Content uploaded by Valdir Marcos Stefenon
Author content
All content in this area was uploaded by Valdir Marcos Stefenon on Aug 02, 2015
Content may be subject to copyright.
Historical gene flow within and among populations of Luehea
divaricata in the Brazilian Pampa
Jordana Carolina Nagel •Denise Ester Ceconi •
Igor Poletto •Valdir Marcos Stefenon
Received: 2 January 2014 / Accepted: 20 February 2015 / Published online: 25 February 2015
ÓSpringer International Publishing Switzerland 2015
Abstract Within and among population gene flow is a
central aspect of the evolutionary history of ecosystems
and essential for the potential for adaptive evolution of
populations. We employed nuclear microsatellite markers
to assess inter- and intra-population gene flow in five nat-
ural populations of Luehea divaricata growing in the
Pampa biome, in southern Brazil. This species occurs in
practically all secondary forests of the Pampa and has
recognized ecological significance for these formations.
The genetic structuring of the studied populations suggests
limited gene dispersal among forest fragments, despite a
homogeneous level of migration among populations.
Notwithstanding the gene flow among populations, sig-
nificant SGS is still found in some fragments. Significant
spatial genetic structure within population was also found
likely as result of limited seed and/or pollen dispersal. The
scattered distribution of the populations and their relatively
high density seem to limit pollen dispersal. Also seed
dispersal by wind is not efficient due to large distances
among forest formations. As conservationist actions to-
wards preserving the genetic resources of L. divaricata and
the Brazilian Pampa, we suggest the protection of the ex-
isting forest formations and the maintenance of the natural
expansion of the forests over the grasslands in the biome.
Keywords Population genetics Spatial genetic
structure Pampa biome Gene dispersal
Introduction
A significant aspect of biodiversity and an important out-
come of the evolutionary history of ecosystems is the or-
ganization of genetic variation within and among
populations. The integration of information on historical
population processes, including the distribution of alleles
and genotypes, is important for the selection of priority
areas for conservation or other nature preservation strate-
gies (Moritz and Faith 1998). The multiplicity of local
genes and genotypes and genetic variation generated over
space and time may considerably increase the potential for
adaptive evolution (Heywood 1991). Gene flow is a critical
factor for the distribution of genetic variation, because high
gene flow tends to homogenize genetic structures, while
low gene flow allows non-random distribution of alleles
and genotypes. Thus, understanding the connectivity of
individuals within and among populations is a primary
aspect of population ecology and evolution.
The dominant view in population genetics is the
Wright’s island model, in which gene flow and genetic drift
are the main forces driving population genetic structure in
nature (Wright 1943,1946). This interpretation defines a
process of isolation by dispersal limitation (IBDL) that
generally leads to a pattern of isolation by distance (IBD).
Electronic supplementary material The online version of this
article (doi:10.1007/s10709-015-9830-9) contains supplementary
material, which is available to authorized users.
J. C. Nagel V. M. Stefenon (&)
Programa de Po
´s-Graduac¸a
˜o em Cie
ˆncias Biolo
´gicas,
Universidade Federal do Pampa, Av. Antonio Trilha 1847,
Sa
˜o Gabriel 97300-000, Brazil
e-mail: valdirstefenon@unipampa.edu.br
D. E. Ceconi
Programa de Po
´s-Graduac¸a
˜o em Engenharia Ambiental,
Universidade Federal de Santa Maria, Bairro Camobi,
Santa Maria 97105-900, Brazil
I. Poletto
Laborato
´rio de Protec¸a
˜o Florestal, Universidade Federal do
Pampa, Av. Antonio Trilha 1847, Sa
˜o Gabriel 97300-000, Brazil
123
Genetica (2015) 143:317–329
DOI 10.1007/s10709-015-9830-9
Recently, conceptual outlines have been added to the dis-
cussion, proposing local genetic adaptation as an important
driver of population genetic structure (e.g. Orsini et al.
2013). These frameworks take in consideration the process
of isolation by environment (IBE), comprised by the iso-
lation by adaptation [(IBA; effective gene flow is reduced
among habitats that show different ecological characteris-
tics; Nosil et al. (2008)] and the isolation by colonization
[(IBC; local genetic adaptation of early colonizing geno-
types results in a reduction in gene flow that promotes the
persistence of founder effects; Nosil et al. (2008)]. The
occurrence of one or another of these processes depends on
the species and on the environment. In addition, isolated or
combined factors may determine the occurrence and the
intensity of gene flow among and within populations.
The Brazilian Pampa is a particular environment located
within the south temperate zone between latitudes 28°000S
and 34°000S and longitudes 49°300W and 58°000W, pre-
senting both subtropical and temperate climates with four
well-characterized seasons (Roesch et al. 2009). This
biome comprises different physiographic formations
(IBGE 2004) with large prevalence of steppe and steppic-
savanna formations (regions which we call Pampa sensu
stricto, see Fig. 1). Forest formations can be found as
gallery forests, as patches of forest within the grassland and
in the transition areas between forest and steppe or steppic-
savanna formations. According to Cordeiro and Hasenack
(2009), the areas of transition in the Brazilian Pampa in-
creased about by 1100 km
2
from 1976 to 2002, as effect of
the expansion of the forest formations over grasslands.
These areas, characterized by a mosaic of grassy-shrub and
forest formations, cover about 6130 km
2
of the Brazilian
Pampa and reflect the natural expansion of the forest
vegetation over the grassy environments (for discussion on
this topic, see Lemos et al. 2014).
Given the fragmented distribution of the forest forma-
tions in the Brazilian Pampa, gene flow among populations
of tree species may be restricted. In addition, the fine-scale
genetic structure of such populations may be strong if
pollen and seed dispersal is limited within small sites,
promoting family structures and inbreeding. The genetic
characteristics of plant species populations (e.g. allelic
richness, heterozygosity, levels of inbreeding, population
differentiation) are highly dependent on reproductive sys-
tems (out-crossing vs. selfing, monoicy vs. dioeicy, seed
and pollen dispersal patterns) and geographical distribution
of populations/individuals (Nybom 2004). Thus, compre-
hending gene flow within and among populations of tree
species is crucial to understand the genetic structure and
the ecological dynamic of the Brazilian Pampa.
