Patterns and drivers of phytodiversity in steppe grasslands of Central Podolia (Ukraine)

Abstract

We asked: (i) Which environmental factors determine the level of α-diversity at several scales and β-diversity in steppic grasslands? (ii) How do the effects of environmental factors on α- and β-diversity vary between the different taxonomic groups (vascular plants, bryophytes, lichens)? We sampled nested-plot series ranging from 0.0001 m² to 100 m² and additional 10-m² plots, covering different vegetation types and management regimes in steppes and semi-natural dry grasslands of Central Podolia (Ukraine). We recorded all terricolous taxa and used topographic, soil, land-use and climatic variables as predictors. Richness-environment relationships at different scales and across taxonomic groups were assessed with multimodel inference. We also fitted power-law species-area relationships (SARs), using the exponent (z-value) as a measure of β-diversity. In general, the richness values in the study region were intermediate compared to those known from similar grasslands throughout the Palaearctic, but for 1 cm² we found seven species of vascular plants, a new world record. Heat index was the most important factor for vascular plants and bryophytes (negative relation), while lichen diversity depended mainly on stone and rock cover (positive). The explanatory power of climate-related variables increased with increasing grain size, while anthropogenic burning was the most important factor for richness patterns at the finest grain sizes (positive effect). The z-values showed more variation at the finest grain sizes, but no significant differences in their mean between scales. The results highlight the importance of integrating scale into ecological analyses and nature conservation assessments in order to understand and manage biological diversity in steppe ecosystems.

Full text

ORIGINAL PAPER
Patterns and drivers of phytodiversity in steppe
grasslands of Central Podolia (Ukraine)
Anna A. Kuzemko
1
•Manuel J. Steinbauer
2
•Thomas Becker
3
•
Yakiv P. Didukh
4
•Christian Dolnik
5
•Michael Jeschke
3
•
Alireza Naqinezhad
6
•Emin Ug
˘urlu
7
•Kiril Vassilev
8
•
Ju
¨rgen Dengler
9,10
Received: 13 October 2015 / Revised: 30 January 2016 / Accepted: 12 February 2016 /
Published online: 24 February 2016
ÓSpringer Science+Business Media Dordrecht 2016
Abstract We asked: (i) Which environmental factors determine the level of a-diversity at
several scales and b-diversity in steppic grasslands? (ii) How do the effects of environ-
mental factors on a- and b-diversity vary between the different taxonomic groups (vascular
plants, bryophytes, lichens)? We sampled nested-plot series ranging from 0.0001 to 100 m
2
and additional 10-m
2
plots, covering different vegetation types and management regimes in
steppes and semi-natural dry grasslands of Central Podolia (Ukraine). We recorded all
terricolous taxa and used topographic, soil, land-use and climatic variables as predictors.
Communicated by Pe
´ter To
¨ro
¨k.
Electronic supplementary material The online version of this article (doi:10.1007/s10531-016-1060-7)
contains supplementary material, which is available to authorized users.
&Anna A. Kuzemko
anya_meadow@i.ua
Ju
¨rgen Dengler
juergen.dengler@uni-bayreuth.de
1
National Dendrological Park ‘‘Sofiyvka’’, National Academy of Sciences of Ukraine, Kyivska Str.
12a, Uman 20300, Ukraine
2
Section Ecoinformatics & Biodiversity, Department of Bioscience, Aarhus University,
Ny Munkegade 116, 8000 Aarhus, Denmark
3
Geobotany, Faculty of Geography and Geosciences, University of Trier, Behringstr. 21,
54296 Trier, Germany
4
M.G. Kholodny Institute of Botany, National Academy of Sciences of Ukraine, Tereschenkivska
Str. 2, Kiev 01601, Ukraine
5
Institute for Natural Resource Conservation, Ecology Centre, University of Kiel, Olshausenstr. 40,
24098 Kiel, Germany
6
Department of Biology, Faculty of Basic Sciences, University of Mazandaran,
P.O. Box. 47416-95447, Babolsar, Mazandaran, Iran
7
Department of Biology Science and Art Faculty, Celal Bayar University, Muradiye, Yag
˘cılar
Campus, 45140 Manisa, Turkey
123
Biodivers Conserv (2016) 25:2233–2250
DOI 10.1007/s10531-016-1060-7

Richness-environment relationships at different scales and across taxonomic groups were
assessed with multimodel inference. We also fitted power-law species-area relationships,
using the exponent (zvalue) as a measure of b-diversity. In general, the richness values in
the study region were intermediate compared to those known from similar grasslands
throughout the Palaearctic, but for 1 cm
2
we found seven species of vascular plants, a new
world record. Heat index was the most important factor for vascular plants and bryophytes
(negative relation), while lichen diversity depended mainly on stone and rock cover
(positive). The explanatory power of climate-related variables increased with increasing
grain size, while anthropogenic burning was the most important factor for richness patterns
at the finest grain sizes (positive effect). The zvalues showed more variation at the finest
grain sizes, but no significant differences in their mean between scales. The results
highlight the importance of integrating scale into ecological analyses and nature conser-
vation assessments in order to understand and manage biological diversity in steppe
ecosystems.
Keywords Biodiversity Bryophyte Lichen Scale dependence Species-area
relationship Species richness
Introduction
Temperate grasslands are known for their high, and in some cases extraordinary, small-
scale diversity of vascular plants (Wilson et al. 2012; Chytry
´et al. 2015) as well as
bryophytes and lichens (Dengler 2005;Lo
¨bel et al. 2006). Extremely high richness values
(‘‘world records’’) have mostly been recorded from semi-natural stands that have been
subjected to low-intensity agricultural management for long periods (Dengler et al. 2014),
but similar species densities are found in natural Palaearctic steppes. For example, Alekhin
(1986) reports 77 species of vascular plants on 1 m
2
and 120 species on 100 m
2
for a
northern forb steppe in the Kursk region of Russia. Along with high phytodiversity, steppe
ecosystems provide refuges for a large number of rare and endangered animal and plant
species, and they can be considered one of the global biodiversity hotspots (Habel et al.
2013). Although the steppes of Eurasia are one of the most threatened biomes worldwide
(Werger and van Staalduinen 2012), we are only starting to understand the relevant
environmental and biological processes that cause this unique biodiversity.
Various factors influence plot-scale (B1000 m
2
) plant species richness in Palaearctic
grasslands, with soil reaction and land use being the most prominent ones (Dengler et al.
2014). Typically, plant species richness (total or vascular plants) in European grasslands
increases linearly with soil pH or peaks around or slightly below the neutral point (Schuster
and Diekmann 2003;Lo
¨bel et al. 2006; Becker and Bra
¨ndel 2007; Olsson et al. 2009;
Merunkova
´et al. 2014). This is largely in accordance with the theories of Pa
¨rtel (2002) and
8
Institute of Biodiversity and Ecosystem Research, Bulgarian Academy of Sciences, Acad G.
Bonchev str., bl. 23, 1113 Sofia, Bulgaria
9
Plant Ecology, Bayreuth Center of Ecology and Environmental Research (BayCEER), University
of Bayreuth, Universita
¨tsstr. 30, 95447 Bayreuth, Germany
10
Synthesis Centre (sDiv), German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-
Leipzig, Deutscher Platz 5e, 04103 Leipzig, Germany
2234 Biodivers Conserv (2016) 25:2233–2250
123

