Eco-Coenotic and Diversity Patterns in Artemisia alba Open Scrubs from Romania within the Context of Similar Communities from Neighbouring Regions

Abstract

No information currently exists on the floristic structure and richness of the Artemisia alba scrubs in Romania and their regional/local environmental drivers. We aimed to fill these knowledge gaps by also considering physiognomically similar communities from Hungary, Serbia and Bulgaria. A total of 89 phytosociological relevés, including 43 performed in Romania, were analysed through clustering, constrained ordination and generalised linear mixed models. The Carpathian and Pontic scrubs were clustered into three distinct groups, which were assigned to as many new syntaxa. Differences in the regional species pool and elevation have the strongest effects on floristic dissimilarities between all studied communities. As opposed to the bare soil fraction, the elevation and slope have positive but no singular effects on species richness in the Pontic-Carpathian coenoses. Species diversity declines steadily with increasing shrub cover in all these communities. The relative cover of annuals has contrasting effects on species richness, positive in the most xerophilous communities and negative in their most mesophilous counterparts. The relative number of annuals is only (negatively) related to overall species richness in the coenoses least affected by moisture deficit. Overall, species diversity is driven mainly by soil water availability and, to a lesser extent, by the relative abundance of shrubs and annuals.

1. Introduction

Artemisia alba Turra is largely distributed in southern Europe and north-western Africa, but restricted to dry, calcareous habitats in central Europe [1]. This dwarf sub-shrub is known for its ‘water-conservation’ strategy that might be regarded as an adaptation to shallow soil conditions [2]. Despite such edaphic specialisation, the relatively large variability of the climatic conditions across its distribution area has led to the diversification of its eco-coenotic affinities. Throughout central and south-eastern Europe, Artemisia alba occurs as a diagnostic and/or constant species in different habitats ascribed to the EUNIS type E1.2, i.e., perennial calcareous grasslands and basic steppes [3].

The grassland/scrub communities featuring Artemisia alba and confined to the E1.2 habitat type were syntaxonomically assigned to many plant associations from different orders i.e., Festucetalia valesiacae in subcontinental Central Europe [4,5,6]), Stipo pulcherrimae—Festucetalia pallentis in central and south-eastern Europe [5], Scorzonero—Chrysopogonetalia in eastern Prealpine, Illyrian and Dinaric regions [7] and Halacsyetalia sendtneri in western Balkans but on ultramafic soils [8]. In Romania, the Artemisia alba scrub communities occur in similar habitats, but their typology is completely unknown, since only few relevés performed in Apuseni Mts. were published, without being interpreted syntaxonomically and phytogeographically [9,10].

To date, no studies exist on the main factors driving the plant species composition and richness of the open coenoses (co)dominated by Artemisia alba throughout Romania. Based on the paradigm of plant species assembly according to a hierarchy of abiotic and biotic filters [11,12], it is expected that these coenoses be mainly structured by the former and, to a lesser extent, by interspecific interactions. This is because the scarce soil moisture is an important limiting factor, given that these open scrubs are restricted to harsh environments characterised by shallow, skeletal but base-rich soils [2]. Therefore, topographic conditions are likely to be among the main drivers of the floristic structure of these scrubs at community scale, since topographically controlled soil moisture (e.g., elevation, slope, aspect) was shown to primarily control local vegetation patterns [13]. Among factors acting at slightly larger scales, the characteristics of the landscape (e.g., heterogeneity or adjacency patterns) in which the Artemisia alba scrub patches are embedded may also exert an influence through the so-called matrix effect [14]. Finally, at regional scale, the mesoclimatic variables are likely to play an increasingly more important role, especially in terms of the amount of precipitation and summer temperatures. Moreover, inter-regional floristic dissimilarities between these dwarf scrubs may arise from dispersal limitations as well as differences in the regional species pool [15] that are largely determined by habitat specialists and geographic vicariants [16,17].

Although biotic filters are expected to play a secondary role in structuring these open, drought-tolerant scrubs, the prevalence of positive or negative interactions between certain plant guilds can modulate species diversity patterns at small scales [18,19,20]. As predicted by the stress gradient hypothesis, the unimodal response of species richness is slightly left-skewed along the full stress gradient [21,22]. Depending on the position of habitats along the aridity gradient and species tolerance to drought, shrubs can outcompete the annuals [23,24,25,26], show neutral effects, or act as nurse neighbours to annuals [24,27,28]. Therefore, in subxeric habitats, the facilitative effects of nurse shrubs on generalist annuals (mainly by ameliorating the microclimatic conditions and soil fertility) usually translate into a larger contribution of the latter to total plant species richness at community scale [20,24,29,30]. In contrast, shrubs usually suppress the annuals both in more benign habitats (e.g., with shorter or cooler dry season) and under high levels of aridity, leading to a low contribution of annuals to total species richness [23,24,25,26].

In order to fill the knowledge gaps mentioned above, we herein aimed: (i) to assess the proper syntaxonomic assignation of the scrubs (co)dominated by Artemisia alba from Romania and their floristic particularities in the context of more or less similar communities from the neighbouring regions; (ii) to determine the effects and relative importance of topographic variables and regional distribution on plant species composition and richness at community level; and (iii) to estimate the contribution of shrub and annual plant abundance/richness to the species diversity of the studied coenoses.

2. Materials and Methods

2.1. Study Areas

The study areas were selected based on the localities where Artemisia alba was previously observed, as indicated on some CL Herbarium sheets and in a few phytosociological relevés [9,10,31]. It is worth mentioning that Artemisia alba was also reported in Oravița Hills [32] and Hășmaș Mts. [33], but it has not been refound there since then [34,35].

The three areas investigated contain all the main landforms (mountains, piedmonts and low hills) on which the habitats of Artemisia alba are distributed throughout Romania. The mountain study area (Cp1 in Figure 1) lies in the Trascău Mountains (part of the Apuseni Mountains) between 650 and 1000 m a.s.l. The piedmont study area (Cp2 in Figure 1) is located between the Vlădeasa and Meseș mountains, between 380 and 420 m a.s.l. The low colline study area is divided into two parts (Pt1 and Pt2 in Figure 1), both located in the Dobrogea province at 60–80 m a.s.l. and not far from the Black Sea coast. In all study areas, the patches of Artemisia alba scrubs are embedded into grasslands or (only in few cases within the Pt2 area) into shrublands/open pine plantations.

The climate of the study areas including the Trascău Mts. and the southern Meseș piedmonts (Carpathian region) is characterised by mean maximum temperatures in the warmest month (July) of 22.9 °C and 24.3 °C, respectively, and average precipitation of 693 mm and 719 mm per year, respectively. The climatic conditions in the third study area from Dobrogea (Pontic region) are quite different, with the mean maximum temperature of 28.0 °C in July, average precipitation of 512 mm per year, and pronounced water deficit in July–August [31].

The geological substrate in all the mentioned study areas consists of limestone underlying shallow, skeletic Leptosols, rich in cations (

Table 1. Extractable cations content and pH of soil samples collected in the investigated sites in the Carpathian (Cp) and Pontic (Pt) regions in Romania (see also Figure 1). The geo-topographic location and corresponding plant associations are indicated for each soil sample. Abbreviations as in Figure 2.

Site	Latitude (Degrees)	Longitude (Degrees)	Elevation (m)	Aspect	Plant Assoc.	pH	Mg (mg/kg)	Ca (mg/kg)	Co (mg/kg)	Cd (mg/kg)	Cu (mg/kg)	Ni (mg/kg)	Zn (mg/kg)	Cr (mg/kg)	Pb (mg/kg)
Colțești (Cp3)	46.25283	23.32494	700	W	Am-Aa	7.36	6.08	180.00	21.40	7.47	41.31	44.63	91.64	55.81	45.65
Remetea (Cp3)	46.44625	23.58159	900	N	Am-Aa	7.60	18.20	210.00	21.33	5.71	40.57	18.26	101.59	51.27	30.38
Jebuc (Cp4)	46.53600	23.06160	394	NW	Op-Aa	7.76	24.30	250.00	12.27	11.64	32.72	19.64	45.29	39.83	4.95
Sfăraș (Cp4)	46.53490	23.06112	390	SW	Op-Aa	7.86	6.08	220.00	17.08	13.43	34.09	29.78	50.00	43.62	12.66
Cotu Văii (Pt1)	43.81048	28.33560	65	NE	Sc-Aa	8.07	24.30	120.00	17.22	1.66	14.11	31.06	45.32	26.71	0.57
Enisala (Pt2)	44.87890	28.84540	75	S	Sc-Aa	8.12	6.10	110.00	21.77	2.75	16.53	55.41	67.07	43.35	9.98

).

