DOI: 10.1111/j.1466-8238.2007.00320.x  2007 The Authors
668 Journal compilation  2007 Blackwell Publishing Ltd www.blackwellpublishing.com/geb
Global Ecology and Biogeography, (Global Ecol. Biogeogr.) (2007) 16, 668­678
RESEARCH
PAPER
Blackwell Publishing Ltd
Plant species richness in continental
southern Siberia: effects of pH and
climate in the context of the species
pool hypothesis
Milan Chytry1
*, Jirí Danihelka1,2
, Nikolai Ermakov3
, Michal Hájek1,2
,
Petra Hájková1,2
, Martin Koçí1
, Svatava Kubeßová1,4
, Pavel Lustyk1
,
Zdenka Otypková1
, Denis Popov5
, Jan Roleçek1
, Marcela Rezníçková1
,
Petr Ímarda1
& Milan Valachoviç6
ABSTRACT
Aim Many high-latitude floras contain more calcicole than calcifuge vascular plant
species. The species pool hypothesis explains this pattern through an historical
abundance of high-pH soils in the Pleistocene and an associated opportunity for
the evolutionary accumulation of calcicoles. To obtain insights into the history of
calcicole/calcifuge patterns, we studied species richness­pH­climate relationships
across a climatic gradient, which included cool and dry landscapes resembling the
Pleistocene environments of northern Eurasia.
Location Western Sayan Mountains, southern Siberia.
Methods Vegetation and environmental variables were sampled at steppe, forest
and tundra sites varying in climate and soil pH, which ranged from 3.7 to 8.6. Species
richness was related to pH and other variables using linear models and regression trees.
Results Species richness is higher in areas with warmer winters and at medium
altitudes that are warmer than the mountains and wetter than the lowlands. In
treeless vegetation, the species richness­pH relationship is unimodal. In tundra
vegetation, which occurs on low-pH soils, richness increases with pH, but it
decreases in steppes, which have high-pH soils. In forests, where soils are more acidic
than in the open landscape, the species richness­pH relationship is monotonic
positive. Most species occur on soils with a pH of 6­7.
Main conclusions Soil pH in continental southern Siberia is strongly negatively
correlated with precipitation, and species richness is determined by the opposite
effects of these two variables. Species richness increases with pH until the soil is very
dry. In dry soils, pH is high but species richness decreases due to drought stress.
Thus, the species richness­pH relationship is unimodal in treeless vegetation.
Trees do not grow on the driest soils, which results in a positive species richness­pH
relationship in forests. If modern species richness resulted mainly from the species
pool effects, it would suggest that historically common habitats had moderate
precipitation and slightly acidic to neutral soils.
Keywords
Calcicole/calcifuge, forest-steppe, plant community, Pleistocene environments,
precipitation, soil acidity, tundra, vascular plants.
*Correspondence: Milan Chytry,
Department of Botany and Zoology,
Masaryk University, Kotlárská 2,
CZ-611 37 Brno, Czech Republic.
E-mail: chytry@sci.muni.cz
1
Department of Botany and Zoology, Masaryk
University, Kotláská 2, CZ-611 37 Brno, Czech
Republic, 2
Institute of Botany, Academy of
Sciences of the Czech Republic, Poíçí 3a,
CZ-603 00 Brno, Czech Republic, 3
Central
Siberian Botanical Garden, Russian Academy of
Sciences, Zolotodolinskaya 101, Novosibirsk,
630090, Russia; 4
Department of Botany,
Moravian Museum, Hviezdoslavova 29a,
CZ-627 00 Brno, Czech Republic, 5
Institute of
Cytology and Genetics, Russian Academy of
Sciences, Lavrent'eva 10, Novosibirsk, 630090,
Russia, 6
Institute of Botany, Slovak Academy of
Sciences, Dúbravská cesta 14, SK-845 23
Bratislava, Slovakia
INTRODUCTION
The distinction between calcicole and calcifuge plant species has
been recognized since the early days of plant ecology (Gigon,
1987). Many floras in the temperate and boreal zone have a
higher proportion of calcicole species, i.e. those associated with
high pH or calcium-rich soils, than of calcifuge species, i.e. those
confined to low-pH soils (Grubb, 1987; Tyler, 1996; Ewald, 2003).
Species richness, pH and climate in southern Siberia
 2007 The Authors
Global Ecology and Biogeography, 16, 668­678, Journal compilation  2007 Blackwell Publishing Ltd 669
Pärtel (2002) demonstrated that at medium and higher latitudes,
species richness of vascular plant communities on a local scale
(further referred to as richness) frequently increases with soil pH.
He related this pattern to the species pool hypothesis (Taylor et al.,
1990), suggesting that the positive richness­pH relationships
result from the existence of larger pools of calcicole than calcifuge
species in modern floras. He hypothesized that this prevalence of
calcicoles is due to the location of most evolutionary centres for
the temperate and boreal zone in the areas with high-pH soils.
Ewald (2003) extended Pärtel's hypothesis to consider not only
species accumulation through speciation of calcicoles but also
hypothetically high extinction of calcifuges in the Pleistocene,
when the climate was cold and dry and high-pH soils were
maintained over huge areas by decreased leaching of cations,aeolian
sedimentation of fine calcium-rich particles and cryoturbation
(Guthrie, 2001; Walker et al., 2001). The modern disparity
between the predominance of acidic soils and a considerably
larger species pool of calcicole than calcifuge species in temperate
Europe may be the result of the Pleistocene bottleneck effects on
calcicole/calcifuge species pools (Ewald, 2003).
Even at medium and high latitudes,however,the local richnesspH
relationships reported in individual studies vary considerably
from negative to positive, and they are often unimodal or nonsignificant
(Schuster & Diekmann, 2003). Interestingly, positive
monotonic relationships are often found in forests, while
grasslands of the same areas exhibit unimodal,non-significant or
even negative relationships (Chytry et al., 2003; Palmer et al., 2003;
Schuster & Diekmann,2003).Besides possible confounding effects
of other environmental variables, this may partly be caused by
the different width of the observation window along the pH
gradient in individual studies. Plants cannot tolerate either too
low pH (due to aluminium toxicity and poor availability of
some nutrients) or too high pH (due to poor solubility of
some essential elements; Tyler, 2003). Therefore the richness­pH
relationship must be inherently unimodal, increasing at low pH
and decreasing at high pH. A positive linear relationship is therefore
more frequently found in studies performed on acidic to
neutral soils, and a negative one is more frequently found in
those performed on neutral to base-rich soils. If the entire range
of pH values compatible with plant life is considered, the appropriate
question is therefore not whether the richness­pH relationship
is positive or negative, but whether it is left-skewed,
symmetrical or right-skewed.
