Original article
Patterns of fine-scale plant species richness in dry grasslands across
the eastern Balkan Peninsula
Salza Palpurina a, *
, Milan Chytrý a
, Rossen Tzonev b
, Jirí Danihelka a, c
, Irena Axmanova a
,
Kristina Merunkova a
, Mario Duchon a
, Todor Karakiev b
a
Department of Botany and Zoology, Masaryk University, Kotlarska 2, CZ-611 37 Brno, Czech Republic
b
Department of Ecology and Environmental Protection, Sofia University ‘St. Kliment Ohridski’, Sofia, Bulgaria
c
Department of Vegetation Ecology, Institute of Botany, Czech Academy of Sciences, Lidicka 25/27, CZ-602 00 Brno, Czech Republic
a r t i c l e i n f o
Article history:
Received 2 September 2014
Received in revised form
22 January 2015
Accepted 2 February 2015
Available online
Keywords:
Alpha diversity
Bulgaria
Romania
Soil chemistry
Species pool
Steppe
a b s t r a c t
Fine-scale plant species richness varies across habitats, climatic and biogeographic regions, but the largescale
context of this variation is insufficiently explored. The patterns at the borders between biomes
harbouring rich but different floras are of special interest. Dry grasslands of the eastern Balkan Peninsula,
situated in the Eurasian forest-steppe zone and developed under Mediterranean influence, are a specific
case of such biome transition. However, there are no studies assessing the patterns of fine-scale species
richness and their underlying factors across the eastern Balkans. To explore these patterns, we sampled
dry and semi-dry grasslands (phytosociological class Festuco-Brometea) across Bulgaria and SE Romania.
In total, 172 vegetation plots of 10 Â 10 m2
were sampled, in which all vascular plant species were
recorded, soil depth was measured, and soil samples were collected and analysed in a laboratory for pH
and plant-available nutrients. Geographic coordinates were used to extract selected climatic variables.
Regression trees and linear regressions were used to quantify the relationships between species richness
and environmental variables. Climatic factors were identified as the main drivers of species richness: (1)
Species richness was strongly positively correlated with the mean temperature of the coldest month:
sub-Mediterranean areas of S and E Bulgaria, characterized by warmer winters, were more species-rich.
(2) Outside the sub-Mediterranean areas, species richness strongly increased with annual precipitation,
which was primarily controlled by altitude. (3) Bedrock type and soil pH also significantly affected dry
grassland richness outside the sub-Mediterranean areas. These results suggest that fine-scale species
richness of dry grasslands over large areas is driven by processes at the regional level, especially by the
difference in the species pools of large regions, in our case the Continental and Mediterranean biogeographic
regions. Local environmental factors are of secondary importance over broad extents, but their
effect on fine-scale species richness increases within climatically and biogeographically homogeneous
regions.
© 2015 Elsevier Masson SAS. All rights reserved.
1. Introduction
Eurasian temperate dry grasslands and steppes locally support
the coexistence of a high number of vascular plant species (Kull and
Zobel,1991; Merunkova et al., 2012; Wilson et al., 2012; Turtureanu
et al., 2014). For several plot sizes smaller than 50 m2
, dry
grasslands of Central and Eastern Europe hold the world records for
richness of vascular plant species (Wilson et al., 2012). However,
not all dry grasslands are so extremely species-rich: their richness
varies considerably, and knowing the relationships of this variation
to environmental factors can help to predict and eventually avoid
biodiversity loss if combined with wise nature conservation
management.
Several observational and experimental studies have been carried
out in temperate grasslands to explore the relationships between
vascular plant species richness (henceforth ‘species
richness’) and environmental factors at the fine scale, i.e. the scale
of vegetation plots (Grace, 1999). Still, there is a lack of a general
* Corresponding author.
E-mail addresses: salza.palpurina@gmail.com (S. Palpurina), chytry@sci.muni.cz
(M. Chytrý), rossentzonev@abv.bg (R. Tzonev), danihel@sci.muni.cz (J. Danihelka),
axmanova@sci.muni.cz (I. Axmanova), merunkova@sci.muni.cz (K. Merunkova),
mario.duchon@gmail.com (M. Duchon), karakiev@abv.bg (T. Karakiev).
Contents lists available at ScienceDirect
Acta Oecologica
journal homepage: www.elsevier.com/locate/actoec
http://dx.doi.org/10.1016/j.actao.2015.02.001
1146-609X/© 2015 Elsevier Masson SAS. All rights reserved.
Acta Oecologica 63 (2015) 36e46
agreement as to which set of factors and which underlying mechanisms
are the most important controls of species richness in dry
grasslands. Local environmental factors, such as site conditions (e.g.
soil properties, water availability, productivity; L€obel et al., 2006;
Chytrý et al., 2007; Merunkova and Chytrý, 2012; Axmanova
et al., 2013), biotic interactions (e.g. competition for light; Kull
and Zobel, 1991; L€obel et al., 2006), and management (Bonanomi
et al., 2013; Turtureanu et al., 2014), were found to be important
controls of fine-scale species richness in different temperate and
(hemi-)boreal grasslands. However, the species richnesseenvironment
relationship is also influenced by historical and evolutionary
processes at the regional level (large-scale migration, extinction,
and speciation), which determine the size and composition of the
species pool (the set of species potentially able to occur within a
given environment; Zobel, 1992). Therefore, patterns of fine-scale
species richness can differ among biogeographic and climatic regions.
The more common a habitat was throughout history, the
more time there was for speciation and large-scale migration,
leading to the evolution of a larger species pool (‘species pool effect’;
Taylor et al., 1990; Ewald, 2003). In many cases, even in the
very species-rich dry grasslands where biotic interactions are
supposed to set a limit to the number of coexisting species
(Bengtsson et al.,1994), fine-scale species richness patterns seem to
be influenced by species pools (P€artel et al., 1996).
Many dry grasslands in the eastern Balkan Peninsula are
considered to be natural remnants of the forest-steppe that has
persisted in the Lower Danube Basin and the Upper Thracian Plain
since the Pleistocene, serving as an important migration corridor
for grassland species between the Pontic, Anatolian, and Pannonian
steppes (Magyari et al., 2008; Tonkov et al., 2011). Because of their
distinct evolutionary and migrational history, diversity patterns
reported from dry grasslands in other regions of Europe cannot be
directly transferred to the eastern Balkan Peninsula. However, finescale
species richness patterns of these dry grasslands have never
been explored. Several previous studies gathered valuable data on
the dry grassland vegetation of this area, but all of them dealt with
smaller regions and focused on vegetation classification (e.g.
Tzonev et al., 2006; Pedashenko et al., 2013; Sopotlieva and
Apostolova, 2014; see Donit¸a et al., 2005 for references on the
Romanian Dobrudzha) and the relationships between species
composition (not richness) and the environment (Pedashenko
et al., 2013).
Dry grasslands in the eastern Balkan Peninsula have developed
under a unique combination of the Mediterranean and
continental climatic influences. Therefore, they are a suitable
model for assessing the relative effects of these two climate types
on fine-scale species richness. Few previous studies have reported
climate as a control of fine-scale species richness patterns
in temperate grasslands (e.g. Adler and Levine, 2007; Chytrý
et al., 2007; Reitalu et al., 2014; Turtureanu et al., 2014),
possibly because most other research projects have not encompassed
large climatic gradients or biogeographic transitions. In
general, factors operating over broad scales such as macroclimate
are considered to be poor predictors of fine-scale species richness
(Pausas and Austin, 2001). Nevertheless, we hypothesize that, if
filtering by local environmental factors is not too strong and the
climatic gradient is large, climate may explain more variation in
fine-scale species richness than local factors (as in Adler and
Levine, 2007; Chytrý et al., 2007). Further, we hypothesize that
climate should have a contrasting effect on the richness of
different life forms, because Mediterranean and temperate
grasslands considerably differ in the participation of different life
forms (Cain, 1950).
