Research Article ii FF o o r r e e s s t t doi: 10.3832/ifor3315-013 Biogeosciences and Forestry vol. 13, pp. 202-208

Typology and synecology of aspen woodlands in the central-southern Apennines (Italy): new findings and synthesis

Giovanni Russo (1), In order to review and complete our knowledge of the typology and synecol- Franco Pedrotti (2), ogy of the aspen communities from the central-southern Apennines, ten origi- (3) nal relevés were performed on the Gargano plateau and a set of 35 relevés as- Dan Gafta signed to four community types (HP: Holco mollis-Populetum tremulae; MP: Melico uniflorae-Populetum tremulae; FP: Fraxino orni-Populetum tremulae; GP: Geranio versicoloris-Populetum tremulae) were assembled from litera- ture. These relevés along with several environmental variables either mea- sured or estimated were involved in cluster and ordination analyses. The relevés from Gargano formed a distinctive cluster and were assigned to a new community type (SP: Stellario holosteae-Populetum tremulae ass. nova), which can be considered an Adriatic synvicariant of HP that is distributed in similar habitats (doline bottoms) but on the Tyrrhenian escarpment. At low levels of floristic similarity, the grouping of relevés in two clusters induces a sharp separation between the aspen communities distributed in the central Apennines (MP and FP) and those from the southern Apennines (SP, HP and GP), which is mainly due to compositional differences in the regional species pool. The ordination scores of relevés were best related to terrain slope, soil nitrogen, elevation, air temperature, light availability and, to a lesser extent, to soil moisture and reaction. Unlike MP and GP that appear the most meso- philous, the FP stands display a slightly more xerophilous and acidophilous character induced by the steeper slopes on which they occur. The HP habitat is the driest and lightest very likely because of the open overlying canopy, in contrast to MP stands featuring a high shrub cover. The highest occurrence of nitrophilous species was observed in SP and MP. The management of these pio- neer woods should be aimed at conservation, as they play an important role in the recovery of forest herb diversity along the ecological succession towards hardwood forests.

Keywords: Differential Species, Distribution Area, Environmental Variable Fit- ting, Gargano, Multivariate Analyses, Phytosociologic Classification, Secondary Succession, Stellario holosteae-Populetum tremulae

Introduction tends across the Eurosiberian and Mediter- changes in the landscape through the de- Cultural landscapes not only represent a ranean regions. Around the middle of the velopment of patches of scrub and pioneer valuable legacy of human activity over mil- 20th century, the profound changes woodlands (Falinski & Pedrotti 1990, Ma- lennia, but also a useful object of study for brought about by industrial development landra et al. 2018, Vitali et al. 2018). They inferring patterns in vegetation dynamics has led to extensive abandonment of both play an important role in vegetation (Foster et al. 2003). In particular, the long arable fields, pasturelands and mown recovery, especially in facilitating the es- tradition of land cultivation and sheep grasslands (Ruggieri 1976). As a conse- tablishment of late successional tree spe- farming in the Apennines has deeply quence, the natural secondary succession, cies throughout the nemoral and boreal shaped the vegetated landscape in these which has been developing over the fol- forest zones (Flinn & Vellend 2005). On the mountainous areas, a pattern which ex- lowing decades, has induced visible positive side, this trend is beneficial for, among other processes, carbon sequestra- tion (La Mantia et al. 2013, Novara et al. (1) Gargano Land Reclamation Authority, Foggia (Italy); (2) Department of Biosciences, 2017), reduction of soil erosion (Erktan et University of Camerino, Camerino (Italy); (3) Department of Taxonomy and Ecology, Babes- al. 2016) and forest connectivity at land- Bolyai University, Cluj-Napoca (Romania) scape level (Hernández et al. 2015). On the other hand, woodland encroachment that @ Dan Gafta ([email protected]) follows the abandonment of traditional land use has also proved a matter of some Received: Dec 06, 2019 - Accepted: Mar 28, 2020 concern with respect to the decline in bio- diversity (Falcucci et al. 2007, Amici et al. Citation: Russo G, Pedrotti F, Gafta D (2020). Typology and synecology of aspen woodlands in 2013, Ascoli et al. 2013). the central-southern Apennines (Italy): new findings and synthesis. iForest 13: 202-208. – doi: Aspen (Populus tremula) stands have a 10.3832/ifor3315-013 [online 2020-05-19] fine-grained, patchy distribution in the cen- tral-southern Apennines, also reaching the Communicated by: Michele Carbognani island of Sicily. The understory of these stands is usually dominated by ubiquitous

