Ecology and floristic composition of heathlands in the Po basin and the Southern Alps (NW Italy)

Abstract

The importance of heathlands as habitats for plants and thus for nature conservation is recognized by European Directive 92/43 (Habitats Directive). However, heathlands are threatened by habitat loss and quality degradation due to several drivers. Temperate Calluna vulgaris communities in the Po basin and in the Southern Alps (NW Italy) are disjunct from the core distribution area in Western Europe and occur at their climatic limits. This study aimed to analyze floristic patterns of heather communities in NW Italy in order to provide detailed recommendations for local conservation needs. Data on plant species composition (phytosociological relevés) and abiotic (environmental and geographical) factors were jointly analyzed using multivariate statistical analyses, to provide a quantitative and statistical interpretation of variation among heathland communities. We show that diversity in species composition was associated with variation in abiotic factors, and we sorted an initial list of “typical species” of the heather’s habitat among indicator species. Several subtypes of heathlands were also recognized and related to habitats “élémentaires”, which require specific conservation measures to preserve their floristic diversity. Finally, we proposed a revised syntaxonomy of heathlands for NW Italy.

Keywords:

• Calluna vulgaris communities
• Habitats Directive
• indicator species
• phytosociology
• heathlands ecology
• typical species

Introduction

Heathlands include temperate shrubby vegetation dominated by heather (Calluna vulgaris (L.) Hull) growing on nutrient-poor soils. In Europe, temperate heathlands are largely distributed in western regions under oceanic influence (Gimingham 1972), where they also exhibit the highest species diversity (González 1998). In some lowland areas of north-western Italy (Lombardy and Piedmont) heathlands were historically so widely distributed that they had a strong influence on toponyms and on socio-cultural heritage for centuries (Brusa and Piazza 2015), similarly as in many other Western Europe countries (Webb 1998). Temperate heathlands are mainly regressive communities that replace woodlands (Ellenberg 1988) and they are mostly anthropogenic in the Po basin and the Southern Alps (Pavari 1927; Tinner et al. 2005; Valese et al. 2014).

Heathlands are part of the landscape and support a large number of species belonging to different systematic groups. The role of heathlands in nature conservation is also recognized by the European Union: “European dry heaths” (cod. Nature 2000: 4030) are listed among habitats of Annex I of European Directive 92/43 (Habitats Directive), which includes all natural habitat types of community value for which conservation special areas of conservation are required to designate in each Member State.

In recent decades the importance of heathland conservation has been stressed as they are threatened by habitat loss and quality degradation due to several drivers (Fagúndez 2012): habitat fragmentation and destruction, pollution and eutrophication, increasing atmospheric CO₂ and climate change, secondary succession and management, and the invasion by alien species. At varying intensities, all these drivers negatively affect heathlands in Southern Alps and in the Po basin (Ascoli and Bovio 2010; Brusa and Piazza 2015; Cerabolini, Ceriani, and De Andreis 1998; Lonati et al. 2009). Moreover, these heathlands are disjunct from the main distribution area of Western Europe heathlands and occur at the climatic limits of Calluna vulgaris.

The Habitats Directive prescribes the protection of habitats against potentially damaging activities. Member States are asked to take appropriate measures to maintain or restore habitat quality and report their conservation state every six years (Evans and Arvela 2011); surveys and monitoring of habitats are necessary to satisfy this requirement. Roughly, most habitats of Annex I can be considered as equivalent to a phytosociological alliance. Some are more narrowly defined while others represent landscape units rather than habitats (Evans 2006). However, the majority of habitat definitions is based on phytosociological syntaxa, often resulting in different interpretations among countries and phytosociological schools (Evans 2010). In addition, monitoring and surveillance of biodiversity in the European Union must include habitats that are not included in the Natura 2000 network. It has been concluded that more supportive information could be developed from existing sources such as the phytosociological literature (Bunce et al. 2013).

The aim of this study was to analyse plant species patterns in heather communities in the Southern Alps and in the Po basin, and to relate these to abiotic (environmental and geographical) factors in order to draw conclusions for conservation. More precisely, we analyzed phytosociological data and tested the possible floristic dependence from previously supposed abiotic gradients (Cerabolini, Ceriani, and De Andreis 1998), with the objective of determining the typical species of “European dry heaths”.

Material and methods

Study area

The study area is located in the Po basin and the Southern Alps and covers two Italian administrative regions (Piedmont, Lombardy) close to the Swiss border (Figure 1). The study area encompasses a latitudinal range of approximately 200 km from the Dora Baltea River (western limit) to the Camonica Valley (eastern limit), and a longitudinal range of approximately 100 km from the alpine Ossola Valley (northward) to the hilly belt near the Po Plain (southward).

Figure 1. The study area. Groups of relevés emerging from the cluster analysis are identified by average values of latitude/longitude.

From the geological point of view, recent deposits of fluvial or fluvial–glacial origin (Holocene) are situated southernmost and at lower altitudes (150–250 m a.s.l.), where gravelly sediments mixed with siliceous sands occur mainly in the valley of the Ticino River (marking the boundary between Piedmont and Lombardy). Older deposits of glacial origin (Pleistocene) are positioned on terraces (200–500 m a.s.l.) on both sides of the Ticino River. Heathlands are located on the most ancient deposits (Mindel) that underwent severe leaching from which clayey soils originated. The hills (400–600 m a.s.l.), where heathlands are limited to the steepest slopes, mainly consist of a conglomerate with siliceous rocks (Pleistocene). The mountains of the northernmost part of the study area cover a wide altitudinal range (300–1200 m a.s.l.). Bedrocks include several types of igneous (e.g. granite) and metamorphic rocks (e.g. gneiss, schist), sharing a common siliceous composition.

