ArticlePDF Available

Environmental and historical controls of floristic composition across the South American Dry Diagonal

Authors:

Figures

Content may be subject to copyright.
ORIGINAL
ARTICLE
Environmental and historical controls of
floristic composition across the South
American Dry Diagonal
1
Danilo M. Neves
1,2
*, Kyle G. Dexter
1,3
,R. Toby Pennington
1
,
Marcelo L. Bueno
2
and Ary T. Oliveira Filho
2
2
1
Royal Botanic Garden Edinburgh, Edinburgh,
Midlothian EH3 5LR, UK,
2
Programa de
P
os-Graduac
ß
~
ao em Biologia Vegetal,
Universidade Federal de Minas Gerais
UFMG, Campus Pampulha, Belo Horizonte,
MG 31270090, Brazil,
3
School of
GeoSciences, University of Edinburgh, King’s
Buildings, Edinburgh, Midlothian EH9 3JN ,
UK
*Correspondence: Danilo M. Neves, Royal
Botanic Garden Edinburgh, 20a Inverleith
Row, Edinburgh, Midlothian EH3 5LR, UK.
E-mail: danilormn@gmail.com
ABSTRACT
Aim The aim of this study was to test the role of environmental factors and
spatially autocorrelated processes, such as historical fragmentation and dispersal
limitation, in driving floristic variation across seasonally dry tropical forests
(SDTFs) in eastern South America.
Location SDTFs extending from the Caatinga phytogeographical domain of
north-eastern Brazil to the Chaco phytogeographical domain of northern
Argentina, an area referred to as the Dry Diagonal.
Methods We compiled a database of 282 inventories of woody vegetation in
SDTFs from across the Dry Diagonal and combined this with data for 14 envi-
ronmental variables. We assessed the relative contribution of spatially autocor-
related processes and environmental factors to the floristic turnover in SDTFs
across the Dry Diagonal using variation partitioning methods. In addition, we
used multivariate analyses to determine which environmental factors were most
important in explaining the turnover.
Results We found that the environmental factors measured (temperature, pre-
cipitation and edaphic conditions) explained 21.3% of the variation in species
composition, with 14.1% of this occurring independently of spatial autocorre-
lation. A spatially structured fraction of 4.2% could not be accounted for by
the environmental factors measured. The main axis of compositional variation
was significantly correlated with a northsouth gradient in temperature regime.
At the extreme south of the Dry Diagonal, a cold temperature regime, in which
frost occurred, appeared to underlie floristic similarities between chaco wood-
lands and southern SDTFs.
Main conclusions Environmental variables, particularly those related to tem-
perature regime, seem to be the most significant factors affecting variation in
species composition of SDTFs. Thus historical fragmentation and isolation
alone cannot explain the turnover in species composition within SDTFs, as is
often assumed. Compositional and environmental heterogeneity needs to be
taken into account both to understand the past distribution of SDTFs and to
manage and conserve this key tropical biome.
Keywords
Caatinga woodlands, chaco woodlands, environmental niche, Pleistocene
arc hypothesis, seasonally dry tropical forests, species turnover, variation
partitioning.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
ª2015 John Wiley & Sons Ltd http://wileyonlinelibrary.com/journal/jbi 1
doi:10.1111/jbi.12529
Journal of Biogeography (J. Biogeogr.) (2015)
J B I 12529
Dispatch: 10.4.15 CE: Malarvizhi
Journal Code Manuscript No.
No. of pages: 11 PE: Pravin Kumar
INTRODUCTION
Seasonally dry tropical forests (SDTFs) occur as fragments of
varying size throughout Latin America, from northern
Argentina and north-east Brazil to north-west Mexico (Pen-
nington et al., 2006). These SDTFs comprise a recognized
biome (Pennington et al., 2000, 2009) that is characterized
by <1600 mm precipitation year
1
, a marked dry season of
56 months and high rates of deciduousness during the dry
season (Bullock et al., 1995; Gentry, 1995; Oliveira-Filho
et al., 2006). SDTFs have received increased attention in
recent years because of both their highly threatened status
(e.g. Mooney et al., 1995; Pennington et al., 2006, 2009; Dir-
zo et al., 2011) and the influential Pleistocene Arc Hypothe-
sis (PAH; Prado & Gibbs, 1993; Pennington et al., 2000).
The PAH postulates that SDTFs had a more widespread and
contiguous distribution in South America during Pleistocene
glaciation phases, including areas that are currently covered
by rain forest or savanna. The PAH is based upon the wide-
spread distributions of multiple, unrelated tree species found
in disjunct SDTF areas from north-east Brazil through cen-
tral and southern Brazil to Paraguay and Argentina and, in
some cases, as far as the dry inter-Andean valleys of western
South America. It is suggested that the putative Pleistocene
Arc, a more continuous area of SDTF, connected these areas
during the cooler and drier climates of Pleistocene glacia-
tions (Prado & Gibbs, 1993; Pennington et al., 2000).
Subsequent research, however, has shown that these wide-
spread species represent a small minority of SDTF tree spe-
cies and that floristic variation between SDTF sites across the
Neotropics is high (e.g. Gillespie et al., 2000; Trejo & Dirzo,
2002; Linares-Palomino, 2006; Lott & Atkinson, 2006; Pen-
nington et al., 2006; Queiroz, 2006; Castillo-Campos et al.,
2008; Linares-Palomino et al., 2011). For instance, Linares-
Palomino et al. (2011) found that less than 2% of 3839
SDTF tree species are found in 10 or more of 21 disjunct
SDTF nuclei across the Neotropics. On the other hand, the
same authors (Linares-Palomino et al., 2011) found higher
floristic similarity (Sørensen similarity index values >0.25)
among SDTF nuclei in Brazil and southern South America
and suggested that this offers some support for the idea that
SDTF was more widespread and continuous within this
region during Pleistocene glacial phases.
When explaining the disjunct distributions of SDTF spe-
cies and floristic variation of SDTFs in general, recent studies
have focused primarily on the role of historical biogeograph-
ical processes, namely habitat fragmentation and dispersal
limitation (e.g. Pennington et al., 2006, 2009). However,
environmental conditions are also expected to affect species
distribution. This study is the first to examine the role of
environmental, primarily climatic, determinants in shaping
the distribution of tree species across a large expanse of
Neotropical SDTF sites.
We aimed to test the role of historical fragmentation and
dispersal limitation versus environmental factors in driving
floristic differentiation among SDTFs across eastern South
America. Understanding the roles of these two processes is
an important and unsolved issue in many systems (e.g. in
the Amazon: Haffer, 1969; ter Steege et al., 2006; 3in the Cer-
rado: Werneck et al., 2012a; in South American subtropical
forests: Oliveira-Filho et al., 2013). Variation partitioning
methods provide a means of assessing the relative contribu-
tion of measured environmental factors and spatially auto-
correlated processes (potentially including unmeasured
environmental factors) in driving species turnover (Legendre
et al., 2012). This approach has been used mostly at small
spatial scales (e.g. within plots; Legendre et al., 2009) to
address the influence of steady-state, small-scale dispersal
limitation. However, it can also be used at much broader
spatial scales (Vetaas & Ferrer-Cast
an, 2008) to address habi-
tat fragmentation and isolation at historical time-scales, and
may be particularly useful in ecosystems with limited palaeo-
ecological data, such as SDTFs.
We compiled a database of SDTF floristic inventories from
sites extending from the caatinga woodlands of north-eastern
Brazil to the chaco woodlands of northern Argentina, an area
referred to as the Dry Diagonal (following Prado & Gibbs,
1993; also see Vanzolini, 1963; Werneck et al., 2012b). This
region was particularly appropriate because it was the main
area studied by Prado & Gibbs (1993) in the development of
the PAH and because it has been the target of previous stud-
ies on SDTF biogeography (de Andrade-Lima, 1954, 1982;
Ratter et al., 1988; Caetano & Naciri, 2011; Werneck et al.,
2011; Collevatti et al., 2012, 2013a,b). If the floristic varia-
tion of SDTFs is solely the result of fragmentation and subse-
quent dispersal limitation between SDTF patches, climatic
and edaphic variables should not be important factors in
explaining the variation in community composition. Alterna-
tively, if previously unrecognized environmental conditions
within SDTFs are the principal factors driving floristic turn-
over, then climatic and edaphic factors should explain a large
proportion of the variation in community composition.