The central aim of this study was to evaluate the dis-
tribution of the genetic variation within and among natural
populations of Luehea divaricata Mart. & Zucc. (Mal-
vaceae) growing in the Brazilian Pampa, in order to en-
hance the knowledge about historical gene flow in this
species. Luehea divaricata is an early secondary tree spe-
cies native to the Brazilian Pampa, growing mainly along
rivers in riparian forests, but also found in forest nucle-
ations within the grasslands. This species can also be found
in other states of South and Southeast regions of Brazil, as
well as in Argentina, Uruguay and Paraguay (Carvalho
2003). It is a heliophyte species, tolerating shading and low
temperatures in the juvenile phase. This characteristics
make L. divaricata an important species for the establish-
ment and expansion of forest formations. Flowers of L.
divaricata are hermaphroditic, pollinated mainly by bees
and hummingbirds (Backes and Irgang 2002), whereas
seeds are wind-dispersed (Paoli 1995). Due to utilization of
its wood in the furniture industry and by farmers, L. di-
varicata has been widely exploited during the last decades.
Fig. 1 Location of the L. divaricata populations studied. aThe Pampa stricto sensu delimitation in Southern Brazil. The sampling area is
highlighted in the figure. bLocation of the five natural populations of L. divaricata plotted on the satellite image of the Pampa biome
318 Genetica (2015) 143:317–329
123
Although not classified as threatened species in official
documents, the expansion of agriculture, livestock and
urban growth in the Pampa biome may lead L. divaricata to
local threat.
The occurrence of L. divaricata in practically all sec-
ondary forests of the Brazilian Pampa and its ecological
significance for these formations make it a noteworthy
species for the study of genetic structure and gene flow in
forest environments of the Pampa. Because several natural
Pampean forest formations comprehend small to mid-sized
fragments that are geographically disconnected, we in-
tended to generate information on population genetics of L.
divaricata and to contribute to the discussion about con-
servation of the Brazilian Pampa, studying the genetic
dynamics of gene flow. Two main hypotheses were tested:
(1) populations of L. divaricata growing within the
Brazilian Pampa exhibit high levels of differentiation,
following an isolation-by-distance (IBD) model and (2)
natural isolation of the secondary forest formations in the
Brazilian Pampa resulted in high inbreeding and high
levels of fine-scale spatial genetic structure (SGS) in
populations of L. divaricata. Testing these hypotheses will
support a better understanding of the genetic consequences
of the current fragmentation of the Pampean forest for-
mations and the planning of conservation and management
strategies for this neglected biome.
Materials and methods
Plant material and DNA extraction
Plant material of L. divaricata was collected in five natural
populations (n =167) occurring in the Brazilian Pampa,
southern Brazil (Fig. 1). The studied area was selected in
order to represent the fragmentation of the forests, i.e.,
presenting small forest fragments, which may closely re-
flect the inter and intra-population gene flow in such dis-
jointed environments. Healthy leaves of at least 30 adult
individuals were collected in each population correspond-
ing to C90 % of the adult individuals in each population.
Populations Camboazinho (96 m asl, 54°030W and
30°200S), Canas (111 m asl, 54°120W and 30°200S), In-
hatinhum (166 m asl, 54°340W and 30°140S) and Cacequi
(105 m asl, 54°380W and 30°060S) grow within riparian
forests along rivers, while population BR290 (126 m asl,
54°330W and 30°150S) grows in an isolated patch along a
roadway, about 2200 m away from the Inhatinhum
population. The five populations are separated from one
another by grassland formations, croplands or urban areas.
Given the scarce occurrence of forest formations in the
Brazilian Pampa (Roesch et al. 2009), it is suggested that
these populations never formed a large continuous
population (Lemos et al. 2014) as in other Brazilian
Biomes like the Atlantic Rain Forest.
Healthy leaves were collected from each sample and
dried in silica gel. About 50 mg of plant material was
washed with 70 % ethanol and distilled water, disrupted in
collection microtubes using the Tissue Lyser (Qiagen
Ò
)
grinder with 3 mm tungsten beads, and total DNA was
extracted using the Invisorb Plant Mini Kit (Invitek
Ò
),
following the instructions of the manufacturer. Isolated
DNA was eluted in 100 mL of elution buffer and deposited
at -20 °C until use.
Microsatellite analysis
For the microsatellite analysis, genomic DNA was elec-
trophoretically separated and stained with GelRed
Ò
(Bi-
otium), checked under UV light and diluted to a
concentration of about 10 ng/lL. Ten microsatellite loci
developed by Ruas et al. (2009) were tested and genotypes
of all samples were scored at five putatively neutral nuclear
microsatellite loci (Ldiv31, Ldiv40, Ldiv48A, Ldiv55 and
Ldiv58), which revealed reliable banding patterns in the
gels. The PCR amplification of all loci was performed as
described by Ruas et al. (2009) in an Eppendorf Thermal
Cycler in 25 lL reactions, comprising 4 lL template DNA,
19PCR buffer, 2.5 mM MgCl
2
, 1.0 U Taq DNA poly-
merase (Invitrogen), 0.2 mM of each dNTP and 10 lMof
each primer. DNA fragments were separated on 10 %
polyacrylamide gels, stained with silver nitrate and pho-
tographed using a digital camera. Alleles were scored from
the digital pictures using the software TotalLab
TM
TL120
(Nonlinear Dynamics).
Analysis of populations’ genetic diversity
Patterns of populations’ genetic diversity were assessed as
effective number of alleles per locus (A
e
), expected and
observed heterozygosity (H
e
and H
o
, respectively) at each
single locus and overall loci, using the software GENALEX
6.5 (Peakall and Smouse 2006,2012).
Analysis of population structure
In order to closely investigate the population structure, a
Bayesian model-based clustering analysis (Pritchard et al.
2000) was implemented for the microsatellite data set. In
this analysis, individual multilocus genotypes are assigned
probabilistically to a defined number of Kclusters with
membership coefficients summing up to one across groups.
Bayesian analysis of population structure was performed
using the frequency independent alleles model with
500,000 Markov Chain Monte Carlo steps and 100,000
burn-in periods for both the admixture and the non-
Genetica (2015) 143:317–329 319
123
admixture models in the software STRUCTURE version 3.2
(Pritchard et al. 2000). The number of Kwas set from two
to twelve and 20 replicates were run for each K. The op-
timum number of clusters Kwas selected using the ap-
proach suggested by Evanno et al. (2005). This method is
based on the computation of DK, the second-order rate of
change of the likelihood function with respect to K, and is
assumed to be reliable when values of ln(X|K) increase
continually with the number of clusters (Evanno et al.
2005).
Additionally, the relationship among populations was
analyzed by means of a cluster analysis using the UPGMA
algorithm based on Nei’s DA genetic distance (Nei et al.