2.2. Data Collection and Transformation

The multiannual climatic means, corresponding to the three study sites in Romania, were calculated based on interpolated monthly data at 30 arc seconds resolution. These data, covering the 1990–2018 period, were retrieved from the CHELSA 2.1 database [36].

The vegetation survey was carried out in 2018 and 2022 by employing the phytosociological method [37]. The relevé plots were delimited in the field based on the (co)dominance of Artemisia alba among the present plant species. The relevé area varied between 10 and 25 m², depending on local site conditions. All the occurring vascular plant species were recorded by visually estimating their relative cover on the Braun-Blanquet ordinal scale. The topographic characteristics (aspect and slope) of the inventoried habitats were measured using an Eclipse 99 compass and clinometer, while the elevation and geographic coordinates were recorded using a GPS unit (Garmin GPSMAP 64S).

Soil sampling was carried out in all six areas where floristic relevés were performed. One soil sample per site was collected from the mineral top-soil (1–10 cm deep) and subsequently analysed using an atomic absorption spectrophotometer, model novAA350 (Analytic Jena). The soil nomenclature followed the latest international soil classification system [38]. The data regarding the geo-topographical characteristics, soil parameters and plant community types associated with each soil sample are reported in Table 1.

Some phytosociological data concerning coenoses (co-)dominated by Artemisia alba were retrieved from the published literature as follows: 5 syntaxonomically unclassified relevés from the southern piedmonts of Meseș Mts. in north-western Romania [10], 10 relevés of Artemisio saxatilis—Festucetum dalmaticae Borhidi 1996 from Mecsek and Villány hills in southern Hungary [39], 21 relevés of Bromo moesiacae—Stipetum epilosae Todorova et Tzonev 2010 from Mt. Vitosha in western Bulgaria [40], 10 relevés of Bromo fibrosi—Artemisietum albae Marković ex Aćić et al. 2014 from Mt. Suvobor in central Serbia [41], 5 relevés of Artemisio albae—Achnatheretum calamagrostis Jovanović-Dunjić et S. Jovanović ex Aćić et al. 2014 from Mts. Goč and Kopaonik in south-western Serbia [42] and 16 relevés of Artemisio albae—Silenetum armeriae Lukušić et Kabaś in Aćić et al. 2014 from Mt. Tara in western Serbia [42]. A general table including all the relevés herein analysed (both novel and collected from literature) is presented in Table S1 (see the Supplementary material). The nomenclature of plant species recorded in the original relevés followed the latest Romanian taxonomic classification [43], whereas the plant names in the relevés performed outside Romania were left unchanged, as originally given by the respective authors. In order to avoid any duplicate taxa, a thorough check for possible synonyms was performed in the combined table of relevés.

In order to reveal biotic-related patterns in species richness, two plant guilds were distinguished: ‘shrubs’ (including subshrubs and tree saplings) and ‘annuals’ (including both annual and biennial herbs). The species abundance classes were converted into mid-class percentage values and subsequently, the latter were square-root transformed for reducing the strong influence of dominant species. The proportion of bare soil within each plot was estimated as the complement of the relative vegetation cover. The terrain aspect was converted into a linear variable ranging from 0 to 2 by using the equation [44]: A′ = cos (45° − A) + 1, where A = aspect measured in degrees and 45° corresponds to the coolest aspect in continental Europe, i.e., north-east. A nominal variable called ’Region’ was created ad hoc on the basis of the geographic proximity of the relevés, so that the effect of the regional species pool could be taken into account in model-based analyses. The five categories of the ‘Region’ factor were defined as follows: Carpathian (mountainous, north-western Romania), Pontic (south-eastern Romania), Pannonian (southern Hungary), Dinaric (south-western Serbia) and Balkan (eastern Bulgaria) (Figure 1).

2.3. Numerical Data Analysis

The classification of the 89 relevés relying on pairwise Bray-Curtis dissimilarities was performed via hierarchical cluster analysis by employing three algorithms: average linkage, beta-flexible and Ward. The dendrogram obtained through the last algorithm was eventually retained, as it delivered the weakest chaining effect and highest agglomerative coefficient (0.88). The optimal number of clusters was inferred based on the location of the maximum value of the Calinski–Harabasz index calculated for all possible solutions with 2–12 clusters. The stability of each retained cluster (expressed in percentages) was assessed by bootstrapping the mean Jaccard similarity of the component relevés. The group-equalised Phi coefficient calculated for real and permutated data was employed to estimate the association strength and significance between the retained clusters of relevés and each individual species. The external validation of the distinguished syntaxa was conducted via a Random Forest classifier by declaring the plant association identity as the response factor and all site variables in hand as predictors. Their importance for the syntaxa differentiation was evaluated through the mean decrease in accuracy.

A constrained and conditioned ordination of relevés in the species reduced space was performed by partial, distance-based redundancy analysis (db-RDA) applied to the same dissimilarity matrix and the vectors of site predictors, while accounting for differences in the regional climate/species pool via the ‘Region’ factor. All possible interactions between single site variables were also considered. The selection of predictors in the final model was done through a stepwise, forward procedure with permutation-based testing. The adjusted R² in the unconditioned db-RDA (i.e., with ’Region’ included among explanatory variables and without interaction terms) assisted in partitioning the variation in floristic dissimilarities due to singular variables or their combinations.

Given the large difference in relevé size (10 to 100 m²), the comparison of plant species diversity between syntaxa was based on sample completeness [45] and involved only five (out of eight) plant associations, which were represented through sampling plots of 10 to 25 m². Species diversity was estimated through the Hill’s number equivalent of order 1 (exponential Shannon entropy), which weights the rare and common species equally, and is less sensitive than species richness to sampling effects [46,47]. An equal sample size of 25 plots was simulated to extend the rarefied diversity curves through extrapolation, so that the sample coverage (completeness) estimated through species incidence would reach over 0.95 in each of the five community types compared.

Species diversity (expressed as Hill’s number equivalent, absolute or relative richness) was fitted against environmental predictors or relative cover/number of shrubs/annuals via generalised linear mixed models (GLMMs) with spatial random effects, but only considering Artemisia alba communities from Romania for which the geographic coordinates were accurately measured herein. Either the negative binomial or beta distribution with the log or logit link function, respectively was used for modelling the response variable. After comparing the variograms corresponding to different covariance structures, the exponential type was retained as the most appropriate in all the spatial mixed models computed. Whenever more than one fixed effect was initially involved in these models, a manual backward selection of predictors was attempted to possibly attain the best parsimonious model based on the ratio of generalised chi-square by degrees of freedom. When appropriate, the covariate ‘plot area’ as well as interaction/quadratic terms involving topographic variables were included in the starting models, but most of them were often removed as non-significant during the selection procedure. The significance of differences in the relative richness of annuals between community types was assessed by contrasting the conditioned least-square means that were produced by a GLMM whose unique fixed predictor was the syntaxon identity.

For complying with statistical assumptions and/or improving the fit of models, some variables were log, square-root or logit transformed prior to numerical analyses. These were carried out using the R packages ‘cluster’ [48], ‘fpc’ [49], ‘randomForest’ [50], ‘indicspecies’ [51], ‘vegan’ [52] and ‘iNEXT’ [53], as well as the GLIMMIX procedure in SAS/STAT 9.4 [54].

3. Results

3.1. Classification of All Relevés under Study

The distribution of the Calinski–Harabasz index by cluster number points to an optimal classification of the 89 relevés in eight clusters, each of them showing a high stability, i.e., over 90% (Figure 2). All relevés pertaining to the five plant associations previously described in neighbouring regions were joined separately in five clusters, whereas the relevés performed in the Apuseni Mountains and Dobrogea (Romania) were grouped into the other three clusters. The two largest clusters produced a clear separation between the Artemisia alba communities from the inner continental zone (containing, among others, the Carpathian, Pannonian and central Balkan regions) and those from its external margins (including the Pontic and Dinaric regions).

3.2. Syntaxonomic Assignation of Artemisia alba Scrubs from the Carpathian and Pontic Regions

The three clusters of Carpathian and Pontic relevés were syntaxonomically assigned to as many new plant associations (Figure 2): Acino majoranifolii—Artemisietum albae (Am-Aa,

Table 2. Acino majoranifolii—Artemisietum albae ass. nova hoc loco (* holotypus).

