Content uploaded by Mario Mairal
Author content
All content in this area was uploaded by Mario Mairal on Aug 21, 2018
Content may be subject to copyright.
Annals of Botany XX: 1–13, 2018
doi: 10.1093/aob/mcy149,
© The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company.
All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
available online at www.academic.oup.com/aob
Unmasking cryptic biodiversity in polyploids: origin and diversication of
Aster amellus aggregate
MarioMairal1,2,3,*, MáriaŠurinová1,2, SílviaCastro4 and ZuzanaMünzbergová1,2
1Department of Botany, Faculty of Science, Charles University, 128 01 Prague, Czech Republic, 2Department of Population
Ecology, Czech Academy of Science, Zámek 1, 252 43 Průhonice, Czech Republic, 3Department of Botany and Zoology,
Stellenbosch University, Private Bag X1, Matieland, 7602, South Africa and 4Centre for Functional Ecology, Department of
Life Sciences of the University of Coimbra and Botanic Garden of the University of Coimbra, Calçada Martim de Freitas s/n,
3000–456 Coimbra, Portugal
*For correspondence. E-mail mariomairal@gmail.com
Received: 7 March 2018 Returned for revision: 16 April 2018 Editorial decision: 15 July 2018 Accepted: 18 July 2018
• Background and Aims The origin of different cytotypes by autopolyploidy may be an important mechanism in
plant diversication. Although cryptic autopolyploids probably comprise the largest fraction of overlooked plant
diversity, our knowledge of their origin and evolution is still rather limited. Here we study the presumed autopoly-
ploid aggregate of Aster amellus, which encompasses diploid and hexaploid cytotypes. Although the cytotypes of
A.amellus are not morphologically distinguishable, previous studies showed spatial segregation and limited gene
ow between them, which could result in different evolutionary trajectories for each cytotype.
• Methods We combine macroevolutionary, microevolutionary and niche modelling tools to disentangle the ori-
gin and the demographic history of the cytotypes, using chloroplast and nuclear markers in a dense population
sampling in central Europe.
• Key Results Our results revealed a segregation between diploid and hexaploid cytotypes in the nuclear gen-
ome, where each cytotype represents a monophyletic lineage probably homogenized by concerted evolution. In
contrast, the chloroplast genome showed intermixed connections between the cytotypes, which may correspond
to shared ancestral relationships. Phylogeny, demographic analyses and ecological niche modelling supported an
ongoing differentiation of the cytotypes, where the hexaploid cytotype is experiencing a demographic expansion
and niche differentiation with respect to its diploid relative.
• Conclusions The two cytotypes may be considered as two different lineages at the onset of their evolutionary
diversication. Polyploidization led to the occurrence of hexaploids, which expanded and changed their ecological
niche.
Key words: Cryptic diversity, autopolyploidy, cytotypes, diversication, ecological niche, reproductive isolation,
Asteraceae.
INTRODUCTION
Polyploidy or whole-genome duplication (WGD) is known as a
major mechanism of adaptation and speciation in evolutionary
history. Numerous WGD events have been detected in the last
500 million years in many eukaryotic taxa (Wendel, 2000; Va n
de Peer, 2017), being especially widespread in plants (Stebbins,
1970; Soltis and Soltis, 2000; Wendel, 2015) in contrast to most
groups of animals (Alix et al., 2017). It has in fact been estab-
lished that all owering plants have experienced episodes of
polyploidization (Masterson, 1994; Wood etal., 2009; Jiao etal.,
2011), with WGD driving plant evolution (Alix etal., 2017).
Polyploids are traditionally classied as either autopoly-
ploids, which arise within a single taxonomic species, or
allopolyploids, which are the product of interspecic hybridi-
zation. While allopolyploidy has been extensively studied
(Müntzing, 1932; Feldman and Levy, 2005; Catalán etal., 2012;
Barker etal., 2016), autopolyploidy has received little attention
in the past, as it was expected to be very rare and maladaptive
in natural populations (Stebbins, 1970; Grant, 1981; Arrigo and
Barker, 2012). Additionally, many autopolyploids have escaped
recognition because they are morphologically similar to their
diploid progenitors (Stebbins, 1947; Soltis etal., 2007, 2010;
Parisod et al., 2010; Husband et al., 2013). The sum of these
factors has suggested that the diversity of autopolyploids has
been underestimated (Soltis etal., 2007; Parisod etal., 2010).
Recently it has been shown that cryptic polyploids may com-
prise the largest fraction of overlooked plant diversity (approx.