For a better understanding of the richness­pH relationship, it
is of key importance to study it across the full range of pH values
tolerated by plants and in areas with different evolutionary
and biogeographical histories. So far, such studies have been rare,
being based either on the meta-analyses of local case studies
(Pärtel, 2002; Schuster & Diekmann, 2003) or on the richness
values extracted from the large phytosociological databases
combined with indicator values used as surrogates for measured
pH (Chytry et al., 2003; Wohlgemuth & Gigon, 2003).
With respect to the species pool hypothesis (Pärtel, 2002;
Ewald, 2003), it is of great theoretical interest to study the
species­pH relationship in dry continental areas with long and
cold winters, which are close modern analogies of Pleistocene
environments, and on geographical transects from such areas to
regions with a more oceanic climate. The modern vegetation
of temperate Europe has its historical roots in the Pleistocene
tundra, steppe and forest-steppe (Berglund et al., 1996), which
has its closest modern analogy in continental southern Siberia
and northern Mongolia (Hilbig, 1995). This region's climate is
locally similar to the Pleistocene climate of central Europe (Frenzel
et al., 1992) and its evolutionary history is also similar, as
evidenced by the geographical ranges of many plant species that
extend from central Europe to southern Siberia (Meusel et al.,
1965­1992). In addition, Last Glacial Maximum biomes reconstructed
from fossil pollen data for southern Siberia and adjacent
areas of Mongolia include steppe and forests, i.e. the same
biomes as today (Tarasov et al., 2000). Similar biomes were
reconstructed from the full-glacial pollen data from eastern
central Europe (Willis et al., 2000; Jankovská et al., 2002).
In the present study, we address the species richness­pH relationship
by analysing terrestrial vegetation along a broad-scale
geographical transect in continental southern Siberia, which
covers a range of different climates and nearly the entire range of
pH values compatible with plant life. Our questions are: (1) what
is the richness­pH relationship in this landscape, which in some
parts preserves conservative features of Pleistocene environments;
(2) does the richness­pH relationship interact with other variables,
in particular climatic ones; and (3) does this relationship differ
between treeless and forest vegetation, and if so, why?
MATERIALS AND METHODS
Study area
The study area is located in southern Siberia (Russia) between
the cities of Abakan and Minusinsk in the north and the RussianMongolian
border in the south (50°43-53°33 N;91°06-93°28 E).
Its northern part includes the Minusinskaya Basin with flat or
gently undulating landforms on Quaternary deposits at altitudes
of 300­600 m. The largest, central part of the study area includes
the Western Sayan Mountains, namely the area adjacent to the
Yenisei river valley. At altitudes of 350­1700 m these mountains
have a rugged topography with steep slopes on metamorphic
rocks, mainly base-rich chloride slates, with igneous rocks and
limestones occupying small areas.Ancient table-lands predominate
at high altitudes (1700­2860 m). Similar landforms are typical of
the Tannu-Ola range (2100­2930 m) in the southern part of the
study area. The area between the Western Sayan and Tannu-Ola
is occupied by the Central Tuvinian Basin with flat landforms at
altitudes of 550­1100 m.
The climate is strongly continental. However, the northern
front ranges of the Western Sayan, which intercept humid northwestern
air masses, have a more humid climate in comparison
with the rest of Siberia. At low to medium altitudes in the north,
January temperature ranges from -11 to -22 °C, July temperature
from +16 to +19 °C and annual precipitation from 500 to
900 mm. Winter snow cover is up to 1.5 m, which protects soil
and ground vegetation from frost. The Central Tuvinian Basin
and the Tannu-Ola range are in the rain shadow area and have an
M. Chytr et al.
 2007 The Authors
670 Global Ecology and Biogeography, 16, 668­678, Journal compilation  2007 Blackwell Publishing Ltd
arid continental climate with annual precipitation of < 400 mm,
of which 88­95% falls from late July to September. January
temperature ranges from -27 to -34 °C, July temperature from
+16 to +18 °C and winter snow cover is shallow and irregular in
this region (Gidrometeoizdat, 1966­1970).
Vegetation formations include steppe on the bottoms of the
Minusinskaya and Central Tuvinian Basins, forest-steppe and
forest at medium altitudes, and alpine tundra above the timberline.
Forests are abundant in the wetter northern part of the
Western Sayan. In contrast, in the cool and dry areas adjacent to
the Central Tuvinian Basin, steppe extends to higher altitudes
and tundra to lower altitudes,while forest is restricted.In the foreststeppe
landscapes, steppe regularly occurs on south-facing
slopes, and forest occurs on north-facing slopes. Cold and dry
lowland steppes are characterized by Artemisia frigida Willd.,
Caragana pygmaea (L.) DC. and Stipa krylovii Roshev., and
mountain steppes in the forest-steppe zone are often dominated
by Artemisia gmelinii Weber, Carex pediformis C. A. Mey. and
Spiraea media Schmidt. Some of the mountain steppes, occurring
on mesic soils, form dense tall-grass species-rich stands, called
meadow steppes. Alpine tundra is mostly dominated by dwarf
shrubs (Betula rotundifolia Spach, Vaccinium myrtillus L. and
V. vitis-idaea L.) on ridges and hillsides, and tall forbs on the
valley bottoms. The forests of the study area can be divided into
taiga and hemiboreal forests (Ermakov et al., 2000). Taiga occurs
on moist to mesic soils and is mostly dominated by Abies sibirica
Ledeb., Picea obovata Ledeb. and Pinus sibirica Du Tour. It
includes Vaccinium myrtillus, V. vitis-idaea and Bergenia crassifolia
(L.) Fritsch in the herb layer and a well-developed moss layer.
Hemiboreal forests, also called sub-taiga, occupy mesic to dry
soils. They contain many shade-intolerant herbs, some of which
occur both in steppe and open forests. Hemiboreal forests in the
driest and coolest areas are dominated by Larix sibirica Ledeb
and are termed ultracontinental forests, while those in the wetter
and warmer areas are dominated by Pinus sylvestris L. or Betula
pendula Roth, the latter often occurring in post-fire successional
stands.
Data sampling
Sampling sites were located on an approximately 300-km long
north­south transect that followed the gradient of increasing
climate continentality and ran roughly perpendicular to theWestern
Sayan range. The transect started in the lowlands, dominated by
steppe and forest-steppe vegetation in the Minusinskaya Basin,
crossed the Western Sayan mountain range, continued to the
Central Tuvinian Basin and ended at the Tannu-Ola range.Along
this transect, we sampled vegetation in the zones of steppe, foreststeppe,
forest and alpine tundra. Sampling units were vegetation
plots of 10 × 10 m, located in central parts of physiognomically
and ecologically homogeneous vegetation stands. Sites affected
by recent disturbances, e.g. early stages of post-fire succession or
human-made habitats, were avoided. Landscape sectors with
different macroclimates, sharply separated by mountain ridges,
were used as the basic sampling strata. Within each stratum, we
placed one plot on each of the following landforms: north-facing
slope, south-facing slope, ridge summit and valley bottom. If the
same landform in a given stratum contained vegetation types
of contrasting physiognomy or with different dominant species
(including different dominants of forest undergrowth), we
placed one plot in each of these vegetation types. In total, 180
plots of treeless vegetation (39 in tundra and 141 in steppe) and
127 plots of forest vegetation (77 in taiga and 50 in hemiboreal
forest) were sampled.