MATERIALS AND METHODS
Study area
The Dry Diagonal stretches from north-eastern Brazil to
Argentina, Paraguay and Bolivia, and occurs primarily in
three phytogeographical domains (IBGE, 1993) that are
named after the main vegetation types that occur within them:
Caatinga (tropical semi-arid thorn woodlands, called caatinga
woodlands in Brazil), Cerrado (seasonal woody savannas,
called cerrado woodlands in Brazil) and Chaco (subtropical/
tropical semi-arid thorn woodlands, called chaco woodlands
in Brazil and elsewhere) (Fig. 1). Prado & Gibbs (1993)
defined three major nuclei of SDTFs in the Dry Diagonal: (1)
the Caatinga nucleus, which occurs in north-eastern Brazil
and is the largest nucleus (essentially the same area as the Ca-
atinga phytogeographical domain); (2) the Misiones nucleus,
which extends from the lower course of the Paraguay and
Paran
a rivers to the upper Uruguay River in Misiones,
Journal of Biogeography
ª2015 John Wiley & Sons Ltd
2
D. M. Neves et al.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
Argentina; and (3) the Piedmont nucleus, which occurs along
the eastern base of the Andes in north-eastern Argentina.
SDTFs also occur in the Chiquitania region of Bolivia (Lin-
ares-Palomino et al., 2011) and in patches scattered through-
out the Cerrado phytogeographical domain (Ratter et al.,
1988; Pennington et al., 2000) in areas of fertile soil.
One environmental element in the definition of SDTFs is
that they are a frost-free vegetation (Prado, 1993a,b; Pen-
nington et al., 2000), which would suggest that the chaco
woodlands, found in northern Argentina, western Paraguay,
south-eastern Bolivia and the extreme western edge of Mato
Grosso do Sul state in Brazil, are not SDTFs because they
suffer regular winter frost. However, large areas delimited as
SDTF in the schematic maps of previous studies (e.g. Pied-
mont and Misiones; Prado & Gibbs, 1993; Oliveira-Filho
et al., 2006; Pennington et al., 2009) also receive significant
frost (see Results). Thus we chose to include the chaco
woodlands in our analyses because they meet other criteria
of SDTF vegetation as defined by the same authors (Prado,
1993a,b; Pennington et al., 2000), such as seasonality of rain-
fall and deciduousness.
The SDTF database
We extracted floristic inventory data from the NeoTropTree
database (available at http://prof.icb.ufmg.br/treeatlan). This
database consists of >2000 georeferenced localities for which
lists of tree species (trees defined here as woody plants
>3 m in height) were compiled from an extensive survey of
published and unpublished literature. A NeoTropTree site is
defined by a single vegetation type contained within a circu-
lar area with a 5-km radius. Therefore, where two or more
vegetation types co-occur in one 10-km area, there may be
two geographically overlapping sites in the NeoTropTree
database, each for a distinct vegetation type. The NeoTrop-
Tree data were derived primarily from inventories of woody
vegetation (i.e. plots, transects and vegetation surveys). Sur-
veys of specimens at major herbaria were then used to aug-
ment the lists for each NeoTropTree site with any tree
species that were collected within the 5-km radius of the ori-
ginal NeoTropTree site and within the same vegetation type.
Moreover, all NeoTropTree species were checked regarding
their taxonomy and synonymies by using the Flora do Brasil
Figure 1 5
Map showing the location of tree species surveys and major vegetation types used for the study of environmental and
historical controls of floristic composition across the South American Dry Diagonal. Circles denote the location of tree species surveys
used in this study (n=282). The phytogeographical domains that form the South American Dry Diagonal are shaded (Ca, Caatinga;
Ce, Cerrado; Ch, Chaco; Cq, Chiquitania) while moist forest phytogeographical domains are white (Am, Amazon rain forest; At,
Atlantic rain forest). The Misiones and Piedmont seasonally dry tropical forest (SDTF) nuclei (Prado & Gibbs, 1993) are delimited by
dashed lines (1, Misiones; 2, Piedmont); the Caatinga nucleus covers the same area as the Caatinga phytogeographical domain. The
phytogeographical domains are named after the major vegetation type that occurs within them (e.g. chaco woodlands, caatinga
woodlands and savanna vegetation; the latter is called Cerrado in Brazil). Note that each phytogeographical domain may contain several
additional vegetation types (e.g. the Cerrado phytogeographical domain contains SDTF and gallery forest in addition to the main
vegetation type of savanna).
LOW RESOLUTION FIG
Journal of Biogeography
ª2015 John Wiley & Sons Ltd
3
Environmental controls of floristic variation in dry forests
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
database for plants that occur in Brazil, and the Tropicos
database for plants that do not occur in Brazil (available at
http://floradobrasil.jbrj.gov.br/ and http://tropicos.org/,
respectively). NeoTropTree does not include occurrence
records without an indication or evidence of vegetation type,
nor sites with very incomplete species lists. The latter is an
important filter because different sampling efforts across sites
may bias their descriptive power.
The 282 NeoTropTree inventories used in this study com-
prised 134 SDTF sites from the Caatinga phytogeographical
domain, 14 from the Misiones nucleus, five from the Pied-
mont nucleus, 91 from SDTF enclaves scattered throughout
the Cerrado phytogeographical domain, six from the Bolivian
Chiquitania and 32 chaco woodland sites from the Chaco
phytogeographical domain. This does not represent an even
geographical coverage of inventories, but we believe our
results are not excessively biased because nuclei covered by
just a few sites often had many species, whereas nuclei with
many sites could only have a few species. For example, the
Misiones nucleus was represented by 398 species from just
14 lists, whereas 891 species were recorded from 134 lists
from the Caatinga phytogeographical domain. The final spe-
cies matrix contained presenceabsence data for 1765 tree
species across the 282 SDTF sites, with a total of 25,650 pres-
ences (Fig. 1).
The NeoTropTree database also included environmental
data for all its sites, derived from multiple sources. Eight
variables were related to edaphic substrate: mean, mini-
mum and maximum monthly soil moisture (%), obtained
from the International Soil Moisture Network (http://
www.ipf.tuwien.ac.at/insitu/); prevailing slope and aspect,
obtained from CGIAR-CSI (2006); and soil fertility (%
base saturation), soil coarseness (% sand) and substrate
rockiness (% surface of rock outcrops), extracted from a
detailed map of soil types produced by EMBRAPA &
IBGE (2003) using the RadamBrasil Soil Survey of the
1970s and 1980s. Climatic variables included the mean
duration (days) and severity (mm) of water-deficit periods,
both extracted from Walter diagrams (Walter, 1985); the
19 bioclimatic variables produced by WorldClim 1.4, a
high-resolution (1-km) database of global climate layers cre-
ated by Hijmans et al. (2005); mean frequency of frosts (days),
cloud cover (%) and cloud interception (mm), obtained from
gridded datasets produced by Jones & Harris (2008); and three
additional variables, potential and actual evapotranspiration
(mm) and an aridity index, derived from WorldClim by Zo-
mer et al. (2007, 2008). A full description with details of the
protocols for NeoTropTree can be found at http://prof.ic-
b.ufmg.br/treeatlan.
Analyses of species composition
We first explored the patterns of similarity in community
composition by conducting a hierarchical clustering analysis
of all 282 sites. We used Jaccard distance as the dissimilarity
metric and unweighted paired groups as the linkage method
(McCune & Grace, 2002). We assessed the confidence for
each node of the hierarchical cluster using multi-scale boot-
strapping (Shimodaira, 2004) in the pvclust package (Su-
zuki & Shimodaira, 2011) in the R statistical environment (R
Core Team, 2012) 4. Because singletons (species found at a
single site) commonly increase the intrinsic noise in most
analyses without contributing information (Lep
s&
Smilauer,
2003), we excluded the 318 singleton species from this and
all subsequent analyses.
We obtained the relative contribution of measured envi-
ronmental factors and unmeasured spatially autocorrelated
factors in explaining variation in community composition
by following methods similar to those proposed by Dray
et al. (2012) and Legendre et al. (2012). This routine com-
prises (1) the compilation of significant spatial and environ-
mental variables through a forward selection method for
canonical correspondence analysis (CCA), with a permuta-
tion-based test for each variable added (Borcard et al.,
2011); (2) an additional and progressive elimination of col-
linear variables based on their variance inflation factor
(VIF) until maintaining only those with VIF <10 (Quinn
& Keough, 2002); and (3) variation partitioning of the
community composition matrix with respect to the signifi-
cant spatial and environmental variables. As spatial vari-
ables, we used Moran’s eigenvector maps (MEMs), which
represent the diagonalization of a centred spatial weighting
matrix (Dray et al., 2012). We tested the overall significance
of the spatial fraction (controlled for measured environmen-
tal variation) and the environmental fraction (controlled for
spatial autocorrelation) by applying a permutation test (999
permutations) for CCA. All variation partitioning analyses
were conducted using the spacemakeR (Dray, 2010), spdep
(Bivand, 2012), tripack (Gebhardt, 2009) and vegan (Oksa-
nen et al., 2012) packages in the R statistical environment
(R Core Team, 2015).