1983) estimated with the software POPULATIONS 1.2.28
(Langella 2002). Bootstrap values were obtained after
1,000 permutations over loci. The pairwise population
differentiation was also calculated as R
ST
taking differences
in allele size into account (Slatkin 1995) using the software
ARLEQUIN 3.01 (Excoffier et al. 2005).
A locus-by-locus non-hierarchical analysis of molecular
variance (AMOVA; Excoffier et al. 1992) as implemented in
ARLEQUIN 3.01 was applied to estimate among-population
differentiation (significance test by 10,000 permutations of
microsatellite genotypes) and within population differen-
tiation. Based on the results obtained from Nei’s DA ge-
netic distance and the Bayesian clustering analyses, a
hierarchical AMOVA analysis was performed, defining two a
posteriori groups of populations (Canas/Camboazinho and
Inhatinhum/Cacequi/BR290).
In order to further evaluate a pattern of IBD, the cor-
relation between genetic differentiation and geographical
distance among populations was evaluated by regressing
the population pairwise genetic differentiation matrix (R
ST
)
against the pairwise geographical distance matrix, using a
Mantel test with 10,000 permutations performed in the
software NTSYS-PC 2.0 (Rohlf 1998).
Characterization of the inter-population gene flow
Rates of migration were assessed using a Bayesian
framework based on coalescent theory (Beerli and
Felsenstein 1999) in the software MIGRATE 2.1.2 (Beerli
2004). Coalescent-based models take into account the
differences in population sizes, allowing the computation
of directional gene flow. The amount of gene flow was
estimated as Nm =M9h
i
/4, where M=m/h9lis the
scaled immigration rate and h
i
=49N
e
9lis the
population size of the recipient population conditional on
the underlying genealogy. It is important to note that the
unknown mutation rate (l) is absorbed into the parameters
hand M, which were initially generated from F
ST
-calcu-
lations. Computations were performed assuming constant
mutation rates for all loci, using a Brownian motion
mutation model. The Markov Chain Monte Carlo simula-
tions were run using 20 short chains (10,000 genealogies
sampled, 500 genealogies recorded per chain) and two long
chains (100,000 genealogies sampled, 5000 genealogies
recorded per chain). An adaptive ‘‘heating scheme’’ was
used to search for additional compatible genealogies using
four chains with start temperatures 1.0, 1.5, 3.0 and
100,000. A matrix with geographic distances between
population pairs was employed to evaluate the occurrence
of landscape barriers to gene flow. The correlation between
the number of immigrants and the geographic distance
between population pairs was evaluated using the Pear-
son’s correlation analysis.
Characterization of the intra-population gene flow
Intra-population gene flow was computed using the soft-
ware SPAGEDI 1.4 (Hardy and Vekemans 2002). The in-
breeding coefficient (f) was estimated for each population
as a kinship coefficient between gene copies within indi-
viduals. Statistical significance of fwas determined by
means of 10,000 permutations of gene copies among in-
dividuals, independently for each locus.
Fine-scale SGS was analyzed in each population using
the kinship coefficient (F
ij
) according to Loiselle et al.
(1995). The relationship of the genetic similarity and
geographic distance between individuals was computed for
each population as the regression slope (b
F
) of kinship
coefficients on log-transformed distances. The standard
errors were estimated using the jackknife method. Addi-
tionally, the Sp-statistic (Vekemans and Hardy 2004) was
computed for each population, based on the regression
slope of kinship coefficients as Sp =-b
F
/(1 -F
1
), where
F
1
is the mean kinship coefficient between individuals
belonging to the first distance class. This measure is ex-
pected to summarize the intensity of SGS, permitting a
quantitative comparison among species or populations
(Vekemans and Hardy 2004). The statistical significance of
F
1
and b
F
was determined through the upper and lower
bounds of the 95 % confidence interval of F
ij
defined after
10,000 permutations of individuals among locations.
Although Backes and Irgang (2002) consider L. di-
varicata as an outcrossing species, the hermaphrodite
flowers may support the occurrence of selfing. Assuming
that pollen dispersal is restricted to short distances, the
relative role of selfing to the overall level of inbreeding
may be estimated from SGS by comparing estimations of
the inbreeding coefficient fwith the kinship coefficient F
1
.
If f[F
1
, it is suggested that selfing is occurring. On the
other hand, if f\F
1
, mating among relatives might ex-
clusively explain the pattern of inbreeding (Vekemans and
Hardy 2004). In order to find evidence of selfing in the
studied populations of L. divaricata, estimations of fand F
1
320 Genetica (2015) 143:317–329
123
were compared in each population. In addition, the soft-
ware Kingroup (Konovalov et al. 2004) was employed to
test for each pair of individuals within each population
whether they were significantly (p\0.05) more likely to
be full-sibs or half-sibs versus unrelated. The relatedness
estimator of Queller and Goodnight (1989) was employed
in this analysis. Confidence levels for the pairwise likeli-
hood values were estimated through 1,000 simulations of
the alternative hypotheses (presence of full-sibs and pres-
ence of half-sibs) following Goodnight and Queller (1999).
Results
Levels of genetic diversity
Levels of gene diversity estimated for L. divaricata
(Table 4) ranged from H
e
=0.53 to H
e
=0.77, while the
observed heterozigosity ranged from H
O
=0.42 to
H
o
=0.57 and the allelic diversity ranged from A
e
=3.07
to A
e
=4.69. Locus, Ldiv40 revealed a particular trend,
with total lack of heterozygote individuals in all popula-
tions (H
o
=0.00 in all populations, Table 4) although at
least two alleles are found in each single population
(Table 4and Supplementary File 1).
Population structure
In the Bayesian analysis of population structure, the non-
admixture model revealed higher values of the log prob-
ability of data [L(K)] than the admixture model (Fig. 2a).
Since the results obtained from both models (admixture
and non-admixture) were quite similar, all data presented
here refer to the analyses performed using the non-ad-
mixture model.
Using the analysis of DK, microsatellite data revealed
higher population structure at K=2 (Fig. 2b), with the
geographically close populations Camboazinho and Canas
forming one group (99.1 and 78.7 % of membership in this
cluster, respectively) and the other populations corre-
sponding to the second cluster, with membership larger
than 99.4 % (Fig. 2c). A second peak at K=4 (Fig. 2b)
was also evident. For K=4, population Camboazinho
clearly shared membership in two clusters (clusters 1 and
2), population Canas in three clusters (clusters 1, 2 and 3)
and population Inhatinhum in two clusters (clusters 3 and
4). Populations Cacequi and BR290 revealed low levels of
admixture with 99 % membership in cluster 4 and 94 %
membership in cluster 3, respectively (Fig. 2d; Table 1).