In the study area, temperature decreases with altitude of approximately 0.5°C every 100 m (Belloni 1975). Mean annual precipitation ranges from 1000 to 2500 mm, with the highest values found in the mountains near Lake Maggiore (Biancotti et al. 1998; Ceriani and Carelli 1999). According to Gimingham (1972), European lowland heathlands occur only in a temperate oceanic climate (type “Cfb” of Köppen’s climate classification system). A summer arid period is lacking even at low altitudes in the study area; nevertheless, the mean temperature is above 22°C in July; therefore a warm climate (Köppen’s class of “Cfa”) is evident over a large part of lowland heathlands, i.e. outside Gimingham’s range.

Data sampling and analysis

Heathlands of the study area were investigated through extensive field surveys over the past 20 years. Heather communities related to coniferous forests (Vaccinio-Piceetea) have been excluded. The analysis was thus restricted to 75 heathlands of the deciduous forest belt (Querco-Fagetea). At least one phytosociological relevé was recorded in each of the 75 sites, whenever a heather community was subjectively recognized as developed. Plot size ranged between 15 and 100 m². The following abiotic (environmental and geographical) factors were assigned to each relevé:

•	latitude (LAT) and longitude (LONG) coordinates of the administrative municipality where the site is located;
•	altitude (ALT), in meters above sea level;
•	geological substrate: sandy and gravelly soils on recent deposits (REC_DEP); clayey soils on old deposits (OLD_DEP); igneous or metamorphic rocks (ROCK); conglomerate rock.

A presence/absence matrix of 187 species × 392 relevés and a quantitative matrix of six abiotic variables × 392 relevés were jointly analyzed using Canonical Correspondence Analysis (CCA; Ter Braak 1987) in order to explore relationships between species composition and abiotic factors (Monte Carlo test, 9999 permutations). Prior to the analysis, species with less than five presences were removed from the floristic data-set, and geological substrates were converted in dummy variables.

Hierarchical clustering (I) of the six abiotic variables rescaled on 0–1 was performed using Euclidean distance and Ward’s method (Murtagh and Legendre 2014). The best cut level in the dendrogram was determined using the “silhouette” method (Rousseeuw 1987). A subsequent multivariate analysis of variance (MANOVA) tested the consistency of cluster groups on the first two CCA axes. Floristic composition in each cluster group of relevés was derived from the matrix species × relevés. After that, Pearson’s phi coefficient of association (Chytrý et al. 2002) was calculated to statistically analyse the association between a species and cluster groups; in this way, indicator species were detected for each group or cluster of groups (Cáceres and Legendre 2009).

Our own groups of relevés (hereafter denoted by letter G and a number between 1 and 12) were then compared with previously published data sets on heathlands from the same study area, namely:

•	Antonietti (1970): seven variants of a phytosociological “association” from the Swiss mountains (Cantone Ticino) and, to a lesser extent, from Italy, between Lake Maggiore and Lake Como;
•	Giacomini (1958): a table of phytosociological relevés on recent deposits near the Ticino River;
•	Guglielmetto Mugion (1996): a phytosociological table of a heathland on clayey soils from Vauda, Piedmont (west of the study area);
•	Hofer (1967): a data-set from the same area investigated by Antonietti (1970);
•	Lonati and Siniscalco (2010): two groups of relevés from the Southern Alps in the province of Biella (Piedmont);
•	Oberdorfer (1964): a table of phytosociological relevés from the Swiss Alps (Cantone Ticino).

In addition, data sets from north-eastern Italy (Karst: Feoli, Pignatti, and Pignatti 1981; other sites in Friuli: Poldini, Oriolo, and Francescato 2004) were included in the analysis to verify possible penetration of Illyrian (Western Balkan) species in the south central and western Alps, as supposed by Cerabolini et al. (2005).

The final data-set consisted of a matrix of 413 species × 27 data sets (including our 12 groups of relevés and 15 published data sets), in which species values were estimated from the midpoint value of percentage class (V = 90, IV = 70, etc.) shown in each data-set. The matrix was submitted to hierarchical clustering using Bray Curtis’ distance and Ward’s method (cluster analysis II), and “silhouette” method to cut the dendrogram. The association between a species and a group of data sets (i.e. checking for indicator species) was performed through the point biserial correlation coefficient, the abundance-based counterpart of the phi coefficient (Cáceres and Legendre 2009). Habitat preference of each species was inferred from the main phytosociological literature as follows: wetlands (Alnetea glutinosae, Isoeto–Nanojuncetea, etc.), mesophilous grasslands (Molinio–Arrhenatheretea), dry grasslands (Festuco–Brometea), acidophilous grasslands (Nardetea strictae), heathlands (Calluno–Ulicetea), woodland herbaceous edges (Trifolio–Geranietea) and forests (Erico–Pinetea and Querco–Fagetea).

Data on Calluna vulgaris were removed from all data sets. Species nomenclature and syntaxonomic names follow respectively Conti et al. (2005) and Biondi et al. (2014). Statistical analyses were computed using some packages in the R software distribution (R Core Team 2015).

Results

In the CCA, 14.9% of the total variance was explained by all the abiotic variables. Monte Carlo test of the overall model, single abiotic variables and axes consistently gave p-values <0.001, providing a statistical validation of the coupling of the floristic data-set and abiotic factors. The proportion of the constrained variance explained by the first two CCA axes (Figure 2) was, respectively, 34.3% and 21.1% (18.4% for CCA axis 3). Latitude, altitude and rocky substrate, and, conversely, old deposits, accounted for most of the variation along the first axis; the two remaining abiotic variables, longitude and recent deposits, explained most of the variation along the second axis.