51 000–61 000 cryptic polyploid species; Barker etal., 2016).
Most of these polyploids would correspond to autopolyploids,
which full most of the species concepts (biological, evolution-
ary, phylogenetic and apomorphic; Soltis et al., 2007; Barker
etal., 2016). Although available data show that autopolyploids
are more numerous than previously thought, our knowledge
of their origin and evolution is still very scarce (Barker etal.,
2016; Kolář etal., 2017; Van de Peer, 2017).
Autopolyploidy could be an important driver of plant evolution
and genetic differentiation (Ramsey etal., 2008; Parisod etal.,
2010; Soltis etal., 2016). However, the effects of autopolyploidy
Downloaded from https://academic.oup.com/aob/advance-article-abstract/doi/10.1093/aob/mcy149/5068832
by University of Stellenbosch user
on 18 August 2018
Mairal etal. — Cryptic diversity in polyploid Aster
2
on evolutionary divergence are barely known and detailed stud-
ies exploring cytotype origins and diversication are still nec-
essary (Kellogg, 2016; Van de Peer, 2017; Kolář etal., 2017).
Polyploid origin may be explained by two alternative models:
the single-origin model and the multiple-origin model. Under
the single-origin model, an n-ploid cytotype is expected to arise
only once from the diploid cytotype. This model also proposes
that the polyploid cytotype originated a long time ago to allow
its spread through its modern range (Halverson etal., 2008). In
contrast, under the multiple-origin model, it is expected that the
cytotypes arose independently several times in different popula-
tions. The relative importance of these two models for the origin
of autopolyploids is, however, notclear.
Here, we reconstruct the phylogenetic relationships, demo-
graphic history and ecological niche of diploid and presumed
autohexaploid cytotypes of Aster amellus, a plant species widely
distributed in central and eastern Europe (Münzbergová et al.,
2011). The genus Aster sensu stricto contains approx. 180 species,
typied by A.amellus and restricted to the Northern Hemisphere
of the Old World (Nesom, 1994). Aster amellus has been dened
as an aggregate grouping diploid and hexaploid cytotypes, with a
few non-fertile minority cytotypes found on very rare occasions
(see Mandáková and Münzbergová, 2006; Castro et al., 2012).
Although both cytotypes grow in close proximity, each natural
population is composed of reproductive plants with only one
ploidy level (based on ow cytometric analyses of >7000 indi-
viduals in 327 populations) and only a mixed population has been
found so far (see Strebersdorf population in Fig.1; Mandáková
and Münzbergová, 2006; Castro etal., 2012); a distribution known
as ‘mosaic parapatry’. Central European populations show a cyto-
geographical structure: while diploid populations are distributed
throughout most of the European area, hexaploid populations are
longitudinally restricted, appearing exclusively east of Germany
(Fig. 1; Castro et al., 2012). In addition, a large contact zone
including populations of both cytotypes exists across Poland, the
Czech Republic, Slovakia, Austria and Romania. Despite this dis-
tribution, fertile intermediate ploidy levels (tetraploids) have not
been detected so far. This may be attributed to their reproductive
isolation due to spatial segregation, strong post-pollination bar-
riers and, to a lesser extent, temporal and behavioural segrega-
tion (Castro etal., 2011, 2012). An additional study using seven
microsatellite loci (Münzbergová etal., 2013) showed no or very
limited gene ow between the cytotypes. Although this suggested
that both cytotypes could be evolving independently, the evolu-
tionary history of the group is not yetclear.
The extensive contact zone of A.amellus together with the
particular spatial segregation of the cytotypes and all the pre-
vious knowledge already accumulated on this system makes
A.amellus an ideal candidate to explore polyploid origin and
its subsequent evolutionary history. Moreover, if the cytotypes
within this aggregate show divergent histories (as suggested in
Münzbergová etal., 2013), it would offer us a unique opportun-
ity to investigate both the origin and the demographic trajecto-
ries of the different cytotypes. To our knowledge, no study has
tested, at the same time, the origin, divergence times, ecological
niche and demographic imprints of an autopolyploid complex.
This knowledge is important to gain insights into the connection
between autopolyploidy and diversication (Kellogg, 2016).