Ewald (2003), who stated that glacial refugia in Europe were dominated by base-rich
substrata, leading to a higher regional pool of species in the current vegetation adapted to
such conditions, which is also reflected at the plot scale. However, there are also a few
cases in which pH was not a determining factor of plot-scale richness of European dry
grasslands (Turtureanu et al. 2014: Transylvania) or even had a negative effect (Palpurina
et al. 2015: in one of the studied Bulgarian regions). Phytodiversity in European grasslands
is further strongly influenced by land use type and intensity. Extremely high richness
values are typically found in semi-natural sites that have been mown regularly over long
periods (Dengler et al. 2014). For example, mown steppe-like grasslands in Transylvania
(Romania) were much richer than the grazed or unused/abandoned counterparts (Tur-
tureanu et al. 2014). Concerning grazing, both Dupre
´and Diekmann (2001) in Sweden and
S
ˇkornik et al. (2010) in Slovenia found highest plot-scale richness at light to moderate
grazing intensity, which declines both towards heavy grazing and towards abandonment, in
agreement with the Intermediate Disturbance Hypothesis (Grime 1973; Connell 1978). In
some studies other factors also had a strong influence, such as heat index and therewith
drought (Turtureanu et al. 2014: negative effect), microrelief (Lo
¨bel et al. 2006: positive
effect) or mesoclimate (Palpurina et al. 2015: positive effect of mild winters).
While it has long been known that the relationships of species composition (e.g. Reed
et al. 1993) and species diversity (e.g. de Bello et al. 2007; Giladi et al. 2011) to envi-
ronmental drivers vary with the grain size of the observed system, this is only rarely taken
into account when comparing results from studies using different plot sizes. For diversity-
environment relationships, Shmida and Wilson (1985) hypothesized that with increasing
area from 0.1 m
2
up to the terrestrial surface of the world, species richness patterns are
determined successively by niche relations, habitat diversity, mass effect and ecological
equivalency. Breaking this conceptual model down to concrete variables, several recent
meta-analyses have provided valuable insights into vegetation diversity patterns. The
relative importance of soil vs. climatic variables increases towards small grain sizes
(Siefert et al. 2012), and the generally positive heterogeneity-diversity relationship
becomes negative towards very small grain sizes (Tamme et al. 2010). However, Tur-
tureanu et al. (2014) is probably the first study that directly analysed the relative impor-
tance of a broad set of different environmental variables for phytodiversity patterns across
seven orders of magnitude of spatial grain. Largely in agreement with the above-mentioned
theoretical and meta-analytical studies, they demonstrated for Transylvanian dry grass-
lands that soil variables were most important at grain sizes of 0.01 m
2
and under, litter
cover and heat index were most important at intermediate grain sizes (0.01–10 m
2
), while
land use and mesoclimate were most important at 1–100 m
2
. However, more datasets
sampled with such a multi-scale approach from grasslands in other regions as well as other
habitat types are needed to assess which of these patterns are generally valid, and which are
specific to the Transylvanian study.
While bryophytes and lichens can make up a large proportion of the overall phytodi-
versity of Palaearctic dry grasslands (Dengler 2005), their diversity-environment rela-
tionships are far less well studied than those of vascular plants and, if so, often in separate
studies (e.g. Mu
¨ller et al. 2012) that make it hard to draw comparisons with vascular plants.
Nevertheless, comparative studies in dry grasslands on the Swedish island of O
¨land (Lo
¨bel
et al. 2006) and in Transylvania (Turtureanu et al. 2014) indicate some general trends
regarding contrasting responses of vascular plant, bryophyte and lichen diversity to
environmental drivers: On O
¨land, soil pH had a unimodal effect on the richness of vascular
plants but a positive effect on that of non-vascular plants, while microtopography was
positive for vascular plants and bryophytes, but not for lichens (Lo
¨bel et al. 2006). In
Biodivers Conserv (2016) 25:2233–2250 2235
123

Transylvania, vascular plant diversity was strongly driven by type of land use and
mesoclimate, while for non-vascular plants in general humus content of the soil and,
particularly for terricolous bryophytes, the rockiness of the soil surface played a bigger role
(Turtureanu et al. 2014). These few examples already indicate that for a general under-
standing of commonalities and differences in diversity-environment relationships of dif-
ferent taxa, more data from other study systems are needed.
One of the countries characterized by a high diversity of grasslands of both natural and
anthropogenic origin is Ukraine (Bilyk 1973a,b; Solomakha 2008). The steppe and forest-
steppe zones cover about 73 % of the total area of Ukraine (Moysienko et al. 2014) and are
assumed to have grasslands as natural vegetation (Bohn et al. 2004). However, the precise
border of the area of natural steppes varies according to the reference (Didukh and She-
lyag-Sosonko 2003; Bohn et al. 2004), and thus also the assessment what is a natural
steppe and what a semi-natural grassland. Today, most Ukrainian grassland, be it natural or
semi-natural, has been transformed into arable fields. High nature value grasslands are
typically preserved only in small patches in protected areas, steep slopes of valleys and
ravines, burial mounds and ancient settlements (Korotchenko and Peregrym 2012; Moy-
sienko et al. 2014). According to rough estimates, natural steppe vegetation today occupies
no more than one percent of the territory of Ukraine, but at the same time, these steppe
fragments conserve about 30 % of the plant and animal species listed in the Red Book of
Ukraine (Burkovsky et al. 2013). This situation highlights the importance of conserving the
remaining steppe areas. A better understanding of the factors causing this high diversity
would help to implement conservation management more effectively.
Podolia is a region of particular importance for the study of biodiversity patterns in
Ukrainian steppe communities, because it is located at the border of the forest zone (where
steppe-like grasslands are all semi-natural) to forest-steppe and steppe zones (which rep-
resent the western margin of the continuous Eurasian distribution of meadow steppes and
grass steppes; Didukh and Shelyag-Sosonko 2003). In a previous paper, Kuzemko et al.
(2014) showed that the mean plant species richness at a grain size of 10 m
2
is relatively
low in the Central Podolian steppes compared to steppe-like semi-natural grasslands in the
Ukrainian Pre-Carpathians (Rolec
ˇek et al. 2014) and Transylvania, Romania (Dengler et al.
2012; Turtureanu et al. 2014) as well as to steppes of central Siberia (Polyakova et al.
unpublished). This ecologically and biogeographically unexpected finding could have
multiple causes and calls for a deeper understanding of the drivers of phytodiversity in
Central Podolian steppes, both from the regional perspective and as a contribution to a
synthetic model of phytodiversity for all Palaearctic grasslands (Dengler et al. 2014).
Moreover, it is not clear whether this relatively lower richness is related to the structure of
the studied communities and whether it holds across spatial scales.
Considering all of the above, we address the following questions in this article:
(i) Which environmental factors determine the level of a-diversity at several scales
(ranging from 0.0001 to 100 m
2
) and b-diversity (expressed as the zvalue of the power-law
SAR) in the region of Central Podolia? (ii) How do a- and b-diversity within the different
taxonomic groups (vascular plants, bryophytes and lichens) differ in their dependence on
environmental factors (soil, topographic, climatic and land-use variables)? This work
complements a previous study using the same data set for vegetation classification
(Kuzemko et al. 2014) by adding an in-depth analysis of the richness-environment rela-
tionships in the nested plot systems in the Central Podolian steppes and placing these in the
broad context of dry grasslands across the Palaearctic.
2236 Biodivers Conserv (2016) 25:2233–2250
123