Figure 2. CCA ordination. Only factors and groups of rlevés (G1–G12) from cluster analysis are shown in the biplot.

Twelve groups were retained after the cluster analysis (I) on abiotic variables (Figure 3) that were well separated (MANOVA, Pillai’s trace, F_22,760 = 151.32, p < 0.001) along the first two CCA axes (Figure 2):

•	G1: on old deposits in the western part of the study area (Piedmont);
•	G2: on recent deposits in central study area (along the Ticino River);
•	G3: on old deposits, like G1, but in the eastern part of the study area (Lombardy);
•	G4: on conglomerate rock;
•	G5–G6: on siliceous rocks in the eastern part of the study area (Lombardy), at low (G6) or high (G5) altitude;
•	G7–G12: on siliceous rocks, like G5–G6, but in the western part of the study area (Piedmont and Lombardy); G7–G12 groups are separated from each other by geographic coordinates and altitude.

Figure 3. Cluster analysis (I) on the abiotic factor data of the relevés. Twelve groups (G1–G12) were recognized in the dendrogram.

Single groups or clusters of groups recognized by cluster analysis (I) on the basis of geological and geographical variables were denoted by one or more indicator species (Table 1). The cluster groups G5–G6 and G7–G12 were split after node 6 (i.e. following the separation from other groups); thus, in checking indicator species, it was far more convenient and practical to consider them as two main groups: they shared several species (Juniperus communis, Polygala chamaebuxus, Solidago virgaurea, etc.) and exhibited more exclusive species (G5–G6: Erica carnea, Phyteuma scheuchzeri, Brachypodium rupestre, etc.; G7–G12: Festuca acuminata, Allium lusitanicum, Euphorbia cyparissias, etc.). Similarly, G1 and G3 shared several species (Frangula alnus, Populus tremula, Gentiana pneumonanthe, etc.) but also exhibited exclusive species (G1: Salix rosmarinifolia, Carex panicea, Serratula tinctoria, etc.; G3: Salix caprea, Juncus tenuis, Lotus pedunculatus, etc.). On the other hand, G2 and G4 had only two species in common (Hypericum perforatum and Aira caryophyllea) and differed by the presence of alien species (G2; Dichanthelium acuminatum, Pinus rigida, Prunus serotina, etc.) or trees (G4: Castanea sativa, Quercus pubescens, Ostrya carpinifolia).

Table 1. Indicator species of groups G1–G12 (cluster analysis I) recognized on the base of the phi coefficient of association. Only the first 12 species are listed for each group or cluster of groups in the dendrogram, if statistically significant. Figures are the percentage of relevés occupied by a given species (those that are statistically significant are shown in bold).

Species	Phicoeff.	Stat.sign.	Groups
			G5–6	G7–12	G3	G1	G4	G2
Erica carnea	0.932	***	90	1	0	0	0	0
Phyteuma scheuchzeri	0.690	***	64	12	0	0	0	0
Brachypodium rupestre	0.673	***	79	9	3	16	9	2
Viola canina	0.667	***	69	13	3	3	0	5
Arctostaphylos uva-ursi	0.617	***	44	1	0	0	0	0
Lathyrus linifolius	0.571	***	41	4	0	0	0	0
Campanula scheuchzeri	0.554	***	36	1	0	0	0	0
Picea abies	0.545	***	41	6	1	0	0	0
Platanthera bifolia	0.522	***	36	0	0	4	0	0
Carlina acaulis	0.503	***	36	7	0	0	0	0
Larix decidua	0.497	***	28	0	0	0	0	0
Viola hirta	0.492	***	38	8	3	0	0	0
Juniperus communis	0.582	***	51	36	0	0	0	0
Polygala chamaebuxus	0.464	***	51	28	1	1	13	0
Solidago virgaurea	0.364	***	46	47	14	13	25	2
Prunella grandiflora	0.309	***	18	9	0	0	0	0
Pseudolysimachion spicatum	0.272	***	10	11	0	0	0	0
Sorbus aucuparia	0.253	**	13	17	1	0	6	0
Sedum montanum	0.233	**	10	6	0	0	0	0
Koeleria pyramidata	0.226	**	10	5	0	0	0	0
Gymnadenia conopsea	0.164	*	5	3	0	0	0	0
Festuca acuminata	0.650	***	0	47	0	0	0	0
Allium lusitanicum	0.390	***	0	18	0	0	0	0
Euphorbia cyparissias	0.369	***	0	16	0	0	0	0
Genista pilosa	0.309	***	0	11	0	0	0	0
Thalictrum minus	0.281	***	0	9	0	0	0	0
Hylotelephium maximum	0.228	**	0	9	0	0	3	0
Rumex scutatus	0.217	**	0	6	0	0	0	0
Potentilla rupestris	0.217	**	0	6	0	0	0	0
Plantago lanceolata	0.203	*	0	7	3	0	0	0
Festuca laevigata	0.198	*	0	5	0	0	0	0
Colchicum alpinum	0.198	*	0	5	0	0	0	0
Tilia cordata	0.198	*	0	5	0	0	0	0
Salix caprea	0.461	***	8	3	44	11	0	2
Juncus tenuis	0.242	**	0	0	11	3	0	2
Lotus pedunculatus	0.252	**	0	0	8	0	0	0
Pinus strobus	0.198	*	0	0	8	3	0	0
Frangula alnus	0.461	***	36	13	91	79	47	48
Populus tremula	0.538	***	21	25	86	91	44	38
Gentiana pneumonanthe	0.511	***	0	7	26	53	0	0
Lysimachia vulgaris	0.312	***	0	0	16	14	0	2
Carex pallescens	0.222	**	0	5	15	9	0	2
Juncus conglomeratus	0.261	**	0	0	10	9	0	0
Lythrum salicaria	0.211	**	0	0	9	4	0	0
Carex demissa	0.292	***	0	0	9	16	0	0
Holcus lanatus	0.198	**	0	3	8	9	0	0
Centaurium erythraea	0.187	*	0	1	6	6	0	0
Viburnum opulus	0.310	***	0	0	1	13	0	0
Rosa gallica	0.341	***	0	0	0	17	0	3
Carpinus betulus	0.272	***	0	0	4	17	6	0
Briza media	0.315	***	3	7	0	20	0	0
Gladiolus imbricatus	0.430	***	0	0	0	21	0	0
Agrostis canina	0.374	***	0	0	10	26	0	0
Nardus stricta	0.271	***	8	15	10	29	0	0
Drosera intermedia	0.538	***	0	0	0	33	0	0
Rhynchospora fusca	0.538	***	0	0	0	33	0	0
Salix rosmarinifolia	0.544	***	0	0	13	46	0	0
Carex panicea	0.843	***	0	0	4	79	0	0
Serratula tinctoria	0.713	***	0	14	23	86	3	0
Castanea sativa	0.471	***	36	30	24	9	84	23
Teucrium scorodonia	0.490	***	5	41	15	6	81	36
Quercus pubescens	0.507	***	5	12	0	0	50	8
Ostrya carpinifolia	0.435	***	0	0	0	0	22	0
Silene nutans	0.313	***	3	7	0	0	22	3
Thesium linophyllon agg.	0.275	**	5	6	0	1	19	0
Hypericum perforatum	0.293	***	21	21	8	3	44	33
Aira caryophyllea	0.304	***	0	0	0	0	16	11
Filago arvensis	0.257	**	0	0	0	0	0	8
Filago minima	0.257	**	0	0	0	0	0	8
Micropyrum tenellum	0.237	**	0	1	0	0	0	8
Erigeron canadensis	0.212	**	0	0	1	0	3	9
Phytolacca americana	0.305	***	0	0	0	0	0	11
Acer campestre	0.254	**	0	0	0	0	3	11
Teesdalia nudicaulis	0.366	***	0	0	0	0	0	16
Dichanthelium acuminatum	0.419	***	0	0	0	0	0	20
Pinus rigida	0.481	***	0	0	8	0	0	34
Prunus serotina	0.611	***	0	0	6	0	0	48