Relevé No.	1	2	3	4	5	6	7	8	9	10 *	11	12	13	14	Frequency (%)
Altitude (m)	680	670	690	680	890	1000	776	733	680	690	670	660	650	660
Aspect	S	SE	E	SE	SV	SE	V	V	S	V	S	V	SE	NE
Slope (degrees)	45	60	55	70	50	10	25	35	25	20	15	30	15	25
Herb cover (%)	85	60	50	45	55	65	75	60	75	80	70	65	75	55
Sample area (sq. m)	25	25	25	25	25	25	25	20	25	15	15	15	15	15
Diagnostic species at association level
Artemisia alba	2.3	3.4	3.4	3.3	1.2	1.2	3.4	2.3	3.4	3.4	3.4	3.4	3.4	3.3	100
Acinos alpinus subsp. majoranifolius	+	+	+.1	+	+.1	+	+	+	.	+	1.2	.	.	.	71
Cytisus albus	.	.	.	+	1.2	1.2	+	.	+	+	.	.	+	+	57
Silene nutans subsp. dubia	+	.	.	.	.	.	.	.	+	+	.	+	+	+	43
Bromo—Festucion pallentis
Phleum montanum	.	.	.	.	+	.	+	+	.	+	+	+	.	+	50
Anthyllis vulneraria subsp. alpestris	.	.	.	.	.	.	1.2	1.2	+	.	.	.	+	+	36
Bromus pannonicus	.	.	.	.	.	.	+	+.1	.	+	+	.	.	+	36
Festuca pallens	.	+	.	+	.	+	.	+	.	1.2	.	.	.	.	36
Seseli osseum	.	.	.	.	.	.	+	+	+	+	.	.	+	.	36
Carduus defloratus subsp. glaucus	.	.	.	.	.	.	+	+	.	.	.	.	.	+	21
Thalictrum minus	.	.	.	.	.	.	.	+	.	+	.	.	.	+	21
Jurinea transsilvanica	.	.	.	+	.	.	.	.	.	.	+	.	.	.	14
Sempervivum marmoreum	+	.	.	+	.	.	.	.	.	.	.	.	.	.	14
Sesleria heufleriana	.	.	.	.	.	+	.	.	.	.	.	.	.	+	14
Stipo—Festucetalia pallentis
Alyssum murale	+	+	+.1	+	.	.	+	+	.	1.2	.	.	.	+	57
Scabiosa columbaria	.	.	.	.	.	.	+	+	.	+	.	+	+	+	43
Anthericum ramosum	.	.	.	.	.	.	+	1.1	+	+	.	.	.	+	36
Melica ciliata	.	.	.	.	.	.	.	+	1.2	1.1	2.3	2.3	.	.	36
Sedum hispanicum	.	.	+	.	.	.	.	.	1.2	+	2.2	+	.	.	36
Stipa pulcherrima	.	.	.	+	.	.	.	.	1.1	.	2.2	1.2	1.1	.	36
Thymus comosus	.	.	.	.	.	.	+	+	.	+	+	.	.	+	36
Erysimum odoratum	.	+	.	.	+	+	+	.	.	.	.	.	.	.	29
Teucrium montanum	.	.	.	.	.	.	1.2	1.1	.	1.2	.	.	2.3	.	29
Helianthemum canum	.	.	.	.	.	.	.	1.3	+	+	.	.	.	.	21
Inula ensifolia	.	.	.	.	+	.	.	.	.	.	.	.	.	.	7
Festuco—Brometea (including Festucetalia valesiacae)
Festuca rupicola	3.3	1.2	1.2	+	1.2	.	.	.	2.3	2.2	+	+	2.2	+	79
Stachys recta	+	+.1	.	+	.	.	+	+.1	+	+	+	+	+	+	79
Helianthemum nummularium subsp. obscurum	.	+	.	+	+	+	.	1.2	+	+	+	+	.	+	71
Koeleria macrantha	1.2	+	+	+	.	.	+	.	+	.	1.2	2.3	+	1.1	71
Teucrium chamaedrys	.	2.3	.	.	.	+	1.1	2.3	+	+	+	+	2.3	1.2	71
Euphorbia cyparissias	+	+	.	.	.	+	.	.	+	+	+	+	+	+	64
Leontodon crispus	.	+	+	+	.	+	.	+	.	+	+	.	+	+	64
Campanula sibirica subsp. divergentiformis	+	+	+.1	+	+	.	+.1	+	.	.	.	.	.	.	50
Carex humilis	.	.	.	.	1.2	.	.	1.1	2.2	2.2	1.2	+	+	.	50
Dianthus carthusianorum	+	.	.	+	.	.	.	.	.	+	+	+	1	+	50
Avenula pratensis	.	.	+	.	.	.	.	+	+	+	.	+	.	1.1	43
Centaurea stoebe subsp. australis	+	.	.	+	.	.	.	+	.	.	.	+	+	+	43
Eryngium campestre	+	+	+	.	.	.	.	.	.	.	+	+	+	.	43
Potentilla cinerea	.	+	.	+	.	.	.	.	2.2	2.2	+	.	2.2	.	43
Alyssum alyssoides	+	+	+	.	.	.	.	.	.	.	.	.	+	+	36
Asperula cynanchica	.	.	.	.	+	.	.	+	.	+	.	+	+	.	36
Potentilla heptaphylla	+	.	.	.	+	.	.	.	.	.	+	+	+	.	36
Allium flavum	.	.	.	+	.	.	.	.	.	+	+	.	+	.	29
Bothriochloa ischaemum	1.1	1.2	.	.	.	.	.	.	+	.	.	.	.	+	29
Bromus erectus	.	.	.	.	.	.	.	+	.	+	+	.	.	+	29
Salvia pratensis	+	.	.	.	.	.	.	+	.	.	.	+	+	.	29
Aster amellus	.	.	.	.	+	.	.	.	.	.	+	.	.	+	21
Centaurea triumfettii	.	.	.	.	+	+	.	+	.	.	.	.	.	.	21
Fragaria viridis	.	.	.	.	.	.	+	.	.	.	.	+	.	+	21
Medicago falcata	.	.	+	.	.	.	.	.	.	.	1.2	.	+	.	21
Phleum phleoides	.	+	+	+	.	.	.	.	.	.	.	.	.	.	21
Sanguisorba minor	.	.	.	.	.	.	+	+	.	.	.	.	+	.	21
Thymus glabrescens	2.2	.	.	.	.	.	.	.	+	.	.	+	.	.	21
Veronica prostrata	.	+	.	.	.	.	.	.	+	.	.	+	.	.	21
Veronica spicata	.	.	.	.	.	.	.	.	+	+	.	+	.	.	21
Astragalus onobrychis	.	.	.	.	.	.	.	1.2	.	.	.	.	+	.	14
Minuartia verna	.	.	.	.	.	.	.	.	+.1	+	.	.	.	.	14
Onobrychis viciifolia	.	.	.	.	.	.	+	1.2	.	.	.	.	.	.	14
Pulsatilla montana subsp. jankae	.	.	.	.	+	.	.	.	.	.	.	.	+	.	14
Bromus riparius	.	.	.	.	.	.	.	+	.	.	.	.	.	.	7
Cephalaria radiata	.	.	.	.	+	.	.	.	.	.	.	.	.	.	7
Pilosella officinarum	.	+	.	.	.	.	.	.	.	.	.	.	.	.	7
Plantago argentea	.	.	.	.	+	.	.	.	.	.	.	.	.	.	7
Prunella grandiflora	.	.	.	.	.	.	.	+	.	.	.	.	.	.	7
Thesium linophyllon	.	.	.	.	.	.	.	+	.	.	.	.	.	.	7
Seslerion rigidae
Helianthemum rupifragum	+	+	+	.	.	.	.	.	+	+	.	.	.	.	36
Sesleria rigida	.	.	.	+	.	+	+	1.1	.	.	.	.	.	2.3	36
Primula veris subsp. columnae	.	.	.	.	.	+	1.2	1.2	.	.	.	.	.	+	29
Seseli gracile	.	.	.	.	.	.	1.1	2.2	.	+	.	.	.	+	29
Helictotrichon decorum	.	.	.	.	2.2	2.2	1.2	.	.	.	.	.	.	.	21
Seseli rigidum	.	.	.	+	.	.	.	.	+	+	.	.	.	.	21
Viola jooi	.	.	.	.	+	+	+	.	.	.	.	.	.	.	21
Centaurea atropurpurea	.	.	.	.	+	.	+	.	.	.	.	.	.	.	14
Centaurea reichenbachii	.	.	.	.	+	.	+	.	.	.	.	.	.	.	14
Dianthus spiculifolius	.	.	.	.	+	+	.	.	.	.	.	.	.	.	14
Ranunculus oreophilus	.	.	.	.	.	+	.	+	.	.	.	.	.	.	14
Geranion sanguinei
Hypericum perforatum	+	+	+	.	+	.	+	.	.	+	.	+	.	.	50
Vincetoxicum hirundinaria	.	.	.	.	+	+	+	+	.	+	+	.	.	+	50
Geranium sanguineum	.	.	.	.	+	+	+	+	.	.	.	+	.	.	36
Verbascum lychnitis	+	+	+	+	.	.	.	.	.	.	.	1.1	.	.	36
Coronilla varia	.	+	+	.	.	.	+	.	+	.	.	.	.	.	29
Trifolium alpestre	.	.	.	.	1.2	.	+	.	.	+	.	+	.	.	29
Cruciata glabra	.	.	.	.	.	.	+	1.1	.	.	.	.	.	+	21
Cnidium silaifolium	.	.	.	.	+	.	.	.	+	.	.	.	.	.	14
Origanum vulgare	.	.	.	.	.	.	+	+	.	.	.	.	.	.	14
Dictamnus albus	.	.	.	.	.	.	.	.	.	.	+	.	.	.	7
Companion
Galium album	.	+	.	+	+	+	1.1	+	.	+	.	+	.	+	64
Bromus arvensis	1.1	+	.	.	.	.	.	.	.	.	.	+	+	.	29
Echium vulgare	.	+	.	.	.	.	+	+	.	.	.	.	+	.	29
Lactuca quercina	+	+	.	+	.	.	.	.	.	.	.	.	.	+	29
Poa angustifolia	.	+	+	+	.	.	.	.	.	.	+	.	.	.	29
Caucalis platycarpos	.	+	.	.	.	.	.	.	.	.	1.1	+	.	.	21
Cytisus nigricans	.	.	.	.	.	.	+	+	.	.	.	.	.	+	21
Orobanche alba	+	.	.	.	.	.	.	.	.	+	.	.	.	+	21
Plantago lanceolata	+	.	.	.	.	.	.	.	.	.	.	+	+	.	21
Vicia pannonica	.	+	.	.	.	.	+	.	.	.	+	.	.	.	21
Asplenium ruta-muraria	.	.	.	.	+	.	.	.	.	+	.	.	.	.	14
Cytisus austriacus	.	.	.	.	.	.	.	+	.	+	.	.	.	.	14
Lepidium campestre	.	+	+	.	.	.	.	.	.	.	.	.	.	.	14
Medicago lupulina	+	.	.	.	.	.	+	.	.	.	.	.	.	.	14
Rhamnus saxatilis subsp. tinctorius	.	.	.	.	.	.	.	+	.	+	.	.	.	.	14
Trifolium pratense	.	+	.	.	.	.	.	.	.	.	.	.	+	.	14
Artemisia campestris	.	.	.	.	.	.	.	.	+	.	.	.	.	.	7
Dianthus giganteus	.	.	.	.	.	.	.	.	.	.	+	.	.	.	7
Peucedanum cervaria	.	.	.	.	.	+	.	.	.	.	.	.	.	.	7
Picris hieracioides	.	.	+	.	.	.	.	.	.	.	.	.	.	.	7
Pilosella hoppeana	.	.	.	.	.	+	.	.	.	.	.	.	.	.	7
Reseda luteola	.	.	.	.	.	+	.	.	.	.	.	.	.	.	7
Trifolium arvense	.	.	.	.	.	.	.	+	.	.	.	.	.	.	7