The UPGMA dendrogram (Fig. 3) reflects a similar re-
lationship among populations as the Bayesian analysis for
K=2, although the bootstrap support was high for the
Camboazinho/Canas cluster (bootstrap =96 %) and only
moderate for the cluster comprising the other three
populations (bootstrap from 55 to 67 %).
In accordance with the Bayesian and the UPGMA ana-
lyses, the pairwise R
ST
values (Table 2), revealed the
lowest differentiation between Camboazinho and Canas
(R
ST
=0.046; p=0.02). The analysis of molecular vari-
ance (AMOVA, Table 3) revealed highly significant
(p\0.001) total population differentiation (F
ST
=0.22).
Assuming a posteriori groups of populations for K=2,
8.95 % (p\0.01) of the total variation was attributed
between groups (Camboazinho/Canas and Inhatinhum/
Cacequi/BR290) and 16.20 % (p\0.001) of the variation
was observed among populations within each group
(Table 3).
In terms of IBD among populations, the Mantel test
showed that the degree of genetic differentiation between
sample locations increased with geographical distance be-
tween populations (r=0.79; p=0.016).
Inter-population gene flow
Gene flow between pairs of populations (Fig. 4) revealed a
homogenous level of migration with a mean of Nm =19
immigrants/generation, ranging from Nm =16.4 immi-
grants/generation (from Camboazinho to BR290) to
Nm =21 immigrants/generation (from Camboazinho to
Canas). The Pearson correlation coefficient between Nm
and population pairwise geographic distance was negative
and significant (r=-0.50; p=0.02), suggesting moder-
ately lower gene flow with the increase of the geographic
distance.
Intra-population genetic structure (SGS)
From 19 to 35 alleles/population were found over all loci,
enabling a consistent estimation of SGS. Three out of the
five populations revealed high (f[0.21) and significant
(p\0.01) estimations of inbreeding coefficients overall
loci (Table 4). Cacequi and BR290 revealed low and non-
significant inferences for the inbreeding coefficient. Con-
sidering single locus analysis, significant estimations of
fwere observed in two populations for locus Ldiv30, in
three populations for locus Ldiv48 and in one population
for locus Ldiv58. Locus Ldiv40 was fixed (H
o
=0.0,
f=1.00) in all populations (Table 4).
Negative values were reported for the regression slope
of F
ij
(b
F
; Table 5) in all populations indicating that, on
average, individuals spatially close are genetically more
similar to each other than individuals separated by larger
distances. Indeed, a pattern of positive F
ij
at short distance
classes (\13 m in populations Canas, Inhatinhum and
BR290) and negative F
ij
at long distance classes ([60 m in
population Canas) is evident in all populations, where a
Genetica (2015) 143:317–329 321
123
near monotonic decrease of the mean kinship coefficient
with the increase of distance is observed (Fig. 5). The
average of F
ij
between individuals at the first distance class
(F
1
) ranged from 0.051 to 0.333 (Table 5). Significant SGS
was detected at short distances in all populations (Fig. 5).
Comparing the populations with significant SGS according
to Sp-statistic (Table 5), the strongest SGS was revealed by
populations Camboazinho and Inhatinhum. However, these
values were not directly correlated with the mean in-
breeding coefficient (f) of each population. For instance,
population Canas revealed an inbreeding coefficient flarg-
er than population Camboazinho. However, the latter
population revealed a much larger Sp value, due to the
higher kinship value found in the first distance class (F
1
)
and the strongly negative log slope (b
F
).
Evidence of selfing was found in populations Canas and
Inhatinhum, in which estimations of f[F
1
, whereas
populations Cacequi, Camboazinho and BR290 revealed
estimations of f\F
1
(Tables 4,5), suggesting occurrence
of mating among relatives. Accordingly, the analysis of
Fig. 2 Determination of the
population structure based on
Bayesian clustering analysis.
aEstimated values of the log
likelihood ln(X|K) as function of
the number of clusters (K)of
population structure for the non-
admixture and the admixture
models. bValues of the second-
order rate of change of ln(X|K)
as function of K. The modal
value of DKrepresents the true
number of populations or the
uppermost level of structure.
cCluster assignments of the 167
individuals of L. divaricata at
the population level for K=2.
dCluster assignments of the
167 individuals of L. divaricata
at the population level for
K=4
Table 1 Membership of the five populations of L. divaricata in each cluster of the Bayesian analysis of population structure considering
existence of two clusters (K=2) or four clusters (K=4)
K=2K=4
Cluster 1 Cluster 2 Cluster 1 Cluster 2 Cluster 3 Cluster 4
Camboazinho 0.991 0.009 0.590 0.406 0.004 0.000
Canas 0.787 0.213 0.069 0.716 0.213 0.001
Inhatinhum 0.001 0.999 0.000 0.002 0.331 0.666
Cacequi 0.006 0.994 0.008 0.000 0.002 0.990
BR290 0.001 0.999 0.001 0.001 0.941 0.057
The membership determines the proportion of individuals from each population that was designed to a particular cluster as function of their
genetic composition
322 Genetica (2015) 143:317–329
123
presence of full-sibs and half-sibs based on the compar-
isons of individual pairs in each population revealed
38.75 % of putatively full-sib pairs in populations Cacequi,
60.13 % in population Camboazinho, 65.37 % in
population BR290, 68.81 % in population Inhatinhum and
73.33 % in population Canas. Concerning putatively half-
sib pairs, the estimation was 26.07 % in populations
Cacequi, 48.36 % in population Camboazinho, 55.56 % in
population BR290, 45.95 % in population Inhatinhum and
56.13 % in population Canas.
Discussion
In this study we assessed gene flow within and among
populations of L. divaricata growing in the Brazilian
Pampa and found congruent results comparing the patterns
of gene flow among populations and within-population
(SGS). As a general trend, patterns of IBDL (Orsini et al.
2013) seem to be the main elements determining the ge-
netic structure of L. divaricata populations in the Brazilian
Pampa.
Inter-population gene flow
Concerning gene flow among populations, the Mantel test
revealed a significantly positive correlation between ge-
netic differentiation (pairwise R
ST
) and geographic dis-
tance, suggesting limitation to gene flow at longer
distances, i.e., a pattern of IBD. In addition, the correlation
between geographic distance and number of immigrants
(Nm) was negative, i.e., the farther the population pairs, the
smaller the migration between them, supporting a pattern
of IBD.