* p≤0.05;

** p≤0.01;

*** p≤0.001.

The floristic relationships among data sets are revealed in the dendrogram produced by the cluster analysis (II), which returned seven groups (A–G), as shown in Figure 4. Floristic features of groups were well supported by indicator species (Table 2). Wetland species were strict indicators of group A. Forest species were shared by all groups, although single species were indicators of one (e.g. group C: Pinus nigra and Sesleria autumnalis), two (groups C and D: Erica carnea; groups A and B: Quercus robur), or three groups (e.g. Fraxinus ornus in C, D, and F). Species of mesophilous grasslands were also shared by all groups; Molinia caerulea subsp. arundinacea occurred in all groups, except in group C. Groups C and F were poor in species of acidophilous grassland. Groups C and E exhibited many species of dry grasslands as indicators, but they shared only Scabiosa columbaria s.l. Heathland species were under-represented in Group A, unlike the other groups. Species of woodland fringes occurred mainly in group C and were rare in groups B and D.

Figure 4. Cluster analysis (II) on data sets from our unpublished relevés (G1–G12) and from publications regarding heathlands in the study area. Seven groups (A–G) were recognized in the dendrogram, correspondingly assigned to five associations: A, Salici rosmarinifoliae–Callunetum; B, Jasiono montanae–Callunetum; C, a nameless Illyrian association; D, Cytiso supini–Antennarietum; E–F, Chamaecytiso hirsuti–Callunetum.

Table 2. Indicator species of cluster analysis (II) recognized on the base of the point biserial correlation coefficient. Species are arranged according to habitat preference. Figures correspond to classes of presence (those that are statistically significant are shown in bold).