Locality and date of surveys: Dealul Cetății, 20 June 2018 (relevés 1–4, 9–12); Colții Trascăului, 7 June 2018 (relevés 5–8, 14); Izvoarele, 20 June 2018 (relevé 13).

), Onosmo pseudoarenariae—Artemisietum albae (Op-Aa,

Table 3. Onosmo pseudoarenariae—Artemisietum albae ass. nova hoc loco (* holotypus).

Relevé No.	15	16	17	18 *	19	20	21	22	23	24	25	26	27	28	29	30	Frequency (%)
Elevation (m)	380	385	390	395	400	380	385	390	395	390	395	410	420	390	385	390
Aspect	SW	W	SW	SW	S	S	S	S	SW	S	SW	S	SW	SW	SW	SW
Slope (degrees)	35	35	65	65	70	25	30	10	30	20	25	25	20	10	25	20
Herb cover (%)	70	70	65	65	75	60	55	55	60	45	65	65	60	50	60	60
Sample area (sq. m)	25	25	20	20	25	10	10	10	10	10	10	10	10	10	10	15
Diagnostic species at association level
Artemisia alba	4.4	4.4	4.4	3.4	4.4	3.5	3.4	3.5	3.5	2.5	3.5	3.5	3.5	2.3	3.4	2.3	100
Onosma pseudoarenaria	+	+	+	1.2	+	+	1.2	+.3	+	1.2	+.2	+.3	+	+	.	+	94
Cephalaria radiata	.	+	+	+	.	.	+	+	.	+	+	+	1.2	+	.	+	69
Asyneuma canescens	.	.	.	+	+	+	+	+	+	+	.	.	.	.	.	.	44
Daphne cneorum	.	.	.	+	+	+	.	.	.	.	+	.	.	+	1.2	+	44
Echinops ritro subsp. ruthenicus	+	.	.	+	+	.	.	.	+	.	.	+	.	1.1	.	+	44
Bromo—Festucion pallentis
Seseli osseum	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	.	94
Linum tenuifolium	+	+	+	+	+	+	+.2	+.2	+	+	+	.	.	1.2	.	+	81
Jurinea transsilvanica	+	+	+	.	+	+	.	.	.	+	+	.	+	.	.	+	56
Scorzonera austriaca	+	+	+	+	.	.	.	+	.	.	+	.	.	.	.	.	38
Phleum montanum	.	.	.	.	+	.	.	.	+	.	.	.	+	.	+	+	31
Sesleria heufleriana	.	+	.	+	.	.	.	.	.	+	.	+	.	.	+	.	31
Festuca pallens	.	.	.	.	.	.	.	.	.	+	.	+	+	.	.	.	19
Euphorbia seguieriana	.	.	.	.	.	.	+	.	.	+	.	.	.	.	.	.	13
Seseli gracile	.	.	.	+	.	.	.	.	+	.	.	.	.	.	.	.	13
Thalictrum minus	.	.	.	.	.	.	+	.	.	.	.	.	.	.	.	+	13
Stipo—Festucetalia pallentis
Anthericum ramosum	+	+	+	+	+	+	+	+	+	+	+	+	.	+	+	+	94
Inula ensifolia	+	+	+	+	+	.	1.3	2.5	+	2.3	+	+	1.3	2.4	1.3	2.3	94
Teucrium montanum	+	+	+	+	.	+	1.2	1.2	+	1.2	+	+	+	+	.	+	88
Helianthemum canum	+	+	+	+	+	+	+	.	.	+	1.2	+	.	+	.	+	75
Thymus comosus	.	+	+	+	+	.	+	+	.	1.2	+	+	.	+	1.2	.	69
Astragalus monspessulanum	.	+	.	+	+	.	+	.	+	+	.	+	.	.	.	+	50
Stipa pulcherrima	+	+	.	.	.	.	.	.	.	.	.	.	.	.	+	.	19
Leontodon crispus	.	+	.	.	.	.	.	.	.	.	.	.	+	.	.	.	13
Festuco—Brometea (including Festucetalia valesiacae)
Carex humilis	1.2	+	+	+	1.2	1.2	1.2	1.3	1.2	1.2	2.3	2.3	2.3	2.5	2.3	3.4	100
Asperula cynanchica	.	+	.	+	+	+	+	+	+	+	+	+	+	+	.	1.2	81
Stipa capillata	+	1.2	.	.	+	1.3	2.3	+	+	2.3	1.1	+	1.3	+	.	2.4	81
Teucrium chamaedrys	+	.	1.2	+	+	+	+	+	+	+	.	+	1.2	.	+	1.3	81
Allium flavum	.	+	+	+	+	1.2	+	.	+	1.2	+	+	1.1	.	+	.	75
Campanula sibirica	+	+	+	+	.	.	+	+.2	+	.	+	+	+	.	.	+	69
Adonis vernalis	.	+	.	.	1.1	1.3	+	.	+	.	.	+	+	.	.	+	50
Bothriochloa ischaemum	1.1	.	.	.	+	.	.	+	+	+	2.3	+	+	.	.	.	50
Dorycnium pentaphyllum subsp. herbaceum	+	+	+	+	+	.	.	.	.	+	+	.	.	.	+	.	50
Eryngium campestre	.	+	.	.	+	.	+	.	+	.	.	+	+	.	+	+	50
Helianthemum nummularium subsp. obscurum	+	.	.	+	+	+	+	.	.	.	.	.	+	.	.	+	44
Sanguisorba minor subsp. minor	.	+	+	+	.	.	+	.	+	.	.	.	+	.	+	.	44
Scabiosa ochroleuca	.	.	+	.	+	.	+	+	+	.	.	+	.	.	+	.	44
Thesium linophyllon	.	.	+	+	.	.	+	.	.	+	.	+	.	+	.	+	44
Potentilla cinerea	.	.	.	.	.	2.3	+	.	2.4	.	.	2.3	1.3	.	.	1.3	38
Aster amellus	.	.	.	+	.	.	.	.	+	.	.	+	.	.	+	+	31
Euphorbia cyparissias	.	.	+	+	.	.	.	.	+	.	.	.	.	.	+	+	31
Festuca rupicola	.	.	.	+	.	+	.	.	.	.	+	+	1.1	.	.	.	31
Plantago argentea	+	.	+	.	+	.	.	.	.	.	.	.	.	+	+	.	31
Salvia pratensis	.	+	+	+	+	.	.	.	.	.	.	.	.	.	+	.	31
Euphorbia glareosa	.	.	+	+	+	.	.	.	.	+	.	.	.	.	.	.	25
Kengia serotina	+	.	+	.	+	.	.	.	+	.	.	.	.	.	.	.	25
Onobrychis viciifolia	.	+	.	.	+	.	.	.	.	.	.	.	.	+	+	.	25
Thymus glabrescens	.	.	.	+	+	.	.	.	.	.	.	+	.	.	.	+	25
Bromus erectus	.	.	.	.	.	.	.	.	.	+	.	+	+	.	.	.	19
Centaurea scabiosa subsp. scabiosa	.	.	.	.	.	.	.	.	+	.	.	.	+	+	.	.	19
Odontites luteus	.	.	.	.	.	.	.	+	.	.	.	.	+	.	+	.	19
Bromus riparius	.	.	.	.	.	.	.	.	.	.	.	.	.	+	.	+	13
Festuca valesiaca	.	.	.	+	.	.	.	+	.	.	.	.	.	.	.	.	13
Koeleria macrantha	.	.	.	.	.	+	.	.	.	.	.	.	+	.	.	.	13
Acinos arvensis	.	.	.	.	.	.	.	.	.	.	.	.	+	.	.	.	6
Avenula pratensis	.	.	.	.	.	.	.	.	.	.	.	.	.	+	.	.	6
Bromus squarrosus	.	.	.	.	.	.	.	+	.	.	.	.	.	.	.	.	6
Cytisus albus	.	.	.	.	.	.	.	.	.	.	.	.	.	.	+	.	6
Polygala major	.	.	.	.	.	.	.	.	.	.	.	.	.	.	+	.	6
Pulsatilla montana	.	.	.	+	.	.	.	.	.	.	.	.	.	.	.	.	6
Stachys recta	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	+	6
Thlaspi perfoliatum	.	.	.	.	.	.	.	.	.	.	.	.	+	.	.	.	6
Veronica spicata	.	.	.	.	.	+	.	.	.	.	.	.	.	.	.	.	6
Geranion sanguinei
Dictamnus albus	.	+	.	+	+	+	.	.	.	.	.	+	.	.	.	.	31
Agrimonia eupatoria	.	.	.	.	+	.	.	+	+	.	.	.	.	.	.	.	19
Hypericum perforatum	.	.	.	.	.	.	.	+	+	.	.	.	.	.	.	.	13
Cirsio—Brachypodion pinnati
Salvia verticillata	.	.	.	.	.	+	+	+	+	+	+	+	+	.	.	1.2	56
Brachypodium pinnatum	.	.	+	+	+	.	+	1.2	1.2	+	.	+	.	.	.	.	50
Linum flavum	.	+	+	.	+	.	.	+.2	.	+	.	.	.	.	.	1.2	38
Gypsophila collina	+	.	.	1.1	.	.	.	.	.	.	+	.	.	.	+	+	31
Prunella grandiflora	.	+	.	.	+	.	.	.	.	.	.	.	.	+	.	.	19
Filipendula vulgaris	.	.	.	+	+	.	.	.	.	.	.	.	.	.	.	.	13
Companion
Echium vulgare	+	.	+	+	+	.	.	+	.	.	+	.	+	.	.	.	44
Cytisus nigricans	+	+	.	+	.	.	+	+	.	.	.	.	.	.	.	.	31
Peucedanum oreoselinum	.	.	.	+	+	.	.	+	.	+	.	.	.	.	.	.	25
Carduus hamulosus	.	.	.	.	.	.	.	.	.	.	+	+	.	.	.	.	13
Linum hirsutum	.	.	+	.	.	.	.	.	.	.	.	.	.	.	+	.	13
Polygala vulgaris	.	.	.	+	+	.	.	.	.	.	.	.	.	.	.	.	13
Lactuca quercina	+	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	6
Quercus pubescens	.	.	.	.	.	.	.	.	.	.	.	.	.	+	.	.	6
Rhamnus saxatilis subsp. tinctorius	.	+	.	.	.	.	.	.	.	.	.	.	.	.	.	.	6