Besides contrasting climates, the Balkan Peninsula is characterized
by a very rich flora, especially in its Mediterranean and subMediterranean
areas (Barthlott et al., 2005), with many endemics
occurring mainly on calcareous substrates, and calcicole species
that immigrated from diversity hotspots (such as Anatolia; Medail
and Quezel, 1997) where calcareous soils prevail (Mücher et al.,
2009). Thus, based on the species pool hypothesis (Taylor et al.,
1990), we expect soil pH to be a strong correlate of fine-scale
species richness in dry grasslands in the eastern Balkan Peninsula.
Positive relationships with soil pH, explained by the prevalence
of calcicole species in the existing species pools (Ewald,
2003), were found under continental, oceanic, and Mediterranean
conditions across Eurasia (Pӓrtel, 2002; Schuster and Diekmann,
2003; Chytrý et al., 2003, 2007, 2010). However, the relationship
between species richness and soil pH could be modified by the
interaction with other factors such as climate (L€obel et al., 2006;
Chytrý et al., 2007). The decrease in species richness on soils with
very high pH might be due to the lower number of species adapted
to the drought stress (low precipitation) associated with very basic
substrates (Chytrý et al., 2007). Thus, we consider dry grasslands in
the eastern Balkan Peninsula, occurring on different bedrock types
and distributed over a large precipitation gradient, to be a good
model for evaluating the species richnessesoil pH relationship in
different biogeographic contexts.
To explore fine-scale species richness patterns in dry grasslands
in the eastern Balkan Peninsula, we sampled vegetation plots along
a broad spatial extent and long climatic and altitudinal gradients,
and assessed the relative importance of these and other measured
variables on species richness. Our main questions were:
1. Is there a geographical pattern in fine-scale species richness in
the eastern Balkan Peninsula coinciding with the transition
between the subcontinental and the sub-Mediterranean
biogeographic regions?
2. Which factors are the most important drivers of the variation in
fine-scale species richness of dry grasslands across the eastern
Balkan Peninsula: soil properties, climate, or management?
3. Do species of specific life forms or biogeographic histories (e.g.
sub-Mediterranean vs. subcontinental) contribute differently to
species richness patterns?
2. Materials and methods
2.1. Study area
The study area is located in the eastern Balkan Peninsula
(41.87e45.15 N, 22.59e29.08 E), and includes two major lowlands
separated by the Stara planina (Balkan) mountain range: the
Dobrudzha (Dobrogea) plateau and the Lower Danube basin in the
north, and the Upper Thracian plain in the south. Dry grasslands
were sampled from the Black Sea coast up to the mid-altitudes,
including the karstic low mountains in western Bulgaria, the
Stara planina, and the Strandzha-Sakar massif (14e1211 m a.s.l.,
Fig. 1). Limestones and other calcareous rocks of different ages and
origins prevail in the study area. An exception is the Upper Thracian
plain, with predominating Quaternary alluvial-diluvial sediments
and Neogene continental deposits, and the Strandzha-Sakar massif,
with predominating volcanic, metamorphic, and granitoid rocks.
But even so, dry grasslands in the latter regions are mainly associated
with outcrops of carbonate bedrocks, such as crystalline
limestone. In the northern Dobrudzha and the Lower Danube basin,
Quaternary loess deposits form a 20e60 km broad band parallel to
the river (Kopralev, 2002).
The climate is temperate-continental (henceforth ‘subcontinental’)
to the north of the Stara planina and in western Bulgaria,
continental-Mediterranean (henceforth ‘sub-Mediterranean’) in
S. Palpurina et al. / Acta Oecologica 63 (2015) 36e46 37
the Strandzha-Sakar massif and along the Black Sea coast, and
transitional in the Thracian plain, the eastern Stara planina, and the
easternmost Dobrudzha (Kopralev, 2002). Continentality decreases
and mediterraneity (i.e. the length and severity of summer drought
and benign winter temperatures) increases from north to south and
with proximity to the Black Sea, and both decrease with altitude
(see Fig. 1). Precipitation decreases with continentality and increases
with altitude (Kopralev, 2002). The precipitation maximum
is in summer in the subcontinental areas, in autumn in the subMediterranean
areas, and in spring and autumn in the transitional
areas. The precipitation minimum is in winter in the subcontinental
areas, in summer in the sub-Mediterranean areas, and
in autumn and summer in the transitional areas. A persistent snow
cover is regularly formed only in the subcontinental areas, whereas
in the sub-Mediterranean areas snow cover lasts normally for less
than 20 days, and usually does not form every year (Kopralev,
2002). Temperature decreases with altitude, with July being the
warmest month, and January the coldest. Summer temperatures
vary little throughout the study area, whereas winters get colder
and longer with increasing continentality.
Forest-steppe is the natural vegetation on chernozems, humuscarbonate
and other soils in the Dobrudzha and along the Danube.
Elsewhere in the lowlands of the subcontinental and transitional
areas, i.e. up to 700 m a.s.l., the potential vegetation would be
thermophilous summer-green oak forests with Quercus robur
subsp. robur (in the river valleys), Quercus robur subsp. pedunculiflora,
Quercus cerris, Quercus frainetto, Quercus pubescens, and Carpinus
orientalis (Kopralev, 2002). Most of the natural vegetation
(steppe, forest-steppe, and thermophilous forests) in these lowlands
was turned into arable land. Nowadays, natural and secondary
grasslands (pastures) are restricted mainly to steep slopes or
stony soils unsuitable for agriculture. In many places where the
forests were cleared, secondary scrub with Paliurus spina-christi,
Syringa vulgaris, Prunus spinosa, and Crataegus monogyna develops
after the cessation of agricultural management, but in a few
places, meadow-steppes remain at regularly mown sites. In the
mountainous areas, above 600 m, mesophilous oak-hornbeam
forests with Quercus dalechampii and Carpinus betulus predominate,
being replaced at altitudes above 1300 m by beech forests
with Fagus sylvatica, except in the Strandzha, where beech forests of
Fagus orientalis occupy humid and cool valleys below the oak
(mostly Quercus polycarpa) forest belt. Within the oak-hornbeam
and beech forest belts, dry grasslands are considered natural only
on steep south-facing slopes with shallow soils over carbonates in
the continental climate of western Bulgaria, where Festuca spp. and
Bromus spp. co-dominate, accompanied by many steppe specialists
and regional endemic species. Grasslands dominated by Chrysopogon
gryllus or Bothriochloa ischaemum, with many generalist
species, are thought to be of a secondary origin, developed at sites
of cleared forests.
2.2. Field survey and data compilation
We wanted to cover the maximum ecological variation of dry
and semi-dry grasslands (phytosociological class Festuco-Brometea)
across a large climatic gradient and different bedrock types.