In each plot, we recorded all species of vascular plants and
measured the following variables (Table 1). (1) Altitude was
measured with GPS and checked in a 1:100,000 topographic
map. (2) Climatic variables (July and January mean temperature
and mean sums of summer and winter precipitation) were
estimated from a climatic model prepared in the ArcGIS 8.2
geographical information system (http://www.esri.com/). The
model was based on a combination of Russian climate station
data and altitudes from a digitized 1:200,000 topographic map.
Temperature values for different altitudes were computed based
on the adiabatic lapse rate of 0.65 °C per 100 m of altitude.
Precipitation was computed from precipitation­altitude charts
compiled for each of the aridity­humidity sectors of the AltaiSayan
region (Polikarpov et al., 1986). (3) Potential solar radiation
above the canopy was calculated from slope aspect, inclination
and latitude, and accounted for shading by the surrounding
topographic features. Topographic shading was measured from
digital hemispheric photographs taken in a vertical direction
at the height of 1.5 m above ground at each site. Calculation
was performed using Gap Light Analyser 2.0 software (http://
www.rem.sfu.ca/forestry/downloads/gap_light_analyzer.htm).
Radiation was estimated as the daily sum of direct and diffuse
radiation for 21 June. (4) Potential solar radiation below the
canopy was calculated for the forest plots only. It was obtained in
the same way as radiation above the canopy, but it additionally
accounted for shading by canopy. (5) Slope inclination was
measured with a clinometer. (6) Soil depth was measured in three
places in each plot and averaged. Soils deeper than 30 cm were
arbitrarily given a value of 30 cm. (7) Soil pH and conductivity of
soil solution were measured from a mixed soil sample from three
places in each plot. Samples were taken from the mineral topsoil
horizon at a depth of 5­10 cm or less if the soil was very shallow.
Each sample was extracted in distilled water for 24 h (weight
ratio of soil/water was 0.4) and measured using the digital
PH114 pH-meter and CM113 conductivity meter (Snail Instruments,
Beroun, Czech Republic). Conductivity caused by H+
ions was subtracted according to Sjörs (1952).
Data analysis
All the statistical analyses were computed in the STATISTICA 7.1
program (http://www.statsoft.com). To simplify the complex
pattern of correlations between environmental variables, the
matrix of variables × plots was subjected to principal components
analysis (PCA), computed on a correlation matrix in
order to remove the effect of different measurement units.
Additionally, least-square linear regressions were calculated for
some univariate relationships that involved pH.
Species richness, pH and climate in southern Siberia
 2007 The Authors
Global Ecology and Biogeography, 16, 668­678, Journal compilation  2007 Blackwell Publishing Ltd 671
In ecological data sets sampled across heterogeneous landscapes,
representation of plots with different combinations of
environmental variables is typically more or less unequal. This
occurs in spite of the use of a stratified sampling scheme such as
that applied in the current study, because some combinations
may be rare or non-existent in some of the studied landscapes.
In traditional statistical analyses, e.g. linear models, such an
unbalanced structure of the data set can be problematic if several
explanatory variables are involved. Therefore, we analysed the
relationships between the response variable (species richness)
and several environmental explanatory variables with regression
trees (CART; Breiman et al., 1984). This method is suitable for
exploratory analysis of complex unbalanced data sets that involve
higher-order interactions and nonlinear relationships between
predictor variables. It is able to reveal ecologically meaningful
relationships that would remain hidden in such data sets if standard
techniques such as linear models were used. The regression tree
is constructed by a hierarchical splitting of the data set into
smaller groups, which minimizes within-group variation in the
response variable. At each split, sampling units are divided into
two groups based on a single explanatory variable.
We applied regression tree analysis separately to the data sets
of treeless and forest vegetation and used the number of vascular
plants as the response variable. To select the optimal tree size
(optimal number of branches, also called nodes or splits), we
used the 10-fold cross-validation method. This method calculates
regression trees on smaller subsamples of the basic data set and
provides the value of cross-validation cost for the tree of each
size, i.e. the error in predicted values relative to the corresponding
tree calculated with the entire data set. The tree with the lowest
cross-validation error is usually accepted as the optimal tree
because it maximizes the variation in the response variable
and simultaneously minimizes the risk of over-learning, i.e. the
danger of erroneous prediction of previously unobserved cases.
For selection of the optimal tree based on the cross-validation
results, we followed SE Rule = 0 (Breiman et al., 1984) because
SE Rule = 1 led to very simple trees. For each node of the tree,
we identified not only the primary splitter variable but also
surrogates, i.e. the variables that are able to allocate cases similarly
as the primary splitter. To consider a variable as a surrogate, we
required that the variable had associated value higher than 0.6.
To assess the relative importance of each predictor variable in the
optimal tree, we calculated their Importance values,which are relative
to the best predictor variable, scored as 100. The importance
values reflect the contribution that each explanatory variable
makes in explaining the variation in the response variable, with
Table 1 Descriptive statistics for variables used in the analysis.
Mean
Standard
deviation Minimum
Lower
quartile
Upper
quartile Maximum
Treeless vegetation (n = 180)
Altitude (m a.s.l.) 1004 480 318 592 1461 2160
July temperature (°C) 14.9 3.1 7.0 14.0 17.0 19.0
January temperature (°C) -28.0 5.5 -36.0 -32.0 -26.5 -17.0
Summer precipitation (mm) 475 347 170 231 559 1341
Winter precipitation (mm) 87 83 28 33 117 369
Radiation (mol m-2
day-1
) 38.0 4.0 26.0 37.0 40.0 42.0
Inclination (°) 19 13 0 9 29 52
Soil depth (cm) 16 10 1 6 30 30a
Soil pH 6.7 1.1 3.7 6.3 7.5 8.6
Soil conductivity (S cm-1
) 120 251 2 49 152 353b
Number of vascular plants 31.6 12.4 7.0 22.5 41.0 64.0
Forests (n = 127)
Altitude (m a.s.l.) 968 414 346 614 1241 1948
July temperature (°C) 13.9 2.2 9.0 12.0 16.0 18.0
January temperature (°C) -25.0 5.5 -37.0 -29.0 -20.0 -17.0
Summer precipitation (mm) 657 267 171 425 821 1283
Winter precipitation (mm) 144 78 28 76 187 371
Radiation above canopy (mol m-2
d-1
) 36.1 3.5 24.7 34.7 39.0 41.0
Radiation below canopy (mol m-2
d-1
) 13.3 4.3 6.0 10.0 15.7 33.0
Inclination (°) 21 12 0 10 29 46
Soil depth (cm) 20 10 1 10 30 30a
Soil pH 5.7 0.9 3.7 5.0 6.3 7.8
Soil conductivity (S cm-1
) 56 45 12 29 65 290
Number of vascular plants 29.3 12.5 6.0 19.0 38.0 69.0
a
Soils deeper than 30 cm were arbitrarily given a value of 30 cm.
b
In one plot an outlying conductivity value of 3350 S cm-1
was measured.