Methods
Study area
The study area covers Central Podolia (48.95°–48.10°N and 27.55°–29.35°E), the central,
lower lying part of the Podolian Upland (73–251 m a.s.l.), which is located in the south-
western part of Ukraine (Fig. 1), at the north-western border of the steppic biogeographic
region of Europe (Liamine 2002a;b). The climate is temperate-subcontinental with a mean
annual temperature of about 7–9 °C and 600–650 mm annual precipitation (Lipinsky et al.
2003). The geobotanical zone of this region is a matter of debate, with Didukh and
Shelyag-Sosonko (2003) assigning the territory of Central Podolia completely to the
Forest-Steppe zone, while Bilyk (1977) and Bohn et al. (2004) consider it largely as part of
the broadleaved forest region. This means that there is no consensus as to which degree the
steppic grasslands of the region are natural. More specifically, the potential natural veg-
etation according to Bohn et al. (2004) is mainly East Polish-Ukrainian lime-pedunculate
oak-hornbeam forests with smaller patches of Podolian-Moldavian thermophilous horn-
beam-pedunculate oak forests and East pre-Carpathian-Moldavian sessile oak-hornbeam
forests. A more detailed description of the study area, including vegetation, can be found in
Kuzemko et al. (2014).
Field sampling
The field sampling was carried out in the second half of July 2010. Sample plots were
selected to represent the variability of dry grassland communities in the study region as
fully as possible (Fig. 1). We sampled 21 nested-plot series covering different regimes of
land-use and vegetation types. The nested-plot series (‘‘biodiversity plots’’) follow the
concept proposed by Dengler (2009b), with square-shaped subplots of 0.0001, 0.001, 0.01,
0.1, 1, and 10 m
2
arranged in two opposite corners of a 100-m
2
(10 m 910 m) plot. In
addition, we sampled 184 additional normal plots of 10 m
2
. We recorded all terricolous
vascular plants, bryophytes and lichens that were superficially present (i.e. shoot presence).
Twenty-three normal plots and five subplot series (i.e. corners) from biodiversity plots
were excluded in some analyses due to missing soil data.
Fig. 1 Location of the studied 19 grassland sites in Central Podolia, within Ukraine
Biodivers Conserv (2016) 25:2233–2250 2237
123

Structural and environmental parameters
For each 10-m
2
plot we estimated the cover of the shrub, herb and moss layers and that of
litter, gravel as well as stones and rocks. We determined geographic coordinates and
altitude with GPS (several different devices, all with an approximate accuracy of 5 m),
aspect with a compass and slope with an inclinometer. The latter two parameters were used
to calculate the heat index according to Olsson et al. (2009). Microrelief was measured as
maximum vertical deviation from an imaginary plane through the plot. We classified the
present land use into rough, quasi-metric categories regarding ‘‘grazing intensity’’ (mostly
by cattle; 0 =not grazed, 1 =low, 2 =medium, 3 =high) and ‘‘burning’’ (0 =no;
1=yes) based on our knowledge of the sites and/or visible signs in the plots.
Mixed soil samples were taken for determination of soil texture type (estimation of
fractions of sand, silt and clay), skeleton content, pH, electric conductivity, organic C and
total N content as well as C
org
/N
tot
ratio (see Kuzemko et al. 2014 for details). We used
climatic variables related to productivity and some potentially limiting bioclimatic factors
(Online Resource 1), extracted from the WorldClim database at an approximate 1-km
resolution (Hijmans et al. 2005;http://www.worldclim.org).
Analyses of richness-environment relationships
Richness-environment relationships were assessed for total species richness separately at
each of the different scales ranging from 0.0001 to 100 m
2
(n=42 for smaller grain sizes;
n=21 for 100 m
2
). The patterns of total, vascular plant, non-vascular plant, bryophyte
and lichen richness were compared at 10-m
2
grain size in a larger number of plots
(n=198; including the nested and the additional individual plots).
Collinearity between the continuous variables was checked using pairwise Pearson
correlations for the 198 10-m
2
plots. Two predictors were considered as collinear when
|r|[0.7, and in such cases we kept the one we considered ecologically more meaningful.
The Percentage of silt was dropped due to high correlation with Percentage of sand
(r =-0.90). Total N was removed in favour to Organic C (r=0.95) and Mean annual
temperature replaced Elevation (r=-0.83) and Temperature of the coldest quarter
(r=0.99). Annual precipitation replaced Temperature seasonality (r=-0.74) and
Precipitation of the driest quarter replaced Precipitation seasonality (r=-0.86) and
Percentage winter precipitation (r=0.88).
Generalized linear models with Poisson error distribution were applied within an
information-theoretical approach that allows estimating variable importance based on
automated model selection within subsets of a supplied ‘‘global’’ model (Burnham and
Anderson 2002 implemented in R-package MuMIn version 1.14; Barton
´2015). MuMIn
estimates variable importance by building models of all possible variable combinations.
The importance of single variables is quantified as the sum of Akaike weights over all
models including the explanatory variable. The approach may become computationally
intensive as the number of independent variables increases. Therefore, prior to each
analysis, all variables unrelated to species richness in individual models were identified
(AIC
c
of model including the variable ?2[AIC
c
of the null model glm (y=1)). For the
analysis of richness of different species groups at the 10-m
2
scale, all variables unrelated to
richness of the group under focus were dropped. In contrast, for the analysis on different
scales only those variables that were unrelated to species richness at all scales were
dropped from the global model in order to compare the importance of the variables
2238 Biodivers Conserv (2016) 25:2233–2250
123