© SISEF https://iforest.sisef.org/ 202 iForest 13: 202-208 Russo G et al. - iForest 13: 202-208 y

r herbs from open habitats that are rem- Marco in Lamis and Cagnano Varano formed in aspen stands throughout the t

s nants of the past seral stages and persist roughly between 600 and 1000 m a.s.l. The central-southern Apennines, were digitised e

r due to the loose foliage of aspen crowns. land topography is typical for karst areas, from published literature as follows: ten o In spite of such apparent floristic homo- with many scattered dolines and lack of relevés (MP01-MP10) of Melico uniflorae- F geneity, the composition of these forests surface waters. At the lowest elevations Populetum tremulae (Pedrotti 1995, Pirone d n varies more or less across ecoregions and (565-570 m) the mean monthly tempera- et al. 2010); eleven relevés (FP01-FP11) of a along environmental gradients, given the ture ranges from 5.4 °C (in January) to 24.1 Fraxino orni-Populetum tremulae (Taffetani s e wide variation in site (local) and geograph- °C (in July), whereas the mean monthly 2000, Pirone et al. 2010); four relevés c

n ical conditions in the central-southern rainfall varies between 37 mm (in August) (HP01-HP04) of Holco mollis-Populetum tre- e i Apennines. As a matter of fact, several for- and 125 mm (in December) summing up to mulae (Rosati et al. 2010); ten relevés c

s est community types (syntaxa) have been a mean of 900 mm per year. According to (GP01-GP10) of Geranio versicoloris-Popule- o described in this geographical area: Melico Pesaresi et al. (2017), the study area is in- tum tremulae (Fascetti et al. 2013). The ge- e g uniflorae-Populetum tremulae (Pedrotti cluded within a climatic unit characterised ographical distribution of the first two and o i 1995), Fraxino orni-Populetum tremulae by a temperate oceanic bioclimate, lower the last two aspen syntaxa reflects to B (Taffetani 2000), Holco mollis-Populetum supratemperate thermotype and upper some extent a regional vicariance (Fig. 1). – tremulae (Rosati et al. 2010) and Geranio subhumid ombrotype. Biogeographically All aspen community types considered in t s versicoloris-Populetum tremulae (Fascetti the area investigated is part of the Medi- this study are jointly presented in a synop- e r et al. 2013). In addition, a survey carried out terranean region, East-Mediterranean sub- tic table (see Tab. S2 in Supplementary ma- o

F recently on the Gargano peninsula (Russo region, Adriatic province and Apulian sec- terial). i & Strizzi 2013) has revealed particular as- tor (Rivas-Martinez et al. 2004). Elevation and terrain slope were mea- pen stands that need to be analysed in the The high plain is covered by secondary sured at each sampled site. Other environ- context of similar communities for a grasslands, abandoned fields and Turkey mental variables were estimated indirectly proper syntaxonomical assignment. To oak-dominated forests (Physospermo verti- by means of ordinal ranks of species eco- date no synthesis exists on the floristic cillati-Quercetum cerridis Aita et al., 1977 logical optima, namely ecological indicator composition and ecology of these aspen em. Ubaldi et al., 1987), the latter repre- values for light, air temperature, edaphic community types. senting the natural potential vegetation. moisture, soil reaction and nitrogen (Pig- In the present study we aimed at: (i) char- Dry grasslands and small Turkey oak wood- natti 2005). acterising syntaxonomically and synecolog- lots occur on the doline slopes, whereas The botanical and syntaxonomical no- ically the aspen stands from Gargano; (ii) their bottoms are covered by aspen woods menclature followed the latest online ver- reviewing all aspen forest types from the or abandoned crops invaded by Pteridium sion of Euro+Med PlantBase (Euro+Med central-southern Apennines; (iii) detecting aquilinum. 2006) and respectively, the “Prodromus of the site/geographical factors that might be the vegetation of Italy” (Biondi & Blasi responsible for the variation in plant spe- Data collection in the field and from 2015). cies composition of these woods. literature The floristic survey of aspen stands from Data analysis Materials and methods Gargano was carried out following the Given the scale of observations, the em- phytosociological approach (Westhoff & phasis on species turnover over species Study area Van Der Maarel 1978). A single relevé (plant dominance, the inevitable bias in species The Gargano peninsula is situated in the community sample) was performed on the cover estimates and the nature of multi- southern part of the Adriatic (eastern) side bottom of each doline that was covered by variate methods subsequently applied on of the Italian peninsula and lies over a geo- a well-developed aspen canopy. In total, a data, the binary community matrix based logical complex exclusively composed of number of ten relevés (SP01-SP10) were only on the presence-absence of species limestones and dolomites. The area sur- performed by visually estimating the rela- was employed in all analyses (Fortin 1997, veyed is located in the core part of the tive cover of each vascular plant species. Podani 2006, Wilson 2012). The community peninsula, that is on the High Plain of San An additional set of 35 relevés, per- dissimilarity matrix was built on the basis

Fig. 1 - Regional distribu- tion of the five aspen woodland types in the cen- tral-southern Apennines (Italy). Abbreviations: SP = Stellario holosteae-Popule- tum tremulae; HP = Holco mollis-Populetum tremulae; MP = Melico uniflorae-Pop- uletum tremulae; FP = Frax- ino orni-Populetum tremu- lae; GP = Geranio versicol- oris-Populetum tremulae.