M. Chytr et al.
 2007 The Authors
672 Global Ecology and Biogeography, 16, 668­678, Journal compilation  2007 Blackwell Publishing Ltd
its contribution 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.
RESULTS
PCA revealed groups of related environmental variables (Table 2).
The main complex environmental gradient associated with PCA
axis 1 is that of decreasing precipitation and increasing soil
pH. The second most important gradient (axis 2) is related to
decreasing altitude and increasing winter temperature, the third
gradient (axis 3) to slope steepness and associated variables such
as soil depth and site insolation, and the fourth gradient (axis 4)
to canopy closure and light availability above the herb layer. An
important relationship revealed by PCA axis 1 is the strong negative
correlation of pH with precipitation; this trend is consistent
in both treeless vegetation and forests, although the mean pH is
lower in the latter (Fig. 1). Another remarkable relationship
revealed by PCA is that precipitation is not correlated with winter
temperature. While winter temperature generally decreases with
increasing altitude, precipitation in both winter and summer is
to a large extent determined by the position on the windward or
leeward side of the main mountain ranges, although at local scale
it also tends to increase with altitude.
Univariate regressions of species richness on pH revealed a
positive relationship for both treeless and forest vegetation
(Table 3). In the treeless vegetation, pH explained only 2.5% of
variation in species richness in linear regression, but after adding
the quadratic term the variation explained increased to 17.0%,
indicating unimodal richness­pH relationship with a peak at pH
6.4. In tundra vegetation, which occurred in the pH range of
3.8­6.1 (-7.4), there was a positive linear relationship between
richness and pH that explained 13.0% of variation (Fig. 2a).
However, in steppe vegetation, which occurred in the pH range
of 5.9­8.1 (-8.6), the linear relationship was negative. In forests,
variation explained by the linear regression was considerably
higher than in treeless vegetation (25.8%), but it was improved
only to 28.1% after adding the quadratic term (although the
quadratic model improved the model significantly; P < 0.05).
The curve of the quadratic relationship was upward convex, with
an indistinct peak occurring at pH 7.2 which did not provide
reliable support for the existence of the unimodal richness­pH
Table 2 Coordinates of environmental variables on axes 1­4 of the
PCA calculated from a matrix of vegetation plots × environmental
variables. As PCA was based on the correlation matrix, the
coordinates are identical with the correlations between PCA axes
and environmental variables. Coordinates  |0.60| are set in bold.
PCA axis 1 2 3 4
Percentage variation explained 34.5 18.8 15.4 9.9
Altitude -0.49 -0.80 0.11 -0.01
July temperature 0.78 0.52 -0.13 0.06
January temperature -0.13 0.87 -0.39 -0.14
Summer precipitation -0.90 0.18 -0.10 -0.27
Winter precipitation -0.88 0.24 -0.18 -0.25
Radiation above canopy 0.30 -0.27 -0.63 -0.49
Radiation below canopy 0.59 -0.28 -0.01 -0.60
Slope inclination -0.09 0.38 0.77 -0.06
Soil depth 0.13 -0.13 -0.62 0.52
Soil pH 0.88 -0.02 0.15 -0.04
Soil conductivity 0.31 0.06 0.22 -0.23
Figure 1 The relationship between pH
and summer precipitation in the study area.
Precipitation values were predicted by the
climatic model and soil pH(H20) was
measured in vegetation plots. Treeless
vegetation: = 0.633; forests: = 0.325.
The slopes of the two regression lines are
significant (P < 0.001), but do not differ
significantly from each other at P < 0.05.
The difference in intercept between them is
significant (P < 0.001). Values of pH also
decrease significantly with winter precipitation
at P < 0.001 (not shown; treeless vegetation:
= 0.594, forests: = 0.332).
Radj
2
Radj
2
Radj
2
Radj
2
Species richness, pH and climate in southern Siberia
 2007 The Authors
Global Ecology and Biogeography, 16, 668­678, Journal compilation  2007 Blackwell Publishing Ltd 673
relationship in forests. After splitting the forest data set into
taiga and hemiboreal forests (Fig. 2b), a positive relationship
only remained in the more acidic taiga (pH 3.7­6.9) while no
relationship was detected in the base-rich hemiboreal forests
(pH 5.2­7.8).
Regression trees were constructed separately for treeless and
forest vegetation. In both of them, January temperature, altitude
and pH were the most important predictors of species richness
(Table 4). For treeless vegetation, the optimal regression tree with
the lowest cross-validation cost had seven terminal nodes (Fig. 3).
This tree explained 56.1% of variation in species richness. In the
first node the tree split based on mean January temperature,
predicting lower species numbers in areas with January means
below -21 °C and at altitudes above 532 m. In low-altitude areas
with warmer winters, species richness increased with altitude, i.e.
with precipitation. The highest richness was found at altitudes
slightly above 500 m and summer precipitation above 479 mm:
Table 3 Standardized regression coefficients (beta), variation
explained ( ) and significances (P) of the linear and quadratic
regression models of the relationship between species richness and
pH for different subsets of vegetation plots. Quadratic regressions
were not considered for tundra, steppe, taiga and hemiboreal forests.
beta-pH beta-(pH)2
P
Treeless vegetation:
both types, linear regression 0.17 -- 0.025 0.025
both types, quadratic regression 3.94 -3.80 0.170 < 0.001
tundra 0.36 -- 0.130 0.022
steppe -0.31 -- 0.091 < 0.001
Forests:
both types, linear regression 0.52 -- 0.258 < 0.001
both types, quadratic regression 2.36 -1.85 0.281 < 0.001
taiga 0.44 -- 0.170 < 0.001
hemiboreal forests 0.09 -- 0.0 0.525
Radj
2
Radj
2
Figure 3 Regression tree for species richness
of treeless vegetation. Numbers at each node
indicate the mean number of vascular plant
species per 100 m2
 standard deviation and
number of plots assigned to that node.
The primary splitter variable and its split value
at each node is given in bold. Surrogates,
i.e. variables that allocate most cases to the
identical group as the primary splitter,
are given in smaller letters below the primary
splitter.
Figure 2 Relationships between number of vascular plant species
per 100 m2
and pH. See Table 3 for regression coefficients. The slope
of the regression line was not significant for hemiboreal forests,
therefore the line was not fitted.
