The study area consists of Mt. Tzena, which is a mountain situated in central Macedonia, Greece, at the borders with the Republic of North Macedonia (Fig. 1). Its highest peak rises at 2,182 m a.s.l. The main geological substrates of the area are acidic. Schists dominate, but calcareous elements like marbles and sipoline marbles cover a significant part of the area (Mountrakis 1985) (see Suppl. material 1, Figure S1). The area, depending on the elevation under consideration, can be classified in the classes BSk (Arid, steppe, cold), Csa (Temperate, dry summer, hot summer), Csb (Temperate, dry summer, warm summer), Dsb (Cold, dry summer, warm summer), Dsc (Cold, dry summer, cold summer), Dfb (Cold, no dry season, warm summer) and Dfc (Cold, no dry season, cold summer) of the Köppen-Geiger classification (Beck et al. 2018) (see also Suppl. material 1, Figure S1). At low elevations, a distinct warm-dry period can be observed during the summer months, in contrast to the high elevations where such period is not observed.

The vegetation of the mountain has been recently described and mapped at an area of 5,326.75 ha by Chasapis et al. (2013). Twelve natural and human-induced land use/land cover units have been identified in the area. In total, deciduous, broad-leaved woodlands cover more than 50% of the mountain area extending to the altitude of 1,800 m a.s.l. The dominant vegetation is the beech forest that covers 30% of the surface area. Heath and gorse of the subalpine and alpine zone also cover a significant surface area, reaching 16% of the total area. Similarly, a recent survey by Chasapis et al. (2020) showed that the flora of the mountain consists of 1,254 taxa.

Regarding the protection regime of the study area, Mt. Tzena is part of the Special Area of Conservation (SAC) of the 92/43/EEC Directive of the European Natura 2000 network under the code GR1240002 – Ori Tzena. It largely coincides with the Special Protection Area (SPA) of the bird’s Directive GR1240007 – Ori Tzena kai Pinovo. SAC GR1240002 consists of two distinct mountains, Mt. Tzena and Mt. Pinovo, the former covering the eastern half of the SAC.

Thirteen habitat types of the 92/43/EEC Directive have been described in the SAC, with one of them, 6230*- Species-rich Nardus grasslands, on siliceous substrates in mountain areas (and submountain areas in Continental Europe), being a priority habitat type. Among these thirteen habitat types, seven are forest habitat types and five of them, 9110 (Luzulo-Fagetum beech forests), 9130 (Asperulo-Fagetum beech forests), 9150 (Medio-European limestone beech forests of the Cephalanthero-Fagion), 91M0 (Pannonian-Balkanic turkey oak –sessile oak forests) and 9280 (Quercus frainetto woods) are under consideration in this paper. The other two, 9270-Hellenic beech forests with Abies borisii-regis and 92C0-Platanus orientalis and Liquidambar orientalis woods (Platanion orientalis), cover a relatively small surface area and were not considered in the analysis. In addition to these aforementioned habitat types, a forest habitat type of national interest that includes the extended Carpinus orientalis and Ostrya carpinifolia forests has been considered. It is coded under the 4-digit code 9254 and described as “Forests of Ostrya, Carpinus orientalis and mixed thermophilous forests” (Dafis et al. 2001) (see Suppl. material 1, Figure S1).

To study the vegetation of Mt. Tzena, 105 relevés were recorded in beech-dominated stands and 59 relevés were recorded in thermophilous deciduous forests, i.e. pure or mixed stands of Quercus pubescens, Q. petraea subsp. polycarpa, Q. frainetto, Carpinus orientalis and Ostrya carpinifolia. The location of the relevés is shown in Figure 1. A preferential sampling was performed, as this is commonly used in phytosociological research and, in certain cases, it is preferable to randomized sampling schemes (Dengler 2016). Especially for the identification of typical species, the preferential sampling on typical localities of each vegetation unit ensures that taxa that indicate degraded conditions will not be included in the typical species inventory. Sampling size was 200 m² (Chytrý and Otýpková 2003). All vascular taxa were recorded in each relevé and their cover-abundance values were estimated using the extended 9-class Braun-Blanquet scale (van der Maarel 2005). All taxa were assigned to three vegetation layers: tree (>5 m), scrub (0.5-5 m) and herb (up to 0.5 m). Additionally, auxiliary variables were recorded in the field, including slope (o), elevation a.s.l. (m), total vegetation cover (%), vegetation cover at each layer (%), maximum vegetation height (m), average height of each layer (m), maximum breast-height diameter (m), and bedrock and orographic position (top, middle or bottom slope).