Fig. 3 UPGMA dendrogram based on Nei’s DA (Nei et al. 1983)
genetic distance between population pairs of L. divaricata. Values at
the nodes are bootstrap support for the clusters
Table 2 Pair-wise genetic differentiation based on R
ST
computations (bellow diagonal) and geographic distance in meters between populations
pairs (above diagonal)
Camboazinho Canas Inhatinhum Cacequi BR290
Camboazinho – 8513 45,501 56,167 43,399
Canas 0.046 – 38,310 49,775 36,224
Inhatinhum 0.347 0.217 – 16,084 2200
Cacequi 0.291 0.213 0.138 – 18,063
BR290 0.199 0.113 0.115 0.138 –
Table 3 Summary of the
analysis of molecular variance
(AMOVA) for all populations and
for a posteriori defined groups
a
Significance level after 10,000
permutations
*** p\0.001; ** p\0.01
Source of variation Variance components
a
Variation (%)
Non-hierarchical analysis
Among all populations 0.46*** 22.41
Within populations 1.61*** 77.59
Hierarchical analysis
Between a posteriori defined groups 0.19** 8.95
Among populations within groups 0.34*** 16.20
Within populations 1.61*** 74.85
Genetica (2015) 143:317–329 323
123
Alternatively, genetic differentiation could be related to
selective adaptation (IBA) or to founder effect (IBC). The
high genetic differentiation among Pampean populations of
L. divaricata (F
ST
=0.22) suggests that gene flow in the
Brazilian Pampa is more restricted than in the Atlantic
Forest, estimated as U
ST
=0.10 for RAPD markers (De
Carvalho et al. 2008) and F
ST
=0.06 for microsatellite
markers (Conson et al. 2013). The significant genetic dif-
ferentiation (U
ST
=0.10, p\0.001) and the differential
response of morphological adaptive traits observed in two
populations of L. diviricata growing in flooded and non-
flooded environments in the Atlantic Rain Forest (De
Carvalho et al. 2008) may represent the IBA model.
However, an analysis of correlation between traits under
potentially divergent selection (Q
ST
) and ecological dis-
tance is needed to confirm this hypothesis. Considering the
high similarity of the environments where the five
evaluated Pampean populations of L. divaricata grow and
the expected neutrality of the nuclear microsatellite
markers, the IBA model should not be primarily evoked as
an alternative to IBD.
On the other hand, patterns of founder effect (IBC) may
be responsible by some characteristics observed in the
Pampean populations studied here. The kingroup analysis
suggests that most individuals are either full- or half-sibs,
which can be an indicative of founder effect. Similarly, the
homozygosity observed for locus Ldiv40 in all populations
might also be effect of a population founded with a few
homozygote individuals for this locus reproducing by
selfing or among relatives (e.g. Schwaegerle and Schaal
1979; Ladizinsky 1985). However, IBC patterns generated
by numerical and adaptive advantage of the first colonizers
(monopolization; De Meester et al. 2002) can be con-
founded with the patterns generated by dispersal limitation
(Orsini et al. 2013).
In general, isolation by adaptation and isolation by
colonization postulate that gene flow among natural
populations is reduced as a consequence of local genetic
adaptation. Patterns of IBA and IBC may exist in the
Brazilian Pampa and should be evaluated in further studies,
in order to detect the most important evolutionary forces
shaping the genetic structure of L. divaricata in this biome,
in addition to IBD.
Following only the IBD model, restricted pollination
and seed dispersion are the causes of the observed patterns
of genetic differentiation among populations. The main
pollinators of L. divaricata are bees and hummingbirds
(Backes and Irgang 2002). However, pollinator’s flight
between forest formations is hampered by grassland and
cropping areas in the Pampa, restricting large-scale dis-
persion. Even if habitat fragmentation does not change the
abundance of pollinator species, it can modify pollinator
foraging behavior and decrease rates of pollen movement
between populations (Sork et al. 1999). In such a case, the
pollinators tend to promote mating just within populations
or among geographically close populations. In the Atlantic
Fig. 4 Graphical representation
of the migration among
populations of L. divaricata
growing in the Brazilian Pampa.
The length of the arrows
represents the geographic
distance between population
pairs while the size of the circle
corresponds to the number of
sampled individuals in the
population. The numbers within
arrows are the number of
immigrants (through seeds or
pollen) from the source
population (gray dashed
circles). Population
Camboazinho was represented
as ‘‘Cam’’ and population
Inhatinhum as ‘‘Inh’’
324 Genetica (2015) 143:317–329
123
Forest biome, the presence of larger forest fragments al-
lows the foraging of the pollinators across extensive areas
and a most efficient transport of pollen among different
populations, reducing the genetic differentiation.
Based on estimations of F
ST
,G
ST
or U
ST
, (0.13, 0.17 and
0.25–0.29, respectively) wind is expected to be an efficient
seed disperser (Nybom and Bartish 2000; Nybom 2004).
Wall (2003) reported distances of seed dispersal through
Table 4 Estimations of the
inbreeding coefficient (f),
effective number of alleles (A
e
)
expected heterozygosity (H
e
)
and observed heterozygosity
(H
o
) in natural populations of L.