203 iForest 13: 202-208 Aspen woodlands in the central-southern Apennines y

of the complement of the Sørensen index, al. 2019), “cluster” (Maechler et al. 2019), Typology of aspen stands from Gargano r t

which has many desirable properties “fpc” (Hennig 2019), “optpart” (Roberts The ten relevés from the cluster SP were s e

(Jongman et al. 1995, Boyce & Ellison 2016) and “indicspecies” (De Caceres & assigned to a new syntaxon named Stel- r

2001). Following many previous findings Jansen 2019). lario holosteae-Populetum tremulae ass. no- o F (Lötter et al. 2013), the flexible beta algo- va hoc loco, within the Aceri obtusati-Popu- d rithm was employed in the hierarchical Results lenion tremulae Taffetani 2000 suballiance, n a clustering of relevés. The output corre- Corylo avellanae-Populion tremulae (Br.-Bl. s sponding to beta = -0.46 was finally re- Classification of all aspen stands ex O. Bolòs, 1973) Rivas-Martínez & Costa, e c

tained as suggested by the largest value of At first glance, the dendrogram of relevés 1998 alliance, Betulo pendulae-Populetalia n e the agglomerative coefficient (0.84). The revealed that these might have been rea- tremulae Rivas-Martínez & Costa, 2002 or- i c

optimal number of clusters was assessed sonably grouped in two to six clusters (Fig. der and Querco-Fagetea Br.-Bl. & Vlieger, s based on the maximum value of the Calin- 2a). The classification in two clusters (dis- 1937 class (Tab. S1 in Supplementary mate- o e ski & Harabasz (1974) index, as all the other tinguished at the nodes marked by black rial). The relevé SP08, displaying both the g o internal validation criteria displayed mono- dots in Fig. 2a) would have matched ex- largest positive silhouette and the highest i B tonic relationships with the number of clus- actly the regional distribution of relevés, mean similarity to the SP cluster, was se- – ters. No further refinement of classification that is in the southern and, respectively, lected as the holotype of the new syntaxon t could be achieved by reallocating the re- the central Apennines (Fig. 1). Neverthe- (Tab. S1). The most important understory s e levés among groups through optimisation less, the maximum value of the Calinski- species in terms of relative cover are r o

algorithms that maximise either the cluster Harabasz index pointed to five clusters as Pteridium aquilinum and Rubus caesius. F average silhouette or the within-cluster / the optimal classification of relevés (Fig. Within the context represented by the i among-cluster similarity ratio (Roberts 2b). Considering also the fair to high stabil- other aspen wood types from the central- 2015). Finally, the cluster stability was esti- ity of clusters, with the exception of HP southern Apennines, several species dis- mated by the level of bootstrap mean simi- cluster that included only four relevés, the playing significant correlation with the larity (BMS), with values larger than 70% solution with five groups was finally re- Stellario-Populetum group of relevés were being considered acceptable for the dis- tained. detected (Tab. 1). Of these discriminant tinction of plant community types. The ho- All the ten relevés performed in Gargano species only three are shade-tolerant un- lotype of the newly described community- were gathered in one cluster (SP), whereas derstory species (Stellaria holostea, Carex type was selected on the basis of the the other four clusters of relevés (HP, MP, depauperata and Allium pendulinum). largest positive silhouette width in the tar- GP and FP) were perfectly circumscribed to The refined list of differential species as- get cluster. The plant species best associ- the aspen wood types described in the lit- sociated with each of the five aspen wood ated with the distinguished community erature (Fig. 2a). types is highlighted in the synoptic table types were identified by means of the group-equalised, point biserial correlation coefficient (r.g), which accounts for differ- ences in sample size (number of relevés) and enables the selection of potential diag- nostic species. The ordination of relevés in the reduced space of species was performed through local non-metric multidimensional scaling (NMDS), as it is a non-parametric, flexible technique making only few assumptions about the nature of the data. A number of three axes was extracted as suggested by the “elbow rule” applied to the distribu- tion of the stress index with increasing di- mensions (Mair et al. 2016). The correlation between the NMDS axes and the species occurrence was estimated through the point biserial coefficient. The importance of the environmental variables in terms of proportion of floristic variance explained (R-square) was assessed separately through non-parametric linear fitting with respect to all NMDS axes by using a modi- fied function (“envfit.iv”), which addresses the bias in the relationship between mean species attributes (i.e., mean ecological in- dicator values) and relevé scores along or- dination axes (Zeleny & Schaffers 2012). The environmental variables displaying only marginal, linear effects on NMDS scores were involved in generalised addi- tive models, in order to fit smooth surfaces through thin-plate splines in the ordination space. The statistical significance of the above Fig. 2 - (a) Dendrogram of the relevés performed in aspen woods from the central- statistics was estimated by 104 random per- southern Apennines. The numbers refer to values of the stability index estimated for mutations. All numerical analyses were car- each of the five clusters distinguished. Abbreviations as in Fig. 1. (b) Distribution of ried out in R ver. 3.6.1 (R Core Team 2019) the Calinski-Harabasz criterion by number of relevé groups distinguished via hierarchi- using the packages “vegan” (Oksanen et cal clustering. The arrow points to the optimal number (5) of clusters. iForest 13: 202-208 204 Russo G et al. - iForest 13: 202-208 y r