To distinguish vegetation units, we applied agglomerative hierarchical clustering. Firstly, relevés were separated into two distinct vegetation data sets on the basis of species dominance: beech and thermophilous broad-leaved forests. These two main datasets reflect the two major groups of forest habitat types that occur in the study area: the temperate beech-dominated forests and the thermophilous Quercus spp. and Carpinus/Ostrya dominated forests. In each data set, we removed taxa that occur in less than 3% of the total number of plots in order to reduce the effect of rare species (McCune and Grace 2002). Plots with less than 5 taxa were also excluded, since plots with a small number of taxa are often identified as outliers in statistical analyses. Cover values of taxa appearing in more than one layer were merged. Species’ cover values were square-rooted to reduce the prominent effect of dominant taxa and allow for taxa with lower cover values to affect the outputs (McCune and Grace 2002). This was decided because the different habitat types, especially beech forest, would have otherwise been identified based only on a few dominant taxa. The vegetation groups were defined by means of a hierarchical cluster analysis. We used the clustering algorithm included in the R package¹ ‘cluster’ (Maechler et al. 2019) using the Bray-Curtis dissimilarity matrix of the square-rooted cover values and the flexible beta as clustering method with the value a=0.625 (McCune and Grace 2002). The choice of this clustering method is based on its compatibility with the Bray-Curtis dissimilarity measure, which shows good results with vegetation data (McCune and Grace 2002). Dissimilarity matrix was calculated by means of the R package ‘vegan’ (Oksanen et al. 2019). Αn important choice in cluster analysis is the assessment of the optimum and meaningful number of clusters (groups) to be formed, and there is a variety of methods that have been proposed for that (Borcard et al. 2011; Kassambara 2017; Tavakoli et al. 2020). Among them, we decided to apply a method presented in McCune and Grace (2002) and it is considered to be well suited for vegetation data. This approach firstly defines the maximum number of clusters (fn) that can be formed running a hierarchical cluster analysis under the restriction that each cluster should include at least three plots. Then, a series of hierarchical cluster analyses is performed, gradually increasing the number of groups to be formed from n=2 to fn. For each step, the number of statistically significant indicator (differential) taxa is calculated using the ‘indicspecies’ package in R (De Cáceres and Legendre 2009; De Cáceres et al. 2010). The hierarchical level with the highest number of indicator taxa was then accepted as the most informative in terms of vegetation differentiation power.

Having selected the final number of clusters, we identified the diagnostic/differential taxa of vegetation units applying: (a) the multilevel analysis for the identification of indicator taxa presented in De Cáceres et al. (2010) using the group adjusted IndVal from the R package ‘indicspecies’ (De Cáceres and Legendre 2009; De Cáceres et al. 2010), and (b) the differential taxa applying the method presented by Tsiripidis et al. (2009). The latter was performed by means of R package ‘goeveg’ version 0.4.2 (Goral and Schellenberg 2018). Applying both methods allow us to take advantages of each method’s features, as shown by Xystrakis (2009).

