DRAINED ASH SWAMP ( VERATRO ALBI-FRAXINETUM ANGUSTIFOLIAE ), A NEW ASSOCIATION IN THE NYÍRSÉG, NE HUNGARY

This paper presents the phytosociological description of a drained swamp community, Veratro albi-Fraxinetum angustifoliae , so far found only in the Nyírség at Nyírábrány “Kis-kőrises”, “Mogyorósi-erdő”; Vámospércs “Jónásrész-Kőrises”; and Vámospércs “Jónás - rész-Buzita”. The habitat of the community is transitional between that of alder swamps ( Fraxino pannonicae-Alnetum glutinosae ), and hardwood riparian forests ( Fraxino pannoni-cae-Ulmetum ). The community is characterised by high proportions of character species of Alnion glutinosae and Molinion coerulei as well as Quercetea pubescentis-petraeae s. l., whereas character species of the order Fagetalia are almost completely absent. It hosts several rare, often threatened species, such as Angelica palustris , Ophioglossum vulgatum , Trollius europaeus and Veratrum album .


INTRODUCTION
Hardwood riparian forests have been the subject of our long-term research on the distribution, composition and geographical variation of the forest vegetation in Hungary.Hardwood riparian forests still occur in most lowland areas of Hungary under different environmental conditions and thus are particularly suitable for studying phytosociological differentiation and biogeographical relationships in the forest vegetation within the Carpathian Basin.During our work in the Nyírség, an extensive area in the northeastern part of Hungary covered with eolic sand, we found several Fraxinus angustifolia dominated forest stands that seemed to be different from all known forest communities described previously in the country (Figs 1-2).No similar forest community is known to exist in the European vegetation either (Braun-Blanquet 1964, Ellenberg 1986, Horvat 1938, Horvat et al. 1974, Mucina et al. 1993, Oberdorfer 1992a, b, Rodwell et al. 2002, Willner and Grabherr 2007a, b).
Because their species composition appeared to be substantially different from that of the Fraxino pannonicae-Ulmetum Soó in Aszód 1935corr. Soó 1963, a community once widely distributed in similarly moist habitats across the country, and Fraxino pannonicae-Alnetum Soó et Járai-Komlódi in Járai-Komlódi 1958, we described it as a novel association under the name Veratro albi-Fraxinetum angustifoliae Kevey et Papp L. (Kevey 2008).
Unfortunately, this association has not been described in more detail, nor has a synoptic table of it been published since then.Here we are going to fill this gap by providing a detailed description of the community based on our sample material of ten relevés.Our primary goal is to substantiate the split of this new association from the rest of the hardwood riparian forests found in Hungary.

Research area
The studied Fraxinus angustifolia dominated stands were found in four different forest areas in the Nyírség: the Mogyorósi-erdő and Kis-kőrises near the town of Nyírábrány, and the Jónásrész-Kőrises and Jónásrész-Buzita near the village of Nyíracsád.The stands were all found along the edge of local depressions where the habitat is characterised by high groundwater levels, which rises above the ground surface only in very wet periods.They either grow at the fringe of genuine alder swamps (Fraxino pannonicae-Alnetum) in the deepest parts of local depressions, or form the transition zone between these swamps and oak-ash-elm forests (Fraxino pannonicae-Ulmetum) growing on higher ground (Fig. 3).Since the ground is normally not covered with water throughout most of the vegetative period in their habitats, the soil contains only small amounts of peat that is generally decaying.The three studied forest areas are all parts of the Natura 2000 network, and are also protected by national law.The Jónásrész "Kőrises" and "Buzita", and "Kis-kőrises" are state reserves, whereas the "Mogyorósi-erdő" is under strict protection.