Fig. 1. Study area (Bulgaria and Romania south of the Danube river) with locations of vegetation plots (n ¼ 172; not all plot symbols are visible due to the coarse scale). The solid line
corresponds to the 0.1C isotherm of January mean temperature. January temperatures increase to the east and south of the isotherm. The dashed line corresponds to the pluvial
continentality index (Icont) equal to 1; Icont increases to the north and west, and with altitude.
S. Palpurina et al. / Acta Oecologica 63 (2015) 36e4638
Therefore, study sites were selected based on the literature and
local expert suggestions with the aim of covering all major types of
dry grasslands (petrophytic steppes, loess steppes, and semi-dry
grasslands). The intensity of sampling in each climatic area (see
Section 2.1.) reflected its size, environmental heterogeneity, and the
range of different dry grassland types. Abandoned, recently
strongly disturbed, or intensively grazed grasslands were avoided.
However, as most of the loess steppes we visited were burnt in the
previous year, we sampled some post-fire sites there.
At each dry grassland site, 10 Â 10 m2
plots were sampled in the
centre of physiognomically uniform vegetation stands with homogeneous
abiotic conditions. Dry grasslands were selected based
on the presence of species characteristic of the phytosociological
class Festuco-Brometea. If a site was considerably topographically or
edaphically variable, and this was reflected in the grassland structure
and composition, we recorded more than one vegetation plot
there. In order to sample dry grasslands at the peak of their growing
season, field work was carried out each year from the end of May till
the end of June, and in the first week of July at higher altitudes.
A total of 172 vegetation plots were sampled in 2010e2012.
Within each 100-m2
plot, all vascular plant species were recorded
(root presence). In addition to total species richness (the number of
all vascular plants per 100-m2
plot), we also counted the number of
species sharing the same life form: annuals, biennials, perennial
hemicryptophytes (henceforth ‘perennials’), geophytes, suffruticose
chamaephytes (henceforth ‘suffruticose’), shrubs, and
trees. Bunch grasses such as Stipa capillata were regarded as perennials,
and rhizomatous grasses such as Elymus repens as geophytes.
Regarding its main centre of origin and distribution in
relation to the study area, each species was classified into one of
eight geoelement categories based on Assyov and Petrova (2012):
Balkan endemic and sub-endemic (main distribution area on the
Balkan Peninsula, or the mountains of SE Europe), subMediterranean
(main distribution area to the south of the study
area), sub-Pontic (main distribution in the steppes north of the
Black Sea), subcontinental (distributed mainly in continental Eurasia),
European (centre of distribution in temperate Europe), subboreal,
cosmopolitan, and adventive.
For each plot, geographical coordinates and altitude were
measured with a GPS device. Slope and aspect were measured by a
clinometer and compass. The cover of bare rock and gravel was
estimated visually in percentages. Soil depth was measured within
each plot at 10 systematically placed points with a 30-cm metal rod.
The mean soil depth was considered as a surrogate for moisture
availability, and the coefficient of variation (CV) of soil depth was
calculated as a measure of soil depth heterogeneity. In the analyses,
soils deeper than 30 cm were considered as 30 cm deep. Soil
samples were taken from each plot at five points (in the centre and
Table 1
Descriptive statistics of the variables used in the analysis based on a dataset of 156 HCR-selected plots. For categorical and ordinal variables, the number of cases (plots) for each
class is given in brackets. Species richness is given as numbers of vascular plant species per 100 m2
. Variables used for the regression tree are in bold.
Variable name Mean ± SD Minimum Maximum
Topography Altitude [m] 306 ± 263 14 1033
Inclination [
] 9.8 ± 6.9 0 34
Radiation (topographic index) 0.92 ± 0.06 0.65 1.01
Heat (topographic index) 0.92 ± 0.07 0.68 1.02
Soil Bedrock type: carbonate (103), loess (26), other sedimentary rock (12), siliceous rock (11), volcanite (4)
Soil depth [cm]a
18.0 ± 9.1 3 30
Soil depth CV [%]a
27.4 ± 21.0 0 101
Cover of gravel and rock [%] 9.7 ± 15.0 0 70
Soil pH (H2O) 7.3 ± 0.9 4.9 8.5
Soil pH CV [%]b
2.1 ± 1.9 0.1 12.0
Soil Ca [mg/kg dry soil] 10207 ± 8277 724 39041
Soil K [mg/kg dry soil] 279 ± 123 85 705
Soil P [mg/kg dry soil] 5.6 ± 8.5 0.0 71.0
Soil Corg [g/kg dry soil]a
65.1 ± 25.7 14.4 175.7
Soil Ntot [g/kg dry soil] 3.4 ± 1.6 1.1 9.6
Soil C/N ratio 19.8 ± 5.0 11.1 41.9
Climate Mean annual temperature [
C] 10.8 ± 1.2 7.1 13.5
Mean January temperature [
C] À0.6 ± 1.4 À3.7 3.8
Mean July temperature [
C] 21.2 ± 1.5 16.6 23.0
Annual temperature range [
C] 21.8 ± 1.3 18.9 24.1
Annual precipitation sum [mm] 578 ± 69 403 711
Spring precipitation sum [mm] 179 ± 31 116 241
Summer precipitation sum [mm] 132 ± 16 98 172
Autumn precipitation sum [mm 147 ± 25 92 218
Winter precipitation sum [mm] 121 ± 19 85 176
Pluvial continentality index [pIII-VIII/pIX-II]b
1.18 ± 0.19 0.61 1.41
Length of cold period [no. of months with t < 0
C]b
0.9 ± 0.8 0 3
Length of summer drought [no. of months with IDM < 20]b
2.7 ± 1.6 0 7
Management type: abandoned (13), burned (14), grazed (120), mown (9)
Species richness All species 52.2 ± 17.6 14 107
Annuals 16.1 ± 12.1 0 54
Biennials 2.9 ± 1.8 0 8
Geophytes 3.0 ± 2.0 0 9
Perennials 24.8 ± 8.4 10 52
Balkan endemic and sub-endemic 2.7 ± 2.2 0 12
European species 2.2 ± 1.9 0 12
Subcontinental species 14.6 ± 6.0 3 31
Sub-Mediterranean species 23.2 ± 12.0 2 68
Sub-pontic species 7.3 ± 4.0 0 19
a
Soils deeper than 30 cm were arbitrarily given a value of 30 cm.
b
Abbreviations: CV e coefficient of variation; IDM e monthly de Martonne index of aridity: drought episodes could be observed within a month in which IDM < 30, while
IDM < 20 indicates a month with a severe drought; p, t e mean monthly precipitation and temperature.
S. Palpurina et al. / Acta Oecologica 63 (2015) 36e46 39
the four corners) at a depth of 3e10 cm, and subsequently air-dried.
Soil pH was measured separately for each soil sample after an
8e10 h extraction in distilled water (2:5 soil:water weight ratio).
The mean value of soil pH per plot and its coefficient of variation
(CV) were used for analyses, the latter being considered as a measure
of soil pH heterogeneity. Nutrients were measured in a
mixture of the five soil samples following standard protocols. Total
nitrogen (Ntot) was determined using the Kjehldahl method. Plantavailable
phosphorus (P), potassium (K), and calcium (Ca) were
extracted by the Mehlich III method. The P content was determined
by spectrophotometer (Spekol 210, Carl Zeiss, Jena, Germany), and
K and Ca contents by atomic absorption spectrophotometer (AAS
933 Plus, GBC Scientific Equipment, Melbourne, Australia). Total
organic carbon (Corg) was determined by loss on ignition at 550C
for 16 h. The C/N ratio was calculated as Corg/Ntot.