t Fig. 3 - NMDS ordina- s

e tion of all the relevés in r the reduced, bidimen- o F

sional space deter- d mined by the axes 1 n a and 2 (a), and by the

s axes 1 and 3 (b). Con- e c vex hulls correspond- n

e ing to each aspen i c woodland type are s

o drawn by dotted lines. e The environmental pre- g

o dictors explaining sig- i

B nificant fractions of

– floristic variation are t

s represented through e vectors. Abbreviations r o as in Fig. 1. F i

Tab. 1 - Group-equalised point biserial correlation coefficients (r.g.) between species and the group of Stellario-Populetum tremulae relevés in the context of all aspen featuring the percentages of occurrence of woodland types from the central-southern Apennines. Only the highest ranked ten species by community type (Tab. S2 in Sup- species in terms of strength of positive correlation are listed. plementary material). The Geranio-Popule- tum and Fraxino-Populetum were poorly represented by differential species com- Differential r.g P-value (>r.g) pared with the other forest types. species Vicia cassubica 0.873 <0.0001 Ordination of all aspen stands Stellaria holostea 0.807 0.0003 Overall, the indirect ordination of relevés Poa trivialis 0.750 <0.0001 in the tridimensional NMDS space revealed Cruciata glabra 0.694 <0.0001 a fair between-group separation that matched the five clusters previously distin- Rubus caesius 0.591 0.0003 guished (Fig. 3). The first NMDS axis was Carex depauperata 0.590 0.0043 best correlated (negatively and positively) Allium pendulinum 0.590 0.0051 with Geum urbanum (r = -0.79) and Junipe- Peucedanum oreoselinum 0.590 0.0052 rus communis (r = +0.57), which pointed to Aristolochia pallida 0.590 0.0054 a nitrogen-based fertility gradient (i.e., from Fraxino-Populetum to Stellario-Pop- Poa pratensis 0.590 0.0063 uletum). The second NMDS axis was best positively correlated with Melica uniflora (r = +0.84) and negatively correlated with Ag- Tab. 2 - Independent testing of the effects of different environmental variables on rimonia eupatoria (r = -0.60) and Dactylis NMDS relevé scores by permutational linear fitting. glomerata (r = -0.59), suggesting a weak, canopy-openness gradient (i.e., from Meli- Response NMDS NMDS NMDS 2 P-value R 2 co-Populetum to Holco-Populetum). Finally, variable axis 1 axis 2 axis 3 (>R ) the NMDS axis 3 was best negatively corre- Elevation -0.373 0.422 -0.827 0.488 <0.0001 lated with Geranium versicolor (r = -0.57) Terrain slope 0.758 0.532 -0.376 0.454 <0.0001 but positively correlated with Acer cam- pestre (r = +0.55), which alluded to an alti- Light -0.088 -0.978 0.189 0.684 0.0025 tudinal gradient (i.e., from Stellario-Popule- Air temperature 0.633 -0.749 0.194 0.727 0.0007 tum/Fraxino-Populetum to Geranio-Popule- Soil moisture -0.292 0.955 -0.051 0.485 0.0673 tum). Soil reaction 0.538 0.839 0.084 0.442 0.1009 Soil nitrogen and terrain slope were lin- early but oppositely related to the NMDS Soil nitrogen -0.871 0.491 -0.024 0.543 0.0342 axis 1, whereas light and temperature were both linearly and concordantly related to Tab. 3 - Summary statistics of the generalised additive models employed for fitting the NMDS axis 2 (Tab. 2, Fig. 3a). Among smooth surfaces referring to indicator values of soil reaction and moisture within the the environmental variables, only elevation space determined by the first two NMDS axes. was linearly and significantly related to the NMDS axis 3 (Tab. 2, Fig. 3b). Although non-linear, the responses of NMDS scores Smooth term Deviance Response Intercept to soil reaction and moisture were statisti- s(NMDS1, NMDS2) Adj-R2 explained variable cally significant and mostly oriented along F value P-value t value P-value (%) the axis 2 (Tab. 3, Fig. 4). Soil 4.685 <0.0001 168.5 <0.0001 0.489 51.6 moisture Discussion The classification of relevés drawn from Soil 5.466 <0.0001 174.0 <0.0001 0.528 57.9 reaction the central-southern Apennines confirmed the compositional separation of the four