divaricata, computed for each
locus individually and for all
loci
*** p\0.001; ** p\0.01;
*p\0.05
ns not significant
Locus Camb Canas Inhatinhum Cacequi BR290
Ldiv31
f-0.600
ns
0.027
ns
0.315** 0.421*** 0.063
ns
A
e
2.11 3.72 4.56 2.90 4.25
H
e
0.55 0.74 0.80 0.67 0.78
H
o
0.86 0.72 0.56 0.39 0.74
Ldiv40
f1.000*** 1.000*** 1.000*** 1.000*** 1.000***
A
e
2.57 2.63 1.80 1.16 2.21
H
e
0.63 0.63 0.45 0.14 0.56
H
o
0.00 0.00 0.00 0.00 0.00
Ldiv48
f0.624*** 0.410*** 0.403** -0.667
ns
-0.329
ns
A
e
3.39 4.39 2.46 1.93 1.84
H
e
0.73 0.78 0.61 0.49 0.46
H
o
0.28 0.47 0.37 0.81 0.61
Ldiv55
f-0.121
ns
-0.056
ns
-0.333
ns
-0.167
ns
-0.364
ns
A
e
4.28 7.81 1.63 2.48 1.66
H
e
0.79 0.88 0.40 0.61 0.41
H
o
0.88 0.94 0.67 0.71 0.55
Ldiv58
f0.261* 0.095
ns
0.185
ns
-0.084
ns
-0.028
ns
A
e
3.64 4.89 4.87 3.52 5.08
H
e
0.75 0.81 0.81 0.73 0.82
H
o
0.56 0.73 0.67 0.79 0.84
All loci
f0.213*** 0.221*** 0.236*** -0.017
ns
0.096
ns
A
e
3.20 4.69 3.07 2.40 3.01
H
e
0.69 0.77 0.62 0.53 0.61
H
o
0.52 0.57 0.42 0.54 0.55
Table 5 Measures of the fine scale spatial genetic structure (SGS) in populations of L. divaricata
b
F
F
1
Sp Kingroup analysis
a
Full-sibs (%) Half-sibs (%)
Camboazinho -0.208 0.333 0.312 60.13 48.36
Canas -0.037 0.148 0.043 73.33 56.13
Inhatinhum -0.108 0.186 0.133 68.81 45.59
Cacequi -0.012 0.051 0.013 38.75 26.07
BR290 -0.014 0.188 0.017 65.37 55.56
b
F
log slope, F
1
kinship coefficient between individuals of the first distance class, Sp Sp-Statistic
a
Percentage of putatively full- and half-sibs in each population. See text for details
Genetica (2015) 143:317–329 325
123
wind ranging from 13.4 to 28.0 meters for four pine species
(Pinus contorta, P. ponderosa, P. jeffreyi and P. lamber-
tiana). These pine species present winged seeds sizing
from 1.51 to 3.86 cm in length. Considering the large
distance among the studied populations of L. divaricata
([2,200 m), although with much smaller seeds (sizing
about 7.8 mm in length and 3.2 mm in width; Paoli 1995),
wind dispersal should not be expected to be effective in this
environment, crossing such gaps among populations.
Despite the pattern of IBD revealed by the Bayesian
structure analysis and the UPGMA dendrogram based on
Nei’s DA genetic distance, the estimations of Nm are
similar in all directions and for all population pairs (mean
of Nm =19 immigrants/generation, ranging from
Nm =16.4 to Nm =21), suggesting absence of significant
geographic barriers to migration. Indeed, low distortion of
the terrain and large areas with grassland or croplands
characterize the study area within the Brazilian Pampa.
Fig. 5 Correlograms of kinship coefficient measures (F
ij
) plotted against the distance between trees. Filled symbols are significant at the 5 %
level. A regression curve and its fitting to the data (R
2
) were calculated for each correlogram
326 Genetica (2015) 143:317–329
123
Intra-population gene flow
Three out of the five studied populations significantly de-
viated from Hardy–Weinberg equilibrium, showing posi-
tive estimations of inbreeding coefficient overall loci.
However, none of the populations showed significant
fvalues for all loci. Although the presence of null alleles
might not be discarded in our analysis (as reported by
Conson et al. 2013), this significant deviation from Hardy–
Weinberg equilibrium may be effect of limited seed and
pollen dispersal, generating family structure and elevated
levels of mating among relatives and even the occurrence
of selfing. The most intriguing fact is the fixation of locus
Ldiv40 in all populations. Ruas et al. (2009) and Conson
et al. (2013) reported only two alleles for this locus, but no
significant deviation from Hardy–Weinberg expectations.
For populations of L. divaricata from the Atlantic Rain-
forest, the significant inbreeding coefficient reported was
attributed to the forest fragmentation (Conson et al. 2013).
Considering the disjointed nature of the Pampean popula-
tions of L. divaricata studied here, such significant deficit
of heterozygotes may also be determined by forest frag-
mentation, which increases the family structure within
populations.
By assuming that the fine-scale SGS within populations
depends on the spatial distribution of seed- and pollen-
mediated gene dispersal events around each parent (Hardy
et al. 2006), the SGS observed in all populations of L.
divaricata studied is also consistent with a pattern of IBD.
However, the Sp-statistics showed that the level of SGS
was low in three out of the five populations (Sp \0.043),
suggesting a more effective gene flow within these sites,
despite the general pattern of IBD. Populations Camboaz-
inho (Sp =0.165) and Inhatinhum (Sp =0.133) exhibit a
more pronounced family structure, implying a com-
paratively more restricted within population gene flow.
Since the estimations of SGS were performed using the
same markers (i.e. equivalent polymorphisms), the same
sampling strategy (covering from the small to the largest
distance within each site) and a significant number of
population pairs in each distance class ([15 pairs), marker-
dependent and sampling-dependent bias can be discarded
as factors affecting the analysis.
Divergent SGS between populations can arise from
ecological factors as differences in the efficiency of seed or
pollen dispersing agents and different population densities.
In addition, if SGS is not at drift–dispersal equilibrium,
divergence can also arise from historical factors (Hardy
et al. 2006). However, the observed non-random spatial
distribution of genotypes at a fine spatial scale in Pampean
populations of L. divaricata probably arises from local
pedigree structures as a result of limited seed dispersal. As
observed in the south Brazilian conifer species Araucaria
angustifolia (Stefenon et al. 2008), such pedigree structures
may diverge among populations as effect of their different
aptitude of seed and pollen dispersal.
Vekemans and Hardy (2004) summarized mean values
of Sp-statistics related to different breeding systems, pollen
and seed dispersal strategies and life forms. These authors
found a mean Sp =0.013 (SD =0.010) for outcrossing
species, Sp =0.010 (SD =0.009) for trees, Sp =0.017
(SD =0.014) for species pollinated through animals and
Sp =0.012 (SD =0.012) for species with wind dispersed
seeds. Although populations Canas, Cacequi and BR290
revealed a level of SGS larger than these values, the SGS
observed in populations Camboazinho (Sp =0.165) and
Inhatinhum (Sp =0.133) is significantly higher than the
mean values summarized for species with the same life
history traits. Actually, these values are similar to selfing
species (Sp =0.143, SD =0.079). Indeed, population In-
hatinhum revealed evidences of selfing in both employed
analyses. Although population Canas also revealed evi-
dences of selfing, the lower value of Sp supports the pos-
sibility of different ecological factors generate SGS.
Moreover, the high Sp value of population Camboazinho
reveals the high impact of mating among relatives in
causing SGS. Also the fixation of locus Ldiv40 may be
effect of selfing and strict mating among relatives. How-
ever, further studies are needed to clarify this fact.
It is possible that population density may play an im-
portant role in the fine-scale SGS of L. divaricata. Both
populations, Camboazinho and Inhatinhum grow within
somewhat species-rich riparian forests, with relative high
overall density (i.e. concerning all tree species in the
fragment). Although the Sp statistic is expected to be in-
versely proportional to the density under IBD, direct
measures of pollen dispersal have shown that pollinator
flight distances decrease when population density increases
(Vekemans and Hardy 2004).