We also performed a detrended correspondence analysis
(DCA) to derive orthogonal eigenvectors representing gradi-
ents in floristic variation and used post hoc correlations to
assess their relationship with environmental factors. We
tested the significance of the post hoc correlation coefficients
by applying a permutation test (999 permutations). There
was a negligible increase in explained variation with the
addition of >2 DCA axes, and we therefore focused on the
first two axes for comparisons with environmental variables.
We also explored the results visually by plotting the site
scores of the DCA axes and the values of the environmental
variables in geographical space. These analyses were per-
formed in the statistical package vegan (Oksanen et al.,
2012) and the maps were designed using the package map-
tools (Lewin-Koh & Bivand, 2012), both in the R statistical
environment.
RESULTS
The hierarchical clustering results were largely congruent
with the previously defined SDTF nuclei. The classification
Journal of Biogeography
ª2015 John Wiley & Sons Ltd
4
D. M. Neves et al.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
segregated the 282 sites into two principal groups, hereafter
termed the northern and southern groups. The northern
group consisted of 232 sites from the Caatinga nucleus,
Cerrado phytogeographical domain SDTFs and Bolivian
Chiquitania, whereas the southern group consisted of 50
sites from the Misiones and Piedmont nuclei and the chaco
woodlands from the Chaco phytogeographical domain
(Fig. 2).
The forward selection procedure retained 14 out of 36
environmental variables for modelling variation in species
composition (Table 1). After partitioning the variation
explained by the selected spatial and environmental predic-
tors, we found that the measured environmental factors
explained 21.3% of the variation in species composition,
with 14.1% of this occurring independently of spatial auto-
correlation. A spatially structured fraction of 4.2% could not
be accounted for by the measured environmental factors;
74.5% of the variation remained unexplained (Fig. 3).
When investigating which environmental variables
explained species turnover across the Dry Diagonal, we
found that the variation in species composition summarized
by the first DCA axis revealed a large-scale gradient (Fig. 4a)
that was correlated with a northsouth gradient in the mean
minimum temperature of the coldest month (Fig. 4c,
Table 1). The first DCA axis was also congruent with a
northsouth segregation in days of frost (dashed black line
in Fig. 4a). The mean minimum of the coldest month in
southern SDTFs ranged from 0.1 °C to 15.4 °C, while some
southern SDTFs, such as those from the Piedmont nucleus,
could experience up to 18 days of frost. In the northern
group, four sites experienced up to 3 days of frost, while the
other 228 sites experienced none. The first DCA axis was also
correlated with a northsouth gradient in isothermality
(Fig. 4e, Table 1), suggesting that the high seasonality,
including both hot summers and winter frosts in the south-
ern group, was a relevant factor driving variation in commu-
nity composition.
The community composition variation summarized by the
second DCA axis revealed a rainfall gradient that was most
strongly correlated with the precipitation of the wettest
month (Fig. 4d, Table 1). In fact, the species composition
variation summarized by the second DCA axis was congruent
Figure 2 Hierarchical clustering of seasonally dry tropical forests (SDTFs) and chaco woodlands across the South American Dry
Diagonal based on their tree species composition. The dissimilarity measure and linkage methods used were Jaccard and unweighted
group average, respectively. Black bullets at the base of clusters indicate approximately unbiased bootstrap values 0.90 (Suzuki &
Shimodaira, 2011). The size of the triangles is proportional to the number of sites, and the floristic nuclei are discriminated by different
colours: red, 134 caatinga woodlands; orange, 91 Cerrado SDTFs; yellow, six SDTFs from Chiquitania; green, 14 SDTFs from Misiones;
light blue, five SDTFs from Piedmont; dark blue, 32 chaco woodlands.
Table 1 Environmental variables selected for the study of
environmental controls of floristic composition across the South
American Dry Diagonal. The variables shown are ordered by the
post hoc correlation coefficient with the first detrended
correspondence analysis (DCA) axis. The post hoc correlation
coefficient of each environmental variable with the second DCA
axis is also shown. VIF, variance inflation factor, obtained using
the r-squared value of the regression of one variable against all
other explanatory variables; n.s., not significant; P<0.05, except
for n.s. correlation coefficients.
Variables
DCA
1
DCA
2 VIF
Isothermality 0.62 0.06 7.64
Minimum temperature of the coldest
month
0.52 n.s. 8.43
Severity of water deficit 0.48 0.02 8.87
Precipitation of the driest month 0.34 0.03 7.57
Cloud cover 0.32 0.12 4.08
Sediment coarseness 0.31 n.s. 4.85
Days of frost 0.21 n.s. 3.08
Soil fertility 0.13 0.02 5.43
Soil water storage capacity 0.09 0.44 9.29
Precipitation of the coldest quarter 0.03 0.04 3.74
Precipitation of the wettest month 0.02 0.43 7.47
Elevation 0.01 0.04 3.69
Surface rockiness 0.01 n.s. 2.02
Cloud interception 0.01 0.02 1.45
COLOR
Journal of Biogeography
ª2015 John Wiley & Sons Ltd
5
Environmental controls of floristic variation in dry forests
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
with most of the precipitation-related variables (Table 1),
segregating a dry group (Caatinga nucleus, chaco woodlands,
Piedmont nucleus and Bolivian chiquitania) from a wet
group (Cerrado phytogeographical domain SDTFs and Misi-
ones nucleus). Furthermore, the second DCA axis correlated
with maximum soil water storage capacity (Fig. 4f), suggest-
ing an interaction between rainfall and soil characteristics in
driving the turnover in species composition summarized by
this axis.
DISCUSSION
Previous large-scale floristic analyses of South American
SDTFs have focused on historical fragmentation and dis-
persal limitation when explaining patterns of species turn-
over within communities (e.g. Pennington et al., 2006,
2009; Linares-Palomino et al., 2011). Our analyses indicate
that climatic and edaphic conditions also merit attention.
Because of the strong environmental correlations we found
(Table 1), it is inappropriate to designate historical frag-
mentation and dispersal limitation as the sole factors shap-
ing species composition in SDTFs. Most of the variation in
species composition that we were able to explain was
assigned to environmental factors that were independent of
spatial autocorrelation (14.1%). On the other hand, another
7.2% of the variation was accounted for by spatially struc-
tured environmental variation, signifying that the predic-
tions made by spatial and niche-based models converged.
However, considering that correlations between community
composition and environmental variables are known to be
important to plants (discussed below), we believe that it is
more parsimonious to attribute most of the spatially struc-
tured environmental variation to niche-based controls (cf.
Legendre et al., 2009).
At a broad scale within the South American tropics, pre-
cipitation regime is likely to be a primary determinant of the
presence versus absence of an SDTF; at a much smaller spa-
tial scale (within north-eastern and central-western Brazil),
Santos et al. (2012) found that precipitation regime was the
chief factor determining the species composition of SDTF
communities. Nevertheless, in analysing SDTF communities
at a larger spatial scale across the entire Dry Diagonal, we
found temperature-related variables to be the most signifi-
cant factors affecting variation in species composition within
SDTFs and chaco woodlands. In agreement with this result,
the fundamental division in our clustering analysis was
between two geographically segregated groups, with the
northern group experiencing mild temperatures in the cold-
est month, no frost (except for three sites with 3 days frost
year
1
) and a low annual temperature range compared with
the southern group.
This fundamental latitudinal division reflects marked dif-
ferences in the frequencies of species from different families.
Species in the Leguminosae, Bignoniaceae, Moraceae and An-
nonaceae are more frequent in the northern group, which
indicates that the northern flora primarily comprise species
incapable of coping with freezing temperatures. Punyasena
et al. (2008) classified these families as temperature-sensitive
because their abundance and richness decrease in low tem-
peratures. Meanwhile, Lauraceae, Asteraceae and Melastom-
ataceae have higher frequencies in the cooler southern group.
Rainfall regime correlated strongly with the second gradi-
ent of species composition, as summarized by the second
DCA axis. The precipitation division reflects the fact that the
relative frequencies of Leguminosae, Capparaceae and Polyg-
onaceae are higher in drier areas, whereas the relative fre-
quencies of Rubiaceae, Moraceae, Arecaceae and Annonaceae
are higher in wetter areas. The lower frequencies of the latter
families in the drier areas match the results of Punyasena
et al. (2008), who showed that their abundance and richness
decrease in areas with marked drought.