Habitat/species		Groups		A			B		C	D			E				F			G
		corr.coeff.	stat.sign.	a	G1	G3	B	G2	c	d	G5	G6	e	f	g	h	G8	i	j	G11	G10	k	G4	G7	G9	G12	l	m	n	o
Wetlands
	Viburnum opulus	1.000	**	I	I	I	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Hypericum humifusum	0.950	**	I	I	I	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	I	.	.	.	.	.	.	.
	Agrostis canina	0.853	*	I	II	I	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Salix rosmarinifolia	0.753	*	I	III	I	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
Mesophilous grasslands
	Bellis perennis	1.000	*	.	.	.	.	.	I	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Galium album s.l.	0.986	*	.	.	I	.	.	III	.	.	I	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Succisa pratensis	0.871	*	III	III	II	.	I	.	.	II	.	.	.	I	.	.	.	.	I	.	I	.	.	I	.	I	.	I	.
	Molinia c. subsp. arundinacea	0.837	*	V	V	V	V	V	.	III	V	V	I	IV	III	IV	V	V	V	III	V	V	V	V	V	IV	V	IV	V	V
	Serratula tinctoria	0.826	**	IV	V	II	.	.	V	II	.	.	.	.	I	I	IV	.	.	.	.	III	I	I	II	.	II	I	I	II
	Carex panicea	0.806	*	II	IV	I	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Genista tinctoria	0.762	*	IV	V	III	III	II	.	II	IV	III	I	I	I	II	I	III	II	I	.	.	.	I	I	.	I	.	II	I
Dry grasslands
	Centaurea jacea subsp. weldeniana	1.000	*	.	.	.	.	.	IV	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Eryngium amethystinum	1.000	*	.	.	.	.	.	IV	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Scorzonera villosa	1.000	*	.	.	.	.	.	I	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Dorycnium pentaphyllum s.l.	0.979	*	.	.	.	.	.	V	II	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Helianthemum nummularium	0.954	*	.	I	.	.	I	IV	I	I	I	.	I	II	I	.	.	.	II	.	.	I	.	.	.	.	.	I	.
	Dianthus carthusianorum	0.924	**	.	I	.	.	.	.	.	.	.	IV	III	III	II	.	.	.	I	.	.	I	.	.	.	I	I	II	I
	Hylotelephium maximum	0.910	**	.	.	.	.	.	.	.	.	.	V	IV	III	II	.	.	.	II	I	.	I	.	.	I	.	I	I	.
	Phleum phleoides	0.832	*	.	.	.	.	.	.	.	.	.	I	I	II	I	.	.	.	I	.	.	.	.	.	.	.	.	.	.
	Artemisia campestris	0.832	*	.	.	.	.	.	.	.	.	.	I	I	II	I	.	.	.	.	.	.	.	I	.	.	.	.	.	.
	Scabiosa columbaria s.l.	0.803	*	.	.	.	.	.	II	.	I	.	I	I	IV	II	.	.	.	I	.	.	I	.	.	.	.	.	I	.
	Galium lucidum	0.803	*	.	I	.	.	I	.	.	.	I	IV	II	III	II	.	.	.	I	.	.	I	II	I	.	II	III	I	II
	Euphorbia cyparissias	0.794	*	.	.	.	.	.	III	I	.	.	II	.	III	I	.	.	.	III	.	.	.	I	I	.	.	I	I	I
	Silene otites	0.728	*	.	.	.	.	.	.	.	.	.	II	.	I	I	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Anthericum liliago	0.726	*	I	II	.	.	I	.	.	IV	II	III	III	II	III	IV	III	II	IV	IV	I	V	I	III	II	III	III	V	III
Acidophilous grasslands
	Festuca acuminata	0.834	**	.	.	.	.	.	.	.	.	.	V	II	III	III	.	.	.	III	IV	I	.	III	III	II	III	V	I	I
	Danthonia decumbens	0.798	**	III	V	IV	.	I	I	IV	IV	II	I	.	II	II	III	II	I	III	V	III	III	III	V	II	II	III	III	III
	Carex pilulifera	0.754	*	IV	V	IV	.	IV	.	.	.	II	.	.	.	II	.	.	I	II	III	IV	III	IV	V	II	I	III	III	III
	Jasione montana	0.742	*	.	I	.	IV	I	.	.	.	.	.	I	I	.	.	.	.	I	.	I	I	.	.	.	I	.	I	.
	Potentilla erecta	0.734	*	V	V	V	IV	III	.	IV	V	IV	.	.	II	II	V	.	II	IV	I	IV	II	II	IV	III	III	III	IV	V
	Festuca tenuifolia	0.726	*	IV	IV	IV	II	II	.	I	V	III	II	III	II	II	.	I	.	V	II	III	III	II	III	II	I	III	III	III
	Silene rupestris	0.721	*	.	.	.	II	I	.	.	I	II	III	II	I	II	.	.	.	III	IV	I	III	III	III	II	II	II	II	I
Heathlands
	Cytisus hirsutus	0.926	*	.	I	.	.	.	V	II	III	.	.	.	.	.	.	I	.	.	.	.	I	.	I	.	.	.	.	.
	Cytisus scoparius	0.891	***	.	I	III	V	V	.	.	.	.	V	V	III	V	.	II	II	I	V	IV	V	II	IV	IV	V	V	V	V
	Cytisus nigricans	0.852	**	.	I	I	.	I	V	IV	V	II	I	.	.	II	II	II	II	II	I	IV	I	II	I	.	II	II	II	II
	Teucrium scorodonia	0.805	**	.	I	I	.	II	.	.	I	.	I	IV	II	III	V	III	IV	I	II	II	V	IV	II	IV	IV	IV	IV	V
	Pteridium aquilinum	0.702	*	I	I	III	I	II	.	III	III	III	I	III	II	IV	IV	III	V	I	.	IV	IV	III	III	IV	V	IV	V	V
Fringe woodlands
	Buphthalmum salicifolium	1.000	*	.	.	.	.	.	I	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Iris graminea	1.000	*	.	.	.	.	.	I	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Tanacetum corymbosum	1.000	*	.	.	.	.	.	IV	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Trifolium rubens	0.996	*	.	.	.	.	.	V	I	.	.	.	.	.	I	.	.	.	I	.	.	.	.	.	.	.	.	.	.
	Rosa gallica	0.988	*	I	I	.	.	I	IV	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Inula hirta	0.984	*	I	I	.	.	.	IV	I	.	.	.	.	.	I	I	.	.	.	.	.	I	.	I	.	.	.	.	.
	Geranium sanguineum	0.884	*	.	.	.	.	.	IV	II	.	.	II	.	III	I	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Cistus salvifolius	0.866	**	.	.	.	.	.	V	II	.	.	III	II	III	IV	.	.	.	.	.	.	.	.	.	.	III	II	II	IV
	Peucedanum cervaria	0.831	**	I	I	.	.	.	V	I	.	.	III	III	IV	I	.	I	.	.	.	.	.	.	.	.	II	III	III	IV
	Vincetoxicum hirundinaria	0.785	**	I	II	.	.	I	.	I	II	III	IV	III	V	V	V	II	II	II	V	V	III	II	III	IV	V	V	IV	V
	Amelanchier ovalis	0.757	*	.	.	.	.	.	.	.	I	II	.	II	.	I	II	V	II	.	.	.	I	I	I	.	I	I	I	I
	Hieracium umbellatum agg.	0.781	*	II	III	IV	.	I	III	I	III	I	.	.	.	III	.	.	.	II	.	.	I	.	I	.	.	.	.	.
Forests
	Lathyrus niger	1.000	*	.	.	.	.	.	III	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Silene italica	1.000	*	.	.	.	.	.	II	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Pinus nigra	0.995	*	.	.	.	.	.	V	.	.	.	.	.	.	.	.	I	I	.	.	.	.	I	.	.	.	.	.	.
	Festuca heterophylla	0.989	*	.	I	.	.	.	III	.	.	.	.	.	.	.	.	.	.	I	.	.	.	.	.	I	.	.	.	.
	Sesleria autumnalis	0.979	*	.	.	.	.	.	V	II	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
	Erica carnea	0.944	**	.	.	.	.	.	V	II	V	IV	.	.	.	I	.	.	.	.	.	.	.	.	.	I	.	.	.	.
	Sorbus aria	0.866	**	.	.	.	.	.	.	.	.	.	.	I	I	II	V	IV	V	II	III	.	IV	I	II	IV	III	II	.	I
	Crataegus monogyna	0.855	**	.	I	.	.	I	III	.	.	II	.	I	II	I	.	.	.	.	.	.	.	I	I	.	I	II	II	I
	Quercus robur	0.835	*	I	IV	IV	II	III	.	.	.	.	.	.	.	.	.	.	.	.	III	.	I	I	.	.	.	.	.	.
	Populus tremula	0.799	*	I	V	V	.	II	.	II	I	II	.	.	I	I	V	V	III	I	II	.	III	II	I	I	I	I	I	.
	Fraxinus ornus	0.795	*	.	.	I	.	I	IV	II	II	IV	.	.	.	.	.	V	V	I	II	.	III	I	.	.	.	.	.	.
	Stachys officinalis	0.794	**	.	II	I	.	I	V	II	III	II	III	I	III	II	I	.	.	II	IV	I	I	.	III	I	I	III	IV	III