M. Chytr et al.
 2007 The Authors
674 Global Ecology and Biogeography, 16, 668­678, Journal compilation  2007 Blackwell Publishing Ltd
grassland vegetation occurring there included mostly tall-grass
meadow steppes, containing on average over 50 species per 100
m2
. In the branch of the tree with low January temperatures,
the most important predictor of species richness was pH. It
discriminated between very poor tundra vegetation on soils with
pH < 5.0 and richer vegetation of both tundra and steppe on
more base-rich soils. In the latter, the poorest vegetation was the
cold short-grass steppe on the flat bottoms of large basins, while
mountain steppe and tundra vegetation was richer. The latter
types of mountain grasslands were poorer at medium altitudes
(594­832 m) where forest forms the landscape matrix and treeless
vegetation mostly occurs in smaller patches surrounded by forest.
The optimal regression tree for forest vegetation had only
three terminal nodes (Fig. 4). First it separated species-poor dark
taiga forests on acidic soils with pH < 5.0 from richer forests on
more base-rich soils.Low pH was weakly related to higher summer
precipitation. The group of species-richer forests on high-pH
soils was further divided into two subgroups. The poorer subgroup
at altitudes above 882 m, with low winter temperatures
and low summer precipitation, contained ultracontinental
forests of L. sibirica and P. sibirica. The richer subgroup at lower
altitudes with warmer winters and higher summer rainfall
included hemiboreal forests and wet Abies sibirica taiga; this
subgroup included the richest forests of the Western Sayan,
containing on average 37 species per 100 m2
. This regression tree
explained 41.2% of variation in species richness. For comparison,
we fitted a multiple least-square regression model with
the two predictors used as primary splitters in this tree: pH and
altitude. This model explained only 38.0% of variation.
Higher mean local species richness in steppe and hemiboreal
forests than in tundra and taiga, revealed by the regression trees,
corresponded to the relative size of species pools for these formations.The
relative size of the species pools was estimated as the total
number of all species occurring in 39 plots selected randomly
from each of these formations. These relative estimations yielded
305 species for steppe, 223 for tundra, 303 for hemiboreal forests
and 212 for taiga (means of 10 random selections).
DISCUSSION
The relationship between pH, precipitation and
species richness
The study area contains bedrocks of varying base status and soils
of different type and depth. Yet, soil pH was shown to depend
predominantly on precipitation: summer precipitation explained
63.3% of variation in soil pH in sites with treeless vegetation (Fig.1).
Such an amount of variation explicable by a single predictor
variable is surprisingly large, given that the precipitation model
used included inaccuracies due to a sparse network of climate
stations. In addition, the moisture status of particular sites
depends not only on precipitation but also on the topographic
position on the valley bottom or the ridge,slope aspect,inclination
and soil properties. The strong pH­precipitation relationship
suggests that on the transition between arid continental and
more humid landscapes, the intensity of leaching of basic cations
from soils is crucial for maintaining high or low pH.
In forests, mean pH was significantly lower than in treeless
vegetation (Fig. 1). This partly reflects the fact that in forestTreeless
vegetation Forests
Variable
Relative
importance Variable
Relative
importance
January temperature 100 Altitude 100
Altitude 82.1 January temperature 97.6
Soil pH 72.3 Soil pH 94.9
Winter precipitation 61.8 July temperature 77.7
Slope inclination 59.3 Soil conductivity 77.1
Summer precipitation 56.9 Summer precipitation 61.7
Soil conductivity 48.8 Winter precipitation 46.2
July temperature 44.7 Soil depth 19.6
Soil depth 29.7 Slope inclination 17.0
Radiation 16.0 Radiation above canopy 15.5
Radiation below canopy 14.2
Table 4 Ranked values of the relative
importance of explanatory variables of the
optimal regression trees for treeless and forest
vegetation. The values reflect the contribution
each explanatory variable makes in explaining
the variation in species richness, with its
contribution 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.The variable with the highest
relative importance is scored as 100 in each
regression tree.
Figure 4 Regression tree for species richness of forests. See Fig. 3
for a detailed explanation.At the first node,the best surrogate,summer
precipitation, had only 30% similarity to the primary splitter.
Species richness, pH and climate in southern Siberia
 2007 The Authors
Global Ecology and Biogeography, 16, 668­678, Journal compilation  2007 Blackwell Publishing Ltd 675
steppe landscapes, forest grows in moister places with a higher
intensity of leaching than steppe vegetation. However, as forest
soils are more acidic than non-forest soils in the precipitationrich
areas as well, more intensive acidification processes probably
occur in forest soils. These may result from: (1) higher moisture
status and more intensive soil leaching under the forest canopy,
which decreases evaporation; (2) larger uptake of bases and their
replacement by H+
cations in the more productive forest vegetation;
(3) higher abundance of mosses contributing to acidification
(Vitt, 2000); or (4) inputs of organic acids from conifer litter
(Ovington, 1953; Finzi et al., 1998). A much lower variation in
pH explained by summer precipitation for forest sites (33.2%)
than for treeless sites (63.3%) is another indication of a stronger
effect of vegetation on soil pH in forests than in open landscape.
In both treeless and forest vegetation, the best correlates of
species richness were January temperature,pH and altitude (Figs 3
and 4, Table 4). Climatic variables such as summer temperatures
and precipitation were correlated with them (Table 2). Other
measured variables were either correlated with these variables or
weakly related to the species richness pattern (Tables 2 & 4). This
corresponds to the alpine vegetation of the Alps, where species
richness is mainly determined by the joint effects of temperature
and pH, while other variables are less important (Vonlanthen
et al., 2006). In the continental climate of the study area with its
very cold winters, frost can be an important ecological factor
limiting plant life, especially in combination with low winter
precipitation and shallow or discontinuous snow cover. Actually,
areas with the strongest winter frosts had the species-poorest
vegetation. Both treeless and forest vegetation were consistently
poorer on soils with pH < 5.0, which corresponds to the value
where toxic Al3+
cations start to solubilize (Tyler, 1996). The
relationship between richness and altitude was more complex in
the study area, reflecting interactions between climatic variables.
At high altitudes, richness was lower due to low winter temperatures,
irrespective of the amount of precipitation, whereas at
lower altitudes, richness increased with precipitation.
The most species-rich treeless vegetation corresponded to
tall-grass meadow steppe, occurring at altitudes slightly above
500 m, which still have comparatively warm winters but higher
precipitation than dry lowlands. The poorest treeless vegetation
was tundra occurring in areas with pronounced winter frosts and
high precipitation, which causes soil leaching and acidification.
The richest forest vegetation occurred in areas where precipitation
was not too high to cause pronounced soil leaching and not
too low to exclude mesic forest species. Such habitats supported
either Pinus sylvestris hemiboreal forest or Abies sibirica wet taiga
on footslopes and valley bottoms. The poorest forest was dark
taiga with Pinus sibirica, Abies sibirica or Picea obovata, occurring
on low-pH soils in precipitation-rich areas.