In this study, we reconstructed the intraspecic-level phylogeny
and phylogeography of the A.amellus aggregate using evidence
from the nuclear ribosomal (nrDNA) internal transcribed spacer
(ITS) region and non-coding chloroplast (pDNA) markers uti-
lizing a large sample of populations – covering most of the spe-
cies range. We combined Bayesian methods, population genetic
analyses, statistical parsimony and ecological niche modelling
tools to disentangle the phylogeographic distribution patterns of
the cytotypes of the A.amellus aggregate. Our aims were to: (1)
reconstruct the relationships and origin(s) of the diploid and hexa-
ploid cytotypes and (2) search for ongoing differentiation in dem-
ography and ecogeographical niche between the cytotypes.
MATERIALS AND METHODS
Cytotype distribution and evidence for autopolyploidy
Previous literature surveys and massive ow cytometric
screening of A. amellus populations across Europe revealed
that most populations present only one cytotype, either dip-
loid (2n=2x=18 chromosomes) or hexaploid (2n=6x=54),
rarely accompanied by minority cytotypes (e.g. triploids, tetra-
ploids, heptaploids and nonaploids) of which no breeding
adults have been found (Mandáková and Münzbergová, 2008;
Münzbergová etal., 2011; Castro etal., 2012).
Several lines of evidence, including data from cytology,
isozymes and morphology, suggest that the hexaploid cyto-
type of A. amellus is of autopolyploid origin (Mandáková and
Münzbergová, 2008; Castro et al., 2012; Münzbergová et al.,
2013). Specically, autopolyploidy is suggested by close mor-
phological resemblance between the diploids and hexaploids
(Mandáková and Münzbergová, 2008). In addition, allozyme
analyses showed that the two cytotypes possess similar arrays
of allozymes at all polymorphic loci and there was no evidence
for xed heterozygosity in the hexaploids [xed heterozygosity
is expected in allopolyploids, although it may be absent in dip-
loidized allopolyploids (Mandáková and Münzbergová, 2008)].
Besides, no signal of allopolyploidy has been detected in the
analyses of the karyotype of the species (Jarolímová V. et al.,
unpubl. res.). Similarly, Münzbergová etal. (2013) also supported
autopolyploidy by very similar microsatellite proles in the two
cytotypes, though they strongly differed in the frequencies of
these microsatellites. Although there are many methods allowing
conrmation of the allopolyploid origin of a species (e.g. Rosato
etal., 2008; Cuadrado etal., 2017), there are no clear-cut methods
allowing such a denitive conrmation for autopolyploids. This
makes proving autopolyploidy extremely difcult, and such def-
inite support is largely impossible in most systems (for an inter-
esting approach, see Catalán et al. 2006). Although, separately
none of our previous data provide unequivocal evidence of auto-
polyploidy in the system, all the evidence together indicate that
this is the most likely explanation of the patterns observed.
Taxon sampling and DNA sequencing
We used cytotype screening for 327 populations (inset in
Fig. 1) performed in our previous studies (Mandáková and
Münzbergová, 2006; Castro etal., 2012) to identify the popu-
lations for this study and to perform ecological niche model-
ling. Based on this knowledge, we collected leaf material for
102 individuals in several eld expeditions throughout central
Downloaded from https://academic.oup.com/aob/advance-article-abstract/doi/10.1093/aob/mcy149/5068832
by University of Stellenbosch user
on 18 August 2018
Mairal etal. — Cryptic diversity in polyploid Aster 3
Europe between 2008 and 2011, representing 72 populations
within the Aster amellus aggregate (Fig.1; Supplementary Data
TableS1) and covering the entire contact area of the cytotypes
and adjacent areas throughout central Europe. Note that while
the distribution range is wider than central Europe, samples
from the eastern part of the distribution range are not available
for this study. As we attempted to cover the contact zone com-
pletely, this does not limit the quality of our data set. Our previ-
ous knowledge, based on microsatellites (Münzbergová et al.,
2013 and an ongoing study) showed very low variation within
populations, and we thus used mainly one individual per popu-
lation. While we acknowledge that one individual per popula-
tion is a low number, the basic patterns detected using nuclear
markers correspond to those previously obtained with a large
sample using microsatellites (Münzbergová etal., 2013). This
suggest that this sampling did not lead to a great loss of infor-
mation. Aspecial sampling effort was made in the Strebersdorf
population (n=22), the only mixed-ploidy population that has
been detected so far (Castro etal., 2012). Additional samples of
a few other populations were also included (see Supplementary
Data TableS1). In all, 596 sequences (ITS + pDNA) for 102
individuals were generated for this study. Species information
and GenBank accession numbers for all the sequences are pro-
vided in Supplementary Data TableS1. DNA was extracted
using the DNeasy Plant Mini Kit (QIAGEN Inc., CA, USA)
following the manufacturer’s instructions from silica gel-dried
leaves obtained from the specimens.