Locality and date of publication/surveys: Jebuc, from Karácsonyi (2017) (relevés 15–19); Sfăraș, 13 August 2020 (relevés 20–30).

) and Saturejo coeruleae—Artemisietum albae (Sc-Aa,

Table 4. Saturejo coeruleae—Artemisietum albae ass. nova hoc loco (* holotypus).

Relevé No.	31	32	33	34	35 *	36	37	38	39	40	41	42	43	Frequency (%)
Elevation (m)	65	65	60	60	65	70	70	65	81	65	65	65	70
Aspect	NW	NE	NE	N	N	NE	NE	E	W	W	W	W	S
Slope (degrees)	8	5	5	8	5	5	3	10	3	3	3	5	10
Herb cover (%)	90	90	90	80	75	70	80	65	60	55	90	80	80
Sample area (sq. m)	10	10	10	10	10	10	10	10	10	10	10	10	10
Diagnostic species at association level
Artemisia alba	2.3	2.5	2.5	4.5	3.5	3.5	4.5	3.5	3.5	1.3	2.5	3.3	2.3	100
Satureja coerulea	4.5	4.5	4.5	1.1	3.5	2.5	2.3	2.4	1.3	4.5	4.5	3.4	2.5	100
Koeleria splendens	+	2.4	.	+	1.2	+	.	2.3	2.3	+	1.3	1.3	.	77
Euphorbia dobrogensis	+	.	.	+	.	.	+	+	+	.	+	+	+	62
Jurinea dobrogensis	.	+	+	+.2	.	+	.	.	+	1.3	+	+	.	62
Agropyron ponticum	.	.	.	+	+	1.2	1.3	+	+	.	+	.	.	54
Hyacinthella leucophaea	+	+	.	.	.	+	.	.	+	.	.	+	.	38
Pimpinello—Thymion zygioidi
Thymus zygioides	+	1.2	1.2	1.2	1.2	2.5	+	1.2	1.2	+	.	.	+.2	85
Potentilla bornmuelleri	+	+	.	.	+	1.2	1.3	+	+	+	+	1.3	.	77
Dianthus pseudarmeria	+	.	.	.	+	.	+	+	+	+	1.2	.	.	54
Astragalus spruneri	.	.	.	.	.	.	.	.	.	.	+	+	.	15
Cytisus jankae	.	.	.	.	.	.	.	.	.	2.3	.	+	.	15
Agropyron brandzae subsp. ciliatum	.	.	.	.	.	.	.	.	.	.	.	.	2.4	8
Dianthus nardiformis	.	.	.	.	.	.	.	.	.	.	.	.	+	8
Euphorbia myrsinites	.	.	.	.	.	.	.	.	.	.	.	.	1.3	8
Koeleria lobata	.	.	.	.	.	.	.	.	.	.	.	.	+	8
Bromo—Festucion pallentis
Linum tenuifolium	+	+	+	+	.	+	.	+	.	.	.	+	+	62
Euphorbia seguieriana	.	.	.	.	.	.	.	.	.	.	.	+	+	15
Thalictrum minus	.	.	.	.	.	.	.	.	.	.	.	+	.	8
Saturejon montanae
Leontodon crispus	+	+	+	+	.	.	.	.	.	.	.	+	.	38
Achillea clypeolata	.	.	+	.	.	.	.	+	.	.	.	+	+	31
Convolvulus cantabrica	.	.	.	.	+	.	.	.	.	.	.	.	.	8
Onobrychis alba	.	.	+	.	.	.	.	.	.	.	.	.	.	8
Festuco—Brometea (including Festucetalia valesiacae)
Festuca valesiaca	1.3	1.3	.	1.3	+	1.2	.	+	.	+	1.3	1.4	1.2	77
Sanguisorba minor	+	.	+	+	+	2.3	1.2	.	1.2	+	+	.	.	69
Filipendula vulgaris	1.4	1.3	1.3	1.4	.	.	.	.	+	+	1.2	1.2	.	62
Polygala major	1.3	+	1.3	+	+.3	+	.	+.2	.	+	.	.	.	62
Adonis vernalis	1.2	.	1.2	.	.	.	.	.	+	+	1.3	+	.	46
Bromus riparius	+	+	.	.	+	.	.	+	+	.	+	.	.	46
Onosma visianii	+	.	+	.	+	.	+	.	.	.	+	+	.	46
Bothriochloa ischaemum	+	+	+	.	.	.	.	.	.	.	+	.	+	38
Campanula sibirica	1.3	+	1.3	+	.	.	.	.	+	.	.	.	.	38
Eryngium campestre	.	+	.	+	.	.	.	.	+	.	+	.	+	38
Salvia nutans	.	.	+	.	.	.	+	.	+	+	+.3	.	.	38
Euphorbia agraria	.	.	+	+	+	+	.	.	.	.	.	.	.	31
Euphorbia glareosa	+	.	+	.	.	+	+	.	.	.	.	.	.	31
Haplophyllum suaveolens	.	.	+	.	.	.	+	.	+	.	.	+	.	31
Paeonia tenuifolia	.	.	.	+	+	+	.	.	.	.	.	+	.	31
Stipa lessingiana	.	.	.	+	+	+	1.2	.	.	.	.	.	.	31
Asperula tenella	.	.	.	+	.	.	.	.	.	.	.	+	+	23
Carex halleriana	.	+	+	+	.	.	.	.	.	.	.	.	.	23
Gypsophila pallasii	.	.	.	.	.	.	+	.	.	+	.	+	.	23
Sideritis montana	.	.	.	+	.	+	.	.	.	.	+	.	.	23
Tanacetum corymbosum	.	.	.	1.2	.	+	.	.	+	.	.	.	.	23
Tanacetum millefolium	.	.	.	.	+	.	.	+	.	.	.	.	1.2	23
Teucrium chamaedrys	.	+	.	+.3	.	.	1.1	.	.	.	.	.	.	23
Anthemis tinctoria	.	.	+	.	.	+	.	.	.	.	.	.	.	15
Crupina vulgaris	.	.	.	.	.	.	+	+	.	.	.	.	.	15
Echinops ritro subsp. ruthenicus	.	.	.	.	.	+	.	.	.	.	.	.	+	15
Erysimum diffusum	+	.	.	+	.	.	.	.	.	.	.	.	.	15
Ferulago confusa	.	.	.	.	+	+	.	.	.	.	.	.	.	15
Jasminum fruticans	.	.	+	.	.	.	.	+	.	.	.	.	.	15
Orlaya grandiflora	.	.	.	+	.	.	.	.	.	.	.	.	+	15
Pilosella officinarum	.	.	.	.	.	.	+	+	.	.	.	.	.	15
Veronica spicata subsp. barrelieri	+	.	+	.	.	.	.	.	.	.	.	.	.	15
Bromus squarrosus	.	.	.	.	.	.	.	.	.	.	.	.	+	8
Cephalaria uralensis	.	.	.	.	.	.	.	.	.	.	.	.	1.2	8
Coronilla varia	.	.	+	.	.	.	.	.	.	.	.	.	.	8