205 iForest 13: 202-208 Aspen woodlands in the central-southern Apennines y r

Fig. 4 - Contour lines of t s

indicator values of soil e moisture (a) and soil r o F

reaction (b) along with

convex hulls of aspen d n

woodland types, over- a

laid on the ordination s e

of relevés within the c n

reduced space deter- e i

mined by the first two c s

NMDS axes. Abbrevia- o

tions as in Fig. 1. e g o i B

–

t s e r o F previously described aspen community communities, which presumably repre- soils underlying the Holco-Populetum i types as well as the floristic distinction of sented an early seral stage. Similar Turkey stands appeared the poorest in bases (and the forest stands inventoried in Gargano. It oak woods, but including few, more ther- also the driest), as inferred from the plant is also worth mentioning that, at low levels mo-xerophilous species, should potentially indicator values. This is in part due to the of compositional resemblance, the group- succeed the Holco-Populetum tremulae sinkholes that have enhanced the water ing of relevés in two clusters induced a stands (Rosati et al. 2010). In contrast, the drainage and base leaching (Ma et al. sharp separation between the aspen com- Melico-Populetum tremulae and Geranio- 2018). On the other side, the Melico-Popule- munities distributed in the central Apen- Populetum tremulae woods should succes- tum woods are the most mesophilous, very nines (Melico-Populetum and Fraxino-Pop- sionally develop toward Fagus sylvatica for- likely because of the finer soil texture de- uletum) and those from the southern ests (Pedrotti 1995, Fascetti et al. 2013), termined by the underlying marls (Verheye Apennines (Stellario-Populetum, Holco-Pop- whereas the Fraxino-Populetum tremulae & De La Rosa 2009). The second most me- uletum and Geranio-Populetum). Such dis- woods should develop towards acido- sophilous aspen forest type is Geranio-Pop- tinction at a large spatial scale was mainly philous Castanea sativa forests (Taffetani uletum of which stands occur on average due to compositional differences in the re- 2000). at higher elevations than the others. gional species pool. Stellario-Populetum tremulae is geomor- Apart from the main environmental driv- The newly described “plant association” phologically and floristically related to Hol- ers responsible for the differentiation of (community type), Stellario holosteae-Pop- co-Populetum tremulae, as both syntaxa the five aspen forest types, there are other uletum tremulae, was assigned to the class develop on the bottom of karstic dolines factors (anthropogenic and developmen- Querco-Fagetea as these stands developed and their understory is dominated by Pter- tal) that have induced more subtle floristic in Gargano were physiognomically tall, al- idium aquilinum. Besides, Stellario-Popule- differences between these communities. most closed canopy woods and hosted tum can be considered an Adriatic synvi- First, a higher amount of soil nitrogen, ei- several mesophilous forest herbs as well as cariant of Holco-Populetum that is distrib- ther remnant from old crops or washed sporadic Quercus cerris, Castanea sativa and uted on the Tyrrhenian escarpment. The through rainfall run-off from the surround- Acer obtusatum saplings. As Stellario-Pop- compositional dissimilarity between these ing pastures, was probably responsible for uletum tremulae belongs undoubtedly to two aspen forest types was mainly deter- the preferential occurrence of nitrophilous the Apenninic suballiance Aceri obtusati- mined by some thermophilous species oc- species such as Rubus caesius and Poa trivi- Populenion tremulae, we had to make ref- curring in the understory and tree canopy alis in Stellario-Populetum stands, or Ae- erence (for the time being) to the upper of the Holco-Populetum stands. Probably gopodium podagraria and Heracleum spho- ranked syntaxon, i.e., the alliance Corylo- on these grounds, Rosati et al. (2010) as- ndylium in Melico-Populetum stands. Sec- Populion, despite it bears a nomen ambi- signed the Holco-Populetum tremulae com- ond, the larger tree canopy openness in guum (Mucina et al. 2016). These authors munity type to a different alliance, i.e., Teu- the Holco-Populetum stands, as opposed to proposed the Astrantio-Corylion avellanae crio siculi-Quercion cerridis. the relatively high cover of Corylus avellana Passarge, 1978 as an alternative valid name Unlike the Melico-Populetum and Geranio- in the Melico-Populetum stands, has deter- but the latter syntaxon encompasses only Populetum stands that possessed a larger mined the exclusive occurrence of shade- scrub communities, being part of the order number of mesophilous species typical of intolerant, thermophilous species such as Prunetalia spinosae Tx., 1952 and the class Fagetalia, the Fraxino-Populetum stands Asphodeline liburnica and Aristolochia lutea Crataego-Prunetea spinosae (syn. Rhamno- displayed a slightly more xerophilous and in the former aspen woods. Prunetea). We argue that the aspen com- acidophilous character induced by the munities of Aceri obtusati-Populenion trem- steeper slopes and expressed by the pref- Conclusions and implications for ulae do not fit into the latter upper ranked erential presence of Carpinus orientalis, Ge- management vegetation units and therefore, the syntax- nista tinctoria and Juniperus communis. The On the basis of current knowledge, five onomical framework herein should be aspen stands, distributed on the north- types of aspen wood can be distinguished maintained until the nomenclatural issues eastern slope of the Turchio Mountain in the central-southern Apennines. Differ- are sorted out. (Abruzzo National Park) and inventoried by ences in topography, edaphic conditions Syndynamically, the stands of Stellario- Pedrotti (1996), are also circumscribed to and light availability are the main factors Populetum tremulae are likely to be re- Fraxino orni-Populetum tremulae. responsible for the observed variation in placed by Turkey oak stands (Physospermo All aspen forest types are distributed on the floristic composition of these woods. verticillati-Quercetum cerridis), which are substrates moderately to rich in base The present findings are based on a rela- acknowledged as the mature stage of that cations, such as mixtures of marls and tively low number of relevés (especially dynamic series (Biondi et al. 2010). The bot- sandstones or limestones, except for Frax- those circumscribed to Holco-Populetum toms of the shallower dolines were cov- ino-Populetum that can also occur on tremulae) and, therefore, some observed ered by Pteridium aquilinum-dominate slightly acidic sandstones. Despite that, the patterns may slightly change once new iForest 13: 202-208 206 Russo G et al. - iForest 13: 202-208 y