between the grain sizes. Explanatory variables were transformed, if necessary, to improve
model residuals (see Online Resource 1). A quadratic term of a variable was additionally
included into the full model if the quadratic term improved model performance in uni-
modal models (DAIC
c
to null model[2). Quadratic terms were only included in models if
the non-quadratic term was also present; thus they can never reach higher importance in the
results. Variable coefficients were integrated over all models using function model.avg in
MuMIn.
Only models for total- and vascular plant species richness showed significant overdis-
persion according to dispersion test in R-package AER version 1.2–4 (Kleiber and Zeileis
2008). Repeating those model runs using a quasi-Poisson error distribution did not change
results qualitatively. Residuals for none of the models showed significant spatial auto-
correlation as tested using function moran.test in R-package spdep version 0.5–88 (Bivand
and Piras 2015).
In order to compare explained variation of models on different scales, McFadden’s
pseudo-R
2
(1—(log likelihood of the full model/log likelihood of the null model)) was
calculated. Note that McFadden’s pseudo-R
2
is not directly comparable to common R
2
measures as it tends to be much smaller than 1.
Analyses of species-area relationships
Species turnover (b-diversity) was assessed via species-area relationships (SARs). SARs
can usually be well approximated via power laws at any spatial scale (Connor and McCoy
1979; Dengler 2009a) and particularly at small grain sizes in continuous vegetation
(Dengler and Boch 2008). The exponent zof the power law can be used as an informative
tool to compare b-diversity between habitats, taxa and scales (Drakare et al. 2006). For
each or the 21 nested-plot series we fitted a power function in double-log space, i.e. log
10
S=log
10
c?z log
10
A, with S=species richness, A=area in m
2
and c=richness on
one unit area (here: 1 m
2
) with linear regression, resulting in an overall z-value. Second,
we calculated ‘‘local’’ z-values, i.e. those for the transition between two subsequent grain
sizes, to test whether b-diversity depends on spatial grain size, i.e. the actual relationship
deviates from a perfect power law (Turtureanu et al. 2014).
Results
Richness values of taxonomic groups
A total of 712 taxa were identified in the dataset of 10-m
2
plots (n=226): 581 vascular
plants (81.6 %), 75 bryophytes, 54 lichens, one macroscopic taxon of algae and one of
cyanobacteria (18.4 % non-vascular plants). At the analysed grain sizes, the proportion of
vascular plants gradually increased from 84.7 % at 0.0001 m
2
to 90.9 % at 100 m
2
. The
richness of bryophytes considerably exceeded that of lichens at all scales (Online Resource
2). Vascular plants in general were fairly evenly distributed among the plots, while the
distribution was more uneven in the case of bryophytes and particularly of lichens (con-
sider the ratios of standard deviations to means in Online Resource 2). Noteworthy is the
complete absence of lichens on our smallest plots of 0.0001 and 0.001 m
2
.
Biodivers Conserv (2016) 25:2233–2250 2239
123

Taxon-dependence of richness-environment relationships
Multimodel inference yielded heat index (negative) and annual mean temperature (uni-
modal) as the most important factors for total species richness of the vegetation at the
10-m
2
grain size, followed by burning (positive), cover of litter (negative), cover of stones
(unimodal) and grazing intensity (positive) (Table 1). Since vascular plants accounted for
Table 1 Predictors of 10-m
2
richness for all taxa and the individual taxonomic groups based on multimodel
inference (n=198)
All
species
Vascular
plants
Non-vascular
plants
Bryophytes Lichens
Topography
Inclination 0.7710.9110.32?
Heat index 1.0021.0021.0021.0020.49-
Microrelief 0.47?0.6210.46?0.681
Soil surface and texture
Cover stones and rocks 0.8510.9611.0010.9910.991
?Cover stones and rocks
2
0.8020.9320.34-0.26-0.44-
Cover gravel 1.0020.982
Cover litter 0.8620.952
Skeleton 0.34?0.8310.891
Clay 0.6610.6810.641
?Clay
2
0.5320.46-0.31-
Soil chemistry
pH 0.26?0.8920.43-0.742
Conductivity 0.29?0.47-0.7320.592
CaCO3 0.28?0.48?0.42?0.29-
C.org 0.33?0.34?
?C.org
2
0.10-0.09-
C/N 0.6520.8720.5120.29-0.532
Mesoclimate
Annual mean temperature 1.0011.0010.40-0.35-0.36-
?Annual mean temperature
2
1.0021.0020.10-0.09?0.10?
Precipitation of driest quarter 0.34-0.5310.5310.7210.651
?Precipitation of driest
quarter
2
0.10-0.22-0.16-0.19?0.26-
Land use
Grazing intensity 0.8410.9010.9520.9320.542
Burning 0.9610.27-0.26?
The table indicates the importance value of each variable and the direction of the relationship. The rela-
tionship is derived from the mean model coefficients across all possible models, weighted by their Akaike
weights (see Methods). Importance values C0.5 (i.e. those occurring in 50 % or more of the plausible
models) are set in bold. Gaps indicate those variables that were excluded prior to analyses as they did not
show a relation with the dependent variable in a univariate model (DAIC
c
to null model\2); this was true
across all five taxa for Sand, Temperature annual range and Annual precipitation
2240 Biodivers Conserv (2016) 25:2233–2250
123

the biggest fraction of total richness, the resulting model for this group did not deviate
much from the overall model. The main differences were that burning had no influence,
soil pH, limestone content and conductivity had a weak positive effect (vs. none for total
richness), while precipitation of the driest quarter had a unimodal relation (vs. negative for
total richness) (Table 1).
Two of the main factors in the overall and vascular model, cover of stones and rocks as
well as heat index, had similar effects also on the diversity of non-vascular plants, although
the unimodal component of rocky surface was less pronounced for these taxa and the heat
index was only included in about one half of the models for lichens. By contrast, some
variables not relevant for vascular plants (and often also not for the total richness), were
influential for the two non-vascular groups (Table 1). Most prominent were inclination and
microrelief, which had always a positive effect. Interestingly, several variables showed
opposite effects for vascular plants and non-vascular plants, namely grazing intensity
(positive for vascular, negative for non-vascular), pH and conductivity (positive for vas-
cular, negative for non-vascular) and annual mean temperature (unimodal for vascular,
negative for non-vascular). The diversity-environment relationships for bryophytes and
lichens were similar, except for surface cover of gravel that only played a role for bryo-
phytes (negative), while skeleton (positive) and clay content (unimodal) were only
included in the lichen model.
The explanatory power of the environmental variables as indicated by McFadden’s
pseudo-R
2
was highest for the richness of lichens (R
2
=0.35) followed by non-vascular
plants (R
2
=0.19), bryophytes (R
2
=0.15) and vascular plants (R
2
=0.10). The model
explaining overall richness reached an R
2
of 0.10.
Scale-dependence of richness-environment relationships
The explanatory power of the derived diversity-environment relationships in the 21
biodiversity plots was good for 100 m
2
, but continuously decreased towards 0.01 m
2
,
with very low pseudo-R
2
values for the smallest three grain sizes (Fig. 2). In general,
Fig. 2 Change of the explanatory power of the richness-environment relationships for all taxa across the
studied spatial scales, ranging from 0.0001 to 100 m
2
. Note that McFadden’s pseudo-R
2
is not comparable to
common R
2
measures as it tends to be small
Biodivers Conserv (2016) 25:2233–2250 2241
123

heat index was the most influential factor for total species richness (negative rela-
tionship) from 0.01 m
2
upwards and was included in nearly all models from 0.1 m
2
upwards (Fig. 3; Table 2). The presence of burning was influential only at the two
smallest grain sizes below 0.01 m
2
(positive). Conversely, the cover of litter had a
strong positive influence only at our coarsest grain size (100 m
2
). The various soil-
related variables (texture and chemistry) generally had only moderate influence and
typically reached their greatest influence for intermediate grain sizes (1 or 10 m
2
)
(Fig. 3). The two mesoclimatic variables of the full model, annual mean temperature
and precipitation of the driest quarter, generally increased in their influence with
increasing grain size (Fig. 3). The influence of microrelief on total diversity was low
across all scales, with changing direction (Fig. 3; Table 2). Some variables exhibited
systematic shifts in the direction of their influence on richness across the scales
(Table 2). For example, while having a big negative effect on richness at the larger
grain sizes, heat index became a (weak) positive factor at 0.001 and 0.0001 m
2
. Cover
of stones and rocks, on the other hand, affected diversity negatively at small scales, but
positively at larger scales.
Fig. 3 Change of relative importance of the 12 variables included in the full models for total species
richness across the studied spatial scales from 0.0001 to 100 m
2
2242 Biodivers Conserv (2016) 25:2233–2250
123