Several species representing hierarchical levels of phyloge-
netic relationships in the order Asterales were selected as alter-
native outgroups in the phylogenetic and dating analyses: Fam.
Goodeniaceae (Goodenia, Scaveola and Verreauxia), Fam.
Calyceraceae (Acicarpa, Boopis, Calycera and Nastanthus)
and Fam. Asteraceae (Aster, Bellis, Calendula, Conyza,
Crinitina, Erigeron, Galatella, Grangea, Myriactis, Solidago
and Tripolium). For this, we downloaded 37 accessions from
GenBank (Supplementary Data TableS1).
A comprehensive pilot study was carried out to select vari-
able regions within the chloroplast genome. We rst tested 17
plastid markers (reported as highly variable in Dong et al.,
2012), of which we selected ve non-coding plastid regions:
atpI–atpH, rps16 3′ exon–trnK (UUU) 5′ exon, rpl32–trnL
and PetN–psbM, which exhibited high levels of genetic vari-
ation, and PsbE–petL with moderate variation (see Table 1).
Amplication of these regions was as follows: 10μL PCRs
Mixed population
152
28
49
16
23
1
38
17
18
19
622
21 46 44
42
20
45
41
29
32 33
53
3130
47 13
12
51 10 4
9
3
7
8
54 2
50
11
52
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
89
26
25
36
515 14
27
35
57
37
34
72
48
95
91–96
97–100
24
2× cytotypes
6x cytoypes
400 km
400 km
F.1. Geographical location of the 72 populations used in the phylogenetic analyses. Numbers correspond to the sampled populations, with codes given in
Supplementary Data TableS1. Pie charts are surrounded in blue for the diploids and in green for the hexaploids, while its circle size is proportional to population
sampling. The colour lling each pie chart shows the chloroplast haplotypes, corresponding to the haplotypes shown in Fig.4A. The star and the inserted box
represent the only mixed-ploidy population of Strebersdorf. The inset 327 populations with known cytotype used in this study, coloured in blue for the diploids
and in green for the hexaploids. The approximate distribution of A.amellus is shaded in purple in the inset. Maps have been modied from GeoMapApp (Ryan
etal., 2009; www.geoMapApp.org).
Downloaded from https://academic.oup.com/aob/advance-article-abstract/doi/10.1093/aob/mcy149/5068832
by University of Stellenbosch user
on 18 August 2018
Mairal etal. — Cryptic diversity in polyploid Aster
4
contained 5μL of QIAGEN Multiplex PCR Master Mix, 0.5μL
of each primer (10mmol L–1 each in initial volume), 3 μL of
double-distilled H2O (ddH2O) and 20ng of DNA dissolved in
1μL of ddH2O. Pre-denaturation for 15 min at 95°C was fol-
lowed by 37 cycles of 95°C/30s, 52°C/30s and 72°C/2min,
and a nal extension step at 72°C for 10min.
Additionally, we sequenced the nuclear marker, ITS. For
amplication, we used ITS4 primer and modied ITS5 primer
as described in Noyes and Rieseberg (1999). Amplication of
ITS regions was as follows: 30μL PCRs contained 15 μL of
QIAGEN Multiplex PCR Master Mix, 0.9μL of each primer
(10 mmol L–1 each in initial volume), 12.2 μL of ddH2O and
20 ng of DNA dissolved in 1μL of ddH2O. Pre-denaturation
for 15min at 95°C was followed by 27 cycles of 95°C/60s,
60°C/60s and 72°C/2min, and a nal extension step at 72°C for
10min. For each sample, both strands were directly sequenced.
Forty-eight individuals of A.amellus revealed double bands
and unreadable electrophoretograms. For these, we followed the
guidelines to obtain reliable ITS sequences in plants (Feliner
and Rosselló, 2007) and used a cloning strategy when neces-
sary to end up with consensual sequences. Cloning proced-
ure included excision of the PCR products from 1 % agarose
gels and purication with the Zymoclean Gel DNA Recovery
kit (Zymoresearch, Orange, CA, USA). These fragments were
cloned using the TOPO TA cloning kit (Invitrogen, Carlsbad,
CA, USA) following the manufacturer’s instructions, but down-
scaled to half-reactions. Twenty colonies per sample were trans-
ferred into 20μL of ddH2O, denaturated at 95 °C for 10 min
and used as templates for subsequent PCR amplications for
sequencing. Sequence variability within clones of one sample
ranged between 0.4 and 1.6 %.All the 20 sequences from the
cloning procedure were merged into one consensus sequence
per individual, which was used for all the phylogenetic analyses.