Unimodal pH­richness relationship in treeless
vegetation
A positive richness­pH relationship has been more frequently
reported from the temperate, boreal and arctic zones than has a
negative one (Pärtel, 2002). In more detail, the sign and shape of
this relationship depends on vegetation types (Chytry et al.,
2003; Schuster & Diekmann, 2003). The relationship tends to
be positive in vegetation types that grow predominantly on
acidic to neutral soils, e.g. in forests and tundra. In contrast,
treeless vegetation on neutral to base-rich soils tends to have a
non-significant or negative richness­pH relationship.
Our study provides a finer and more complete picture than
most others because it covers nearly the whole range of soil pH
compatible with plant life. The existence of upper and lower
physiological limits to plant growth on the pH gradient implies
that the relationship is inherently unimodal. This is reported by
some studies covering broad ranges of pH values, especially
in grasslands (e.g. Tyler, 2000; Löbel et al., 2006) and mires (e.g.
Glaser et al., 1990). We found a pronounced unimodal richnesspH
relationship in treeless vegetation but not in forests.
Within treeless vegetation, tundra occurred on low-pH soils
in precipitation-rich areas, and steppe on high-pH soils in
precipitation-poor areas. After splitting these two vegetation
types, the former showed a positive linear relationship to soil pH
and the latter showed a negative linear relationship. The positive
relationship in the southern Siberian alpine tundra is consistent
with a number of earlier studies from the arctic tundra, where
increasing pH was identified as one of the major factors enhancing
species richness (Gough et al., 2000; Walker et al., 2001; van der
Welle et al., 2003). In contrast, steppe vegetation, which occurred
on soils with pH 5.9­8.6, exhibited a weak yet significant negative
linear relationship between richness and pH. This relationship
appears to be consistent with earlier studies from North
American prairies (Palmer et al., 2003). Our data suggest that the
decreasing species richness of steppe vegetation on high-pH soils
could be caused by climatic stress instead of or in addition to the
physiological effect of high pH. The highest pH values are found
in the driest soils with strongly limited or no cation leaching
(Fig. 1). However, these habitats experience not only pronounced
summer droughts but also strong winter frosts without protective
snow cover. As most species of the study area are not adapted to
such harsh conditions, the negative richness­pH correlation
appears. The richest vegetation occurs where soil is not too dry
and not too acidic at the same time. Our data suggest that for the
treeless vegetation of the study area it is at pH 6­7.
The putative role of drought stress in limiting species richness
on high-pH soils is supported by the comparison with richnesspH
relationships reported for wetlands and mires. In these
habitats, high pH is usually due to cation saturation from
mineral springs and it is not associated with drought stress.
In such environments, species richness often increases up to pH
7 or even beyond (Glaser et al., 1990; Hájková & Hájek, 2003;
Güsewell et al., 2005). However, no base-rich wetlands or mires
were included in the current study because they are very rare in
the study area.
Positive pH­richness relationship in forests
Unlike in treeless vegetation, the richness­pH relationship in forests
is mostly reported as monotonic positive (Dupré et al., 2002;
Schuster & Diekmann, 2003), while unimodal relationships
M. Chytr et al.
 2007 The Authors
676 Global Ecology and Biogeography, 16, 668­678, Journal compilation  2007 Blackwell Publishing Ltd
are rare and negative relationships are exceptional. The latter are
usually attributed to the confounding effects of factors other
than pH (Dupré et al., 2002). Species richness in forests usually
increases steeply in acidic soils, but the curve becomes less steep
above pH 5.5 and often levels off at a pH value close to 7.0. A
positive relationship is documented both from deciduous forests
of Europe (e.g. Brunet et al., 1997; Dumortier et al., 2002; Chytry
et al., 2003; Wohlgemuth & Gigon, 2003; Coudun & Gégout,
2005; Schuster & Diekmann, 2005) and North America (e.g. Peet
& Christensen, 1980; Brosofske et al., 2001; Peet et al., 2003) and
from coniferous forests in both continents (Rey Benayas, 1995;
Brosofske et al., 2001; Zinko et al., 2006). Palmer et al. (2003)
pointed to the contrasting pH­richness relationships in Oklahoma
forests and prairies, positive in the former and negative in the
latter. The same pattern, based on entirely different species pools,
was found in forests and steppes in our study area (Fig. 2).
We hypothesize that the predominance of the positive richnesspH
relationship in forests mainly results from the truncated pH
gradient at its base-rich end (Fig. 5). There are two reasons for
this truncation. First, forest cannot grow at extremely dry sites
that only support grassland, but these sites tend to have the
highest pH, at least in the semi-arid areas such as forest-steppe.
Second, productive vegetation with trees tends to decrease soil
pH (Fig. 1; Tyler, 2003). Truncation of the pH gradient in forests
is also apparent in the meta-analysis of German case studies
(Schuster & Diekmann 2003), where grasslands were often
recorded on soils with pH 7­8, but very few studies reported
forests on soils with pH > 7. Because the truncation removes
habitats where species pool sizes have already started to decrease
due to drought stress, most richness­pH relationships reported
in the literature from forests are monotonic positive. Besides this
main explanation, we suggest two other mechanisms that can
explain why the richness­pH curve of forest vegetation does not
usually decrease at high-pH conditions (Fig. 5). First, the species
richness of forests on soils with pH values of about 7 can be
partly increased by the mass effect at the forest-steppe ecotone.
Forests on such soils, which are usually rather dry, contain
several species typical of adjacent dry grasslands. These lightdemanding
species find a suitable habitat under the open canopy
formed by pine or larch. In contrast, no such mass effect can
enhance the species richness of grasslands at pH 7.5­8.5 because
there are no sufficiently large species pools of other vegetation
types available for this pH range in the study area. Second, in dry
areas such as the forest-steppe zone, the forest canopy maintains
higher air humidity in the understorey than in treeless habitats
(Moore et al., 2000). Therefore, forest richness may not be so
much limited by drought stress as is steppe richness on soils with
pH values of about 7.0­7.5.
Species pools and the Pleistocene bottleneck
The species pool hypothesis predicts that the species richness of
particular habitats is correlated with the historical habitat commonness
and associated opportunities for species accumulation
over evolutionary time (i.e. speciation minus extinction; Aarssen
& Schamp, 2002). Over most of today's temperate zone of Eurasia,
the Pleistocene climate was not only cool but was also dry (Frenzel
et al., 1992). Such a climate supported the development of dry
calcium-rich soils over vast areas of cold steppe, the habitat of
large Pleistocene herbivores (Guthrie, 2001; Walker et al., 2001).