Management of the dry grasslands was assessed in the field, and
assigned to the following four categories: grazed (parts of plants
missing, animal dung present, livestock present at the site or
nearby), mown (no or little litter accumulated), burned (charcoal
present, no litter), and abandoned (litter accumulated).
Radiation and heat-load indices (McCune and Keon, 2002, eq. 3
in their Table 2) were calculated based on latitude, slope, and
aspect. The radiation index has the highest value of 1 on the steep
south-facing slopes and the lowest value of 0 on the steep northfacing
slopes; for the heat load index, the highest and lowest
values are shifted to SW- and NE-facing slopes, respectively.
Bedrock types were extracted from the geological maps of
Bulgaria and Romania (Anonymous, 1968; Cheshitev et al., 1989)
and cross-checked in the field. The categories given in the maps
were combined into five broad categories: carbonate (including
crystalline limestone and flysch), loess, other sedimentary rock
(sandstone, siltstone, and conglomerate), siliceous rock (e.g. granitoid,
rhyolite, and quartzite schist), and volcanite (basalt and
andesitobasalt).
The mean monthly precipitation sum and mean monthly air
temperature were extracted from the WorldClim database at a
spatial resolution of ca.1 km (Hijmans et al., 2005; www.worldclim.
com). In order to approximate the degree of continentality or
mediterraneity of the climate, the pluvial continentality index, Icont,
was calculated as the precipitation of the six warmest months
(March to August) divided by the precipitation of the six coldest
months (September to February; Tuhkanen, 1980). In the study
area, September is warmer than March, but we included it in the
colder half-year, since higher precipitation in September is an
indication of sub-Mediterranean conditions. Thus, Icont > 1 indicates
continental type, and Icont < 1 Mediterranean type of precipitation
seasonality. Additionally, we calculated (1) the length of
the cold period, expressed as the number of months with mean
monthly temperature below 0C; and (2) the length of the drought
period, as the number of months with IDM < 20. IDM is the de
Martonne index of aridity (mm/C), calculated for each month as
12p/(t þ 10), where p and t are the monthly mean precipitation
(mm) and air temperature (C). Within the study area, drought
episodes could be observed within a month in which IDM is lower
than 30; value below 20 indicates severe drought (Tuhkanen,1980).
A summary of the descriptive statistics for the variables used in
the analysis is provided in Table 1.
2.3. Data analysis
Besides environmental factors and biogeographic context,
grassland species richness is also influenced by management. In
particular, abandoned sites tend to be species-poorer than regularly
managed sites. As we were able to obtain only rough data on the
recent management of all sampled sites, we decided to also analyse
a subset of those plots that were the species-richest in each area,
removing those that became species-poor due to possible suboptimal
management. We overlaid the study area by a 10 Â 10 km
grid, and within each cell we selected the plot with the highest
richness. To reduce the effects of oversampled regions, we made
another selection using grid cells of 40 Â 50 km cells. In each grid
cell that contained at least five plots, we selected up to five plots
representing various species compositions following the
heterogeneity-constrained random (HCR) resampling procedure
(Lengyel et al., 2011) using the JUICE program (Tichý, 2002). We
made parallel analyses with the whole dataset (n ¼ 172), the
richest-plot selection (n ¼ 95), and the HCR-resampled plots
(n ¼ 156). As the results were almost identical between the full
dataset and its subsets, we describe and discuss here only the results
from the HCR-resampled dataset.
We chose regression trees as an effective exploratory tool for
describing the relationships between the response and explanatory
variables (De'ath and Fabricius, 2000). To estimate the prediction
error of the trees, we used the 10-fold cross-validation procedure.
We ran 100 series of the 10-fold cross-validation, and from each
series selected the best tree size according to the minimum error
rule. From the distribution of the 100 ‘best’ tree sizes, the most
frequently occurring size was chosen (De'ath and Fabricius, 2000).
For each split of the tree, we also selected the first five surrogate
variables, i.e. those that allocate the most cases similarly to the
allocation by the primary splitter. As a criterion for including a
variable as a surrogate, we required its association value to the
main splitter to be higher than 0.7. As the trees may be unstable if
there is a strong correlation among the explanatory variables,
collinearity between continuous predictor variables was checked
by pairwise Pearson correlations. Two predictors were considered
to be collinear when Pearson's jrj > 0.70. In such cases, we kept only
the predictors that we considered more ecologically meaningful for
explaining richness. Details on the correlations between explanatory
variables are given in Appendix A.
Table 2
Ranked values of the relative importance of explanatory variables based on the
optimal regression tree for species richness. The relative importance of a given
variable is calculated as the average of the overall improvement of the model
(measured as a deviance explained) by this variable stemming both from the variable's
role as a primary splitter of a node, and its role as a surrogate to any of the
primary splitters. Note that due to rounding the sum of relative importances across
individual variables or groups slightly exceeds 100.
Category Individual variables Relative
importance (%)
Climate January temperature 37.1
Topography Altitude 12.0
Climate Annual temperature range 11.4
Soil Bedrock type 8.8
Climate July temperature 7.4
Soil Soil pH 4.7
Climate Annual precipitation 4.6
Soil C/N ratio 3.2
Soil Soil Ca 2.5
Soil Soil K 2.4
Soil Soil pH CV 1.6
Topography Radiation 1.6
Soil Soil N 1.4
Soil Cover of rock
and gravel
0.9
Management Management 0.5
By group
Climate 60.5
Soil 25.5
Topography 13.6
Management 0.5
Total 100.0
S. Palpurina et al. / Acta Oecologica 63 (2015) 36e4640
Linear models were built separately for the relationship
involving the richness of all vascular plants and soil pH and other
selected environmental variables. Where a unimodal response was
theoretically expected, we tested for the significance of the negative
quadratic term. The response and explanatory variables were
analysed on their original, non-transformed, scale.
To assess the strength of the relationship between the richness
of separate life forms/geoelements and environmental variables,
Spearman's rank correlation coefficient was used. The Monte-Carlo
permutation test was used with 1000 permutations to test the
significance of the relationship.
The statistical analyses were carried out in the R environment (R
Core Team, 2012) using the package ‘vegan’ (Oksanen et al., 2013)
for the linear regression models, and ‘rpart’ (Therneau and
Atkinson, 2012) and ‘partykit’ (Hothorn and Zeileis, 2013) for
building and visualizing the regression trees.
3. Results
Total species richness varied between 14 and 107 vascular plant
species per 100 m2
(both for the whole dataset and the subset of
HCR-selected plots). An overview of species richness by life forms
and geoelements is given in Appendix B.
Climatic variables explained most of the variation in total species
of dry grasslands in the study area, followed by soil properties
and topography (Table 2). January temperature was the strongest
predictor, with its split reducing the variation in species richness by
more than a third (Fig. 2). Species richness increased strongly with
January temperature, reaching the highest values at sites with
positive January means (Fig. 2, Appendix C). Within the study area,
January temperature is strongly negatively correlated with the
length of the cold period and the pluvial continentality index (see
Appendix A). Thus, positive January temperature indicates mild and
precipitation-rich winter, typical of the sub-Mediterranean climate.