Galium flavescens	.	.	.	.	.	.	.	+	.	.	.	.	.	8
Medicago minima	+	.	.	.	.	.	.	.	.	.	.	.	.	8
Stipa capillata	.	.	.	.	.	.	.	.	.	.	+	.	.	8
Teucrium polium subsp. capitatum	.	.	.	.	.	.	.	.	.	.	.	+	.	8
Companion
Lappula marginata	.	.	.	.	.	.	.	.	.	.	+	.	+	15
Peucedanum arenarium	.	.	.	.	.	.	+	+	.	.	.	.	.	15
Salvia ringens	.	.	.	.	.	+	.	1.2	.	.	.	.	.	15
Aegilops cylindrica	.	.	.	.	.	+	.	.	.	.	.	.	.	8
Ajuga laxmannii	.	.	.	+	.	.	.	.	.	.	.	.	.	8
Allium rotundum	.	.	.	.	.	.	.	.	.	.	.	.	+.3	8
Aster oleifolius	.	.	.	.	.	.	.	+	.	.	.	.	.	8
Carthamus lanatus	.	.	.	.	.	.	.	.	.	.	.	.	+	8
Centaurea orientalis	.	.	.	.	.	.	.	.	.	.	.	+	.	8
Consolida regalis	.	.	.	.	.	.	.	.	.	.	.	.	+	8
Dactylis glomerata	.	.	.	+	.	.	.	.	.	.	.	.	.	8
Hypericum elegans	.	.	.	.	.	.	+	.	.	.	.	.	.	8
Iris pumila	+	.	.	.	.	.	.	.	.	.	.	.	.	8
Jurinea tzar-ferdinandi	.	.	.	.	.	.	+	.	.	.	.	.	.	8
Linaria genistifolia	.	.	.	.	.	.	.	.	.	.	.	.	+	8
Linum austriacum	+	.	.	.	.	.	.	.	.	.	.	.	.	8
Nonea atra	.	.	+	.	.	.	.	.	.	.	.	.	.	8
Ononis pusilla	.	.	.	+	.	.	.	.	.	.	.	.	.	8
Potentilla pedata	.	.	+	.	.	.	.	.	.	.	.	.	.	8
Tragopogon dubius	.	.	.	.	.	.	.	.	.	.	.	.	+	8
Xeranthemum annuum	.	.	.	.	.	.	.	.	.	.	.	.	+	8

Locality and date of surveys: Cotu Văii, 2 June 2022 (relevés 31–42); Enisala, 3 June 2022 (relevé 43).

). Within the context of the 89 relevés considered, each of the three mentioned community types stands out through a group of statistically significant discriminant species (

Table 5. Discriminant species best associated with each of the three syntaxa distinguished in the Carpathian and Pontic regions (only the first five species with the highest correlations per group are shown). All coefficients are significant at 0.01% alpha probability.

Species Name	Phi Coefficient
Acino majoranifolii—Artemisietum albae
Acinos alpinus subsp. majoranifolius	0.770
Festuca rupicola	0.664
Cytisus albus	0.641
Scabiosa columbaria	0.629
Silene nutans subsp. dubia	0.629
Onosmo pseudoarenariae—Artemisietum albae
Onosma pseudoarenaria	0.870
Seseli osseum	0.796
Cephalaria radiata	0.725
Inula ensifolia	0.706
Stipa capillata	0.699
Saturejo coeruleae—Artemisietum albae
Satureja coerulea	0.911
Thymus zygioides	0.782
Potentilla bornmuelleri	0.776
Euphorbia dobrogensis	0.764
Jurinea dobrogensis	0.704

) and diagnostic species (

Table 6. Reduced synoptic table including the percentage frequencies (%) of diagnostic species only, for each of the eight syntaxa distinguished. The values in bold are retained as high enough to distinguish between the community types. Abbreviations as in Figure 2.

Diagnostic Species at Association Level	Am-Aa	Op-Aa	Sc-Aa	Aa-Fd	Aa-Ac	Aa-Sa	Bf-Aa	Bm-Se
Acinos alpinus subsp. majoranifolius	71	.	.	.	.	.	.	.
Cytisus albus	57	6	.	.	.	.	.	.
Silene nutans subsp. dubia	43	.	.	.	.	.	.	.
Onosma pseudoarenaria	.	94	.	.	.	.	.	.
Cephalaria radiata	7	69	.	.	.	.	.	.
Asyneuma canescens	.	44	.	.	.	.	.	.
Daphne cneorum	.	44	.	.	.	.	.	.
Echinops ritro subsp. ruthenicus	.	44	15	.	.	.	.	.
Satureja coerulea	.	.	100	.	.	.	.	.
Koeleria splendens	.	.	77	.	60	.	.	.
Euphorbia dobrogensis	.	.	62	.	.	.	.	.
Jurinea dobrogensis	.	.	62	.	.	.	.	.
Agropyron ponticum	.	.	54	.	.	.	.	.
Hyacinthella leucophaea	.	.	38	.	.	.	.	22
Festuca dalmatica	.	.	.	100	.	.	.	78
Thymus praecox subsp. clivorum	.	.	.	67	.	.	.	.
Achnatherum calamagrostis	.	.	.	.	100	8	.	.
Silene armeria	.	.	.	.	20	100	.	.
Galium corrudifolium	.	.	.	.	.	69	.	.
Koeleria pyramidata	.	.	.	.	20	15	100	.
Stipa epilosa	.	.	.	.	.	.	.	100
Thymus striatus	.	.	.	.	.	.	.	89
Hypericum rumeliacum	.	.	.	.	.	.	.	89
Pimpinella tragium subsp. lithophila	.	.	.	.	.	.	.	89
Asphodeline taurica	.	.	.	.	.	.	.	89
Bromus moesiacus	.	.	.	.	.	.	.	56
Tragopogon balcanicus	.	.	.	.	.	.	.	56

).