G1–G12: groups recognized in the cluster analysis (I).

a: Guglielmetto Mugion (1996); b: Giacomini (1958); c: Poldini, Oriolo and Francescato (2004); d: Feoli, Pignatti and Pignatti (1981); e–f–g: Antonietti (1970), “Gryllo–Callunetum” (subass. Allium senescens, var. 1–3–2); h: Hofer (1967); i–j: Lonati and Siniscalco (2010), var. edaph. C. gryllus, var. edaph. B. pendula; k: Oberdorfer (1964); l–m–n–o: Antonietti (1970): “Gryllo–Callunetum” (subass. typical, var. 2–1 and subass. Melampyrum, var. 1–2).

Discussion

Plant species composition and indicator species

In the present study, we detected variation in floristic composition even if the groups of relevés were not primarily based on species but on abiotic factors. The identification of diagnostic species (e.g. indicator species) by means of statistical analysis is an important step in vegetation science, because these species can be used to characterize and indicate plant community types (Chytrý and Tichý 2003). If relevés are independently classified from floristic data, as in the present study, the significance of statistical tests performed on the indicator species will be meaningful, i.e. the indicator species can thus be fully considered as indicators, in the true sense of the word (Borcard, Gillet, and Legendre 2011).

Instead of describing plant communities using concepts based on subjectivity, e.g. the assertion of character and differential species (Chytrý and Tichý 2003), phytosociologists should refer to statistical methods in data analysis, even for the most simple studies (Kent 2012) and where possible include more rigorous sampling designs (Chiarucci 2007). Preferential relevés made using the phytosociological approach (i.e. a preferential sampling) can be analyzed by statistical tests (Botta-Dukát et al. 2007), although results must be interpreted with caution (Michalcová et al. 2011). In order to avoid any possible bias, our floristic-independent groups of relevés were compared with each other and then with published data sets by means of statistical tests. In this way, differences in species composition (i.e. the indicator species in Tables 1 and 2) can be considered robust as it is based on the “modern” concept of fidelity (Chytrý et al. 2002).

Our results have practical consequences for the implementation of the Habitats Directive. A habitat is considered to have a favorable conservation status when the “typical species” they support themselves grow under favorable conditions. The term “typical species” is not defined by the Habitats Directive although conceivably typical species should be selected to reflect favorable structure and functioning of the habitat type. Within a given country, different “typical species” may be needed for different parts of the habitat range or for different subtypes of this habitat (Evans and Arvela 2011). The selection of “typical species” and habitat subtypes are primarily based on floristic data (Maciejewski 2010). In the present study, indicator species of each cluster group reflect particular ecological conditions or even biogeographic patterns in heathland habitat. The loss of indicator species could reveal change in habitat quality (e.g. decrease of wetland species in response to drainage or decreased rainfall) and therefore deterioration of its conservation status. An initial list of “typical species” could be derived from indicator species but, it seems unreasonable to consider that indicator species can include all typical species. For example, Gentiana pneumonanthe was found to be more abundant at sites with the endangered butterfly Maculinea (Phengaris) alcon (Maes and Dyck 2005), therefore it could be considered a “typical species” for wet heathlands. Indeed, G. pneumonanthe was selected as an indicator species by cluster analysis of relevés (Table 1, G1 and G3), but our analysis failed to find it when other data sets were evaluated (in Table 2, G. pneumonanthe was not reported, as the point biserial correlation coefficient equaled 0.826 for group A but was at limit of the statistical significance, p = 0.063). Finally, additional criteria could be considered for the determination of “typical species”, but indicator species should be the best choice on a local scale.