Species-area relationships
The overall z-values (all taxa, across all grain sizes) as measure of b-diversity ranged from
0.185 to 0.340, with a mean of 0.243. They were not particularly well explained by the
measured environmental factors, with cover of litter being the most consistent positive
predictor across the plausible models (Table 2). The local zvalues showed more variation
at the smallest grain sizes, but no significant differences in their mean between scales
(repeated measures ANOVA: p=0.145).
Table 2 Parameter estimates for the seven spatial grain sizes of a-diversity and of the overall z-value as
measure of b-diversity based on multimodel inference (n=21 for 100 m
2
and 37 for all smaller grain sizes)
0.0001 m
2
0.001 m
2
0.01 m
2
0.1 m
2
1m
2
10 m
2
100 m
2
z
Topography
Heat index 0.22?0.21?0.46-0.9020.9520.9420.9420.20-
Microrelief 0.20?0.25-0.23-0.24-0.20-0.25?0.16?0.20-
Soil surface and texture
Cover stones and rocks 0.33-0.50-0.31?0.25-0.32?0.6310.22?0.26-
?Cover stones and
rocks
2
0.10-0.14-0.11-0.06?0.06-0.17-0.06-0.03?
Cover litter 0.28?0.26?0.29?0.25-0.20-0.31-0.8110.831
Skeleton 0.20?0.22-0.21-0.19?0.33?0.24?0.16?0.15-
Clay 0.34?0.40?0.29?0.42-0.7220.32-0.12-0.17?
?Clay
2
0.07-0.08-0.09-0.09-0.29-0.05-0.02?0.03-
Soil chemistry
Conductivity 0.44?0.5410.22?0.24?0.21?0.53?0.14-0.15-
C
org
0.36-0.28?0.43?0.53?0.51?0.33?0.18?0.16?
?C
org
2
0.14-0.06-0.28-0.41-0.36-0.16-0.06-0.02?
C/N 0.21?0.30?0.22-0.25?0.47-0.31-0.17-0.17-
Mesoclimate
Annual mean
temperature
0.36?0.24-0.26?0.29?0.36?0.35?0.64-0.24-
?Annual mean
temperature
2
0.09-0.05?0.06-0.07-0.19-0.17-0.54?0.03?
Precipitation of driest
quarter
0.31-0.28?0.28?0.42?0.48?0.55?0.17-0.19?
?Precipitation of driest
quarter
2
0.08-0.08-0.08-0.10-0.14-0.35-0.03-0.02?
Land use
Burning 0.7510.48?0.22?0.26?0.26?0.34?0.15?0.15-
The table indicates the importance value of each variable and the direction of the relationship. The rela-
tionship is derived from the mean model coefficients across all possible models, weighted by their Akaike
weights (see Methods). Importance values C0.5 (i.e. those occurring in 50 % or more of the plausible
models) are set in bold. The following variables were excluded prior to analyses as they did not show a
relationship with the dependent variable in univariate models at any scale (DAIC
c
to null model \2):
Inclination, Cover gravel, Sand, pH, CaCO
3
, Temperature annual range, Annual precipitation, Grazing
intensity
Biodivers Conserv (2016) 25:2233–2250 2243
123

Discussion
Richness values of taxonomic groups
Comparing our results of the species richness of Central Podolian steppes with those of
other dry and steppic grasslands across the Palaearctic (Table 3) revealed that we have
recorded a new ‘‘world record’’ at 0.0001 m
2
grain size. While the previously known
maximum was five species of vascular plants (Wilson et al. 2012; Chytry
´et al. 2015), we
found seven species on 1 cm
2
(shoot presence) twice in Ukraine: between the villages
Chetvertynivka and Mytkivka (Trostyanets district, Vinnytsia region) and near Faihorog
village (Kryzhopil district, Vinnytsia region). In the first case it was a mesoxeric grassland
on granite assigned to the Fissidens viridulus-Festuca rupicola community (order
Brachypodietalia pinnati); in the second case a xeric grassland on limestone of the Salvia
nutans-Carex humilis community (order Festucetalia valesiacae; for details on syntaxo-
nomic placement, see Kuzemko et al. 2014). For 0.001 m
2
, our maximum value was also
quite high: with 11 species of vascular plants we found a higher diversity than any other
study of Table 3and just one species less than the world record in a limestone grassland in
Sweden (Wilson et al. 2012). For the grain sizes from 0.01 m
2
upwards, the Central
Podolian mean and maximum vascular plant richness values were clearly below those for
the semi-natural steppic grasslands of Transylvania and slightly below those of the true
steppes in South Siberia (Khakassia), but they were similar to those in Bulgarian dry
grasslands and above those of Mediterranean grasslands of Sicily (Table 3).
In contrast to the other taxonomic groups, we found no lichen species at plot sizes below
0.01 m
2
. Fruticose as well as foliose lichens occurred only sparsely in the denser grassland
vegetation, building clumped stands in suitable microhabitats, e.g. gaps and surrounding
small rocky outcrops. Due to the small number of replicates, the chances of encountering
lichens in the small subplots were therefore quite small. By contrast, at least some bryo-
phytes reached high cover values (see Kuzemko et al. 2014), leading to a higher chance of
their being present at the smallest scales. Thus, lichens contribute proportionally more to
species richness at medium to larger scales.
Taxon-dependence of richness-environment relationships
The strong inverse relationship of the species richness to the heat index of total, vascular,
bryophyte and to a lesser extent also lichen richness is consistent with the findings of
Turtureanu et al. (2014) for Transylvania. Steep south-west facing slopes under summer-
warm conditions provide a quite hostile environment that seemingly excludes many spe-
cies from the stands, mainly because of drought stress. The same factor is considered
responsible for the sharp decline in species richness of the steppe communities at the
transition from the northern forb steppe to the southern grass steppes (Alekhin 1986).
Unlike the situation in central and northern Europe (Schuster and Diekmann 2003;Lo
¨bel
et al. 2006), soil pH (and related soil factors) played only a minor positive role for vascular
plant species richness. This could be related to a rather short pH gradient and a high mean
(Online Resource 1) for the Ukrainian study sites, which is typical for continental regions.
Similarly, Turtureanu et al. (2014) in Romania did not find a pH effect on vascular plant
species richness and Palpurina et al. (2015) in one region of Bulgaria even a negative
relationship. Richness of non-vascular plants showed a clear negative relationship to soil
pH in strong contrast to the situation in O
¨land, where Lo
¨bel et al. (2006) found a strong
2244 Biodivers Conserv (2016) 25:2233–2250
123