Patches of coniferous forests also occurred in some areas, mainly
in the mountains and wet lowland sites; they have been reconstructed
from the full-glacial fossil data both from the RussianMongolian
border area (Tarasov et al., 2000) and from central
Europe (Willis et al., 2000; Jankovská et al., 2002). Tundra was
confined to places with higher precipitation or lower evapotranspiration,e.g.in
the mountains or at non-glaciated higher latitudes,
often in the close vicinity of steppe (Tarasov et al., 2000).
The present-day landscape in some parts of southern Siberia
and Mongolia is probably the closest modern analogy to the
Pleistocene environments that existed in the non-glaciated part
of northern Eurasia. It has to be pointed out that modern
analogies are never perfect. Predictions based on analogies must
not be taken too seriously until there is independent and reliable
fossil evidence. However, if modern analogies exist, they may
Figure 5 Conceptual model of species richness in treeless
vegetation and forests across the combined gradient of decreasing
precipitation and increasing pH. On low-pH soils in precipitationrich
areas, pools of species adapted to toxic effects of Al3+
and H+
ions are limited and species richness is low. With decreasing
precipitation and increasing pH, richness first increases but
subsequently decreases, because of a limited pool of species adapted
to very dry conditions. In contrast, forest richness does not decrease
or decreases only slightly in precipitation-poor areas for the
following reasons: (1) forests do not occur in the driest areas, so
the extreme end of the precipitation­pH gradient is truncated;
(2) forests occurring on their ecological limit have an open
canopy and their herb layer is saturated by steppe species pools;
and (3) the protective effect of forest canopy maintains higher air
humidity and reduces drought stress.
Species richness, pH and climate in southern Siberia
 2007 The Authors
Global Ecology and Biogeography, 16, 668­678, Journal compilation  2007 Blackwell Publishing Ltd 677
provide valuable insights and enhance our understanding of
ecological patterns and processes in the past environments.
The current study included vegetation types with the same
leading species as reconstructed from the full-glacial central
Europe (Willis et al., 2000; Jankovská et al., 2002), and parts of
the study area had a climate corresponding to the cool periods
of the Pleistocene (Frenzel et al., 1992). In southern Siberia, the
richest vegetation is found on soils with a pH of 6­7 (Fig. 2),
which are associated with moderate precipitation (Fig. 1). If the
modern species richness of different habitats resulted from
speciation and extinction in the historically prevailing environments
(species pool effects; Aarssen & Schamp, 2002; Pärtel,
2002; Ewald, 2003), this pattern would suggest that such slightly
acidic to neutral soils were common in the pre-Holocene
environments of temperate Eurasia. In contrast, very dry calciumrich
steppe or wet acidic tundra probably never supported
species-rich vegetation.
Increasing precipitation in the Holocene, especially in
areas with an oceanic climate, caused large-scale soil acidification
(Roberts, 1998). The species-pool hypothesis predicts that today's
acidic soils, although prevailing in large areas, are poor in species
because of the lack of calcifuge species, which may have become
largely extinct during the Pleistocene (Ewald, 2003). Such acidified
habitats are represented in the wetter parts of the study area by
taiga forests (in warmer climate) or by tundra (in cooler climate).
Both of these habitats are generally species poor, but exhibit
positive richness­pH relationships, for which the species pool
hypothesis provides a plausible explanation.
ACKNOWLEDGEMENTS
We thank the local staff of our expeditions for their extensive
support during the fieldwork and two anonymous referees
for useful comments on the previous version of this paper.
This research was supported from projects IAA6163303 from
the Grant Agency of the Academy of Sciences of the Czech
Republic and long-term research plans MSM0021622416 and
AVOZ60050516.
REFERENCES
Aarssen, L.W. & Schamp, B.S. (2002) Predicting distributions of
species richness and species sizes in regional floras: applying the
species pool hypothesis to the habitat templet model. Perspectives
in Plant Ecology, Evolution and Systematics, 5, 3­12.
Berglund, B.E., Birks, H.J.B., Ralska-Jasiewiczowa, M. &
Wright, H.E. (1996) Palaeoecological events during the last
15,000 years. John Wiley & Sons, Chichester.
Breiman, L., Friedman, J.H., Olshen, R.A. & Stone, C.G. (1984)
Classification and regression trees. Wadsworth International
Group, Belmont, CA.
Brosofske, K.D., Chen, J. & Crow, T.R. (2001) Understory vegetation
and site factors: implications for a managed Wisconsin
landscape. Forest Ecology and Management, 146, 75­87.
Brunet, J., Falkengren-Grerup, U., Rühling, A. & Tyler, G. (1997)
Regional differences in floristic change in South Swedish oak
forests as related to soil chemistry and land use. Journal of
Vegetation Science, 8, 329­336.
Chytry, M., Tichy, L. & Roleçek, J. (2003) Local and regional patterns
of species richness in Central European vegetation types
along the pH/calcium gradient. Folia Geobotanica, 38, 429­
442.
Coudun, C. & Gégout, J.-C. (2005) Ecological behaviour of
herbaceous forest species along a pH gradient: a comparison
between oceanic and semicontinental regions in northern
France. Global Ecology and Biogeography, 14, 263­270.
Dumortier, M., Butaye, J., Jacquemyn, H., Van Camp, N., Lust, N.
& Hermy, M. (2002) Predicting vascular plant species richness
of fragmented forests in agricultural landscapes in central
Belgium. Forest Ecology and Management, 158, 85­102.
Dupré, C., Wessberg, C. & Diekmann, M. (2002) Species richness
in deciduous forests: Effects of species pools and environmental
variables. Journal of Vegetation Science, 13, 505­516.
Ermakov, N., Dring, J. & Rodwell, J. (2000) Classification of continental
hemiboreal forests of North Asia. Braun-Blanquetia,
28, 1­131.
Ewald, J. (2003) The calcareous riddle: why are there so many
calciphilous species in the Central European flora? Folia
Geobotanica, 38, 357­366.
Finzi, A.C., Canham, C.D. & van Breemen, N. (1998) Canopy
tree­soil interactions within temperate forests: species effects
on pH and cations. Ecological Applications, 8, 447­454.
Frenzel, B., Pécsi, M. & Velichko, A.A. (eds) (1992) Atlas of
paleoclimates and paleoenvironments of the Northern
Hemisphere. Geographical Research Institute, Budapest &
Gustav Fischer Verlag, Stuttgart.
Gidrometeoizdat (1966­1970) Spravochnik po klimatu SSSR
[Reference books on the climate of the USSR]. Gidrometeoizdat,
Leningrad.
Gigon, A. (1987) A hierarchic approach in causal ecosystem
analysis: the calcifuge-calcicole problem in alpine grasslands.
Ecological Studies, 61, 228­244.
Glaser, P.H., Janssens, J.A. & Siegel, D.I. (1990) The response of
vegetation to chemical and hydrological gradients in the Lost
River peatland, northern Minnesota. Journal of Ecology, 78,
1021­1048.
Gough, L., Shaver, G.R., Carroll, J., Royer, D.L. & Laundre, J.A.