r data are added and analysed. bined effects of soil, root traits and plant com- ysis basics and extensions. R package version t

s Aspen stands have little economic value munity characteristics. Plant and Soil 398: 121- 2.1.0, user manual, pp. 82. [online] URL: http:// e 137. - doi: 10.1007/s11104-015-2647-6 CRAN.R-project.org/package=cluster r because of the low density of their timber. o As a consequence, the forest management Euro+Med (2006). Euro+Med PlantBase - the in- Mair P, Borg I, Rusch T (2016). Goodness-of-fit F should be aimed at conservation, as these formation resource for Euro-Mediterranean assessment in multidimensional scaling and un- d n pioneer woods play an important role in plant diversity. Botanic Garden and Botanical folding. Multivariate Behavioral Research 51: a the recovery of forest herb diversity along Museum, Berlin-Dahlem, Germany, web site. 772-789. - [online] URL: http://www.tandfonlin s e the ecological succession towards hard- [online] URL: http://ww2.bgbm.org/EuroPlus e.com/doi/full/10.1080/00273171.2016.1235966 c

n wood forests. To address the concerns re- Med/ Malandra F, Vitali A, Urbinati C, Garbarino M e i garding the decline in biodiversity due to Falcucci A, Maiorano L, Boitani L (2007). (2018). 70 years of land use/land cover changes c

s aspen woodland encroachment, an ade- Changes in land-use/land-cover patterns in Italy in the Apennines (Italy): a meta-analysis. For- o quate fraction of meadows and pastures and their implications for biodiversity conserva- ests 9: 551. - doi: 10.3390/f9090551 e g should be continuously maintained in the tion. Landscape Ecology 22: 617-631. - doi: Mucina L, Bültmann H, Dieren K, Theurillat JP, o i landscape and managed appropriately. 10.1007/s10980-006-9056-4 Raus T, Carni A, Willner W, Dengler J, Gavilán B

Falinski JB, Pedrotti F (1990). The vegetation and García R, Chytry M, Hájek M, Di Pietro R, Iaku- – dynamical tendencies in the vegetation of shenko D, Pallas J, Daniëls FJA, Bergmeier E, t Acknowledgements s GR and FP carried out the field survey. DG Bosco Quarto, Promontorio del Gargano, Italy. Guerra AS, Ermakov N, Valachovič M, Scham- e r conceived the study, performed the nu- Braun-Blanquetia 5: 1-31. inée JHJ, Lysenko T, Didukh YP, Pignatti S, Rod- o