Table 3 Comparison of mean and maximum richness values of vascular plants (shoot presence) found in Central Podolia with those found in other natural and semi-natural
dry grasslands of the Palaearctic
Country Region Reference Number of replicates Statistics 0.0001 m
2
0.001 m
2
0.01 m
2
0.1 m
2
1m
2
10 m
2
100 m
2
Ukraine Central Podolia This paper 42-226-21 Max 71113 21 42 64 86
Mean 2.5 4.0 7.3 13.8 24.4 37.2 66.8
Germany Upper Franconia (Hopp and Dengler 2015) 2-2-1 Max 4 9 19 31 43 55 65
Mean 4.0 6.5 14.0 25.0 37.0 47.5 65.0
Romania Transylvania Dengler et al. (2012) 40-82-20 Max 5 8 18 43 79 98 127
Mean 2.3 4.2 9.6 21.1 37.5 49.7 83.3
Bulgaria NW Bulgarian
mountains
Pedashenko et al. (2013) 30-98 -15 Max 6 9 14 25 36 60 87
Mean 2.3 3.9 7.6 13.3 22.8 34.1 56.7
Italy Sicily Dembicz et al. (unpublished) 42-67-21 Max 4 9 14 27 39 68 98
Mean 1.7 3.2 6.4 12.8 21.0 35.4 55.4
Russia Khakassia (Polyakova et al.
unpublished)
78-132-39 Max 5 9 17 28 52 72 94
Mean 2.1 4.0 8.2 17.3 29.7 43.9 65.3
Maxima within this comparison are shown in bold, minima in italics and current world records are underlined. The number of replicates is given in the sequence \10 m
2
–
10 m
2
–100 m
2
Biodivers Conserv (2016) 25:2233–2250 2245
123

positive relationship. It might be that the relationship in Podolia was not caused by soil
reaction itself but rather the fact that the few sites with low pH in Podolia were those close
to granite outcrops, where the herb layer was rather sparse and thus allowed a relatively
dense cryptogam layer to develop. Grazing intensity had a strong effect on the diversity of
all taxonomic groups. However, this was not as suggested by the Intermediate Disturbance
Hypothesis (Grime 1973; Connell 1978) and found, for example, in grasslands by S
ˇkornik
et al. (2010). Instead, we found a positive relationship for vascular plants and negative ones
for bryophytes and lichens. An explanation might be that even the most intensively grazed
plots in our study were still only moderately grazed when viewed in a wider context. By
contrast, cryptogams in pastures might be more affected by trampling of herbivores than by
their grazing, and this could exclude sensitive and slow-growing bryophytes and lichens at
much lower land use intensities. These results are similar to the finding of Mu
¨ller et al.
(2012) that land use negatively affects bryophyte diversity in (mostly mesic) grasslands.
Scale-dependence of richness-environment relationships
For several of the analysed factors, we found that their relevance for diversity patterns
differed considerably across spatial scales. This is generally in agreement with the
expectation that diversity-environment relationships are not constant across grain sizes
(Shmida and Wilson 1985; Siefert et al. 2012). Specifically, we could corroborate Siefert
et al.’s (2012) prediction and result from their meta-analysis that mesoclimatic variables
increase in relative importance with grain size. Our results also agree in many respects with
findings for the dry grasslands of Transylvania (Turtureanu et al. 2014), in particular with
regard to the dominant role of the heat index and litter cover at the larger grain sizes. While
in Transylvania, land use type (i.e. mown vs. either grazed or non-used) was the most
important factor at all the grain sizes from 0.1 m
2
upwards, this factor was not relevant in
Central Podolia. The reason could simply be that in Central Podolia we had no mown
grasslands, but only unused and grazed types of different intensity, which also in Tran-
sylvania hardly differed in their richness. While in Transylvania, humus content was the
most important factor at the three smallest grain sizes with a unimodal relationship, such an
influential unimodal relationship towards humus content (organic C) was found in Central
Podolia at intermediate grain sizes only (0.1–1 m
2
).
One important finding is the different explanatory power of our models for the different
scales, which increased towards larger grain sizes (Fig. 2). One could argue that this could
be caused by a mismatch between the spatial scales used for sampling of richness and for
environmental variables, because we used the same set of environmental factors for
modelling of richness across scales, most determined at the 10-m
2
scale and the mesco-
climatic variables at the 1-km
2
scale. While this mismatch could explain the decrease of
explanatory power towards grain sizes smaller than 10 m
2
, this is not consistent with the
strong further increase of explanatory power from 10 to 100 m
2
. It would also not explain
that the influence of burning increases towards the smallest grain sizes. While it would
have been advantageous to measure all environmental variables at the same grain sizes as
biodiversity, this would have come with much higher workload (for soil and topographic
parameters) or have been impossible (for climate variables). We therefore believe that our
approach to use environmental variables from the 10-m
2
scale as an approximation of the
conditions at the smaller grain sizes might have introduced some additional noise (and thus
likely lower explanatory power), but it should not have masked or changed actual patterns.
The very low explanatory power at the smallest grain sizes might also be due to the fact
that here co-occurrence of species is more ‘‘stochastic’’, i.e. less determined by
2246 Biodivers Conserv (2016) 25:2233–2250
123

environmental factors but by biological processes like dispersal, lateral spread or species–
species interactions.
Species-area relationships
The average total zvalue (in log Sspace) in Podolian dry grasslands (0.243) is at the upper
margin of what was found in a review of European dry grasslands (Dengler 2005), but not
as high as in Transylvania (0.275; Turtureanu et al. 2014). The z-values are a conventional
and easily comparable measure of b-diversity (Drakare et al. 2006), but it is still not well
understood what drives the variation in small-scale b-diversity among different grassland
types in different regions.
Like Dengler and Boch (2008) in their study of Estonian dry grasslands, we did not find
any scale dependency of the zvalues across the seven orders of magnitude studied, which
contrasts to the pronounced scale-dependency (with a peak around 0.01–0.1 m
2
) reported
from Transylvania (Turtureanu et al. 2014). Our findings support the view of Dengler
(2009a) that power laws are in most cases very good approximations for species-area
relationships at small scales in continuous vegetation, with z-values being more or less
constant over many orders of magnitude.
The reason why pronounced deviations from this general pattern are found in some
exceptional cases, such as the Transylvanian dry grasslands, needs to be further explored
with SAR data from other biogeographic and ecological contexts. The same is true for
understanding the variation in z-values between different study regions. One possible
research direction may be spatial scaling of environmental heterogeneity. The richness-
environment relationship at one spatial grain size may be related to spatial heterogeneity of
the focal environmental variable at the next smaller grain. Changes in local zvalues are
thus particularly expected if spatial heterogeneity at a particular grain size is missing or
removed (e.g. by land use). For future studies addressing scale dependent richness-envi-
ronment relationships, the accumulating pool of highly standardised data from the EDGG
Research Expeditions (Vrahnakis et al. 2013) could be a major source.
Conclusions
We have shown that the last remnants of Ukrainian steppe vegetation are particularly
species rich at small scales, with a new world record of species richness at 0.0001 m
2
grain
size. The number of vascular plants is above average in meso-xeric grassland types with
less intensive heat development, while the occurrence of stones and more pronounced
microrelief provide habitat for a relatively higher number of bryophyte and lichen species.
Analysis of scale-dependence of richness-environment relationships revealed that the cli-
mate factors increase in importance with grain size while land use/soil-related variables
decrease. Almost all identified drivers varied in importance or even direction with scale,
and there were also pronounced differences in how the three studied taxonomic groups
reacted to these drivers. This calls for caution when conclusions for ecological theory or
conservation applications are drawn from studies involving only a single spatial scale or a
single taxon. Instead, we recommend multi-scale and multi-taxon studies as an approach to
gain more comprehensive and reliable insights, not only when designing conservation
approaches for grassland vegetation. One important insight in our case was that land-use
intensity (grazing, burning), albeit on an overall low level, had contrasting effects on the
Biodivers Conserv (2016) 25:2233–2250 2247
123