The distinctiveness of the chaco woodlands
Previous authors have excluded the chaco woodlands from
the SDTF biome, citing floristic differences, with the expla-
0 102030405060708090100
% variation
Environmental
variation
Spatial
structure
(a)
14.1
(b)
7.2
(c)
4.2
Unexplained
74.5
Figure 3 Variation partitioning by canonical correspondence analysis to determine how much of the spatial variation in floristic
composition across the South American Dry Diagonal was accounted for by the environmental variables measured. Fraction (a)
represents non-spatial environmental variation, fraction (b) spatially structured environmental variation, i.e. the overlap between
environmental and spatial components, and fraction (c) the spatial structure not explained by the measured environmental variables. All
values are expressed as percentages.
Journal of Biogeography
ª2015 John Wiley & Sons Ltd
6
D. M. Neves et al.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
nation that SDTF occurs only in frost-free areas (Prado,
1993a,b; Pennington et al., 2000; Werneck et al., 2011).
However, we found that the sites we analysed in the Chaco
domain were compositionally similar to southern SDTFs
(Fig. 4a; additionally some Chaco domain sites clustered
with SDTFs from the Cerrado domain and vice versa in
Fig. 2). In fact, the chaco woodlands from our dataset shared
82% of their species with at least one of the SDTF nuclei of
the Dry Diagonal. These results, combined with the finding
that many sites considered as SDTF experience significant
frost (e.g. in the Piedmont and Misiones nuclei), suggest that
the chaco woodlands may not be abruptly distinct from
SDTFs and they may represent one extreme of a floristic gra-
dient driven primarily by temperature. In support of the idea
of chaco woodlands being at one end of an SDTF tempera-
ture gradient, a previous study (Sarkinen et al., 2011) found
that 8.4% of SDTF specialist species occurred in the Chaco
phytogeographical domain. However, further analyses of sites
across the Chaco domain and surrounding areas are needed
to address its distinctness versus similarity compared with
(a) (b)
(c) (d)
(e) (f)
Figure 4
6Geographical variation in
seasonally dry tropical forest (SDTF) and
chaco woodlands characteristics across the
South American Dry Diagonal. The
northern group from the hierarchical
clustering analysis is discriminated by
triangles and the southern group by bullets
(see Figs 1 and 2 and Results). Values are
illustrated by the colour of the symbols:
warmer colours indicate higher values.
Numbers in the header for each panel
indicate minimum and maximum values.
(a) Scores from the first axis of the
detrended correspondence analysis (DCA)
analysis of species community composition.
The black dashed line delimits frost-free
(northern SDTFs) and frost-affected
(southern SDTFs and chaco woodlands)
areas; (b) scores from the second DCA axis;
(c) mean minimum temperature of the
coldest month; (d) precipitation of the
wettest month; (e) isothermality; and (f)
maximum soil water storage capacity.
LOW RESOLUTION COLOR FIG
Journal of Biogeography
ª2015 John Wiley & Sons Ltd
7
Environmental controls of floristic variation in dry forests
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
Neotropical SDTFs, in particular analyses that focus on the
potentially distinct edaphic conditions often present in the
Chaco phytogeographical domain (i.e. hypersaline soils)
(Prado, 1993a).
Affinities of the arboreal caatinga
Within northern SDTFs, Santos et al. (2012) classified the
arboreal caatinga (sites that occur along the border with
the Cerrado phytogeographical domain) as part of the Ca-
atinga nucleus. However, performing the hierarchical clus-
tering with a larger database we found that the arboreal
caatinga (shown as red within the orange Cerrado phyto-
geographical domain cluster in Fig. 2) is more similar to
SDTFs found in the Cerrado phytogeographical domain
(support value 0.9 in Fig. 2), which could be related to
edaphic similarities between arboreal caatinga and the the
Cerrado phytogeographical domain SDTFs (calcareous, rich
soils). The soil fertility (measured as a percentage of base
saturation), for instance, of the arboreal caatinga and Cer-
rado phytogeographical domain SDTFs from our database
were 57 2 (mean SD) and 46 13, respectively,
whereas other caatinga woodlands had poorer soils
(35 14).
Unexplained variation
While the environmental correlates of species composition
in SDTF were fairly clear and straightforward to interpret,
we still failed to explain a large fraction (74.5%) of the
variation in species composition. There are many poten-
tially important factors in determining the species composi-
tion of assemblages that we have not accounted for
adequately, such as: (1) ecological drift (cf. Hubbell, 2001)
driving stochastic rearrangements of species distribution
ranges through time; (2) biotic processes that were not
measured (e.g. competition and natural enemies); (3) false
absences in the surveys (e.g. researcher error and misiden-
tification of species); (4) unmeasured descriptors of habitat
quality (e.g. succession stage and degree of land degrada-
tion); (5) unsaturation, i.e. when species do not occupy all
patches of suitable habitat (Titeux et al., 2004); and (6)
other spatially and non-spatially structured environmental
factors that were not measured. Moreover, a high propor-
tion of unexplained variation, ranging from about 33% to
75% (e.g. Legendre et al., 2009; Oliveira-Filho et al., 2013;
reviewed in Soininen, 2014), is a common outcome in
studies of floristic composition over similar spatial scales,
as statistical noise is very high in analyses with species
presenceabsence data (ter Braak, 1986; Guisan et al.,
1999).
Concluding remarks
Our demonstration that the floristic composition of SDTFs
is highly correlated with temperature regime (variation in
species composition summarized by the first DCA axis in
Table 1) suggests that SDTF formations would have been
affected by Pleistocene climatic changes (cf. Prado & Gibbs,
1993; Pennington et al., 2000), assuming that species occu-
pied similar climatic niches in the Pleistocene as they do at
present. However, these climatic relationships may have had
more complex effects than producing a simple Pleistocene
arc of SDTF. Palaeoecological studies are beginning to reveal
this complexity (Behling & Lichte, 1997; Burbridge et al.,
2004; Whitney et al., 2011, 2013) and more such studies are
needed. Palaeodistribution modelling also holds promise for
understanding the past distribution of SDTFs (cf. Werneck
et al., 2011) but, given the high floristic variation of SDTFs
documented here, SDTFs clearly cannot be treated as a single
unit (as in Werneck et al., 2011). Rather, the distributions of
floristically homogeneous subunits of SDTFs and/or individ-
ual species should be modelled, both to understand the past
distribution of SDTF and its future under global climate
change.
Regarding the identity of SDTFs of the Dry Diagonal,
our results showed that the chaco woodlands are not as
floristically distinct from the other SDTFs of the Dry
Diagonal as has been assumed previously. Also, the second
DCA axis supports the view that these southern sites have
floristic similarities with various sites of the northern
group (i.e. the Misiones nucleus grouping with the Cerra-
do phytogeographical domain SDTFs and chaco woodlands
grouping with the Caatinga nucleus; Fig. 4b). Therefore,
we suggest that the identity of the chaco woodlands and
how they relate to surrounding SDTFs merits further
attention.
Finally, we highlight the necessity of revisiting conserva-
tion priorities for dry forests in Brazil. Our results demon-
strate the floristic similarity between the arboreal caatinga
and SDTFs from the Cerrado phytogeographical domain,
suggesting they have equal conservation merit in terms of
their flora. However, under the current Brazilian environ-
mental law the arboreal caatinga receives full protection,
whereas other SDTFs that are equally important, such as
the Cerrado domain SDTFs, are unprotected and highly
threatened.
ACKNOWLEDGEMENTS
D.M.N. thanks the Brazilian government funding agency
CAPES for supporting a 12-month study period at the Royal
Botanic Garden Edinburgh (grant BEX 2415/11-9), where
most of the ideas for this study were developed. D.M.N. and
A.O.F. received additional funding from Conselho Nacional
de Pesquisa Cient
ıficaBrazil (CNPq). D.M.N. and R.T.P.
were supported by the National Environmental Research
Council (grant NE/I028122/1). K.G.D. was funded by an
NSF International Research Fellowship (OISE-1103573) dur-
ing the time this research was completed. We are indebted
to Ole Vetaas and the anonymous referees for their valuable
contributions to the manuscript.
Journal of Biogeography
ª2015 John Wiley & Sons Ltd
8
D. M. Neves et al.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
REFERENCES
de Andrade-Lima, D. (1954) Contribution to the study of the
flora of Pernambuco, Brazil. MSc Thesis, State University
of New York, New York.
de Andrade-Lima, D. (1982) Present-day forest refuges in
north-eastern Brazil. Biological diversification in the tropics
(ed. by G.T. Prance), pp. 245251. Plenum Press, New
York.
Behling, H. & Lichte, M. (1997) Evidence of dry and cold
climatic conditions at glacial times in tropical southeastern
Brazil. Quaternary Research,48, 348358.
Bivand, R. (2012) spdep: spatial dependence. Weighting
schemes, statistics and models. R package version 0.551/
r458. Available at: http://R-Forge.R-project.org/projects/
spdep/.