(2000) Vascular plant species richness in Alaskan arctic tundra:
the importance of soil pH. Journal of Ecology, 88, 54­66.
Grubb, P.J. (1987) Global trends in species-richness in terrestrial
vegetation: a view from the northern hemisphere. Organization
of communities: past and present (ed. by J.H.R. Gee and P.S.
Giller), pp. 99­118. Blackwell Science, Oxford.
Güsewell, S., Bailey, K.M., Roem, W.J. & Bedford, B.L. (2005)
Nutrient limitation and botanical diversity in wetlands: can
fertilisation raise species richness? Oikos, 109, 71­80.
Guthrie, R.D. (2001) Origin and causes of the mammoth steppe:
a story of cloud cover, woolly mammal tooth pits, buckles,
and inside-out Beringia. Quaternary Science Reviews, 20, 549­
574.
Hájková, P. & Hájek, M. (2003) Species richness and aboveground
biomass of poor and calcareous spring fens in the
M. Chytr et al.
 2007 The Authors
678 Global Ecology and Biogeography, 16, 668­678, Journal compilation  2007 Blackwell Publishing Ltd
flysch West Carpathians, and their relationships to water and
soil chemistry. Preslia, 75, 271­287.
Hilbig, W. (1995) The vegetation of Mongolia. SPB Academic
Publishing, Amsterdam.
Jankovská, V., Chromy, P. & Niznianská, M. (2002) Íafárka --
first palaeobotanical data of the character of Last Glacial
vegetation and landscape in the West Carpathians (Slovakia).
Acta Palaeobotanica, 42, 39­50.
Löbel, 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 Geobotanica, 41, 377­393.
Meusel, H., Jäger, E.J., Weinert, E. & Rauschert, S. (1965­1992)
Vergleichende Chorologie der zentraleuropäischen Flora I­III.
Gustav Fischer Verlag, Jena.
Moore, K.E., Fitzjarrald, D.R., Sakai, R.K. & Freedman, J.M.
(2000) Growing season water balance at a boreal jack pine
forest. Water Resources Research, 36, 483­493.
Ovington, J.D. (1953) Studies of the development of woodland
conditions under different trees. I. Soil pH. Journal of Ecology,
41,13­52.
Palmer, M.W., Arévalo, J.R., del Carmen Cobo, M. & Earls, P.G.
(2003) Species richness and soil reaction in a northeastern
Oklahoma landscape. Folia Geobotanica, 38, 381­389.
Pärtel, M. (2002) Local plant diversity patterns and evolutionary
history at the regional scale. Ecology, 83, 2361­2366.
Peet, R.K. & Christensen, N.L. (1980) Hardwood forest vegetation
of the North Carolina piedmont. Veröffentlichungen des
Geobotanischen Institutes ETH, Stiftung Rübel, Zürich, 69, 14­
39.
Peet, R.K., Fridley, J.D. & Gramling, J.M. (2003) Variation in
species richness and species pool size across a pH gradient
in forests of the southern Blue Ridge Mountains. Folia
Geobotanica, 38, 391­401.
Polikarpov, N.P., Chebakova, N.M. & Nazimova, D.I. (1986)
Klimat i gornye lesa Sibiri [Climate and mountain forests of
Siberia]. Nauka, Novosibirsk.
Rey Benayas, J.M. (1995) Patterns of diversity in the strata of
boreal montane forest in British Columbia. Journal of Vegetation
Science, 6, 95­98.
Roberts, N. (1998) The Holocene: an environmental history, 2nd
edn. Blackwell Science, Oxford.
Schuster, B. & Diekmann, M. (2003) Changes in species density
along the soil pH gradient -- evidence from German plant
communities. Folia Geobotanica, 38, 367­379.
Schuster, B. & Diekmann, M. (2005) Species richness and
environmental correlates in deciduous forests of Northwest
Germany. Forest Ecology and Management, 206, 197­205.
Sjörs, H. (1952) On the relation between vegetation and electrolytes
in north Swedish mire waters. Oikos, 2, 241­258.
Tarasov, P.E., Volkova, V.S., Webb, T., III, Guiot, J., Andreev, A.A.,
Bezusko, L.G., Bezusko, T.V., Bykova, G.V., Dorofeyuk, N.I.,
Kvavadze, E.V., Osipova, I.M., Panova, N.K. & Sevastyanov, D.V.
(2000) Last glacial maximum biomes reconstructed from
pollen and plant macrofossil data from northern Eurasia.
Journal of Biogeography, 27, 609­620.
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,
239­250.
Tyler, G. (1996) Soil chemistry and plant distributions in rock
habitats of southern Sweden. Nordic Journal of Botany, 16,
609­635.
Tyler, G. (2000) Integrated analysis of conditions accounting
for intersite distribution of grassland plants. Nordic Journal of
Botany, 20, 485­500.
Tyler, G. (2003) Some ecophysiological and historical approaches
to species richness and calcicole/calcifuge behaviour --
contribution to a debate. Folia Geobotanica, 38, 419­428.
van der Welle, M.E.W., Vermeulen, P.J. Shaver, G.R. & Berendse, F.
(2003) Factors determining plant species richness in Alaskan
arctic tundra. Journal of Vegetation Science, 14, 711­720.
Vitt, D.H. (2000) Peatlands: ecosystems dominated by bryophytes.
Bryophyte biology (ed. by A.J. Shaw and B. Goffinet), pp. 312­
343. Cambridge University Press, Cambridge.
Vonlanthen, C.M., Kammer, P.M., Eugster, W., Bühler, A. &
Veit, H. (2006) Alpine vascular plant species richness: the
importance of daily maximum temperature and pH. Plant
Ecology, 184, 13­25.
Walker, D.A., Bockheim, J.G., Chapin, F.S., III, Eugster, W.,
Nelson, F.E. & Ping, C.L. (2001) Calcium-rich tundra, wildlife,
and the `Mammoth Steppe'. Quaternary Science Reviews, 20,
149­163.
Willis, K.J., Rudner, E. & Sümegi, P. (2000) The full-glacial forests
of central and southeastern Europe. Quaternary Research, 53,
203­213.
Wohlgemuth, T. & Gigon, A. (2003) Calcicole plant diversity in
Switzerland may reflect a variety of habitat templets. Folia
Geobotanica, 38, 443­452.
Zinko, U., Dynesius, M., Nilsson, C. & Seibert, J. (2006) The role
of soil pH in linking groundwater flow and plant species
density in boreal forest landscapes. Ecography, 29, 515­524.
Editor: Martin Sykes
BIOSKETCH
Milan Chytry is an associate professor of botany in the
Department of Botany and Zoology, Masaryk University,
Brno, Czech Republic. His research, as well as research
by the co-authors, is focused on vegetation ecology and
large-scale diversity patterns of multi-species assemblages
of vascular plants.