taxonomic groups studied: While vascular plant species richness seemed to benefit from
burning and more intense grazing, both were negative for bryophytes and lichens. This
indicates that the current species richness of the vegetation cannot be maintained with a
single ‘‘best’’ management practice, but rather needs a mosaic of different land uses.
Comparing the identified drivers of phytodiversity in the steppe grasslands of Central
Podolia with those in other regions allowed us to determine recurring patterns and
mechanisms and to differentiate these from regional idiosyncrasies.
Acknowledgments The study was jointly planned by J.D. (methodology) and A.A.K. (site selection and
field work). A.A.K, T.B., Y.P.D., A.N. E.U., K.V. and J.D. were involved in the field sampling, while C.D.
and M.J. determined critical cryptogams, and T.P. analysed the soil samples. M.J.S. (multimodel inference)
and J.D. (species-area relationships) carried out the statistical analyses. The text was drafted by A.A.K, J.D.
and M.J.S. while all authors critically revised it. We thank Ioana Violeta Ardelean, Ute Becker, Monica
Beldean, Ina Cultasov, Thomas Haberler, Yulia Kazmirova, Igor Kuzemko, Aslan U
¨nal, Evgeniy I. Vorona
and Olena H. Yavorska for their invaluable help with the field sampling, the Fo
¨rderkreis fu
¨r Allgemeine
Naturkunde (Biologie) (www.fan-b.de) and the European Dry Grassland Group (EDGG; www.edgg.org) for
financial support of the research expedition as well as the International Association for Vegetation Science
(IAVS; http://www.iavs.org) and BAYHOST (http://www.uni-regensburg.de/bayhost), which made the
three short research stays of the first author in the groups of the last author in Hamburg and Bayreuth
possible, during which the analyses were conducted and the paper drafted. Laura Sutcliffe kindly polished
our English language usage. Finally, we thank the co-ordinating editor, Pe
´ter To
¨ro
¨k, and two anonymous
reviewers for their thoughtful comments, which contributed significantly to the improvement of the
manuscript.
References
Alekhin VV (1986) Theoretical problems of the phytocenology and steppe science. Moscow University
Publishing House, Moscow
Barton
´K (2015) MuMIn: Multi-Model Inference. R package version 1.14.0. https://r-forge.r-project.org/
scm/viewvc.php/*checkout*/www/MuMIn-manual.pdf?revision=347&root=mumin. Accessed 1 Jan
2016
Becker T, Bra
¨ndel M (2007) Vegetation-environment relationship in a heavy metal-dry grassland complex.
Folia Geobot 42:11–28
Bilyk HI (1973a) Basic patterns of distribution of the steppes vegetation of USSR. In: Barbarich AI (ed)
Vegetation of the UkrSSR. Steppes, rocky outcrops, sands. Naukova Dumka, Kyiv, pp 14–18
Bilyk HI (1973b) Mesoxerophytic grassland. In: Barbarich AI (ed) Vegetation of the UkrSSR. Steppes,
rocky outcrops, sands. Naukova Dumka, Kyiv, pp 33–94
Bilyk HI (1977) Euro-Siberian forest-steppe region. Geobotanical zonation of the USSR. Naukova Dumka,
Kyiv, pp 140–195
Bivand R, Piras G (2015) Comparing implementations of estimation methods for spatial econometrics. J Stat
Softw 63:1–36
Bohn U, Gollub G, Hettwer C, Neuha
¨uslova
´Z, Raus T, Schlu
¨ter H, Weber H, Hennekens S (eds) (2004)
Map of the natural vegetation of Europe. Scale 1 : 2 500 000. Interactive CD-ROM: explanatory text,
legend, maps. Bundesamt fu
¨r Naturschutz, Bonn
Burkovsky OP, Vasyliuk OV, Yena AV, Kuzemko AA, Movchan YI, Moysienko II, Sirenko IP (2013) Last
steppes of Ukraine: to be or not to be. Geoprynt, Kyiv
Burnham KP, Anderson DR (2002) Model selection and multimodel inference: a practical information-
theoretic approach, 2nd edn. Springer, New York
Chytry
´M, Draz
ˇil T, Ha
´jek M, Kalno
´kova
´V, Preslerova
´Z, S
ˇibik J, Ujha
´zy K, Axmanova
´I, Berna
´tova
´D
et al (2015) The most species-rich plant communities in the Czech Republic and Slovakia (with new
world records). Preslia 87:217–278
Connell JH (1978) Diversity in tropical rain forests and coral reefs. Science 199:1302–1310
Connor EF, McCoy ED (1979) The statistics and biology of the species-area relationship. Am Nat
113:791–833
de Bello F, Leps
ˇJ, Sebastia
`MT (2007) Grazing effects on the species-area relationship: variation along a
climatic gradient in NE Spain. J Veg Sci 18:25–34
2248 Biodivers Conserv (2016) 25:2233–2250
123