Borcard, D., Gillet, F. & Legendre, P. (2011) Numerical ecol-
ogy with R. Springer, New York.
ter Braak, C.J.F. (1986) Canonical correspondence analysis: a
new eigen vector technique for multivariate direct gradient
analysis. Ecology,67, 11671179.
Bullock, S.H., Mooney, H.A. & Medina, E. (1995) Seasonally
dry tropical forests. Cambridge University Press, Cam-
bridge, UK.
Burbridge, R.E., Mayle, F.E. & Killeen, T.J. (2004) Fifty-thou-
sand-year vegetation and climate history of Noel Kempff
Mercado National Park, Bolivian Amazon. Quaternary
Research,61, 215230.
Caetano, S. & Naciri, Y. (2011) The biogeography of season-
ally dry tropical forest in South America. Seasonally dry
tropical forests: ecology and conservation (ed. by R. Dirzo,
H.S. Young, H.A. Mooney and G. Ceballos), pp. 2344.
Island Press, Washington, DC.
Castillo-Campos, G., Halffter, G. & Moreno, C.E. (2008) Pri-
mary and secondary vegetation patches as contributors to
floristic diversity in a tropical deciduous forest landscape.
Biodiversity and Conservation,17, 17011714.
CGIAR-CSI (2006) NASA Shuttle Radar Topographic Mission
(SRTM) (data set available as 3 arc-second DEMs). Avail-
able at: http://srtm.csi.cgiar.org/.
Collevatti, R.G., Terribile, L.C., Lima-Ribeiro, M.S., Nabout,
J.C., Oliveira, G., Rangel, T.F., Rabelo, S.G. & Diniz-Filho,
J.A.F. (2012) A coupled phylogeographic and species dis-
tribution modeling approach recovers the demographic
history of a Neotropical seasonally dry forest tree species.
Molecular Ecology,21, 58455863.
Collevatti, R.G., Lima-Ribeiro, M.S., Diniz-Filho, J.A.F.,
Oliveira, G., Dobrovolski, R. & Terribile, L.C. (2013a) Sta-
bility of Brazilian seasonally dry forests under climate
change: inferences for long-term conservation. American
Journal of Plant Sciences,4, 792805.
Collevatti, R.G., Terribile, L.C., Oliveira, G., Lima-Ribeiro,
M.S., Nabout, J.C., Rangel, T.F. & Diniz-Filho, J.A.F.
(2013b) Drawbacks to palaeodistribution modelling: the
case of South American seasonally dry forests. Journal of
Biogeography,40, 345358.
Dirzo, R., Young, H.S., Mooney, H.A. & Ceballos, G. (2011)
Seasonally dry tropical forests. Island Press, Washington,
DC.
Dray, S. (2010) spacemakeR: spatial modelling. R package ver-
sion 473 0.05/r101. Available at: http://R-Forge.R-pro-
ject.org/projects/sedar/.
Dray, S., Pelissier, R., Couteron, P., Fortin, M.-J., Legendre,
P., Peres-Neto, P.R., Bellier, E., Bivand, R., Blanchet, F.G.,
de Caceres, M., Dufour, A.-B., Heegaard, E., Jombart, T.,
Munoz, F., Oksanen, J., Thioulouse, J. & Wagner, H.H.
(2012) Community ecology in the age of multivariate mul-
tiscale spatial analysis. Ecological Monographs,82, 257275.
EMBRAPA & IBGE (2003) Mapa de solos do Brasil (escala
1:5.000.000). Centro nacional de pesquisas de solos da Em-
presa Brasileira de Pesquisa Agropecuaria and Instituto
Brasileiro de Geografia e Estatistica, Brasilia, Brazil.
Gebhardt, A. (2009) tripack: triangulation of irregularly spaced
data. R package version 1.34. Available at: http://
CRAN.R-project.org/package=tripack.
Gentry, A.H. (1995) Diversity and floristic composition of
neotropical dry forests. Seasonally dry tropical forests (ed.
by S.H. Bullock, H.A. Mooney and E. Medina), pp. 146
194. Cambridge University Press, Cambridge, UK.
Gillespie, T.W., Grijalva, A. & Farris, C.N. (2000) Diversity,
composition, and structure of tropical dry forests in Cen-
tral America. Polish Journal of Ecology,147,3747.
Guisan, A., Weiss, S.B. & Weiss, A.D. (1999) GLM versus
CCA spatial modeling of plant species distribution. Plant
Ecology,143, 107122.
Haffer, J. (1969) Speciation in Amazonian forest birds. Sci-
ence,165, 131137.
Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G. & Jarvis,
A. (2005) Very high resolution interpolated climate sur-
faces for global land areas. International Journal of Clima-
tology,25, 19651978.
Hubbell, S.P. (2001) The unified neutral theory of biodiversity
and biogeography. Princeton University Press, Princeton,
NJ.
IBGE (1993) Mapa de vegetac
ß
~
ao do Brasil. Fundacao Instituto
Brasileiro de Geografia e Estatistica, Rio de Janeiro, Brazil.
Jones, P. & Harris, I. (2008) CRU Time Series (TS): high res-
olution gridded datasets. NCAS British Atmospheric Data
Centre, London.
Legendre, P., Mi, X., Ren, H., Ma, K., Yu, M., Sun, I. & He,
F. (2009) Partitioning beta diversity in a subtropical
broad-leaved forest of China. Ecology,90, 663674.
Legendre, P., Borcard, D. & Roberts, D.W. (2012) Variation
partitioning involving orthogonal spatial eigenfunction
submodels. Ecology,93, 12341240.
Lep
s, J. &
Smilauer, J.P. (2003) Multivariate analysis of eco-
logical data using CANOCO. Cambridge University Press,
Cambridge, UK.
Lewin-Koh, N.J. & Bivand, R. (2012) maptools: tools for read-
ing and handling spatial objects. R package version 0.817/
r238. Available at: http://R-Forge.R project.org/projects/
maptools/.
Journal of Biogeography
ª2015 John Wiley & Sons Ltd
9
Environmental controls of floristic variation in dry forests
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
Linares-Palomino, R. (2006) Phytogeography and floristics
of seasonally dry tropical forests in Peru. Neotropical
savannas and seasonally dry forests: plant biodiversity, bio-
geography and conservation (ed. by R.T. Pennington, J.A.
Ratter and G.P. Lewis), pp. 249272. CRC Press, Boca
Raton, FL.
Linares-Palomino, R., Oliveira-Filho, A.T. & Pennington,
R.T. (2011) Neotropical seasonally dry forests: diversity,
endemism and biogeography of wood plants. Seasonally
dry tropical forests: ecology and conservation (ed. by R. Dir-
zo, H.S. Young, H.A. Mooney and G. Ceballos), pp. 321.
Island Press, Washington, DC.
Lott, E. & Atkinson, T. (2006) Mexican and Central Ameri-
can seasonally dry tropical forests: Chamela-Cuixmala,
Jalisco, as a focal point for comparison. Neotropical savan-
nas and seasonally dry forests: plant biodiversity, biogeogra-
phy and conservation (ed. by R.T. Pennington, J.A. Ratter
and G.P. Lewis), pp. 315342. CRC Press, Boca Raton, FL.
McCune, B. & Grace, J.B. (2002) Analysis of ecological com-
munities. MjM Software Design, Gleneden Beach, OR.
Mooney, H.A., Bullock, S.H. & Medina, E. (1995) Introduc-
tion. Seasonally dry tropical forests (ed. by S.H. Bullock,
H.A. Mooney and E. Medina), pp. 18. Cambridge Uni-
versity Press, Cambridge, UK.
Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., Minchin,
P.R., O’Hara, R.B., Simpson, G.L., Solymos, P., Stevens,
M.H.H. & Wagner, H. (2012) vegan: community ecology
package. R package version 2.03. Available at: http://
CRAN.R-project.org/package=vegan.
Oliveira-Filho, A.T., Jarenkow, J.A. & Rodal, M.J.N. (2006)
Floristic relationships of seasonally dry forests of eastern
South America based on tree species distribution patterns.
Neotropical savannas and seasonally dry forests: plant diver-
sity, biogeography and conservation (ed. by R.T. Penning-
ton, G.P. Lewis and J.A. Ratter), pp. 159192. CRC Press,
Boca Raton, FL.
Oliveira-Filho, A.T., Budke, J.C., Jarenkow, J.A., Eisenlohr,
P.V. & Neves, D.M. (2013) Delving into the variations in
tree species composition and richness across South Ameri-
can subtropical Atlantic and Pampean forests. Journal of
Plant Ecology,6,123.
Pennington, R.T., Prado, D.E. & Pendry, C.A. (2000) Neo-
tropical seasonally dry forests and Quaternary vegetation
changes. Journal of Biogeography,27, 261273.