Comparing the genetic characteristics of the Pampean
populations of L. divaricata with populations of this spe-
cies growing in the Atlantic Rain Forest (De Carvalho et al.
2008; Conson et al. 2013) and given the absence of sig-
nificant landscape obstacles in the Brazilian Pampa, the
main barrier to gene flow in these populations is likely
composed by biotic elements.
The large grassland areas and crop plantations among
forest formations in the Brazilian Pampa limit the foraging
area of pollinators (bees and hummingbirds), restricting
among populations gene flow. In the same way, each iso-
lated forest formation tends to increase population density,
concentrating the foraging area of the pollinators and de-
creasing gene flow distance within populations.
Concerning seed dispersion, although species with seeds
dispersed by wind tend to have an efficient distribution
through the landscape (Heuertz et al. 2003), L. divaricata
Genetica (2015) 143:317–329 327
123
populations seem to present a different pattern in the
Brazilian Pampa. When seed dispersion is more restricted
than pollen dispersion, kinship coefficients decrease faster
with the logarithm of the distance at short distances than at
large distances (Heuertz et al. 2003). The almost linear
regression of the kinship coefficients over short and long
distances observed for the five populations of L. divaricata
suggests similar distances for both seed and pollen dis-
persal, i.e., limited pollen and seed dispersion across the
landscape.
Remarks on species and biome conservation
Considering the conservation of the species genetic re-
sources, the particularities of the Brazilian Pampa have to
be taken into consideration. The Pampa is a quite neglected
biome (Overbeck et al. 2007; Roesch et al. 2009) and few
studies have discussed the ecological dynamic and genetic
structure of its forest species. Despite the grassland pre-
dominance in the physiognomy of the Pampa, this biome
presents climatic and edaphic conditions propitious for
forest development. However, human interference in the
environment and the resilience of the grassy formations
seem to be the main factors precluding forest expansion in
the Brazilian Pampa (Lemos et al. 2014).
By expanding croplands and grazing areas, the human
interference favors the existence of scattered forest for-
mations in the Brazilian Pampa and high genetic diver-
gence among populations. Despite this high genetic
divergence, Pampean populations of L. divaricata revealed
a high number of alleles (mean of 4.88 alleles/locus,
ranging from 3.8 in population Cacequi to 7.0 in population
Canas) in comparison to populations growing in the At-
lantic Rain Forest (mean of 5.00 alleles/locus, ranging from
4.78 to 5.22; Conson et al. 2013). Allelic richness is a
crucial factor for population adaptability in environments
under constant climatic and/or demographic changes. De-
spite the current high allelic richness of the L. divaricata
populations, the maintenance of their adaptability is not
assured, given the negative effects of genetic drift and in-
breeding in isolated populations. Hence, if we consider the
expansion of the forest formations over grasslands as the
natural demographic dynamic of the Pampa‘s vegetation,
conservation of L. divaricata genetic resources demands
protection of the existing forest formations and the main-
tenance of the natural expansion of the forests in the
Brazilian Pampa.
Historically, the Pampean forests have been retained as
fragmented formations due to the expansion of agricultural
and grazing areas (Roesch et al. 2009). Even considering
conservation initiatives, particular attention is given to
grasslands, whereas forests are neglected (e.g. Pillar 2003;
Overbeck et al. 2007). Luehea divaricata is a quite
important species in the secondary forest formations and
has been frequently recommended in reforestation pro-
grams of riparian forests in Brazil. Considering the typical
occurrence of this species in riparian forest formations in
the Brazilian Pampa, its conservation is essential at both
ecologically and economically levels. The species flooding
tolerance (De Carvalho et al. 2008) turns L. divaricata into
an interesting species for recuperation of degraded areas in
different environments. However, testing for environmen-
tal adaptation at larger scales with common gardens ex-
periments, and using quantitative genetic analyses may
help to understand the adaptability of this species across
diverging habitats.
Acknowledgments We would like to thank CNPq (Processes
471812/2011-0 and 474758/2012-5) and UNIPAMPA (PROPESQ
and PROPG) by the financial support.
References
Backes P, Irgang B (2002) A
´rvores do Sul. Guia de identificac¸a
˜oe
interesse Ecolo
´gico. As principais espe
´cies nativas sul-brasi-
leiras, Instituto Souza Cruz
Beerli P (2004) Migrate: documentation and program, part of
LAMARC Version 2.0. http://evolution.ge.washington.edu/
lamarc.html
Beerli P, Felsenstein J (1999) Maximum-likelihood estimation of
migration rates and effective population numbers in two
populations using a coalescent approach. Genetics 152:763–773
Carvalho PER (2003) Espe
´cies arbo
´reas brasileiras ed. Embrapa
Informac¸a
˜o Tecnolo
´gica, Brası
´lia, Brası
´lia
Conson ARO, Ruas EA, Vieira BG, Rodrigues LA, Costa BF,
Bianchini E, Prioli AJ, Ruas CF, Ruas PM (2013) Genetic
structure of the Atlantic Rainforest tree species Luehea di-
varicata (Malvaceae). Genetica 141:205–215
Cordeiro JLP, Hasenack H (2009) Cobertura vegetal atual do Rio
Grande do Sul. In: Pillar VP, Mu
¨ller SC, Castilhos ZMS, Jacques
AVA (eds) Campos Sulinos: conservac¸a
˜o e uso sustenta
´vel da
biodiversidade. Brası
´lia, MMA, pp 285–299
De Carvalho MCCG, Da Silva DCG, Ruas PM, Medri ME, Ruas EA,
Ruas CF (2008) Flooding tolerance and genetic diversity in
populations of Luehea divaricata. Biol Plant 52:771–774
De Meester L, Gomez A, Okamura B, Schwenk K (2002) The
monopolization hypothesis and the dispersalgene flow paradox
in aquatic organisms. Acta Oecol 23:121–135
Evanno G, Regnaut S, Goudet J (2005) Detecting the number of
clusters of individuals using the software Structure: a simulation
study. Mol Ecol 14:2611–2620
Excoffier L, Smouse PE, Quattro JM (1992) Analysis of molecular
variance inferred from metric distances among DNA haplotypes:
application to human mitochondrial DNA restriction data.
Genetics 131(479–4):91
Excoffier L, Laval G, Schneider S (2005) Arlequin ver. 3.0: an
integrated software package for population genetics data
analysis. Evol Bioinform Online 1:47–50
Goodnight KF, Queller DC (1999) Computer software for performing
likelihood tests of pedigree relationship using genetic markers.