Each relevé was assigned to an already described syntaxon at the association level. We used the works of Bergmeier and Dimopoulos (2001), and of Tsiripidis et al. (2007), as the basis to assign relevés of beech forests. Similarly, the works of Bergmeier and Dimopoulos (2008) provided the basis to assign relevés of thermophilous broadleaved forests. In order to assign each relevé in a syntaxon, we adopted a similar approach with the one proposed by Tsiripidis et al. (2018) for the assignment of relevés in habitat types of the 92/43/EEC Directive. Taxa in relevés were relativized on the basis of their abundance values, i.e. abundances of taxa were divided by the sum of abundances in each relevé. Then, the phi coefficient of diagnostic/differential taxa for beech syntaxa was obtained by Tsiripidis et al. (2007) and was calculated for the diagnostic taxa identified by Bergmeier and Dimopoulos (2001) and by Bergmeier and Dimopoulos (2008). For comparison, we used the original version of the phi coefficient (Bruelheide 2000) because this version was applied by Tsiripidis et al. (2007). Having relativized the relevé data and having estimated the phi coefficients of diagnostic taxa for each syntaxon, a weighted averaging was performed using the relativized abundance values of taxa in each and every relevé in our data set, as well as the phi coefficient of every diagnostic/differential species of each syntaxon described by Bergmeier and Dimopoulos (2001), Tsiripidis et al. (2007) and by Bergmeier and Dimopoulos (2008). The sum of phi coefficients for all of the aforementioned combinations was then calculated. Each relevé was finally assigned to the syntaxon with the highest weighted averaged summary of phi coefficient. All calculations were performed in R (R Core Team 2021). Taxonomic nomenclature follows Bergmeier and Dimopoulos (2001) for beech associations, Bergmeier and Dimopoulos (2008) for thermophilous associations and Mucina et al. (2016) for syntaxa of higher rank.

In order to refine the typical species inventory, vegetation relevés have to be first assigned to a habitat type, or eventually to one of the vicariants of habitat types that were identified by Tsiripidis et al. (2018). The assignment was achieved through the estimation of floristic similarity of the vegetation relevés with one of the existing field-sheets that were prepared by Tsiripidis et al. (2018). Each elaborated field sheet by Tsiripidis et al. (2018) has a unique typical species inventory and each of the taxa in this inventory is attributed a mean cover-abundance class at the Braun-Blanquet scale. Therefore, each protocol can be considered as a vegetation relevé and (dis-)similarity indices can be applied. We thus estimated the similarity between each vegetation relevé and each evaluation protocol by applying the Bray-Curtis dissimilarity index, which takes into consideration the square rooted cover-abundance values of taxa. Moreover, each relevé was assigned to the habitat type protocol with the highest weighted averaged summary of phi coefficient, estimated with the same procedure that was followed in the affiliation with existing syntaxa, and which is described by Tsiripidis et al. (2018). To minimize potential errors, an initial draft elimination of habitat types that cannot possibly occur in the area took place. Analyses were performed using the R package ‘vegan’ (Oksanen et al. 2019). Each relevé was assigned to that habitat type reflecting the minimum dissimilarity.

To refine the typical species inventory and form site-specific evaluation protocols, we (a) narrowed down the typical taxa in the assessment protocol proposed by Tsiripidis et al. (2018) by deleting those that do not occur at the site, and (b) added to the inventory taxa of high constancy values (III-V), taxa with high indicator values (De Cáceres and Legendre 2009; De Cáceres et al. 2010; Dufrene and Legendre 1997) and taxa identified as differential by the algorithm developed by Tsiripidis et al. (2009). Taxa with high indicator values were estimated by means of the package ‘indicspecies’ R package (De Cáceres and Legendre 2009; De Cáceres et al. 2010) applying the method with group and without group combinations. Combining different methods to conclude over the differentiation significance of taxa allows for taking advantage of the various positive aspects of each method. To illustrate this, on the one hand the algorithm proposed by Tsiripidis et al. (2009) has a sound phytosociological background but on the other hand, indicator taxa as estimated by De Cáceres et al. (2010) take advantage of abundances, and not only presence/absence. Taxa that characterize degraded ecosystems were manually excluded from the typical species inventory, even if the analyses identified them as constant or differential.