Pennington, R.T., Ratter, J.A. & Lewis, G.P. (2006) An over-
view of the plant diversity, biogeography and conservation
of neotropical savannas and seasonally dry forests. Neo-
tropical savannas and seasonally dry forests: plant biodiver-
sity, biogeography and conservation (ed. by R.T.
Pennington, J.A. Ratter and G.P. Lewis), pp. 129. CRC
Press, Boca Raton, FL.
Pennington, R.T., Lavin, M. & Oliveira-Filho, A.T. (2009)
Woody plant diversity, evolution, and ecology in the tro-
pics: perspectives from seasonally dry tropical forests.
Annual Review of Ecology, Evolution, and Systematics,40,
437457.
Prado, D.E. (1993a) What is the Gran Chaco vegetation in
South America? I. A review. Contribution to the study of
flora and vegetation of the Chaco. V. Candollea,48, 145172.
Prado, D.E. (1993b) What is the Gran Chaco vegetation in
South America? II. A redefinition. Contribution to the
study of flora and vegetation of the Chaco. VII. Candollea,
48, 615629.
Prado, D.E. & Gibbs, P.E. (1993) Patterns of species distribu-
tion in the dry seasonal forests of South America. Annals
of the Missouri Botanical Garden,80, 902927.
Punyasena, S.W., Eshel, G. & McElwain, J.C. (2008) The
influence of climate on the spatial patterning of Neotropi-
cal plant families. Journal of Biogeography,35, 117130.
Queiroz, L.P. (2006) The Brazilian Caatinga: phytogeographi-
cal patterns inferred from distribution data of the Legumi-
nosae. Neotropical savannas and seasonally dry forests: plant
biodiversity, biogeography and conservation (ed. by R.T.
Pennington, J.A. Ratter and G.P. Lewis), pp. 121157.
CRC Press, Boca Ratan, FL.
Quinn, G.P. & Keough, M.J. (2002) Experimental design and
data analysis for biologists. Cambridge University Press,
Cambridge, UK.
R Core Team (2015) R: a language and environment for sta-
tistical computing, version 3.1.0. R Foundation for Statisti-
cal Computing, Vienna. Available at: http://www.R-
project.org/.
Ratter, J.A., Pott, A., Pott, V.J., da Cunha, C.N. & Haridasan,
M. (1988) Observations on woody vegetation types in the
Pantanal and at Corumba, Brazil. Notes of the Royal Botan-
ical Garden Edinburgh,45, 503525.
Santos, R.M., Oliveira-Filho, A.T., Eisenlohr, P.V., Queiroz,
L.P., Cardoso, D.B.O.S. & Rodal, M.J.N. (2012) Identity
and relationships of the arboreal Caatinga among other
floristic units of seasonally dry tropical forests (SDTFs) of
north-eastern and central Brazil. Ecology and Evolution,2,
409428.
Sarkinen, T., Iganci, J.R.V., Linares-Palomino, R., Simon,
M.F. & Prado, D.E. (2011) Forgotten forests: issues and
prospects in biome mapping using seasonally dry tropical
forests as a case study. BMC Ecology,11, 27.
Shimodaira, H. (2004) Approximately unbiased tests of
regions using multistep multiscale bootstrap resampling.
Annals of Statistics,32, 26162641.
Soininen, J. (2014) A quantitative analysis of species sorting
across organisms and ecosystems. Ecology,95, 32843292.
ter Steege, H., Pitman, N.C.A., Phillips, O.L., Chave, J., Saba-
tier, D., Duque, A., Molino, J., Prevost, M., Spichiger, R.,
Castellanos, H., von Hildebrand, P. & Vasquez, R. (2006)
Continental-scale patterns of canopy tree composition and
function across Amazonia. Nature,443, 444447.
Suzuki, R. & Shimodaira, H. (2011) pvclust: hierarchical clus-
tering with P-values via multiscale bootstrap resampling.R
package version 1.22. Available at: http://CRAN.Rpro-
ject.org/package=pvclust.
Titeux, N., Dufrene, M., Jacob, J.P., Paquay, M. & Defourny,
P. (2004) Multivariate analysis of a fine-scale breeding bird
Journal of Biogeography
ª2015 John Wiley & Sons Ltd
10
D. M. Neves et al.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
atlas using a geographical information system and canoni-
cal correspondence analysis: environmental and spatial
effects. Journal of Biogeography,31, 18411856.
Trejo, I. & Dirzo, R. (2002) Floristic diversity of Mexican
seasonally dry tropical forests. Biodiversity and Conserva-
tion,11, 20632084.
Vanzolini, P.E. (1963) Problemas faunisticos do Cerrado.
Simp
osio sobre o Cerrado (ed. by M. Ferri), pp. 305321.
Editora da Universidade de Sao Paulo, Sao Paulo, Brazil.
Vetaas, O.R. & Ferrer-Cast
an, D. (2008) Patterns of woody
plant species richness in the Iberian Peninsula: environ-
mental range and spatial scale. Journal of Biogeography,35,
18631878.
Walter, H. (1985) Vegetation of the earth and ecological sys-
tems of the geo-biosphere, 3rd edn. Springer-Verlag, Berlin.
Werneck, F.P., Costa, G.C., Colli, G.R., Prado, D.E. & Sites,
J.W., Jr (2011) Revisiting the historical distribution of Sea-
sonally Dry Tropical Forests: new insights based on palae-
odistribution modelling and palynological evidence. Global
Ecology and Biogeography,20, 272288.
Werneck, F.P., Nogueira, C., Colli, G.R., Sites, J.W., Jr & Costa,
G.C. (2012a) Climatic stability in the Brazilian Cerrado:
implications for biogeographical connections of South
American savannas, species richness and conservation in a
biodiversity hotspot. Journal of Biogeography,39, 16951706.
Werneck, F.P., Gamble, T., Colli, G.R., Rodrigues, M.T. &
Sites, J.W., Jr (2012b) Deep diversification and long term
persistence in the South American ‘dry diagonal’: integrat-
ing continent-wide phylogeography and distribution mod-
eling of geckos. Evolution,66, 30143034.
Whitney, B.S., Mayle, F.E., Punyasena, S.W., Fitzpatrick,
K.A., Burn, M.J., Guillen, R., Chavez, E., Mann, D., Pen-
nington, R.T. & Metcalfe, S.E. (2011) A 45 kyr palaeocli-
mate record from the lowland interior of tropical South
America. Palaeogeography, Palaeoclimatology, Palaeoecology,
307, 177192.
Whitney, B.S., Mayle, F.E., Burn, M.J., Guillen, R., Chavez,
E. & Pennington, R.T. (2013) Sensitivity of Bolivian sea-
sonally-dry tropical forest to precipitation and temperature
changes over glacial-interglacial timescales. Vegetation His-
tory and Archaeobotany,23,114.
Zomer, R.J., Bossio, D.A., Trabucco, A., Yuanjie, L., Gupta,
D.C. & Singh, V.P. (2007) Trees and water: smallholder
agroforestry on irrigated lands in northern India. IWMI
Research Report 122. International Water Management
Institute, Colombo, Sri Lanka.
Zomer, R.J., Trabucco, A., Bossio, D.A., van Straaten, O. &
Verchot, L.V. (2008) Climate change mitigation: a spatial
analysis of global land suitability for clean development
mechanism afforestation and reforestation. Agriculture Eco-
systems and Environment,126,6780.
DATA ACCESSIBILITY
A full description with details of data accessibility for Neo-
TropTree can be found at http://prof.icb.ufmg.br/treeatlan
BIOSKETCH
Danilo M. Neves is a Brazilian Research Fellow at the
Royal Botanic Garden Edinburgh, UK. He is interested in
tropical ecology and evolutionary biology of plants, with an
emphasis on niche evolution of Neotropical trees.
Author contributions: D.M.N., K.G.D. and A.O.F. designed
the paper; D.M.N., A.O.F. and M.L.B. assembled the data-
base; D.M.N. and K.G.D. analysed the data; A.O.F. and
M.L.B. commented on earlier versions of the manuscript;
D.M.N., K.G.D. and R.T.P. led the writing. All authors read
and approved the final manuscript.
Editor: Ole Vetaas
Journal of Biogeography
ª2015 John Wiley & Sons Ltd
11
Environmental controls of floristic variation in dry forests
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
... Fig. 2A) most of the Cerrado can be classified as humid (AI>0.65). This pattern also holds if only sites of dry formations are considered (i.e., the local climate of specific sites with xeromorphic vegetation; Neves et al. 2015). However, the Cerrado becomes arid to semiarid during the austral winter (Fig. 2C). ...