Methods
Our sampling procedure followed the traditional quadrate method of the Zürich-Montpellier phytosociological school (Becking 1957, Braun-Blanquet 1964).Sample plots were designated visually by selecting the parts of a stand that seemed to be the most homogeneous in habitat characteristics, vegetation structure and species composition, and showed no signs of human impact (including forest management).Because the traditionally used 400 m 2 plot size does not satisfy the requirement for minimal area (see Du Rietz 1921), our sample plots were 1600 m 2 in size except for one (1200 m 2 ).This size suffices the requirement of representativity in temperate deciduous forests (Kevey 2008).
Because forests in similarly mesic habitats tend to exhibit large phenological changes during the vegetative period, we sampled each stand twice (spring and summer) using the same plots.During sampling, we recorded all species within the sample quadrate and estimated their projected cover.We also estimated the height of each vegetation layer, and the trunk diameter of trees.
The raw data were compiled and arranged in a synoptic table by the NS software program (Kevey and Hirmann 2002), which also was used to calculate constancy values of each species, and proportions of species characteristic of a particular syntaxon.To assess the syntaxonomic relationship of the studied stands, we compared them to a representative material of the spatially adjacent Fraxino pannonicae-Alnetum and Fraxino pannonicae-Ulmetum, and all previously described communities growing in similar habitats in Hungary: Ophioglosso-Betuletum pubescentis (Vértesalja: Riezing andSzollát 2008-2009: 6 relevés); Molinio-Alnetum glutinosae (Tengelici-homokvidék: Kevey 2008: 20 relevés); Molinio-Salicetum cinereae (Szigetköz: Kevey 2008: 25 relevés).In doing so, we performed pairwise comparisons between sample sets and deter- mined the set of differential species (species that differed in their constancy value by at least two steps) and the proportions of character species.We also carried out binary cluster and principal coordinates analyses (PCoA) with the help of the Syntax 2000 package (Podani 2001).The method of grouping in the cluster analyses was complete link, and the similarity coefficient in both types of analyses was that of Baroni-Urbani and Buser.
The order of syntaxa in the synoptic and statistical tables follows the modified syntaxonomic system of Soó (1980) according to the suggestions and results of Borhidi et al. (2012), Kevey (2008), Mucina et al. (1993) and Oberdorfer (1992a, b).

Physiognomy and structure
In the studied stands, the forest canopy was structured into two distinct layers.The upper layer was situated at about 20-28 m height and was rather dense with high (60-80%) projected cover.The most abundant (A-D: 3-4) tree species in this layer were Fraxinus angustifolia subsp.danubialis and Populus alba.They were also constant species across the samples.
The lower canopy layer was at the height of 12-20 m.The projected cover was rather variable among samples.It was made up of mostly tree-sized shrubs and young individuals of trees.The most abundant species in this layer was Fraxinus angustifolia subsp.danubialis.
Shrubs in the samples were 1.5-3.5 m tall, and formed a moderately dense layer with 25-70% cover value.It was composed of Cornus sanguinea, Crataegus monogyna, Frangula alnus, Fraxinus angustifolia subsp.danubialis, and Ligustrum vulgare.Only Cornus sanguinea and Ligustrum vulgare had rather high cover values.The layer of saplings was greatly variable in projected cover (1-50%).

Frequency distribution of constancy classes
The ten samples included 22 constant (K: V) and 12 sub-constant K: IV) species.The number of accessorial (K: III), sub-accessorial (K: II), and accidental (K: I) species in the samples was 19, 37, and 69, respectively (see Table 1).The frequency distribution of species in the five constancy categories, from I to V, follows a characteristic pattern.The highest frequency value at I steeply drops to II, then decreases further with diminishing increments to IV, where it reaches its minimum.At V, its value is again higher (Fig. 4).This patterns is only typical of phytosociological samples that are representative of the sampled vegetation unit (see Kevey 2008).

Proportion of character species
As usual in temperate deciduous forests in Central Europe, the species characteristic of the Querco-Fagetea class or syntaxa within it play a substantial role (30.8%) in the species composition of these Fraxinus angustifolia dominated stands.They are followed by the character species of SE European dry oak woods, Quercetea pubescentis-petraeae s. l. (17.1%) despite the rather mesic habitat, and Alnetea glutinosae s. l., Molinio-Juncetea s. l., Salicetea s. l., and Phragmitetea s. l. in decreasing order.Within the Querco-Fagetea class, Fagetalia s. l. and Alnion incanae species attain the highest proportions.
In comparison to other communities, it is noteworthy that character species proportions in all but two syntaxa were typically intermediate between those in Fraxino pannonicae-Alnetum and