The outputs of the cluster analysis allowed for the identification of four vegetation groups in the beech dominated forests of the study area (Table 1). The sorted vegetation table of beech dominated units, in which differential species using the approaches described in the methods section is shown, can be found in Suppl. material 3, Table S2.

The assignment of the relevés to the syntaxa (units) that are described by Tsiripidis et al. (2007) and by Bergmeier and Dimopoulos (2001) for beech forests is shown in Table 2.

The site characteristics of the relevés forming each group are shown in Figure 2.

Table 2.

Number of relevés of beech dominated forests assigned to each syntaxonomic unit described by Tsiripidis et al. (2007) and by Bergmeier and Dimopoulos (2001). Green-shaded cells indicate the cumulative assignment of more than 50% of relevés in each group.

Among the indicator species of this group, taxa of the alliance Quercion frainetto such as Primula veris, Quercus frainetto, Trifolium pignantii are in abundance. Furthermore, it can be observed that relevés of this group occur in the lowest elevational zone of the examined Fagus forests (Figure 2). Thus, the assignment of all relevés of group 1 to the habitat type 9280 (Quercus frainetto woods) is expected. This habitat type includes the thermophilous beech forests appearing in the transition between Fagus- and Quercus-dominated forests and are characterized by significant presence of Quercion frainetto species (Dafis et al. 2001). The updated typical species inventory of habitat type 9280 that includes constant and differential species of this group is shown in Suppl. material 4, Table S3

Group 2 Group 2 is related to unit 7 of Tsiripidis et al. (2007) which includes the association Lamiastro montani-Fagetum sylvaticae Bergmeier et Dimopoulos 2001 of the alliance Fagion sylvaticae Luquet 1926, of the order Fagetalia sylvaticae Pawłowski 1928 of the class Carpino-Fagetea sylvaticae Jakucs ex Passarge 1968. This association includes the mesophilous beech forests of Mt. Tzena and Mt. Voras, growing on moist, nutrient-rich soils. Tree height in these relevés is considerably high, when compared to other groups (Figure 2).

Relevés of this group should be included in the habitat type 9130 (Asperulo fagetum beech forests) (European Commission 2013). This habitat type includes beech forest at considerably high elevation on neutral or near-neutral soils with mild mull-type humus that form a considerably rich herb layer (European Commission 2013). Seven relevés from this group are assigned to habitat type 9280, indicating the relatively thermophilous character of this group as well. The updated typical species inventory of habitat type 9130 that includes constant and differential species of this group is shown in Suppl. material 4, Table S3. Urtica dioica has been removed from the typical species inventory of habitat type 9130 because it is often associated with degraded habitats.

Group 3 is a large, rather heterogeneous group. It consists of 37 relevés, the majority of which (36) occurs on siliceous bedrock. It reflects an intermediate character between cool and warm habitats. Relevés of this group are associated with various units described by Tsiripidis et al. (2007), but their majority is associated with units 12 and 11 (units of the Lathyro alpestris-Fagetum sylvaticae) (Table 2). On the other hand, when units by Bergmeier and Dimopoulos (2001) are considered, this group shows tight affiliation with the Orthilio secundae-Fagetum (subcom. with Luzula luzuloides). The latter unit includes beech forests of north central Greece on acidic bedrock, which are characterized by the presence of species like Avenella flexuosa and Luzula luzuloides. These two taxa have also been identified as indicators of this group in our data set (Table 1). Based on the tighter and clearer relation of this group with the association Orthilio secundae-Fagetum Bergmeier 1990, we propose the inclusion of our relevés to this association. This association belongs to the alliance Fagion sylvaticae Luquet 1926 of the order Fagetalia sylvaticae Pawłowski 1928 of the class Carpino-Fagetea sylvaticae Jakucs ex Passarge 1968. Forests of the Orthilio secundae-Fagetum often border and form transitional communities with the acidophilous vicariants of Lathyro alpestris-Fagetum which generally occur at lower elevations than the former alliance (Bergmeier and Dimopoulos 2001). This could explain the observed similarity with the Lathyro alpestris units described by Tsiripidis et al. (2007).