... Partially due to this lack of climatic identity, de Queiroz et al. (2017) argued that the term "Dry Diagonal" is conceptually equivocal to designate the eSADD. While most references cited above refer to the eSADD a composed of the Catinga, Cerrado and Chaco domains, Neves et al. (2015) also included the Pantanal, a seasonally flooded savanna formation located between Cerrado and Chaco (Junk and Nune da Cunha 2016) and the Chiquitania, a mosaic of savanna, savanna wetland and dry forests in eastern Bolivia (Killeen et al. 1990), in their assessment of the eSADD. ...
... Prado (2000) suggested that observed distribution patterns indicate that areas of the Cerrado had been inhabited by taxa currently distributed only in the Caatinga and Chaco domains in the recent geological past. Current absence of taxa in the Cerrado may thus be due to more humid conditions and more pronounced rainfall seasonality (Neves et al. 2015;see Figs. 2 and 3) that also favor the occurrence of fires (Miranda et al. 2009), but general climate of the area currently occupied by Cerrado appears to have been drier during Quaternary glacial periods (Collevatti et al. 2020, Oliveira et al. 2020). Examples of other species distributed across the eSADD are also available for anurans (Santos et al. 2009;Vasconcelos et al. 2019), geckos (Werneck et al. 2012), lizards (Fonseca et al. 2018;Ledo et al. 2020), birds (Rocha et al. 2020), and arthropods (Zanella 2011, Bartoleti et al. 2018, though many of these groups show connections to the neighboring domains, especially the Atlantic Forest. ...
... However, contrary to the effect on richness of IAT, species composition in urban areas was mostly affected by variables acting at broader scales (i.e., climate variables, PCNMs 1, 2, and 3). It suggest that climate determines composition of IAT in strongly modified ecosystems, as well as affecting native species composition in non-urban ecosystems (Neves et al. 2015;Rezende et al. 2018). Thus, urbanization can be considered as a filter at local scales (Aronson et al. 2016), affecting the richness of IAT from a regional pool of species previously filtered by climate variables acting at broader scales. ...
... Climate variables are vastly recognized as important drivers of exotic and native species distribution at broader scales (Field et al. 2009;Neves et al. 2015;Arruda et al. 2017;Rezende et al. 2018;Heringer et al. 2019a), while we found climate also affect the occurrence of IAT in urban areas. Species composition was grouped by climate zones, mainly differentiated by seasonality and precipitation, which are known important environmental filters for plants (Neves et al. 2015;Gastauer et al. 2017;Rezende et al. 2020). ...
... Climate variables are vastly recognized as important drivers of exotic and native species distribution at broader scales (Field et al. 2009;Neves et al. 2015;Arruda et al. 2017;Rezende et al. 2018;Heringer et al. 2019a), while we found climate also affect the occurrence of IAT in urban areas. Species composition was grouped by climate zones, mainly differentiated by seasonality and precipitation, which are known important environmental filters for plants (Neves et al. 2015;Gastauer et al. 2017;Rezende et al. 2020). For instance, high climate seasonality may prevent the establishment of IAT not adapted to temperature variation at a regional scale. ...
Article
Full-text available
Urban areas are strongly transformed by the human population and these environmental changes can affect the richness and composition of invasive alien trees (IAT) across cities. In this study, we investigate the role of urbanization and climate variables on richness and species composition of IAT in cities from Brazil. We investigated (i) the relative contribution and how variables related to urbanization affects the patterns of richness and composition; and (ii) the relative contribution and how variables related to climate affects richness and composition of IAT. Thus, we used occurrences of IAT in 93 urban areas in Brazil gathered from 130 references. We tested the variation partitioning using three models with variables related to urbanization, climate, and spatial correlation for richness and species composition of IAT. Next, we tested the significance of each set of variables on the richness and species composition of IAT. Urbanization variables explained 18% of the variation in species richness of IAT, whereas climate explained 18% of the variation in species composition of IAT. Demographic density (habitants per km2) positively affected the richness of IAT, whereas the percentage of urban forest influenced species composition. Köppen climate zones affected richness while the covariate temperature annual range had a negative effect on richness. Temperature annual range, annual precipitation, precipitation of driest quarter, and precipitation of warmest quarter affected the species composition of IAT. We confirmed urban variables have a relevant influence on species richness, but climate variables are still important to understand the composition of IAT across cities. Thus, species richness is affected at a local scale by urbanization, and species composition by climate at broader scales.
... Since the beginning of the twenty-first century, interest in the diversity and biogeography of NSDF has increased (Pennington et al. 2009;Powers et al. 2009;Williams-Linera and Lorea 2009;Linares-Palomino et al. 2011;Hernández-Stefanoni et al. 2012;Banda-R et al. 2016). The physiognomy and floristic composition of NSDF are related to environmental factors (such as temperature, precipitation, and edaphic conditions) at large spatial scales (Santos et al. 2012;Neves et al. 2015). There is also considerable inter-and intra-annual variability in rainfall within and among NSDF, including the Colombian Caribbean (Allen et al. 2017). ...
... Rainfall regime is a main driver of the current distribution and species composition of dry forests in the Caatinga, Brazil (Santos et al. 2012). Similarly, rainfall and other environmental factors (temperature and solar radiation) explained significant amounts of the variation in species composition in dry forests of the Caatinga and Chaco (Neves et al. 2015). At larger spatial scales, precipitation may be one of the most important factors driving species distributions, and thus the community structure, of NSDF (Mercado Gómez et al. 2020). ...
Article
Background Montes de María is the best-preserved tropical dry forest fragment in the Colombian Caribbean, making it a good model to relate environmental and geographic factors to woody plant community structure. Aims We related alpha and beta diversity of woody plant communities to geographic distance and bioclimatic factors to understand the underlying factors of community structure. Methods We compared species composition among seven sites and calculated alpha (using effective numbers of species) and beta diversity using Whitaker and Bray–Curtis dissimilarity index. Canonical correspondence analysis (CCA) and a Mantel test were used to quantify how community structure was related to environmental and/or geographic distance. Results We found that Montes de María is as diverse as other dry forest fragments in Colombia. We detected three groups of communities which were shaped mainly by turnover associated with both precipitation and geographic distance. Conclusions The high beta diversity of the dry forest of Montes de María is related to a mixture of environmental variation and geographic distance.
... This result is very similar (26%) to the meta-analysis conducted by Soininen (2014), which included organisms other than plants such as microorganisms, invertebrates and vertebrates. It also resembles the results found for plants in Baldeck et al. (2016;24%), Neves et al. (2015;21%), or Arellano et al. (2016b;20%). We found, however, wide ranges in the variation explained by soil (from 1 to 58%) or climatic variables (from 0.2 to 82%). ...
Article
Full-text available
A vast literature indicates that environment plays a paramount role in determining floristic composition in tropical forests. However, it remains unclear which are the most important environmental factors and their relative effect across different spatial scales, plant life forms or forest types. This study reviews the state of knowledge on the effect of soil and climate on floristic composition in tropical forests. From 137 publications, we collated information regarding: (1) spatial scale, continent, country, life form, and forest type; (2) proportion of variance in floristic composition explained by soil and climatic variables and how it varies across spatial scales; and (3) which soil and climate variables had a significant relationship on community composition for each life form and forest type. Most studies were conducted at landscape spatial scales (67%) and mainly in South America (74%), particularly in Brazil (40%). Studies majorly focused on trees (82%) and on lowland evergreen tropical forests (74%). Both soil and climate variables explained in average the same amount (14% each) of the variation observed in plant species composition, although soils appear to exert a stronger influence at smaller spatial scales while climate effect increases toward larger ones. Temperature, precipitation, seasonality, soil moisture, soil texture, aluminum, and base cations-calcium and magnesium-and their related variables (e.g., cation exchange capacity, or base saturation) were frequently reported as important variables in structuring plant communities. Yet there was variability when comparing different life forms or forest types, which renders clues about certain ecological peculiarities. We recommend the use of standardized protocols for collecting environmental and floristic information in as much as possible, and to fill knowledge gaps in certain geographic regions. These actions will be especially beneficial to share uniform data between researchers, conduct analysis at large spatial scales and get a better understanding of the link between soils and climate gradients and plant strategies, which is key to propose better conservation policies under the light of global change.
... A deeper investigation of the exact origin of the hybrids detected in this study would require additional analyses based on whole genome data in order to carry out a proper test of these two scenarios. Biogeographic History of Tillandsioideae and Opportunities for Hybridization Kessous et al. (2020) inferred a broad ancestral area for the core Tillandsioideae, spreading across the Atlantic Forest (AF), the Andes and the Chacoan dominion (i.e., the South American "Dry Diagonal, " Neves et al., 2015). The authors hypothesized that tectonic and climatic events during the Miocene (the formation of the Paranean Sea and the Dry Diagonal) likely caused the vicariance between the Andean lineages (Tillandsieae + Cipuropsidinae) and the Vrieseinae in the AF (Kessous et al., 2020). ...