Syntaxonomic outcomes

The diversity in floristic composition recognized in the present study allows the syntaxonomy of heathlands in the Po basin and the Southern Alps to be revised. We can propose two improvements: a division of the heathland habitat into different subtypes and the assessment of the corresponding “typical species”. The subtypes correspond to “habitats élémentaires” (elementary habitats; Bensettiti 2001–05), which require specific conservation measures to preserve their floristic diversity. However, as pointed out (Table 2), most species emerging as indicator species are not strictly heathland species. The reasons are probably the intrinsic nature of heathlands, i.e. a transient stage in the ecological succession of plant communities, and mostly to the high impact of traditional use in the study area (Brusa and Piazza 2015). For the latter reason, several “typical species” are vanishing, especially at low altitude.

Group A. Three data sets are included: Guglielmetto Mugion (1996) and two groups of our relevés (G1 and G3). All characterize heathlands on clayey soils from old deposits, where waterlogging occurs in the topsoil layers. “Typical species” are wetland species such as Viburnum opulus, Hypericum humifusum, Agrostis canina, Salix rosmarinifolia (from Table 2) and Lotus pedunculatus, Gentiana pneumonanthe, Lysimachia vulgaris, Juncus conglomeratus, Lythrum salicaria, Carex demissa, Drosera intermedia, Rhynchospora fusca, and Carex panicea (from Table 1). Wet heathlands can thus be recognized as a separate community. Salix rosmarinifolia, a bush willow that grows scattered throughout the same layer of heather, appears to be the best candidate species for naming the new association: Salici rosmarinifoliae–Callunetum vulgaris (Genisto–Vaccinion alliance).

Group B. Two data sets are included: Giacomini (1958) and one group of our relevés G2, from gravelly deposits mixed with sands near the Ticino River. In contrast with the lowland heathlands of group A, they exhibit porous soils. Short lived species, flowering in spring before summer drought, can be retained as “typical species”: Filago arvensis, Filago minima, Micropyrum tenellum, Teesdalia nudicaulis, Aira caryophyllea (Table 1), and Jasione montana (Table 2); the endemic Euphrasia cisalpina, recently retrieved in the study area (Martignoni 2014), should be considered as another typical species of group B. Oberdorfer (1964) reported these heathlands as Cytiso hirsuti–Callunetum vulgaris. However, his original description clearly referred to a heathland on bedrock (Montorfano, i.e. our G7 groups of relevés) and has been questioned (Cerabolini, Ceriani, and De Andreis 1998; Hofer 1967). Our analysis supports the distinctiveness of our group from Oberdorfer’s association, so that we propose a new association: Jasiono montanae–Callunetum vulgaris. In addition, the lack of mountain species (Arctostaphylos uva-ursi, Festuca acuminata, Genista pilosa, Phyteuma scheuchzeri, Polygala chamaebuxus, etc.) establishes its inclusion in the alliance Genistion tinctorio-germanicae

Group C. This group includes only one data-set from north-eastern Italy and was assigned to the Genisto–Callunetum illyricum (Feoli, Pignatti and Pignatti 1981). Due to the lack of typical Balkan species, Poldini, Oriolo, and Francescato (2004) considered it as Chamaecytiso hirsuti–Callunetum vulgaris (Oberdorfer 1964), the south-eastern variant of the central European Cytiso supini–Callunetum vulgaris O. Bolós 1956 (=Cytiso supini–Antennarietum Preising 1953, according to Ellmauer 1993). However, the heathland includes exclusive Illyrian species (Centaurea jacea subsp. weldeniana, Eryngium amethystinum, Scorzonera villosa, Sesleria autumnalis) and many species of dry grasslands. The group is confirmed as an outlier in our analysis and its attribution to Oberdorfer’s associations is very unreliable. Its classification should be amended in an analysis involving Illyrian scrublands.

Group D. Three data sets are included in this group. Our two groups of relevés (G5 and G6) are from the eastern part of the study area, thus providing a floristic link to Friuli heathlands (Poldini, Oriolo and Francescato 2004) classified as Chamaecytiso hirsuti–Callunetum vulgaris. However, Group D is clearly separated from Group C (Karst heathlands) as reported above; additionally, it is different from western heathlands (groups E–F, including the true Chamaecytiso hirsuti–Callunetum vulgaris). The “typical species” are Cytisus nigricans, Erica carnea, Fraxinus ornus (Table 2) and Phyteuma scheuchzeri, Brachypodium rupestre, Viola canina, Arctostaphylos uva-ursi, Lathyrus linifolius, Campanula scheuchzeri (Table 1). Group D occurs on dry, sunny and warm mountains, and it presents floristic affinities with northern Apennine heathlands (Angiolini et al. 2007; Nowak 1987) which are assigned to the Erico herbaceae–Genistetum pilosae Oberdorfer and Hofmann 1967 ex Nowak 1987;. However, Deschampsia flexuosa, Genista pilosa, and Vaccinium myrtillus are infrequent in group D and in the original table of the association (Oberdorfer and Hofmann 1967); in addition, the Mediterranean Erica arborea is lacking. Surprisingly, there is a strong similarity with the floristic list of the “Calluno–Ericetum” described from the Austrian Southern Alps (Onno 1933) and currently included p.p. within the Cytiso supini–Antennarietum dioicae (Ellmauer 1993). Pending a complete phytosociological review of the Alpine heathlands, it seems more reasonable to provisionally assign group D to the latter association.