Deschampsia caespitosa III I
Populus tremula III I  V-Fr Fr-U

Constant species
Angelica sylvestris V -  Fraxino pannonicae-Ulmetum, which reflects the transitional habitat of Veratro albi-Fraxinetum angustifoliae between the former two (Table 3, Figs 5-10).The two exceptions are the Quercetea pubescentis-petraeae s. l. and Molinio-Juncetea s. l. species.The proportion of the former is the highest not only among the three, but among all six communities, whereas that of the latter is the highest among the three hardwood riparian forest associations, but is the smallest among all drained swamp communities.The distribution of character species proportions in Veratro albi-Fraxinetum pannonicae are rather different from the latter communities.The largest differences are found in the character species of Phragmitetea s. l., Molinio-Juncetea s. l., Galio-Urticetea s. l., Epilobietea s. l., Salicetea purpureae s. l., Alnetea glutinosae s. l., Querco-Fagetea s. l., Fagetalia, Alnion incanae s. l. and Quercetea pubescentis-petraeae s. l. (Table 4).The occurrence of introduced aliens (Fraxinus pennsylvanica, Acer negundo, Celtis occidentalis, Echinocystis lobata, Parthenocissus inserta, Robinia pseudoacacia, Vitis riparia) in the association is apparent (Table 1), but compared to other associations in the region, their proportion is low (1.9%).

Number of differentiating species
The number of differentiating species (species for which the difference between their constancy values in the compared two communities equals or exceeds two) in Veratro albi-Fraxinetum angustifoliae is greater than 40 in all pairwise comparisons.The highest number of differentiating species (68) was   5-8).

Similarity relations in multivariate analyses
In the cluster analysis with complete linkage algorithm, the ten samples of Veratro albi-Fraxinetum angustifoliae grouped with the samples of Fraxino pannonicae-Ulmetum.The rest of the samples formed the sister cluster of this group.In this cluster, samples of the alder swamp (Fraxino pannonicae-Alnetum) were the sister cluster of the three drained swamp communities (Ophioglosso-Betuletum pubescentis, Molinio-Alnetum, Molinio-Salicetum cinereae) (Fig. 11).The result in the analysis with the group-average algorithm differed only in the placement of one of the drained swamp communities (Ophioglosso-Betuletum pubescentis) as a sister group of the Veratro albi-Fraxinetum and Fraxino pannonicae-Ulmetum cluster (Fig. 12).The dissimilarity level of the Veratro albi-Fraxinetum and Fraxino pannonicae-Ulmetum was very similar, though slightly smaller than those between the Ophioglosso-Betuletum and Molinio-Alnetum (complete linkage), and Molinio-Alnetum and Molinio-Salicetum cinereae (group average).
The result of the PCoA is in agreement with the above.In the plane of axes one and two, the samples of Veratro albi-Fraxinetum were adjacent to both, Fraxino pannonicae-Ulmetum and Ophioglosso-Betuletum pubescentis (Fig. 13).However, the position of the latter changed substantially in the plane of axes one and three, while the spatial relation of the samples of Veratro albi-Fraxinetum and Fraxino pannonicae-Ulmetum essentially did not change (Fig. 14).

DISCUSSION
Field observations on habitat characteristics of Veratro albi-Fraxinetum angustifoliae suggested its transitional nature between Fraxino pannonicae-Alnetum and Fraxino pannonicae-Ulmetum.It is likely that this community develops from Fraxino pannonicae-Alnetum as organic and inorganic deposits accumulate and the habitat gradually dries out.During this process, most species of the Lemno-Potametea s. l. class, and also many Phragmitetea s. l. and Alnetea glutinosae s. l. species disappear or are replaced by Molinietalia s. l., Querco-Fagetea and Quercetea pubescenti-petraeae species.This genealogic relationship would manifest itself in similarities in species composition and character species proportions to both associations.
It also seemed reasonable to assume that this association may be closely related to drained swamp communities based on their similarities in habitat conditions.The habitat of these communities may be covered with water in wet periods, but the soil is typically not saturated with water most of the time.This allows the establishment of Molinietalia and Quercetea species in relatively high proportions.These communities also share a number of additional features including the relatively high proportions of Phragmitetea and Alnetalia glutinosae, and the low proportion of Fagetalia elements.Despite these similarities, the Veratro albi-Fraxinetum angustifoliae cannot be identified with any of the studied associations, but is best recognised as a novel association.Its distinctiveness is supported by a suite of evidence, including the number of differentiating species, the distribution of character species proportions, and the dissimilarity in floristical composition.In our opinion, the amount of differences in these features are sufficiently high to designate this community as a distinct association.Whereas the Veratro albi-Fraxinetum angustifoliae is best treated as a new association, its syntaxonomic affinity is rather difficult to ascertain.Owing to its intermediate characteristics in many respects, it could be placed either in the Alnetea or the Querco-Fagetea class.The Molinio-Betuletea class by Passarge and Hofmann (1968), which includes strongly acidophilic communities distributed over the more humid western and northern parts of Europe, may be excluded for this reason.Based on preliminary data, Kevey (2008)