Article
Full-text available
The tropics hold at least an order of magnitude greater plant diversity than the temperate zone, yet the reasons for this difference are still subject to debate. Much of tropical plant diversity is in highly speciose genera and understanding the drivers of such high species richness will help solve the tropical diversity enigma. Hybridization has recently been shown to underlie many adaptive radiations, but its role in the evolution of speciose tropical plant genera has received little attention. Here, we address this topic in the hyperdiverse Bromeliaceae genus Vriesea using genome skimming data covering the three genomic compartments. We find evidence for hybridization in ca. 11% of the species in our dataset, both within the genus and between Vriesea and other genera, which is commensurate with hybridization underlying the hyperdiversity of Vriesea, and potentially other genera in Tillandsioideae. While additional genomic research will be needed to further clarify the contribution of hybridization to the rapid diversification of Vriesea, our study provides an important first data point suggesting its importance to the evolution of tropical plant diversity.
... The most accepted hypothesis about the formation of these Atlantic Forest enclaves is that in the Pleistocene, climatic changes resulted in the advance of the tropical forest to the interior of the continent, establishing connections between the Amazon and the Atlantic Forest. After this period, the forest vegetation retreated to what is currently seen to be dominant, except for the locations that managed to preserve the old humidity pattern, which is mainly dominant in windward mountains and valleys with permanent sources of water (AB'SABER, 1985;PENNINGTON, PRADO and PENDRY, 2000;TABARELLI and SANTOS, 2004;SANTOS et al., 2007;NEVES et al., 2015;BASTOS et al., 2016;THOMÉ et al., 2016;LEDO and COLLI, 2017;LAVOR et al. , 2018). ...
Article
The National Forest Inventory (Inventário Florestal Nacional-IFN) is a large initiative that uses standardised methods to survey Brazilian forestry resources. One target of the IFN is the Cerrado, which contains one of the richest floras in the world. The aim of this study was to assess the contribution of the IFN to the knowledge of Cerrado woody flora. We analysed data from field-collected vouchers sampled by the IFN Cerrado. We restricted our analyses to IFN collections of native trees and shrubs, including palms, which were identified at the species level. Habitat of each collection was obtained by overlaying specimens’ geographic coordinates with land cover maps available in the Mapbiomas platform. Our final dataset comprised 28,602 specimens distributed in 2,779 sites (conglomerates) in Bahia, Distrito Federal, Goiás, Maranhão, Mato Grosso, Mato Grosso do Sul, Minas Gerais, Piauí, São Paulo and Tocantins. Collections were located in the following habitats: savannas (40.5%), forests (30.2%), anthropic areas (25.6%), grasslands (3.5%), and water (0.2%). We recorded 1,822 species belonging to 543 genera and 105 families, representing 34% of Cerrado woody species recorded on Flora do Brasil 2020. Fabaceae had the largest number of species, while Tapirira guianensis and Matayba guianensis were the most collected species. We highlight 60 potentially new records of occurrence for several states and 64 new records for the Cerrado, primarily in riparian forests where species from other biomes occur. In addition, 232 recorded species are Cerrado endemics, while 36 are cited in the CNCFlora’s red list as endangered. The systematic sampling carried out by the IFN enabled vegetation sampling in remote and poorly known areas, which expanded the geographic range of many woody species and contributed to the knowledge of plant diversity in the Cerrado.
Article
The Sapindales comprise one of the hallmark orders in the Neotropics, being an important component of forested and open formations, especially in Brazil. In recent decades, there have been many efforts made to unravel the Brazilian flora and its distribution; among them is the Brazilian Flora 2020 Project, the main goal of which was to produce updated monographs about Brazilian plant groups. In this work, we present a synthesis of the occurrence and distributions of species and genera of Sapindales in Brazil, noting species endemic to the country, as well as the prevalence of the order in Brazilian phytogeographic domains and vegetations. We have shown that Brazil has 1003 native species in 119 genera of six families of Sapindales, with 388 (43%) species endemic to the country, and the Rutaceae and Sapindaceae families being the richest in species, genera and endemics. Diverse species and life forms are found in virtually all vegetations and phytogeographic domains, but mostly in rainforests of the Amazonian and Atlantic Forest domains. Life forms vary greatly across different families, with a prevalence of lianas in Sapindaceae and of trees in Rutaceae. We discuss our results in light of the current patterns of the families and their assemblage in different vegetation and pytogeographic domains of Brazil.
Article
Full-text available
Invasive alien species are responsible for a high economic impact on many sectors worldwide. Nevertheless, there is a scarcity of studies assessing these impacts in Central and South America. Investigating costs of invasions is important to motivate and guide policy responses by increasing stakeholders’ awareness and identifying action priorities. Here, we used the InvaCost database to investigate (i) the geographical pattern of biological invasion costs across the region; (ii) the monetary expenditure across taxa and impacted sectors; and (iii) the taxa responsible for more than 50% of the costs (hyper-costly taxa) per impacted sector and type of costs. The total of reliable and observed costs reported for biological invasions in Central and South America was USD 102.5 billion between 1975 and 2020, but about 90% of the total costs were reported for only three countries (Brazil, Argentina and Colombia). Costs per species were associated with geographical regions (i.e., South America, Central America and Islands) and with the area of the countries in km ² . Most of the expenses were associated with damage costs (97.8%), whereas multiple sectors (77.4%), agriculture (15%) and public and social welfare (4.2%) were the most impacted sectors. Aedes spp. was the hyper-costly taxon for the terrestrial environment (costs of USD 25 billion) and water hyacinth ( Eichhornia crassipes ) was the hyper-costly taxon for the aquatic environment (USD 179.9 million). Six taxa were classified as hyper-costly for at least one impacted sector and two taxa for at least one type of cost. In conclusion, invasive alien species caused billions of dollars of economic burden in Central and South America, mainly in large countries of South America. Costs caused by invasive alien species were unevenly distributed across countries, impacted sectors, types of costs and taxa (hyper-costly taxa). These results suggest that impacted sectors should drive efforts to manage the species that are draining financial sources.
Article
Full-text available
ABSTRACT The tree flora of seasonally dry forests (SDTF) of eastern tropical and subtropical South America was investigated according to two main aspects: (a) the variations in floristic composition were analyzed in terms of geographical and climatic variables by performing multivariate analyses on 532 existing floristic checklists; and (b) the links among different seasonally dry forest formations, Amazonian forests and cerrados (woody savannas) were assessed. Analyses were performed at the species, genus and family levels. There was a strong spatial pattern in tree species distribution that only receded and allowed clearer climate-related patterns to arise when either the geographical range was restricted or data were treated at the genus and family levels. Consistent floristic differences occurred between rain and seasonal forests, although these were obscured by Strong regional similarities which made the two foresttypes from the same region closer to each other floristically than they were to their equivalents in different regions. Atlantic rain and seasonal forests were floristically closer to each other than to Amazonian rain forests but north-east rain and seasonal forests were both closer to Amazonian rain forests than each other, though only at the generic and familial levels. Atlantic seasonal forests also share a variable proportion of species with caatingas, cerrados and the chaco, and may represent a transition to these open formations. Increasing periods of water shortage, with increases in soil fertility and temperature are characteristic of a transition from semideciduous to deciduous forests and then to the semi-arid formations, either caatingas (tropical) or chaco forests (subtropical), while increasing fire frequency and decreasing soil fertility lead from seasonal forests to either cerrados (tropical) or southern campos (subtropical). The SDTF vegetation of eastern South America may be classified into three floristic nuclei: caatinga, chaco and Atlantic forest (sensu latissimo). Only the last, however, should be linked consistently to the residual Pleistocenic dry seasonal flora (RPDS). Caatinga and chaco represent the extremes of floristic dissimilarity among the three nuclei, also corresponding to the warm-dry and warm-cool climatic extremes, respectively. In contrast to the caatinga and chaco nuclei, the Atlantic SDTF nucleus is poor in endemic species and is actually a floristic bridge connecting the two drier nuclei to rain forests. Additionally, there are few grounds to recognize the Atlantic nucleus flora as a clearly distinct species assemblage, since there is a striking variation in species composition found throughout its wide geographical range. Nevertheless, there is a group of wide-range species that are found in most regions of the Atlantic nucleus, some of which are also part of the species blend of the Caatinga and Chaco floras, though the latter plays a much smaller part. We propose that it is precisely this small fraction of the Atlantic nucleus flora that should be identified with the RPDS vegetation.