The most species-poor dry grasslands were at sites with negative
January temperatures (i.e. outside the sub-Mediterranean
areas) on soils over loess or siliceous bedrock, characterized by a
low Ca and N content and, in a few cases, by burning. These
grasslands are distributed in the subcontinental plains as indicated
by the low altitude and low precipitation. Elsewhere in the subcontinental
lowlands, where the precipitation was higher, or the
bedrock was calcium-richer, dry grasslands were species-richer. In
general, outside the sub-Mediterranean areas, richness tended to
increase with both altitude and precipitation (surrogates in the
bedrock and annual temperature range splits), with dry grasslands
in mountainous and precipitation-rich areas being comparably rich
as some rocky steppes in the sub-Mediterranean areas. An exception
to this pattern were the dry grasslands on very basic soil
(pH > 8), which were more species-poor than other dry grasslands
at higher altitudes.
Fig. 2. The optimal regression tree for the total species richness cross-validated using the minimum rule. Numbers at each split indicate the mean number of vascular plant species
per 100m2
± standard deviation. The primary splitter variable is in bold, surrogates (variables that allocate most cases to the same group as the primary splitter) are in smaller
letters below the primary splitter. Number of plots assigned to each node, and the deviance explained are in smaller font below the main splitter name.
S. Palpurina et al. / Acta Oecologica 63 (2015) 36e46 41
After taking the climate effect into account (Fig. 3), a decrease in
species richness with soil pH was observed in the foothills and the
mountains. In the sub-Mediterranean areas and subcontinental
plains no pattern was detected. Soil depth had contrasting effects
on species richness in different regions. In the foothills and the
mountains, species richness increased with soil depth; in the subMediterranean
areas, the relationship was unimodal (Fig. 3).
Within the study area, higher species richness was associated
with a strong increase in the number of sub-Mediterranean species,
most of them being annuals (Fig. 4). The total species richness was
not much related to the number of perennials, while the numbers
of geophytes, biennials, and suffruticose species per plot were low
and did not vary much. Most of the plots contained no trees or
shrubs (Fig. 4, Appendix B). The positive relationship between total
species richness and January temperature arose mainly from the
strong increase in the richness of sub-Mediterranean (and annual)
species with January temperature (Table 3, Appendix C). The
numbers of sub-Mediterranean and annual species were moderately
to strongly negatively correlated with the length of the cold
period, the pluvial continentality index, and the annual temperature
range (i.e their richness increased strongly with mediterraneity
and decreased with continentality), and were weakly
correlated with topographic or soil factors (Table 3). The other most
abundant life form, perennials, showed no strong relationship with
either local factors or climate. Within a plot, subcontinental and
sub-Pontic species were on average the second and third most
represented geoelements after sub-Mediterranean species
(Appendix B). Similarly to sub-Mediterranean species, the number
of sub-Pontic species increased strongly with the mediterraneity of
the climate (i.e. decreased with the length of the cold period and
increased with January temperature, characteristic for the mild
winter along the coast), but was also strongly positively correlated
with soil factors such as pH and Ca content (Table 3). The number of
subcontinental species was not correlated with climate, but was
strongly correlated with soil depth (positively) and with the cover
of rock and gravel (negatively).
4. Discussion
4.1. Species richness changes at the transition of two biogeographic
regions: Macroclimatic control or migration constraints?
Our data suggest that climate is the most important driver of
fine-scale species richness patterns across dry grasslands in the
eastern Balkan Peninsula. Based on the strong decrease of subMediterranean
species with decreasing January temperature
(Appendix C) and the absence of many of them from the subcontinental
areas (Appendix E), we suggest that the zero January
Fig. 3. The relationships between total species richness and annual precipitation, soil depth, and soil pH within climatic areas. Climatic areas were delineated based on the January
temperature in the first split of the optimal regression tree, and the annual temperature range in the last split (Fig. 2) as follows: sub-Mediterranean areas (January
temperature ! 0.1C); subcontinental areas (January temperature < 0.1C, and the annual temperature range ! 21.8C); mountains and foothills (January temperature < 0.1C, and
annual temperature range < 21.8C). Only significant trends were fitted. Significance levels: ***: P < 0.001, **: P < 0.01, *: P < 0.05. The number before the significance level is
adjusted R2
.
S. Palpurina et al. / Acta Oecologica 63 (2015) 36e4642
isotherm creates the main border between the richer (sub-Mediterranean)
and poorer (subcontinental) dry grasslands by limiting
mainly the distribution of sub-Mediterranean species. The limiting
effect of January temperature on species distribution can be a direct
one e due to the physiological effect of frost (Normand and Treier,
2009). However, it is difficult to separate the effect of physiological
constraints from dispersal limitation in determining species ranges,
especially in mountainous areas such as the Balkan Peninsula. It is
possible that the mountain ranges in the study area influenced
species migrations indirectly through their effect on climate patterns,
rather than directly as migration barriers. The Stara planina
and Rhodopes Mountains did not stop the northward migration for
many sub-Mediterranean species, which were able to migrate
northwards along the Black Sea coast, probably aided by milder
winters near the sea. Additionally, the stability of climatic conditions
(Davis et al., 2003) and the continuous persistence of foreststeppe
in the area (Magyari et al., 2008; Tonkov et al., 2011) may
have facilitated continuous migrations between the subcontinental
and sub-Mediterranean areas. Still, given the large beta-diversity of
the dry grasslands across the study area (not shown), large-scale
dispersal limitation at the sub-Mediterranean/subcontinental
boundary is not improbable. Dispersal limitation might explain
the locally lower species richness in the dry grasslands of the
western Bulgarian limestone mountains, as their connection with
the sub-Mediterranean areas is not as open as in the mountains
near the Black Sea.
Although some studies suggest that high temperature and
drought can limit species ranges in some areas (Cahill et al., 2014),
the greater summer aridity, associated with the sub-Mediterranean
climate, appeared as a limiting factor only for the number of Balkan
endemics and sub-endemics, probably because most of these species
originated in the precipitation-richer and colder mountainous
areas. Otherwise, the length of the summer drought was not
associated with any decline in species richness of particular life
forms or geoelements in the study area. An explanation could be
that the summer drought period is an intrinsic attribute of the dry
grassland habitat, with most species of continental origin reaching
the peak of their biomass production before the summer drought
period. It is interesting to note that many typical dry grassland
species of continental origin, e.g. Filipendula vulgaris, Salvia nemorosa
and Salvia nutans, extend their range into sub-Mediterranean
areas but not further south to the typical Mediterranean areas.
Although the species pool effects seem to have a dominant influence
on species richness patterns within our study system, a
difference in the species pool may not be the full explanation why
dry grasslands in the sub-Mediterranean areas are species-richer
locally, in small plots. Local biotic processes such as asymmetric
competition can reduce diversity (Kull and Zobel, 1991; L€obel et al.,
2006). Theoretically, competitively superior perennial species
might outcompete most of the sub-Mediterranean species that are
annuals, but this does not happen in the eastern Balkan subMediterranean
dry grasslands. An explanation can be that earlier
growth and seed production allow annuals and geophytes to avoid
competition with perennials (Shmida and Wilson, 1985), which
produce most of their biomass later in the growing season. In
addition, grasslands at sites with long summer droughts in areas
with stronger Mediterranean climatic features provide more
regeneration gaps for germination in autumn-winter (Ortega et al.,
1997) favourable for the establishment of annuals (Huber, 1994).