Likewise in group 1 and group 2, a considerable number of relevés are similar to habitat type 9280, which includes beech forests with a significant presence of floristic elements of the alliance Quercion frainetto (Dafis et al. 2001; European Commission 2013). Nevertheless, some of the most characteristic species of the Quercion frainetto forests like Quercus frainetto or Q. pubescens are negative differential of group 3 (Table 1). Therefore, we preferred to assign this group to the habitat 9110 (Luzulo-Fagetum beech forests) because (a) a significant percentage of relevés of this group (38%) is assigned to the habitat type 9110 that includes mountainous acidophilous beech forest of Greece, (b) stands of the association Ortillio secundae-Fagetum Bergmeier 1990, in which the relevés of this group are assigned, are included in this habitat (Dafis et al. 2001). The updated typical species inventory of habitat type 9110 that includes constant and differential species of this group is shown in Suppl. material 4, Table S3.

Group 4 consists of 23 relevés, the majority of them being located on calcareous substrate. It is strongly affiliated with units 10-forests on calcareous substrate at high elevations (Cardamine graeca-Fagus sylvatica) and 11-forests on calcareous substrate at lower elevations (subcom. with Fraxinus ornus of the Lathyro alpestris-Fagetum sylvaticae)) of Tsiripidis et al. (2007) (Table 2). Interestingly, it is associated with the acidophilous unit 5-Orthilio secundae-Fagetum (subcom. with Luzula luzuloides) described by Bergmeier and Dimopoulos (2001). Among the indicator species of group 4, some calcicolous taxa like Arabis collina, Cotoneaster nebrodensis, Euonymus verrucosus and Carex humilis are found. This demonstrates that group 4 is differentiated by calcicolous plant taxa. Nevertheless, units 10 and 11 by Tsiripidis et al. (2007) are vegetation units that are endemic to Mt. Olympus, corresponding to the Cardamine graeca-Fagus sylvatica community by Bergmeier and Dimopoulos (2001). The latter mostly includes the calcareous beech forests in the association Lathyro alpestris-Fagetum sylvaticae Bergmeier 1990. It is argued here that the calcicolous beech forests of north central Greece are understudied from a phytosociological perspective and further research is required. The assignment of the relevés of group 4 to an association should be a result of a broader research including relevés from extended geographical regions that will include adjacent mountain ranges.

The majority of the relevés of this group are associated with the habitat types 9150 (Medio-European limestone beech forests of the Cephalanthero-Fagion) that include a large part of beech forests developing on calcareous substrates in Greece (Dafis et al. 2001). The updated typical species inventory of this habitat type that includes constant and differential species of this group is shown in Suppl. material 4, Table S3.

Comparing the updated typical species inventories with the respective inventories presented in the original protocols, it is made obvious that substantial changes took place. Out of 29 typical species in the existing evaluation protocol of the habitat type 9110, only 11 are also included in the new inventory, which counts 26 taxa. Regarding habitat type 9130, out of 36 species, 19 are also present in the updated protocol (in these taxa Urtica dioica is also included). Thirty-five taxa are considered as new addition to the updated inventory. Regarding habitat type 9150, 16 out of 44 taxa of the existing evaluation protocol are considered in the updated inventory while 20 more taxa have been included additionally. Finally, in habitat type 9280, 31 out of 53 taxa are included in the updated inventory while 28 new taxa are added. These changes are mostly related to the extended sampling that took place in this NATURA 2000 area. The existing typical species inventories were compiled using a limited number of relevés from each NATURA 2000 site.

The outputs of the cluster analysis justified the acceptance of three groups in the data set that includes the relevés dominated by thermophilous deciduous broadleaved species (Table 3). The sorted vegetation table of the thermophilous deciduous broadleaved forests, in which differential species using the approaches described in the methods section is shown, can be found in Suppl. material 5, Table S5.