Dengler J (2005) Zwischen Estland und Portugal: Gemeinsamkeiten und Unterschiede der Phytodiver-
sita
¨tsmuster europa
¨ischer Trockenrasen. Tuexenia 25:387–405
Dengler J (2009a) Which function describes the species-area relationship best? a review and empirical
evaluation. J Biogeogr 36:728–744
Dengler J (2009b) A flexible multi-scale approach for standardised recording of plant species richness
patterns. Ecol Indic 9:1169–1178
Dengler J, Boch S (2008) Sampling-design effects on properties of species-area curves: a case study from
Estonian dry grassland communities. Folia Geobot 43:289–304
Dengler J, Becker T, Ruprecht E, Szabo
´A, Becker U, Beldean M, Bita-Nicolae C, Dolnik C, Goia I, Peyrat
J, Sutcliffe LME, Turtureanu PD, Ug
˘urlu E (2012) Festuco-Brometea communities of the Transyl-
vanian Plateau (Romania): a preliminary overview on syntaxonomy, ecology, and biodiversity.
Tuexenia 32:319–359
Dengler J, Janis
ˇova
´M, To
¨ro
¨k P, Wellstein C (2014) Biodiversity of Palaearctic grasslands: a synthesis.
Agric Ecosyst Environ 182:1–14
Didukh YP, Shelyag-Sosonko YR (2003) Geobotanic zoning of Ukraine and adjacent areas. Ukr Bot J
60(1):6–17 (in Ukrainian)
Drakare S, Lennon JJ, Hillebrand H (2006) The imprint of the geographical, evolutionary and ecological
context on species-area relationships. Ecol Lett 9:215–227
Dupre
´C, Diekmann M (2001) Differences in species richness and life-history traitsbetween grazed and
abandoned grasslands in southern Sweden. Ecography 24:275–286
Ewald J (2003) The calcareous riddle: why are there so many calciphilous species in the Central European
flora? Folia Geobot 38:357–366
Giladi I, Ziv Y, May F, Jeltsch F (2011) Scale-dependent determinants of plant species richness in a semi-
arid fragmented agro-ecosystem. J Veg Sci 22:983–996
Grime JP (1973) Competitive exclusion in herbaceous vegetation. Nature 242:344–347
Habel JC, Dengler J, Janis
ˇova
´M, To
¨ro
¨k P, Wellstein C, Wiezik M (2013) European grassland ecosystems:
threatened hotspots of biodiversity. Biodivers Conserv 22:2131–2138
Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A (2005) Very high resolution interpolated climate
surfaces for global land areas. Int J Climatol 25:1965–1978
Hopp D, Dengler J (2015) Scale-dependent species diversity in a semi-dry basiphilous grassland (Bromion
erecti) of Upper Franconia (Germany). Bull Eurasian Dry Grassl Group 28:10–15
Kleiber C, Zeileis A (2008) Applied econometrics with R. Springer, New York
Korotchenko I, Peregrym M (2012) Ukrainian steppes in the past at present and in the future. In: Werger
MJA, van Staalduinen MA (eds) Eurasian steppes. Ecological problems and livelihoods in a changing
world. Springer, Dordrecht, pp 173–196
Kuzemko AA, Becker T, Didukh YP, Ardelean IV, Becker U, Beldean M, Dolnik C, Jeschke M, Naqinezhad
A, Ug
˘urlu E, U
¨nal A, Vassilev K, Vorona EI, Yavorska OH, Dengler J (2014) Dry grassland vegetation
of Central Podolia (Ukraine): a preliminary overview of its syntaxonomy, ecology and biodiversity.
Tuexenia 34:391–430
Liamine N (ed) (2002a) The continental biogeographical region: agriculture, fragmentation and big rivers.
Eur Environ Agency. http://www.eea.europa.eu/publications/report_2002_0524_154909/biogeogra
phical-regions-in-europe/continental_biogeografical_region.pdf. Accessed 28 Sept 2015
Liamine N (ed) (2002b) The steppic region: the plains of Europe. Eur Environ Agency. http://www.eea.
europa.eu/publications/report_2002_0524_154909/biogeographical-regions-in-europe/continental_bio
geografical_region.pdf. Accessed 28 Sept 2015
Lipinsky VM, Diachuk VA, Babichenko VM (eds) (2003) Climate of Ukaine. Vyd-vo Rayevs’kogo, Kyiv
(in Ukrainian)
Lo
¨bel 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 41:377–393
Merunkova
´K, Preislerova
´Z, Chytry
´M (2014) Environmental drivers of species composition and richness in
dry grasslands of northern and central Bohemia, Czech Republic. Tuexenia 34:447–466
Moysienko II, Zachwatowicz M, Sudnik-Wo
´jcikowska B, Jabłon
´ska E (2014) Kurgans help to protect
endangered steppe species in the Pontic grass steppe zone, Ukraine. Wulfenia 21:83–94
Mu
¨ller J, Klaus VH, Kleinebecker T, Prati D, Ho
¨lzel N, Fischer M (2012) Impact of land-use intensity and
productivity on bryophyte diversity in agricultural grasslands. PLoS One 7:e51520. doi:10.1371/
journal.pone.0051520
Olsson PA, Ma
˚rtensson LM, Bruun HH (2009) Acidification of sandy grasslands: consequences for plant
diversity. Appl Veg Sci 12:350–361
Biodivers Conserv (2016) 25:2233–2250 2249
123

Palpurina S, Chytry
´M, Tzonev R, Danihelka J, Axmanova
´I, Merunkova
´K, Duchon
ˇM, Karakiev T (2015)
Patterns of fines-scale plant species richness in dry grasslands across the eastern Balkan Peninsula.
Acta Oecol 63:36–46
Pa
¨rtel M (2002) Local plant diversity patterns and evolutionary history at the regional scale. Ecology
83:2361–2366
Pedashenko H, Apostolova I, Boch S, Ganeva A, Janisova
´M, Sopotlieva D, Todorova S, U
¨nal A, Vassilev
K, Velev N, Dengler J (2013) Dry grasslands of NW Bulgarian mountains: first insights into diversity,
ecology and syntaxonomy. Tuexenia 33:309–346
Reed RA, Peet RK, Palmer MW, White PS (1993) Scale dependence of vegetation-environment correla-
tions: a case study of a North Carolina piedmont woodland. J Veg Sci 4:329–340
Rolec
ˇek J, C
ˇornej II, Tokarjuk AI (2014) Understanding the extreme species richness of semi-dry grasslands
in east-central Europe: a comparative approach. Preslia 86:13–34
Schuster B, Diekmann M (2003) Changes in species density along the soil pH gradient: evidence from
German plant communities. Folia Geobot 38:367–379
Shmida A, Wilson MV (1985) Biological determinants of species diversity. J Biogeogr 12:1–20
Siefert A, Ravenscroft C, Althoff D, Alvarez-Ye
´piz JC, Carter BE, Glennon KL, Heberling JM, Jo IS,
Pontes A, Sauer A, Willis A, Fridley JD (2012) Scale dependence of vegetation-environment rela-
tionships: a meta-analysis of multivariate data. J Veg Sci 23:942–951
S
ˇkornik S, Vidrih M, Kaligaric
ˇM (2010) The effect of grazing pressure on species richness, composition nd
productivity in North Adriatic Karst pastures. Plant Biosyst 144:355–364
Solomakha VA (2008) The syntaxonomy of vegetation of the Ukraine. The third approximation, Phy-
tosociocentre, Kyiv (in Ukrainian)
Tamme R, Hiiesalu I, Laanisto L, Szava-Kovats R, Pa
¨rtel M (2010) Environmental heterogeneity, species
diversity and co-existence at different spatial scales. J Veg Sci 21:796–801
Turtureanu PD, Palpurina S, Becker T, Dolnik C, Ruprecht E, Sutcliffe LME, Szabo
´A, Dengler J (2014)
Scale- and taxon-dependent biodiversity patterns of dry grassland vegetation in Transylvania. Agric
Ecosyst Environ 182:15–24
Vrahnakis MS, Janis
ˇova
´M, Ru
¯sin¸ a S, To
¨ro
¨k P, Venn S, Dengler J (2013) The European Dry Grassland
Group (EDGG): stewarding Europe’s most diverse habitat type. In: Baumbach H, Pfu
¨tzenreuter S (eds)
Steppenlebensra
¨ume Europas: Gefa
¨hrdung, Erhaltungsmaßnahmen und Schutz. Thu
¨ringer Ministerium
fu
¨r Landwirtschaft, Forsten, Umwelt und Naturschutz, Erfurt, pp 417–434
Werger MJA, van Staalduinen MA (eds) (2012) Eurasian steppes. Ecological problems and livelihoods in a
changing world. Springer, Dordrecht
Wilson JB, Peet RK, Dengler J, Pa
¨rtel M (2012) Plant species richness: the world records. J Veg Sci
23:796–802
2250 Biodivers Conserv (2016) 25:2233–2250
123