This is also in agreement with the fact that many species, especially
annuals, respond positively to dry conditions in the previous year
(Adler and Levine, 2007). Thus, beside frost, the observed decrease
in species richness (especially of annuals) with continentality and
altitude in the study area can be explained by a reduced availability
Fig. 4. Proportion of life forms and geoelements in plots with different species richness (n ¼ 156). The range of species richness within each class is in brackets below the bars
(squared bracket e value in the interval, rounded bracket e value not in the interval). The number of plots within a given class is above the corresponding bar. Shading intensities
and sections of the bars correspond to the combinations of life forms and geoelements.
S. Palpurina et al. / Acta Oecologica 63 (2015) 36e46 43
Table 3
Spearman rank correlations between fine-scale species richness (number of species per 100 m2
) of the entire dry grassland community or respective life forms/geoelements, and environmental variables (HCR-selected plots,
n ¼ 156). Significance levels:***: P < 0.001,**: P < 0.01,*: P < 0.05, ns e not significant. Significance was tested using Monte Carlo test with 1000 permutations. Strong relationships (r ! j0.5j) are in bold. ann e annual, bie e biennial,
geo e geophyte, per e perennial, suffr e suffruticose, balk e Balkan (sub-)endemic; pont e sub-Pontic, cont e continental, med e sub-Mediterranean, eur e European, bor - subboreal.
All species Life forms Geoelements
ann bie geo per suffr shrub tree balk pont cont med eur bor
Topography Altitude ns ns À0.28** ns ns 0.14* ns 0.15* 0.45*** À0.28*** À0.19** ns ns 0.19**
Inclination ns ns ns À0.20** ns 0.26*** ns ns ns ns À0.28*** ns ns ns
Radiation À0.15* ns À0.14* À0.21** À0.23*** ns ns ns ns À0.18** À0.14* ns À0.15* ns
Heat ns ns À0.15* À0.18* À0.15* ns ns ns ns ns ns ns À0.17* ns
Soil Soil depth ns À0.22** ns 0.14* 0.15* À0.37*** ns 0.13* À0.39*** À0.30** 0.40*** À0.31*** 0.39*** 0.27**
Cover of gravel and rock ns 0.19** ns À0.14* À0.14* 0.33*** ns ns 0.32*** 0.27*** À0.42*** 0.32*** À0.37*** À0.25**
Soil pH (H2O) ns ns ns À0.20* ns 0.51*** ns À0.21** 0.18* 0.40*** À0.20** ns À0.32*** À0.37***
Soil pH CVa
ns ns ns ns ns À0.36*** ns ns À0.21** À0.49*** 0.17* ns 0.23** 0.16*
Soil Ca 0.19* 0.14* 0.24*** ns ns 0.44*** ns ns 0.17* 0.53*** ns 0.21** À0.26** À0.31***
Soil K ns ns ns ns ns 0.23*** ns ns ns 0.18* ns 0.17* À0.15* À0.24***
Soil P À0.15* ns ns ns À0.14* À0.24*** ns ns À0.24** ns ns À0.21** ns ns
Soil Corg 0.22** ns ns 0.19* 0.15* 0.22** ns ns 0.33*** 0.19* ns 0.24** ns ns
Soil Ntot 0.20** ns ns 0.22** 0.15* 0.25** ns ns 0.37*** 0.24** ns 0.23** ns ns
Soil C/N ratio ns ns ns À0.21** ns ns ns ns À0.19** À0.15* 0.15* ns ns ns
Climate Mean annual temperature 0.22** 0.35*** 0.28*** 0.18* ns À0.16* ns ns À0.19** 0.25** 0.14* 0.26*** ns ns
Mean January temperature 0.47*** 0.51*** 0.34*** 0.29*** 0.16* ns 0.13* ns ns 0.46*** 0.21** 0.50*** ns ns
Mean July temperature ns 0.15* 0.16* ns À0.24*** À0.21** À0.26** ns À0.34*** ns ns ns À0.13* À0.13*
Annual temperature range ¡0.52*** À0.27*** ns À0.39*** À0.39*** À0.25** À0.31*** À0.14* À0.49*** À0.27*** ns À0.48*** ns À0.14*
Annual precipitation sum ns ns À0.18* ns ns À0.17* ns 0.27*** 0.23** À0.44*** ns ns 0.21** 0.18**
Spring precipitation sum À0.25** À0.25*** À0.23*** À0.16* ns ns ns 0.15* ns ¡0.55*** ns À0.26*** 0.18* 0.16*
Summer precipitation sum À0.27*** À0.27*** À0.18* À0.20** ns ns ns ns ns À0.46*** ns À0.30*** 0.15* ns
Autumn precipitation sum 0.49*** 0.37*** ns 0.40*** 0.25*** ns 0.21** 0.24** 0.52*** ns ns 0.50*** 0.16* ns
Winter precipitation sum 0.20* 0.18* À0.16* 0.20** ns À0.19* ns 0.26*** 0.33*** À0.39*** ns 0.24*** 0.18* 0.19*
Pluvial continentality index À0.47*** À0.42*** ns À0.37*** À0.19** ns À0.18** ns À0.37*** À0.34*** ns ¡0.55*** ns ns
Length of cold period ¡0.51*** ¡0.58*** À0.36*** À0.30*** À0.17* ns ns ns ns À0.47*** À0.24** ¡0.55*** ns ns
Length of summer drought 0.16* 0.22** 0.21** ns ns ns ns ns À0.25*** 0.41*** ns 0.18* ns ns
a
CV e coefficient of variation.
S.Palpurinaetal./ActaOecologica63(2015)36e4644
of regeneration gaps at higher water availability (that is in summer),
increasing with both continentality and altitude.
4.2. The importance of bedrock and soil pH in subcontinental areas:
Is it only a species pool effect?
In subcontinental areas, dry grasslands over loess and siliceous
bedrock contained the fewest species. While we could explain the
low richness of acidic grasslands with a limited number of species
adapted to the double stress of soil acidity and drought, it is more
difficult to explain why grasslands developed on fertile chernozem
soils over loess were so species-poor. A combination of drought
with the burning of some of the loess grasslands might be one of
the reasons, as this management can have negative short-term
effect on the species richness of temperate grasslands (Ryser
et al., 1995).
Contrary to our expectations, the relationship between species
richness and soil pH in dry grasslands across the eastern Balkan
Peninsula was weak, but the unimodal shape with an optimum at
pH about 7 was consistent with previous studies (e.g. Schuster and
Diekmann, 2003; L€obel et al., 2006). Nevertheless, after taking
climate in account (Fig. 3), a strong decrease in richness on soils
with pH > 8 in the mountainous areas was observed. This pattern
can be due to fewer species adapted to nutrient uptake limitation
associated with very basic substrates (Tyler, 1994). However,
considering that the few plots on high-pH soils are all from the
karst mountains of western Bulgaria, this pattern can be locally
idiosyncratic.
In the sub-Mediterranean areas, soil pH varied over a broad
range, but in spite of that no relationship with species richness was
detected. Both low-pH and high-pH soils were species-rich, indicating
that probably most sub-Mediterranean species in the study
area are generalists regarding soil pH. This is in contrast to the
positive relationship of species richness to pH reported for the
typical Mediterranean vegetation (Chytrý et al., 2010).
4.3. Species richness increases with precipitation: Moisture
availability or spatial mass effect?