These three groups correspond to the i) Ostrya carpinifolia-Carpinus orientalis-Quercus pubescens, ii) Quercus pubescens-Carpinus orientalis, and iii) Quercus frainetto and Q. petraea forest formations respectively. The assignment of the relevés to the syntaxa (units) identified in Greece by Bergmeier and Dimopoulos (2008) for the thermophilous deciduous forests is shown in Table 4.

In terms of site characteristics (Figure 3), the identified groups show a clear gradient in tree height and cover of the tree layer. There is a remarkable homogeneity regarding the cover of the tree layer and slope’s inclination in group 3. In contrast, values of elevation, tree cover for group 1 and of slope inclination in group 2 reflect high variability.

Group 1 consists of Ostrya carpinifolia-Carpinus orientalis mixed stands, with the former taxon being rather dominant in most relevés. Quercus pubescens is a constant species in this group, but it participates in the tree cover with considerably low values. Relevés of this group are assigned to the Dryopterido pallidae–Ostryetum carpinifoliae Bergmeier 1990 association (Bergmeier and Dimopoulos 2008) of the Carpinion orientalis Horvat 1958 alliance. This alliance includes the low-elevation calcareous thermophilous oak and oriental hornbeam forests (Mucina et al. 2016). This alliance belongs to the Quercetalia pubescenti-petraeae Klika 1933 order of the Quercetea pubescentis Doing-Kraft ex Scamoni et Passarge 1959 class. Bergmeier and Dimopoulos (2008) consider this alliance to be synonym with Fraxino orni-Ostryion but Čarni et al. (2009), Mucina et al. (2016) and Škvorc et al. (2017) treat these two alliances distinctly. Both alliances occur in the eastern Mediterranean region, and their distinction is mostly based on the fact that Fraxino orni-Ostryion has a more (sub)Mediterranean character than Carpinion orientalis, which is characterized by the abundance of heliophilous, montane-Mediterranean taxa (Čarni et al. 2009). Despite this, Bergmeier and Dimopoulos (2008) provide extensive argumentation against the use of macroclimatic gradients for the characterization of the thermophilous deciduous forest vegetation of Greece. One can argue that the presence of Ostrya carpinifolia is indicative of a submontane character, but the fact that (a) Quercus pubescens is also a positive differential of group 1 (Table 3), and (b) group 1 is characterized by a considerably open canopy and trees of low height (Figure 3), an indication of marginal conditions of dry and poor sites, allows for the inclusion of this group in the Carpinion orientalis Horvat 1958 alliance (Mucina et al. 2016) with the strict sense. In any case, there is a clear need for a syntaxonomical revision of the thermophilous deciduous forests in the southern Balkan peninsula. The work of Bergmeier and Dimopoulos (2008) can form the basis for the classification, but it must be further ameliorated by new data from additional studies. These include data from local research (Gerasimidis and Korakis 2005; Korakis 2003) or data sets covering broader geographical areas (e.g. (Čarni et al. 2009; Stupar 2016).

Relevés of group 1 should be included in the habitat type 925A (forests of Ostrya, Carpinus orientalis and mixed thermophilous forests) (Dafis et al. 2001). This is a habitat type of national interest that includes forests of Carpinion orientalis dominated by Carpinus spp. and Ostrya carpinifolia. Since it is not a habitat type of community interest, an inventory of typical species does not exist, therefore the typical species shown in Suppl. material 4, Table S4 is the first attempt to create one.

Group 2 has a similar character with group 1, but Quercus pubescens is rather dominant when compared with Carpinus orientalis and Ostrya carpinifolia, with the latter having low cover values (<3) in all relevés but three. Relevés of this group are assigned to the Dryopterido pallidae-Ostryetum association Bergmeier 1990 (Bergmeier and Dimopoulos 2008) of the Carpinion orientalis Horvat 1958. The inclusion of relevés of group 2 in the Carpinion orientalis (Mucina et al. 2016), in the strict sense, is even clearer when compared with relevés of group 1.

