 2005); 7-20: from tables 1 and 2 by Poldini and Sburlino (2018)     Mesophilous woods can be found in sites where submersion is very occasional. This is a community with a more evident C-European character, highlighted by the lowest percentage of Mediterranean species and the highest percentages of European and European-Mediterranean ones (Fig. 2). According to the EIV analysis (Fig. 1), the woods dominated by Carpinus betulus show ecological requirements similar to those of Quercus robur communities, except for a slightly prefer-ence for lower temperature.
Distribution in Tuscany (Fig. 3): Cerbaie (Arrigoni 1997(Arrigoni , 1998. Syntaxonomy: The association Asparago tenuifolii-Carpinetum betulii was described by Arrigoni (1997) for the Cerbaie hills, in north-western Tuscany. Arrigoni (1997; Table 2) reported two different associations: Polygonato multiflori-Quercetum roboris Sartori 1984(Sartori 1984Brullo and Spampinato 1999) and the new association Asparago tenuifolii-Carpinetum betulii (Arrigoni 1997). According to our analysis, and also to some important floristic differences (i.e. the lack of Convallaria majalis, diagnostic for Polygonato-Quercetum roboris, see Brullo and Spampinato 1999), the totality of the relevés in question can be assigned to the association Asparago tenuifolii-Carpinetum betulii and the association Polygonato-Quercetum roboris is to be excluded from Tuscany. These forests were found in the small plain sites of the lower parts of the Cerbaie hills. This association was assigned to the Crataego-Quercion and could be considered the wettest association of this alliance. Vagge ex Biondi, 1999. Holotypus: table 20, rel. 3 in Vagge and -Elm interdune groves (   Poldini and Sburlino (2018) with the exception of Lim-Fx, data from this paper.

Ecology and Chorology:
Recently established woods, located on wet clay soils rich in nutrients. This is the most typical Mediterranean vegetation type among those investigated: here we found the highest percentage of Mediterranean species (Figure 2). According to the EIV analysis (Fig. 1), the groves with Ulmus minor show a negative relation to Soil Moisture (U).
Distribution in Tuscany (Fig. 3): This vegetation type was found in the coastal systems of Tuscany. Tenuta di Migliarino Sani et al. 2011), Tombolo di Tirrenia (AA.VV. 2005, Tombolo di Cecina . Syntaxonomy: Ulmus minor groves of Tuscany were referred to the association Periploco graecae-Ulmetum minoris , described for the coastal areas of Tuscany from Migliarino to Cecina . This vegetation encompasses the woods that are located along riparian areas, placed in sites where the influence of rivers is marginal, like abandoned meanders or alluvial sectors in sedimentation. In the wet season, these communities are subject to even prolonged periods of submersion. This is a heterogeneous vegetation type, in which patches dominated by trees like poplars were found mixed with tall willow trees and shrubs. This vegetation is characterized by chorotypes with wide distribution (Eurosibiric, Cosmopolitan/Circumboreal) (Fig. 2). Unfortunately, from the conservation point of view, these vegetation types are rich in alien species. According to the EIV analysis (Fig. 1), the woods dominated by Populus sp.pl., and Salix alba are positively related to N (Nutrient).

DIOSCOREO COMMUNIS-POPULETUM NIGRAE
Distribution in Tuscany (Fig. 3): Macchia Lucchese , Selva di San Rossore ), Lago di Montepulciano and Lago di Chiusi , Lago di Santa Luce (Bertacchi et al. 2005), Lago di San Floriano, Piana di Firenze and Valdarno superiore (Viciani and Raffaelli 2003). Many other sites with this vegetation type are probably present in other parts of Tuscany in addition to those reported here.
Syntaxonomy: In the past, similar coenoses have been attributed by many Italian authors (e.g. Pedrotti and Gafta 1996;Venanzoni and Gigante 2000;Pirone et al. 2003) to Populetum albae (Br.-Bl. 1931) Tchou or to Salici-Populetum nigrae (Tüxen 1931) Meyer-Drees 1936, mainly in relation to the dominant species. Very recently, some authors (Poldini et al., 2017; faced this issue and found that many Salici-Populetum nigrae communities sensu Auct. Ital. can be attributed to a recent established association named Dioscoreo communis-Populetum nigrae, which encompasses typical aspects dominated by Salix alba and Populus nigra, together with aspects dominated by Populus alba. Dioscoreo communis-Populetum nigrae is attributed to a new alliance, named Dioscoreo communis-Populion nigrae. Our relevés belonging to group 6 ( Table 7) are not easy to classify, due to their heterogeneous floristic composition (common in these coenoses) but can be provisionally attributed to this syntaxon. The relevés 1-7, dominated by Populus alba, can be attributed   . We add to these the groves dominated by Ulmus minor that can be found in Mediterranean coastal areas of central-southern Tuscany. Furthermore, we added the poplar and willow vegetation that, with the only exception of coastal areas, was also found in the areas around the large lakes of Chiusi and Montepulciano ).

Nature conservation remarks
Our results clarified the phytosociological classification of lowland woody communities in Tuscany, that is crucial to attribute them to the appropriate habitat type of conservation concern, listed in the Directive 92/43/EEC. The coenoses here described represent, in general, threatened plant communities that deserve strict conservation measures. Accordingly, based on Biondi et al. (2009Biondi et al. ( , 2012 and the European Commission (2013), the floodplain woods of Tuscany can be referred to Natura 2000 habitats of Community interest as follows.