F merical analyses and wrote the first draft. Fascetti S, Pirone G, Rosati L (2013). The vegeta- well JS, Capelo J, Weber HE, Solomeshch A, Di- i All authors contributed to refining the ini- tion of the Maddalena Mountains (southern mopoulos P, Aguiar C, Hennekens SM, Tichy L tial draft. Italy). Plant Sociology 50: 5-37. (2016). Vegetation of Europe: hierarchical floris- The authors are grateful to John Akeroyd Flinn KM, Vellend M (2005). Recovery of forest tic classification system of vascular plant, bryo- for critically commenting on and editing plant communities in post-agricultural land- phyte, lichen, and algal communities. Applied the latest draft of the manuscript, and to scapes. Frontiers in Ecology and the Environ- Vegetation Science (Suppl 1) 19: 3-264. - doi: Murrja Edmir for providing the colour back- ment 3: 243-250. - doi: 10.1890/1540-9295(2005) 10.1111/avsc.12257 ground of the distribution map. 003[0243:ROFPCI]2.0.CO;2 Novara A, Gristina L, Sala G, Galati A, Cresciman- Fortin MJ (1997). Effects of data type on vegeta- no M, Cerdà A, Badalamenti E, La Mantia T Conflict of interest tion boundary delineation. Canadian Journal of (2017). Agricultural land abandonment in Medi- The authors declare that they have no Forest Research 27: 1851-1858. - doi: 10.1139/x97- terranean environment provides ecosystem conflict of interest. 156 services via soil carbon sequestration. Science Foster D, Swanson F, Aber J, Burke I, Brokaw N, of Total Environment 576: 420-429. - doi: 10.101 References Tilman D, Knapp A (2003). The importance of 6/j.scitotenv.2016.10.123 Amici V, Santi E, Filibeck G, Diekmann M, Geri F, land-use legacies to ecology and conservation. Oksanen J, Blanchet FG, Friendly M, Kindt R, Le- Landi S, Scoppola A, Chiarucci A, Vetaas O BioScience 53: 77-88. - doi: 10.1641/0006-3568 gendre P, McGlinn D, Minchin PR, O’Hara RB, (2013). Influence of secondary forest succes- (2003)053[0077:TIOLUL]2.0.CO;2 Simpson GL, Solymos P, Stevens MHH, Szoecs sion on plant diversity patterns in a Mediter- Hennig C (2019). Package ‘fpc’: flexible proce- E, Wagner H (2019). Package ‘vegan’: commu- ranean landscape. Journal of Biogeography 40: dures for clustering. R package version 2.2-3, nity ecology package. R package version 2.5-6, 2335-2347. - doi: 10.1111/jbi.12182 user manual, pp. 161. [online] URL: http://cran.r- user manual, pp. 296. [online] URL: http:// Ascoli D, Lonati M, Marzano R, Bovio G, Cav- project.org/web/packages/fpc/fpc.pdf cran.r-project.org/web/packages/vegan/vegan. allero A, Lombardi G (2013). Prescribed burning Hernández A, Miranda M, Arellano EC, Saura S, pdf and browsing to control tree encroachment in Ovalle C (2015). Landscape dynamics and their Pedrotti F (1995). The trembling poplar woods of southern European heathlands. Forest Ecology effect on the functional connectivity of a the central Apennines. Studi Trentini di Scienze and Management 289: 69-77. - doi: 10.1016/j.for Mediterranean landscape in Chile. Ecological In- Naturali - Acta Biologica 70: 99-105. [in Italian] eco.2012.09.041 dicators 48: 198-206. - doi: 10.1016/j.ecolind.20 Pedrotti F (1996). The trembling poplar (Populus Biondi E, Blasi C (2015). Prodromus of the vege- 14.08.010 tremula) colonising the abandoned fields in the tation of Italy. Ministry of the Environment and Jongman RHG, Braak CJF, Van Tongeren OFR Abruzzo National Park. In: “Phytodynamics - of the Protection of the Territory and the Sea, (1995). Data analysis in community and land- The Different Aspects of Vegetation Dynamics” Rome, Italy, web site. [in Italian] [online] URL: scape ecology. Cambridge University Press, (Biondi E, Gehu JM eds). Colloques Phytosoci- http://www.prodromo-vegetazione-italia.org/ Cambridge, UK, pp. 299. [online] URL: http:// ologiques 24, J. Cramer, Berlin, pp. 111-121. [in Biondi E, Casavecchia S, Beccarini L, Marchiori S, books.google.com/books?id=Oy8F3oVGPeIC Italian] Medagli P, Zuccarello V (2010). The vegetation La Mantia T, Gristina L, Rivaldo E, Pasta S, No- Pesaresi S, Biondi E, Casavecchia S (2017). Biocli- series in the Region Apulia. In: “The Vegetation vara A, Rühl J (2013). The effects of postpas- mates of Italy. Journal of Maps 13: 955-960. - of Italy” (Blasi C ed). Palumbi Editore, Rome, ture woody plant colonization on soil and doi: 10.1080/17445647.2017.1413017 Italy, pp. 391-409. aboveground litter carbon and nitrogen along Pignatti S (2005). Bioindicator values of vascular Boyce RL, Ellison PC (2001). Choosing the best a bioclimatic transect. iForest 6: 238-246. - doi: plants of the flora of Italy. Braun-Blanquetia 39: similarity index when performing fuzzy set or- 10.3832/ifor0811-006 1-97. [in Italian] dination on binary data. Journal of Vegetation Lötter MC, Mucina L, Witkowski ETF (2013). The Pirone G, Frattaroli AR, Biondi E, Casavecchia S, Science 12: 711-720. - doi: 10.2307/3236912 classification conundrum: species fidelity as Pesaresi S (2010). The forest vegetation of the Calinski T, Harabasz J (1974). A dendrite method leading criterion in search of a rigorous method Gran Sasso and Monti della Laga National Park. for cluster analysis. Communications in Statis- to classify a complex forest data set. Commu- L’Italia Forestale e Montana 65: 699-735. [in tics 3: 1-27. - doi: 10.1080/03610927408827101 nity Ecology 14: 121-132. - doi: 10.1556/ComEc.14. Italian] - doi: 10.4129/IFM.2010.6.03 De Caceres M, Jansen F (2019). Package ‘indic- 2013.1.13 Podani J (2006). Braun-Blanquet’s legacy and species’: relationship between species and Ma M, Gao Y, Song X, Green SM, Xiong B, Dun- data analysis in vegetation science. Journal of groups of sites. R package version 1.7.8, user gait JA, Peng T, Quine TA, Wen X, He N (2018). Vegetation Science 17: 113-117. - doi: 10.1111/j.16 manual, pp. 32. [online] URL: http://cran.r-proj Migration and leaching characteristics of base 54-1103.2006.tb02429.x ect.org/web/packages/indicspecies/indicspecies cation: indicating environmental effects on soil R Core Team (2019). R: A language and environ- .pdf alkalinity in a karst area. Environmental Science ment for statistical computing. R Foundation Erktan A, Cécillon L, Graf F, Roumet C, Legout C, and Pollution Research 25: 20899-20910. - doi: for Statistical Computing, Vienna, Austria. [on- Rey F (2016). Increase in soil aggregate stability 10.1007/s11356-018-2266-x line] URL: https://www.R-project.org/ along a Mediterranean successional gradient in Maechler M, Rousseeuw P, Struyf A, Hubert M, Rivas-Martinez S, Penas A, Diaz TE (2004). Bio- severely eroded gully bed ecosystems: com- Hornik K (2019). Package ‘cluster’: cluster anal- geographic map of Europe. Scale 1:16.000.000.