We found that outside of the sub-Mediterranean area of the
eastern Balkan Peninsula species richness increased with precipitation,
which in turn increased with altitude. This is consistent with
patterns found in other temperate grasslands: in southern Siberia,
meadow steppes at higher altitudes (with higher precipitation)
were richer than dry lowland steppes (100-m2
plots; Chytrý et al.,
2007), or in North America, species richness increased with precipitation
in temperate grasslands (1-m2
plots; Adler and Levine,
2007). The existence of fewer species in the regional species
pools adapted to drought-stress conditions was proposed as an
explanation for this pattern (Chytrý et al., 2007).
Another explanation for the decrease in fine-scale species
richness in dry grasslands from the mid-altitudes to the more
continental lowlands could be the increased unpredictability and
severity of summer droughts. Unpredictability and severity of
climate increase the chance for local extinctions of species at the
limit of their range or in isolated populations (Huston, 1979). This
could be particularly valid for dry grasslands in the subcontinental
lowlands in the eastern Balkan Peninsula, as most of them occur in
isolated smaller patches, which implies limited species immigration
and increased risk of extinction (MacArthur and Wilson, 1963).
An additional explanation for the higher species richness at
higher altitudes could be related to the spatial mass effect (Shmida
and Wilson, 1985). Unlike dry grasslands in the continental lowlands,
scattered as isolated patches among agricultural fields, those
in the mountainous areas are surrounded by a mosaic of mesic
grasslands and forests, and thus more mesophilous species can
coexist with dry grasslands species.
5. Conclusions
This is the first study on fine-scale species richness patterns in
dry grasslands of the eastern Balkan Peninsula. Our results suggest
that climate is the most important factor influencing fine-scale
species richness in these grasslands, possibly due to species-pool
effects. Fine-scale species richness patterns within the study area
exhibit strong geographical patterns in relation to climate: dry
grasslands in the sub-Mediterranean areas are species-richer,
mainly due to the higher number of sub-Mediterranean species,
especially annuals. When focussing on different climatic regions
separately, species richness increases with precipitation outside the
sub-Mediterranean areas. In the subcontinental lowlands, lower
precipitation in combination with specific substrate properties (e.g.
greater acidity) or unfavourable management (burning) results in
the low species richness of dry grasslands. Elsewhere in the subcontinental
areas, species richness increases with precipitation but
decreases if the substrate is too basic (pH > 8). Thus, local factors
gain importance mainly within climatically and biogeographically
homogeneous regions.
Statement of authorship
SP and MC designed the study. All authors participated in field
sampling. SP prepared the dataset, performed all analyses in
collaboration with MC, and wrote the manuscript. All authors discussed
the results and were involved in revisions of the manuscript.
Acknowledgements
We are grateful to Chavdar Gussev, Hristo Pedashenko, Iva
Apostolova, Kalina Lazarova, and Simeon Marin for providing data
on the distribution of dry grasslands in Bulgaria; Svetlana Bancheva
for critical revision of the Centaurea specimens; the Persina Nature
Park Directorate, Petar Todorov, Vassil Kirov, Vesselina Kadreva, and
Yordan Palpurin, for logistics and technical assistance in the field;
and Anne Johnson, Radomíra Bednarova, and Robert Helan for
language correction of the text (service provided as part of Academic
Writing course at Masaryk University). This research was
funded by the Czech Science Foundation (project no. 14-36079G,
Centre of Excellence PLADIAS).
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.actao.2015.02.001.
References
Adler, P.B., Levine, J.M., 2007. Contrasting relationships between precipitation and
species richness in space and time. Oikos 116, 221e232.
Anonymous, 1968. Harta Geologica a Rom^aniei, Scara 1:200 000. Institutul Geologic
al Rom^aniei, Bucures¸ ti.
Assyov, B., Petrova, A. (Eds.), 2012. Conspectus of the Vascular Flora in Bulgaria.
Bulgarian Biodiversity Foundation, Sofia, p. 489.
Axmanova, I., Chytrý, M., Danihelka, J., Lustyk, P., Kocí, M., Kubesova, S., Horsak, M.,
Cherosov, M.M., Gogoleva, P.A., 2013. Plant species richnesseproductivity relationships
in a low-productive boreal region. Plant Ecol. 214, 207e219.
Barthlott, W.B., Utke, J.M., Rafiqpoor, D., Kier, G., Holger, K., Kreft, H., 2005. Global
centers of vascular plant diversity. Nova Acta Leopold. 83, 61e83.
Bengtsson, J., Fagerstr€om, T., Rydin, H., 1994. Competition and coexistence in plant
communities. Trends Ecol. Evol. 9, 246e250.
Bonanomi, G., Incerti, G., Allegrezza, M., 2013. Assessing the impact of land abandonment,
nitrogen enrichment and fairy-ring fungi on plant diversity of
Mediterranean grasslands. Biodivers. Conserv. 22, 2285e2304.
Cahill, A.E., Aiello-Lammens, M.E., Fisher-Reid, C.M., Hua, X., Karanewsky, C.J.,
S. Palpurina et al. / Acta Oecologica 63 (2015) 36e46 45
Ryu, H.Y., Sbeglia, G.C., Spagnolo, F., Waldron, J.B., Wiens, J.J., 2014. Causes of
warm-edge range limits: systematic review, proximate factors and implications
for climate change. J. Biogeogr. 41, 429e442.
Cain, S.A., 1950. Life-forms and phytoclimate. Bot. Rev. 16, 1e32.
Cheshitev, G., Kancev, I., Valkov, V., Marinova, R., Shilyafova, J., Russeva, M., Iliev, K.,
1989. Geological Map of Bulgaria, Scale 1:500 000. Committee of Geology,
Sofia.
Chytrý, M., Danihelka, J., Axmanova, I., Bozkova, J., Hettenbergerova, E., Li, C.-F.,
Rozbrojova, Z., Sekulova, L., Tichý, L., Vymazalova, M., Zelený, D., 2010. Floristic
diversity of an eastern Mediterranean dwarf shrubland: the importance of soil
pH. J. Veg. Sci. 21, 1125e1137.
Chytrý, M., Danihelka, J., Ermakov, N., Hajek, M., Hajkova, P., Kocí, M., Kubesova, S.,
Lustyk, P., Otýpkova, Z., Popov, D., Rolecek, J., Rezníckova, M., Smarda, P., 2007.
Plant species richness in continental southern Siberia: effects of pH and
climate in the context of the species pool hypothesis. Glob. Ecol. Biogeogr. 16,
668e678.
Chytrý, M., Tichý, L., Rolecek, J., 2003. Local and regional patterns of species richness
in Central European vegetation types along the pH/calcium gradient. Folia
Geobot. 38, 429e442.
Davis, B.A.S., Brewer, S., Stevenson, A.C., Guiot, J., Data Contributors, 2003. The
temperature of Europe during the Holocene reconstructed from pollen data.
Quat. Sci. Rev. 22, 1701e1716.
De'ath, G., Fabricius, K.E., 2000. Classification and regression trees: a powerful yet
simple technique for ecological data analysis. Ecology 81, 3178e3192.
Donit¸a N., Popescu A., Pauca-Comanescu M., Mihailescu S., Biris¸ I.A. (Eds.), 2005,
Habitatele din Rom^ania vol. 1. Bucures¸ ti: Silvica Tehnica, p. 496.
Ewald, J., 2003. The calcareous riddle: why are there so many calciphilous species in
the Central European flora? Folia Geobot. 38, 357e366.
Grace, J.B., 1999. The factors controlling species density in herbaceous plant communities:
an assessment. Perspect. Plant Ecol. Evol. Syst. 2, 1e28.
Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G., Jarvis, A., 2005. Very high resolution
interpolated climate surfaces for global land areas. Int. J. Climatol. 25,
1965e1978.
Hothorn, T., Zeileis, A., 2013. Partykit: a Toolkit for Recursive Partytioning, R Package
Version 0.1-5. http://CRAN.R-project.org/package¼partykit.
Huber, R., 1994. Changes in plant species richness in a calcareous grassland
following changes in environmental conditions. Folia Geobot. Phytotaxon. 29,
469e482.
Huston, M., 1979. A general hypothesis of species diversity. Am. Nat. 113, 81e101.
Kull, K., Zobel, M., 1991. High species richness in an Estonian wooded meadow.
J. Veg. Sci. 2, 715e718.
Kopralev, I.E. (Ed.), 2002. Geography of Bulgaria. Physical Geography. Socioeconomic
Geography. ForComPublishers, Sofia.
Lengyel, A., Chytrý, M., Tichý, L., 2011. Heterogeneity-constrained random resampling
of phytosociological databases. J. Veg. Sci. 22, 175e183.
L€obel, S., Dengler, J., Hobohm, C., 2006. Species richness of vascular plants, bryophytes
and lichens in dry grasslands: the effects of environment, landscape
structure and competition. Folia Geobot. 41, 377e393.
MacArthur, R.H., Wilson, E.O., 1963. An equilibrium theory of insular zoogeography.
Evolution 17, 373e387.
Magyari, E.K., Chapman, J.C., Gaydarska, B., Marinova, E., Deli, T., Huntley, J.P.,
Allen, J.R.M., Huntley, B., 2008. The ‘oriental’ component of the Balkan flora:
evidence of presence on the Thracian Plain during the Weichselian late-glacial.
J. Biogeogr. 35, 865e883.
McCune, B., Keon, D., 2002. Equations for potential annual direct incident radiation
and heat load. J. Veg. Sci. 13, 603e606.
Medail, F., Quezel, P., 1997. Hot-spots analysis for conservation of plant biodiversity
in the Mediterranean Basin. Ann. Mo. Bot. Gard. 84, 112e127.
Merunkova, K., Chytrý, M., 2012. Environmental control of species richness and
composition in upland grasslands of the southern Czech Republic. Plant Ecol.
213, 591e602.
Merunkova, K., Preislerova, Z., Chytrý, M., 2012. White Carpathian grasslands: can
local ecological factors explain their extraordinary species richness? Preslia 84,
311e325.
Mücher, C.A., Hennekens, S.M., Bunce, R.G.H., Schaminee, J.H.J., Schaepman, M.E.,
2009. Modelling the spatial distribution of Natura 2000 habitats across Europe.
Landsc. Urban Plan. 92, 148e159.
Normand, S., Treier, U., 2009. Importance of abiotic stress as a range-limit determinant
for European plants: insights from species responses to climatic gradients.
Glob. Ecol. Biogeogr. 18, 437e449.
Oksanen, J., Guillaume, B.F., Kindt, R., Legendre, P., Minchin, P.R., O'Hara, R.B.,
Simpson, G.L., Solymo, P., Stevens, M., Henry, H., Wagner, H., 2013. Vegan:
Community Ecology Package, R Package Version 2.0-7. http://CRAN.R-project.
org/package¼vegan.
Ortega, M., Levassor, C., Peco, B., 1997. Seasonal dynamics of Mediterranean pasture
seed banks along environmental gradients. J. Biogeogr. 24, 177e195.
P€artel, M., 2002. Local plant diversity patterns and evolutionary history at the
regional scale. Ecology 83, 2361e2366.
P€artel, M., Zobel, M., Zobel, K., van der Maarel, E., 1996. The species pool and its
relation to species richness: evidence from Estonian plant communities. Oikos
75, 111e117.
Pausas, J.G., Austin, M.P., 2001. Patterns of plant species richness in relation to
different environments: an appraisal. J. Veg. Sci. 12, 153e166.
Pedashenko, H., Apostolova, I., Boch, S., Ganeva, A., Sopotlieva, D., Todorova, S.,
Ünal, A., Vassilev, K., Velev, N., Dengler, J., 2013. Dry grasslands of NW Bulgarian
mountains: first insights into diversity, ecology and syntaxonomy. Tuexenia 33,
309e346.
R Development Core Team, 2012. R: a Language and Environ for Statistical
Computing. R Foundation for Statistical Computing, Vienna. ISBN 3-900051-07-
0. http://www.R-project.org.
Reitalu, T., Helm, A., P€artel, M., Bengtsson, K., Gerhold, P., Rosen, E., Takkis, K.,
Znamenskiy, S., Prentice, H.C., 2014. Determinants of fine-scale plant diversity
in dry calcareous grasslands within the Baltic Sea region. Agric. Ecosyst. Environ.
182, 59e68.
Ryser, P., Langenauer, R., Gigon, A., 1995. Species richness and vegetation structure
in a limestone grassland after 15 years management with six biomass removal
regimes. Folia Geobot. 30, 157e167.
Schuster, B., Diekmann, M., 2003. Changes in species density along the soil pH
gradient e evidence from German plant communities. Folia Geobot. 38,
367e379.
Shmida, A., Wilson, M.V., 1985. Biological determinants of species diversity.
J. Biogeogr. 12, 1e20.
Sopotlieva, D., Apostolova, I., 2014. Dry grassland vegetation in the transition zone
between two biogeographic regions. Hacquetia 13, 79e120.
Taylor, D.R., Aarssen, L.W., Loehle, C., 1990. On the relationship between r/K selection
and environmental carrying capacity: a new habitat templet for plant life
history strategies. Oikos 58, 239e250.
Tichý, L., 2002. JUICE, software for vegetation classification. J. Veg. Sci. 13, 451e453.
Therneau, T.M., Atkinson, B., 2012. rpart: Recursive Partitioning, R package version 3.1-
51. http://mayoresearch.mayo.edu/mayo/research/biostat/splusfunctions.cfm.
Tuhkanen, S., 1980. Climatic parameters and indices in plant geography. Acta Phytogeogr.
Suec. 67, 1e105.
Tonkov, S., Beug, H.-J., Filipova-Marinova, M., Bozilova, E., Jungner, H., 2011. Palaeoecological
studies at the Kaliakra area, northeastern Bulgarian Black Sea coast:
6000 years of natural and anthropogenic change. Veg. Hist. Archaeobot. 20,
29e40.
Turtureanu, P.D., Palpurina, S., Becker, T., Dolnik, C., Ruprecht, E., Sutcliffe, L.M.E.E.,
Szabo, A., Dengler, J., 2014. Scale- and taxon-dependent biodiversity patterns of
dry grassland vegetation in Transylvania (Romania). Agric. Ecosyst. Environ. 182,
15e24.
Tyler, G., 1994. A new approach to understand the calcifuge habit of plants. Ann. Bot.
73, 327e330.
Tzonev, R., Roussakova, V., Dimitrov, M.A., 2006. The western-pontic steppe vegetation
in Bulgaria. Hacquetia 5, 5e23.
Wilson, J.B., Peet, R.K., Dengler, J., P€artel, M., 2012. Plant species richness: the world
records. J. Veg. Sci. 23, 796e802.
Zobel, M., 1992. Plant species coexistence: the role of historical, evolutionary and
ecological factors. Oikos 65, 314e320.
S. Palpurina et al. / Acta Oecologica 63 (2015) 36e4646