Groups E and G. These two groups are treated together due to the strong similarities in their species composition. Many species of dry grasslands are present in group E, in contrast with species of acidophilous grasslands occurring mainly in group G. Accordingly, “typical species” of group E include Dianthus carthusianorum, Hylotelephium maximum, Artemisia campestris, Phleum phleoides, Galium lucidum, Scabiosa columbaria s.l., Silene otites; and for group G, Danthonia decumbens, Potentilla erecta, and Carex pilulifera. However, Festuca acuminata, an endemic species mainly from the western Alps (Wallossek 1999) is the most prominent “typical species” of both groups. The groups include the provisional association “Gryllo–Callunetum” in all its variants (Antonietti 1970) and data sets reported in Hofer (1967) and in Oberdorfer (1964). There is a continuum in the floristic composition of the western heathlands on bedrock, largely due to anthropogenic factors, rendering irrelevant any attempt to subdivide the groups into new associations. We thus assign these groups to the association Chamaecytiso hirsuti–Callunetum vulgaris described by Oberdorfer (1964); however, a new subassociation could be recognized where F. acuminata occurs. Nevertheless, a group of our relevés (G4) lacks F. acuminata because it includes heathlands on sedimentary rocks at low altitude; it could simply correspond to a floristic impoverishment due to local ecological factors.

Group F. This group includes three data sets from the western area of the Southern Alps. Two of them represent relevés including the western European Erica cinerea reported by Lonati and Siniscalco (2010), who referred to the heathlands as floristic races within the Chamaecytiso hirsuti–Callunetum vulgaris. Our analysis confirmed these authors’ interpretation: group F includes our group G8 where Erica cinerea is missing (the species was not detected as a typical species for group F). In conclusion, group F is included in a broader definition of the western alpine association Chamaecytiso hirsuti–Callunetum vulgaris (including groups E and G), in which many floristic races of biogeographical significance (respectively with Erica cinerea and Festuca acuminata) could be detected hitherto.

Syntaxonomic scheme

The syntaxonomic proposal is in accordance with the guidelines of the International Code of Phytosociological Nomenclature (Weber, Moravec, and Theurillat 2000).

Calluno vulgaris–Ulicetea minoris Braun-Blanq. & Tüxen ex Klika in Klika & Hadač 1944

Vaccinio myrtilli–Genistetalia pilosae R.Schub. 1960

Genisto pilosae–Vaccinion uliginosi Br.-Bl. 1926

Cytiso supini–Antennarietum dioicae Preising 1953

Chamaecytiso hirsuti–Callunetum vulgaris Oberdorfer 1964

ericetosum cinereae Lonati and Siniscalco 2010

festucetosum acuminatae subass. nova, holotypus (Montecrestese, VB, Piedmont; 13/08/1998): Calluna vulgaris 4, Festuca acuminata 2, Quercus pubescens 2, Agrostis capillaris +, Allium lusitanicum +, Anthericum liliago +, Carex pilulifera +, Cytisus scoparius +, Danthonia decumbens +, Festuca stricta subsp. trachyphylla +, Fraxinus ornus +, Genista germanica +, Juniperus communis +, Molinia caerulea subsp. arundinacea +, Quercus robur +, Sempervivum arachnoideum +, Silene rupestris +, Vincetoxicum hirundinaria +, Stachys officinalis r

Genistion tinctorio-germanicae de Foucault 2008

Jasiono montanae–Callunetum vulgaris ass. nova hoc loco, holotypus (Cameri, NO, Piedmont; 04/07/1996): Calluna vulgaris 4, Molinia caerulea subsp. arundinacea 2, Rubus fruticosus agg. 2, Betula pendula 1, Cytisus scoparius 1, Dichanthelium acuminatum 1, Pinus sylvestris 1, Teucrium scorodonia 1, Agrostis capillaris +, Frangula alnus +, Hypericum perforatum +, Jasione montana +, Luzula multiflora +, Micropyrum tenellum +, Quercus pubescens +, Quercus robur +, Rumex acetosella +, Solidago gigantea +, Teesdalia nudicaulis +, Filago arvensis r, Hypochoeris radicata r

Salici rosmarinifoliae–Callunetum vulgaris ass. nova hoc loco, holotypus (Candelo, BI, Piedmont; 17/06/1996): Calluna vulgaris 5, Molinia caerulea subsp. arundinacea 3, Carex panicea 1, Frangula alnus 1, Populus tremula 1, Salix rosmarinifolia 1, Agrostis canina +, Carex pilulifera +, Danthonia decumbens +, Festuca filiformis +, Gentiana pneumonanthe +, Potentilla erecta +, Serratula tinctoria +, Carex demissa r, Carpinus betulus r, Drosera intermedia r, Genista tinctoria r, Succisa pratensis r

Notes on contributors

Bruno E. L. Cerabolini, PhD, Associate Professor of Environmental and Applied Botany (University of Insubria). Contributions: field surveys for data collection, data analysis, writing text.

Guido Brusa, PhD, Freelance biologist and institutional collaborator (University of Insubria). Contributions: field surveys for data collection, data analysis, writing text.

Roberta M. Ceriani, PhD, Biologist at an institutional body of nature conservation (Centro Flora Autoctona). Contributions: field surveys for data collection, refining the manuscript.

Stefano Armiraglio, PhD, Naturalist and curator of Botany (Brescia Civic Museum of Natural Sciences). Contributions: field surveys for data collection.

Cristina De Molli, PhD student, Master of Science (University of Insubria). Contributions: field surveys for data collection.

Simon Pierce, PhD, Assistant Professor of Environmental and Applied Botany Botany (University of Milan). Contributions: refining the manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.