In agreement with their grouping within the dendrogram (Figure 2) and the frequencies of the characteristic species of the upper syntaxa (see Table S1 in Supplementary Materials), Am-Aa and Op-Aa, on one side and, Sc-Aa, on the other side, were assigned to different alliances and orders within the Festuco-Brometea class, as shown below.

   

3.3. Floristic and Habitat Features of the Carpathian and Pontic Communities (Co)Dominated by Artemisia alba

3.3.1. Acino majoranifolii—Artemisietum albae ass. nova hoc loco

Holotypus: relevé 10 (Table 2)

These communities are distributed in the submontane and montane belt of the Trascău Mountains (650–1000 m a.s.l.), where they form open, scattered scrubs on rocky, and sunny slopes. The underlying soils are hyperskeletic, calcaric Leptosols with a weak alkaline reaction (pH = 7.4–7.6). Along with the dominant Artemisia alba, which reaches on average a relative cover of 30%, a series of regional, diagnostic species for this syntaxon (Acinos alpinus subsp. majoranifolius, Silene nutans subsp. dubia, Cytisus albus) are present (Table 5). A specific particularity of the floristic composition is the occurrence of some Carpathian subendemic species that are characteristic of the alliance Seslerion rigidae (Sesleria rigida, Helictotrichon decorum, Viola jooi, Seseli gracile, Dianthus spiculifolius) (Table 2 and Table S1 in Supplementary Materials).

3.3.2. Onosmo pseudoarenariae—Artemisietum albae ass. nova hoc loco

Holotypus: relevé 18 (Table 3)

These basiphilous, open scrubs develop on the ridges and slopes of the calcareous hills (380–400 m a.s.l.), adjacent to the southern edge of the Meseș Mountains. Soils are classified as skeletic, cambic Leptosols with a weak alkaline reaction (pH = 7.6–7.9). The relative cover of Artemisia alba is 35% on average. The prevalent occurrence of two endemic taxa (Cephalaria radiata and Onosma pseudoarenaria) makes them good diagnostic species for this plant association (Table 5). The sub-xerophilous character of these coenoses is indicated, among others, by the prevalent presence of Seseli osseum and Inula ensifolia (Table 3 and Table 5).

3.3.3. Saturejo coeruleae—Artemisietum albae ass. nova hoc loco

Holotypus: relevé 35 (Table 4)

These xerophilous scrubs have a higher overall relative cover and are codominated by Artemisia alba and Satureja coerulea. They are distributed on the mild but rocky slopes of hills from Dobrogea and, very likely, in north-eastern Bulgaria (western side of the Black Sea). The bedrock consists of limestones underlying hyperskeletic, calcaric Leptosols with alkaline reaction (pH > 8). Several (sub)endemic/diagnostic species such as Euphorbia dobrogensis, Jurinea dobrogensis and Agropyron ponticum (Table 5), together with some thermophilous, south-European species (e.g., Koeleria splendens and Hyacinthella leucophaea), occur frequently in these coenoses that stand out through these phytogeographic particularities (Table 4). These were actually determinants for assigning these communities to the regional alliance Pimpinello—Thymion zygioidi.

3.4. Predictors Contribution to Floristic Dissimilarities between Communities

The ranking of topographic predictors in terms of their importance in classifying the 89 relevés in the eight distinguished plant associations shows that elevation has the highest mean decrease accuracy (Figure 3). The predictive power of all the topographic variables taken together is rather good, since the classification error rate of the corresponding Random Forest model is 13.48%.

The partition of variance in species composition among all predictors revealed a relatively large contribution of ‘Region’ and much lower importance of topographic variables (Figure 4). Among the latter, elevation has a contribution three times higher than slope and aspect. About 46% of the total variance is explained by all predictors based on their linear combination.

Following the forward selection of environmental predictors that should partly explain the variation in species composition, while controlling for the relevé grouping in different regions, all topographic variables in hand show significant but no singular effects, because of their interactions (

Table 7. Output of the forward selection of site variables and their interactions in the partial db-RDA, conditioned by the factor ‘Region’, involving all the 89 relevés under study. Elevation was square-rooted prior to analysis.

Predictors	AIC	F-Value	Prob(>F)
Elevation (m)	259.83	5.528	0.0001
Slope (degrees)	258.42	3.205	0.0001
Elevation × Slope	256.92	3.245	0.0002
Aspect	256.61	2.109	0.0057
Aspect × Slope	255.79	2.543	0.0010

). The topographically constrained and spatially conditioned ordination of relevés reveals a more or less sharp distinction of communities assigned to different associations, except for the relevés of Sc-Aa, Aa-Fd and Bm-Se that are intermixed and display low scores on both axes (Figure 5). The Aa-Sa coenoses appear as the sunniest compared to all the others, whereas the Bf-Aa communities are separately grouped at the positive end of the slope gradient. No consistent patterns with respect to the altitudinal gradient are reflected in the disposal of relevés in the ordination space.

3.5. Patterns in Species Diversity

When controlling for spatial correlations, the elevation, slope and fraction of bare soil display positive and respectively, negative effects on species richness in the scrub communities (Am-Aa, Op-Aa and Sc-Aa) from the Carpathian and Pontic regions (

Table 8. Standardised coefficient estimates of the fixed effects involved in the GLMM fitting the log (species richness) against site variables in all Artemisia alba-(co)dominated communities (n = 43) from the Carpathian and Pontic regions (Romania). Elevation and slope were square-root transformed prior to analysis.

Predictors	Coefficient Estimate	t Value	Prob (>t)	Goodness-of-Fit Statistics
Intercept	3.3256	118.64	<0.0001	Chi-sq./DF = 1
Elevation (m)	1.9706	3.55	0.0011
Slope (degrees)	2.3010	2.93	0.0059
Bare soil (decimals)	−0.4643	−2.24	0.0312
Elevation × Slope	−2.6791	−2.34	0.0251

). Elevation and slope have only simple effects because their (negative) interaction term is also significant. This is mainly due to the positive correlation between the two topographic variables (Spearman’s rho = +0.641; p < 0.0001).

At 10 to 25 m² scale, the Am-Aa coenoses are significantly more species diverse than the other four plant associations under comparison (Figure 6). Op-Aa and Sc-Aa do not differ significantly in terms of species diversity, but are both significantly more diverse than Aa-Fd and Aa-Sa. Finally, the latter two plant associations differ significantly, with Aa-Sa being the least species diverse.

Species diversity, expressed as Hill’s number equivalent of order 1, decreases steadily with increasing shrub relative cover in the three community types (Sc-Aa, Op-Aa and Am-Aa) distinguished in the Carpathian and Pontic regions (Figure 7).

The fraction (relative number) of annuals at community level is significantly lower in Op-Aa than in both Am-Aa and Sc-Aa (Figure 8). The latter includes a slightly higher fraction of annuals than Am-Aa, but the difference is not significant. The relative number of annual species is not related to overall species richness, except for the significant negative relationship detected in Am-Aa coenoses (Figure 9).

A significant increase in species richness with increasing annual relative cover is only noticeable in Sc-Aa (Figure 10). In contrast, a significant decrease is observed in Am-Aa, whereas in Op-Aa the positive slope of the regression line is not significantly different from null (Figure 10).

4. Discussion

4.1. Floristic and Syntaxonomic (Dis)Similarities

Compared with similar communities from the Pannonian, Dinaric and central Balkan regions, those (co)dominated by Artemisia alba from Romania stand out through the presence of several basiphilous, (sub)endemic species (Onosma pseudoarenaria, Cephalaria radiata, Acinos alpinus subsp. majoranifolius, Silene nutans subsp. dubia, Euphorbia dobrogensis, Jurinea dobrogensis, Agropyron brandzae, Dianthus spiculifolius and Viola jooi), most of which are good diagnostic species. Unlike in the physiognomically similar grasslands/scrubs from the neighbouring regions, most of the characteristic species of the alliances Potentillion visianii [42], Saturejion montanae [5,40] and Chrysopogono—Festucion dalmaticae [55] are missing.