Relevés of group 2 should be included in the habitat type 91M0 (Pannonian-Balkanic turkey oak- sessile oak forests) due to the relative dominance of Quercus pubescens. Habitat type 91M0 has replaced the habitat type 924A-eastern Mediterranean and Balkan thermophilous oak-dominated forests (Dimopoulos et al. 2018). This habitat type included all Greek deciduous Quercus spp. formations with the exception of the (a) Quercus trojana subsp. trojana formations which are included in habitat type 9250 - Quercus trojana woods, (b) Quercus ithaburensis formations which are included in the habitat type 9350 - Quercus macrolepis forests, and (c) Quercus brachyphylla formations that are often associated with Q. macrolepis or Q. ilex in the Aegean islands which are included in the 9310 - Aegean Quercus brachyphylla forests (Dafis et al. 2001; European Commission 2013). The latter needs a critical revision because it is associated with Quercus brachyphylla Kotschy, which is a synonym for Q. pubescens Willd. There is a need to examine floristic and ecological differentiation between the 9310 and the Quercus pubescens dominated vicariants of 91M0 habitat type. Nevertheless, such a comparison is not within the scope of this paper.

The updated inventory of typical species that includes differential and constant taxa of group 2 is shown in Suppl. material 4, Table S4. This should be considered as the typical species of the habitat type 91M0 (Quercus pubescens vicariant).

Group 3 is clearly differentiated from the previous group. Quercus petraea subsp. polycarpa and Q. frainetto dominate the canopy and it is differentiated by a large number of taxa. When Q. petraea subsp. polycarpa dominates, relevés of this group are assigned to the Genisto carinalis-Quercetum petraeae. When Q. frainetto dominates, relevés of this group are assigned to the Verbasco glabrati-Quercetum frainetto Gamisans et Hebrard 1979, Huetio cynapioidis-Quercetum frainetto Raus ex Raus Bergmeir 2008, or the Digitali viridiflorae-Quercetum frainetto Gamisans et Hebrard 1980 associations (Table 4). We decided to affiliate relevés of group 3 to the association Verbasco glabrati-Quercetum frainetto Gamisans et Hebrard 1979 on the basis of its already recorded presence in the area by Bergmeier and Dimopoulos (2008). This association belongs to the alliance Quercion confertae Horvat 1958 of the class Quercetalia pubescenti-petraeae Klika 1933 of the order Quercetea pubescentis Doing-Kraft ex Scamoni et Passarge 1959. Nevertheless, there should be a critical analysis in order to eventually conclude over the presence of the association Digitali viridiflorae-Quercetum frainetto Gamisans et Hebrard 1980 in the area due to the large number of relevés from this study that are associated with this association. There is only one relevé which is assigned to the association Huetio cynapioidis-Quercetum frainetto Bergmeier et Dimopoulos 2008, probably due to the considerably high cover of Fagus sylvatica subsp. sylvatica and the accordingly low cover of Q. frainetto in the canopy. The geographical distribution (eastern Thessaly in central Greece in south-central Greece) and the ecological characteristics (pasture woods) of this association (Bergmeier and Dimopoulos 2008; Samaras et al. 2021) do not justify its occurrence in the area, therefore this relevé should be eventually assigned to the association Digitali viridiflorae-Quercetum frainetto Gamisans et Hebrard 1980, which is the syntaxon with the second largest sum of fidelity values for this plot.

Relevés of group 3 should be included in the habitat type 91M0 (Pannonian-Balkanic turkey oak- sessile oak forests), but a distinct vicariant of the Quercus frainetto-Quercus petraea subsp. polycarpa dominated formations should be formed. The updated inventory of typical species is shown in Suppl. material 4, Table S4.

Since there is not an existing evaluation protocols for the habitat type 925A or the two vicariants of the 91M0 habitat type for the area, it is not possible to perform any comparisons with existing typical species inventories.