Mol Ecol 8:1231–1234
Hardy OJ, Vekemans X (2002) SPAGeDi: a versatile computer
program to analyse spatial genetic structure at the individual or
population levels. Mol Ecol Notes 2:618–620
328 Genetica (2015) 143:317–329
123
Hardy OJ, Maggia L, Bandou E, Breyne P, Caron H, Chevallier M-H
et al (2006) Fine-scale genetic structure and gene dispersal
inferences in 10 Neotropical tree species. Mol Ecol 15:559–571
Heuertz M, Vekemans X, Hausman J-F, Palada M, Hardy OJ (2003)
Estimating seed versus pollen dispersal from spatial genetic
structure in the common ash. Mol Ecol 12:2483–2495
Heywood JS (1991) Spatial analysis of genetic variation in plant
populations. Annu Rev Ecol Syst 22:335–355
IBGE (2004) Mapa de Biomas do Brasil. http://www.ibge.gov.br.
Accessed 20 Dec 2013
Konovalov DA, Manning C, Henshaw MT (2004) KINGROUP: a
program for pedigree relationship reconstruction and kin group
assignments using genetic markers. Mol Ecol Notes 4:779–782
Ladizinsky G (1985) Founder effect in crop-plant evolution. Econ Bot
39:191–199
Langella O (2002) Populations (Version 1.2.28). Centre National de
la Recherche Scientifique, France
Lemos RPM, D’Oliveira CB, Rodrigues CR, Roesch LFW, Stefenon
VM (2014) Modeling distribution of Schinus molle L. in the
Brazilian Pampa: insights on vegetation dynamics and conser-
vation of the biome. Ann For Res. doi:10.15287/afr.2014.272
Loiselle BA, Sork VL, Nason J, Graham C (1995) Spatial genetic
structure of a tropical understory shrub, Psychotria officinalis
(Rubiaceae). Am J Bot 82:1420–1425
Moritz C, Faith DP (1998) Comparative phylogeography and the
identification of genetically divergent areas for conservation.
Mol Ecol 7:419–430
Nei M, Tajima F, Tateno Y (1983) Accuracy of estimated phyloge-
netic trees from molecular data. J Mol Evol 19:153–170
Nosil P, Egan SP, Funk DJ (2008) Heterogeneous genomic differen-
tiation between walking-stick ecotypes: ‘isolation by adaptation’
and multiple roles for divergent selection. Evolution 62:316–336
Nybom H (2004) Comparison of different nuclear DNA markers for
estimating intraspecific genetic diversity in plants. Mol Ecol
13:1143–1155
Nybom H, Bartish IV (2000) Effects of life history traits and sampling
strategies on genetic diversity estimates obtained with RAPD
markers in plants. Perspect Plant Ecol Evol Syst 3:93–114
Orsini L, Vanoverbeke J, Swillen I, Mergeay J, De Meester L (2013)
Drivers of population genetic differentiation in the wild:
isolation by dispersal limitation, isolation by adaptation and
isolation by colonization. Mol Ecol 22:5983–5999
Overbeck GE, Muller SC, Fidelis A, Pfadenhauer J, Pillar VD, Blanco
CC, Boldrini II, Both R, Forneck ED (2007) Brazil’s neglected
biome: the south Brazilian Campos. Perspect Plant Ecol Evol
Syst 9:101–116
Paoli AAS (1995) Morfologia e Desenvolvimento de Sementes e
Pla
ˆntulas de Luehea divaricata Mart. et Zucc. (Tiliaceae). Rev
Bras Sem 17:120–128
Peakall R, Smouse PE (2006) GENALEX 6: genetic analysis in
Excel. Population genetic software for teaching and research.
Mol Ecol Notes 6:288–295
Peakall R, Smouse PE (2012) GenAlEx 6.5: genetic analysis in Excel.
Population genetic software for teaching and research—an
update. Bioinformatics 28:2537–2539
Pillar VD (2003). Dina
ˆmica da expansa
˜o florestal em mosaicos de
floresta e campos no sul do Brasil [Dynamic of the forest
expansion in mosaics of forest and grasslands in South Brazil].
In: Claudino-Sales V (Ed.) Ecossistemas Brasileiros: Manejo e
Conservac¸a
˜o, Expressa
˜o Gra
´fica e Editora [a Brazilian publish-
er], Fortaleza [Brazil], pp. 209–216
Pritchard JK, Stephens M, Donnely P (2000) Inference of population
structure using multilocus genotype data. Genetics 155:945–959
Queller DC, Goodnight KF (1989) Estimating relatedness using
genetic markers. Evolution 43:258–275
Roesch LFW, Vieira FCB, Pereira VA, Schu
¨nemann AL, Teixeira IF,
Senna AJT, Stefenon VM (2009) The Brazilian Pampa: a fragile
biome. Diversity 1:182–198
Rohlf FJ (1998) NTSYS-pc: numerical taxonomy and multivariate
analysis system (version 2.0). State University of New York,
New York
Ruas EA, Conson ARO, Costa BF, Damasceno JO, Rodrigues LA,
Reck M, Vieira AOS, Ruas PM, Rua CF (2009) Isolation and
characterization of tem microsatellite loci for the tree species
Luehea divaricata Mart. (Malvaceae) and intergeneric transfer-
ability. Conserv Genet Resour 1:245–248
Schwaegerle KE, Schaal BA (1979) Genetic variability and founder
effect in the pitcher plant Sarracenia purpurea L. Evolution
33:1210–1218
Slatkin M (1995) A measure of population division based on
microsatellite allele frequencies. Genetics 139:457–462
Sork VL, Nason J, Campbell DR, Fernandez JF (1999) Landscape
approaches to historical and contemporary gene flow in plants.
Trends Ecol Evol 14:219–224
Stefenon VM, Gailing O, Finkeldey R (2008) The role of gene flow in
shaping genetic structures of the subtropical conifer species
Araucaria angustifolia. Plant Biol 10:356–364
Vekemans X, Hardy OJ (2004) New insights from fine-scale spatial
genetic structure analyses in plant populations. Mol Ecol
13:921–935
Wall SBV (2003) Effects of seed size of wind-dispersed pines (Pinus)
on secondary seed dispersal and the caching behavior of rodents.
Oikos 100:25–34
Wright S (1943) Isolation by distance. Genetics 28:139–156
Wright S (1946) Isolation by distance under diverse systems of
mating. Genetics 31:39–59
Genetica (2015) 143:317–329 329
123