The coenoses of Am-Aa do host few characteristic species of the Balkan alliance Saturejion montanae, but the prevalence of a consistent group of species characteristic of Bromo—Festucion pallentis, was decisive for its syntaxonomic assignation. This decision is also supported by the presence of several montane species in Am-Aa that are characteristic of the east-Carpathian alliance Seslerion rigidae Zolyomi 1936. Besides, similar semi-dry communities from the western Carpathians were also assigned to the Bromo—Festucion pallentis alliance [56,57].

The association Op-Aa differentiates itself through a group of heliophilous species, typical for the semi-dry grasslands of Cirsio—Brachypodion pinnati, and its distribution within the colline belt. The floristic composition of Op-Aa partially resembles that of Artemisio—Festucetum dalmaticae Borhidi 1996, which was originally described in southern Hungary but assigned to a distinct alliance, i.e., Chrysopogono—Festucion dalmaticae [55,58].

Compared to the Artemisia alba communities from the Carpathian region, those assigned to Sc-Aa stand out through a series of Pontic and Balkan species and, the absence of characteristic species of the order Stipo—Festucetalia pallentis. Therefore, unlike Mucina et al. [59] who ascribed the alliance Pimpinello—Thymion zygioidi to the order Stipo pulcherrimae—Festucetalia pallentis, we considered the former as part of Festucetalia valesiacae, as initially proposed by Dihoru and Doniță [60]. The Sc-Aa association displays certain floristic affinities with the other three associations from the Dinaric region (Aa-Ac, Bf-Aa and Aa-Sa), but these develop on ultramafic soils and host several serpentinophytes, which determined their inclusion in a distinct alliance (Potentillion visianii) and order (Halacsyetalia sendtneri) [8,42].

4.2. Regional and Local Site Effects on Species Composition/Diversity

The compositional dissimilarity between the two largest clusters of relevés, segregated geographically as Carpathian, Pannonian and Balkan vs. Pontic and Dinaric, may be induced, among others, by certain climatic differences. In fact, this geographical separation follows rather well a continentality gradient, i.e., from more to less continental, respectively [61].

The region of distribution explains by far most of the variation in species composition among single communities. Nevertheless, this strong pattern was artificially enhanced by the aggregated distribution of the studied relevés in different regions. Most of the observed compositional variation is probably induced by geographic vicariants and habitat specialists [16,17]. For instance, long-distance dispersal limitations and mountainous barriers (the Carpathian and Balkan chains) may have played an important role in the compositional distinctiveness of Sc-Aa communities from the Pontic region, while ultramafic substrates are responsible for the occurrence of serpentinophytes in Artemisia alba communities from the Dinaric region.

As reported in other environments and vegetation contexts [13,62,63], the topographic conditions controlling soil moisture prove to be important drivers of the floristic structure of these scrub communities. The effects of the three topographic variables (elevation, slope and aspect) on floristic composition are definitely not singular i.e., independent one from another, given their significant interaction terms. That is very likely the reason for observing some inconsistencies in the ordination space, where—for instance—the Bm-Se and Sc-Aa coenoses are disposed in the middle instead at the two ends of the altitudinal gradient. The ecological compensation resulting from the combined effects of these topographic variables may also lead to similar soil moisture conditions in contrasting habitats, e.g., higher elevation, steep, sunny slopes vs. lower elevation, mild, shady slopes.

The most important topographic variable explaining the compositional variation between both single communities and syntaxa is elevation which, however, shares a part of its explanatory fraction with the factor ‘Region’. This is because, in our data, there are larger inter-region than intra-region differences in sampled site elevation. The relatively high importance of elevation derives obviously from its positive relationship with the amount of precipitation. Although the other two topographic variables (slope and aspect) explain lower fractions of compositional variance, both are important for the floristic distinctiveness of particular plant associations. Thus, the Aa-Sa communities occur exclusively on sunny slopes, whereas the Bf-Aa and Sc-Aa coenoses develop only on steep and mild slopes, respectively.

While accounting for the other site variables, the terrain aspect does not affect the species richness in Artemisia alba scrub communities in the Carpathian and Pontic regions. As expected, elevation and bare soil fraction show positive and negative effects, respectively. These can be easily explained, respectively, by the decreasing water stress toward higher elevations and by the reduction in the net habitat area used once the overall vegetation cover becomes lower and lower. Similar trends along short (low to medium levels) altitudinal gradients and vegetation cover range were generally documented in semi-dry environments [64,65]. The simple positive effect of terrain slope on species richness is totally unexpected at first sight, but explainable after taking into account its positive correlation with elevation and their negative interaction term. In other words, toward higher elevations, the slopes become steeper, but their combined effect on species richness is negative.

The highest species diversity revealed in Am-Aa communities is surely determined by the larger amount of precipitation but, to some extent, may also be the consequence of the matrix effect (sensu Cook et al. [14]), as the surrounding landscape is composed of successionally more advanced plant communities, such as meso-xerophilous closed grasslands and mixed Quercus petraea forest patches [66]. Despite the stronger water stress under the steppic climate in Dobrogea, the Sc-Aa coenoses display a relatively high diversity (undistinguishable from that of Op-Aa counterparts) that may be in part explained by the larger regional (Pontic) species pool [43] and the higher evenness in the distribution of species relative abundance. The even lower diversity of the Aa-Fd communities is probably determined by the high summer temperatures and severe drought in the Pannonian basin [67]. Finally, the lowest estimated diversity observed in the Aa-Sa coenoses is undoubtedly related to the underlying low fertility ultramafic soils [42].

4.3. Annual and Shrub Contribution to Species Diversity

Even in the most diverse (Am-Aa) communities, the relative contribution of annual plant species is rather modest (about 10% on average), probably due to the negative net effect from shrubs. In the Artemisia alba-dominated coenoses pertaining to Romania, the relative number of annual species does not display a positive relationship with overall plant species diversity, as observed worldwide in Mediterranean shrublands [30]. On the contrary, in the least water-stressful conditions, i.e., in Am-Aa communities, plant species richness declines with the increasing fraction of annual species. Such a pattern is probably determined by the presence of annual (competitive) generalists in more benign habitats.

The role of annual abundance in structuring the studied communities from the Carpathian and Pontic regions seems to be very different, depending on the level of water stress. In fact, an increasing relative cover of annuals leads to a decreasing vs. increasing species diversity in the least stressful (Am-Aa) vs. the harshest (Sc-Aa) habitats among those under study. These patterns should be related to the presumed differentiation of strategies (more competitive vs. more stress-tolerant) in annual plants growing in the two contrasting habitats.

The observed decline in overall species diversity with increasing shrub relative cover in each of the three communities (Am-Aa, Op-Aa and Sc-Aa) closely matches the global pattern of woody encroachment in dry grasslands [68,69,70]. This suggests that these dwarf scrub communities may have originated successionally from open, steppe-like grasslands, possibly as a result of release from livestock grazing [70,71]. Although not direct evidence, the negative effect of shrub relative abundance on local species diversity is in accordance with the refined stress gradient hypothesis, according to which asymmetric competition is driven by resource-based (e.g., water) stress [22,72].

4.4. Limitations and Concluding Remarks

As any study, the present one has some limitations determined by inexorable assumptions and overlooked factors. First, we assumed that the negative relationships between species diversity and abundance strictly mirror the asymmetric competition for water, although other kinds of biotic interactions (e.g., allelopathic, trophic) or third-party factors (e.g., eutrophication) can induce similar patterns. Second, data on land-use history and landscape structure, which may have non-negligible effects on plant species composition and richness, were not available. Finally, it must be acknowledged that the observed contribution of shrub and annual abundance to local taxonomic diversity is implicitly a function of the specific traits of species composing the two plant guilds.

In conclusion, the Artemisia alba-(co)dominated community types from Romania are well distinguished both floristically (through dissimilarities between regional species pools) and ecologically (especially through differences in the level of water stress induced by variation in regional climate and elevation). Species diversity declines steadily with increasing shrub relative cover in all the Carpathian-Pontic communities studied. The fraction of annual species is always low but only becomes negatively related to overall species richness under more benign moisture conditions (i.e., in Am-Aa coenoses). The effect of annual relative abundance on species richness depends on the magnitude of water stress, i.e., being positive in the most xerophilous communities (Sc-Aa) and negative in their most mesophilous counterparts (Am-Aa).

Although grassland “invasion” by Artemisia alba may appear as an unfavourable process for the preservation of local plant diversity, the resulting scrubs may play an important protective role against soil erosion and, very likely, represent a seral stage toward the natural recovery of the former xerophilous woodlands. When the main objective is the conservation of grassland species, the use of traditional management (e.g., by grazing cattle) should help limit the abundance of Artemisia alba and other shrubs.