207 iForest 13: 202-208 Aspen woodlands in the central-southern Apennines y

Cartographic Service, University of Leon, Spain. ques Phytosociologiques 29, J. Cramer, Berlin, sometimes represents a plant community as r t

Roberts DW (2015). Vegetation classification by Germany, pp. 577-603. [in Italian] well as species abundances do, or better. Jour- s

two new iterative reallocation optimization al- Taffetani F (2000). Vegetation series of the geo- nal of Vegetation Science 23: 1013-1023. - doi: e r

gorithms. Plant Ecology 216: 741-758. - doi: morphological system of Mount Ascensione 10.1111/j.1654-1103.2012.01430.x o F 10.1007/s11258-014-0403-2 (central Italy). Fitosociologia 37: 93-151. [in Ital- Zeleny D, Schaffers AP (2012). Too good to be d

Roberts DW (2016). Package ‘optpart’: optimal ian] true: pitfalls of using mean Ellenberg indicator n a partitioning of similarity relations. R package Verheye WH, De La Rosa D (2009). Mediter- values in vegetation analyses. Journal of Vege- s

version 2.3-0, user manual, pp. 48. [online] ranean soils. In: “Encyclopedia of Land Use, tation Science 23: 419-431. - doi: 10.1111/j.1654- e c

URL: http://cran.r-project.org/web/packages/op Land Cover and Soil Sciences” - vol. VII, part 2 1103.2011.01366.x n e

tpart/optpart.pdf (Verheye WH ed). UNESCO-EOLSS Publishers, i c

Rosati L, Filibeck G, De Lorenzis A, Lattanzi E, Oxford, UK, pp. 96-120. [online] URL: http:// Supplementary Material s Surbera S, Fascetti S, Blasi C (2010). The forest books.google.com/books?id=IorTCwAAQBAJ o e

vegetation of the Alburni Mts. (Cilento National Vitali A, Urbinati C, Weisberg PJ, Urza AK, Gar- Tab. S1 - Phytosociological table of the re- g o

Park, Campania, southern Italy): syntaxonomy barino M (2018). Effects of natural and anthro- levés performed in aspen stands of Gar- i B and phytogeography. Fitosociologia 47: 17-55. pogenic drivers on land-cover change and tree- gano (Stellario holosteae-Populetum tremu- – [in Italian] line dynamics in the Apennines (Italy). Journal lae ass. nova hoc loco). t

Ruggieri M (1976). The abandoned fields: a new of Vegetation Science 29: 189-199. - doi: 10.1111/ s e

component of the landscape. Bolletino della jvs.12598 Tab. S2 - Synoptic table of aspen wood r o

Società Geografica Italiana 5: 441-464. [in Ital- Westhoff V, Van Der Maarel E (1978). The Braun- types distinguished in the central-southern F i ian] Blanquet approach. In: “Classification of Plant Apennines. Russo G, Strizzi C (2013). The vegetation of the Communities” (Whittaker RH ed). Junk, The Gargano National Park (Promontory of Gar- Hague, Netherlands, pp. 287-399. - doi: 10.1007/ Link: [email protected] gano and Tremiti Islands). In: Proceedings of 978-94-009-9183-5_9 the “Stelvio Congress” (Pedrotti F ed). Collo- Wilson JB (2012). Species presence/absence

iForest 13: 202-208 208