All reactions were run on Eppendorf Mastercycler Pro S
(Eppendorf, Hamburg, Germany). PCR products were puri-
ed using the QIAquick PCR purication kit (Qiagen, Hilden,
Germany) and sequenced (Seqme, Dobříš, Czech Republic).
The ve plastid DNA loci selected, together with the nuclear
ribosomal ITS, were successfully amplied using the primers
listed in Supplementary Data TableS2. Sequences were edited
in Geneious 10.1.3, and aligned using the ‘global alignment’
option with free end gaps, and manually adjusted when neces-
sary following alignment rules described by Kelchner (2000).
To address different objectives, we constructed three data
sets: (1) the ‘pDNA outgroups data set’ (n=21), which included
accessions of the concatenated pDNA sequences for the
Asterales outgroups plus several accessions of Aster (A.alpinus,
A.amellus, A.foliaceus, A.lavandulifolius and A.tongolensis);
(2) the ‘amellus ITS data set’ (n=102), which included at least
one ITS accession of each population within the A. amellus
aggregate plus one sequence of A.alpinus as outgroup [identied
as the sister species of A.amellus (Li etal., 2012)]; and (3) the
‘amellus pDNA data set’ (n=100), which included at least one
concatenated pDNA accession of each population within the
A.amellus aggregate, plus one sequence of A.alpinus.
Phylogenetic inference and divergence time estimation
Phylogenetic relationships were estimated for each marker sep-
arately using Bayesian inference, implemented in MrBayes 3.2.2
(Ronquist etal., 2012). Choice of substitution models was based
on the Akaike information criterion implemented in JModelTest
2.2 (Posada, 2008): the models selected for each region are speci-
ed in Supplementary Data TableS3. Two independent analyses
of four chains each were run for a minimum of 10 million gener-
ations, sampling every 1000th chain. Convergence was assessed
by monitoring cumulative split frequencies. After discarding the
rst 25 % of samples as burn-in, we pooled the remaining trees
to construct a 50 % majority rule consensus tree. Additionally,
maximum likelihood analyses were run in the software RAxML
(Stamatakis et al., 2008) using the online tool (http://embnet.
vital-it.ch/raxml-bb/). The ‘amellus ITS data set’ and the ‘amel-
lus pDNA data set’ were rooted using A. alpinus as outgroup.
Before concatenating the plastid genes into a combined data set,
we checked for topological congruence in the inferred relation-
ships by examining the Bayesian consensus trees and searching
for well-supported clades [posterior probability (PP) >0.95]. For
the nal analyses, we concatenated the ve plastid regions into
a combined pDNA matrix: the ‘amellus pDNA data set’. As the
‘amellus ITS data set’ did not support the same signal, we ana-
lysed it separately. The concatenated data matrix was analysed
under the GTR + G model, partitioned by gene and allowing
the overall mutation rate to differ among partitions. To support
further the results based on our single nuclear marker, we con-
structed a Neighbor–Joining dendogram using the nuclear micro-
satellite data set developed in our previous study (Münzbergová
etal., 2013). While the data have already been published, they
were not previously used for dendrogram construction. Here we
wanted to clarify whether the results based on the microsatellites
correspond to those based on theITS.
Simultaneous dating of plastid and nuclear genomes, which
might have very different evolutionary rates, may lead to poten-
tial artefacts (Wolfe et al., 1987). To avoid this and because
the ITS marker may be subjected to concerted evolution, we
performed the dating analysis only for the plastid data set. To
T1. Summary statistics of the chloroplast and nuclear regions analysed here for the Aster amellus markers (no outgroups)
Aster amellus markers rps16–trnKatpI–atpHpetN–psbMpsbE–petLrpl32–trnL ITS
Unaligned length (bp) 874–919 995–1042 617–618 400–623 775–792 540–641
Aligned length (bp) 919 1042 618 623 814 642
Constant sites 902 1021 610 622 800 626
Variable sites 17 21 8 1 14 16
Haplotype diversity 0.48 0.483 0.497 0.503 0.533 0.74
Nucleotide diversity 0.0034 0.0035 0.0038 0.0013 0.0039 0.00705
Fragment length is given in bp; alignment length includes the indels.
Downloaded from https://academic.oup.com/aob/advance-article-abstract/doi/10.1093/aob/mcy149/5068832
by University of Stellenbosch user
on 18 August 2018
Mairal etal. — Cryptic diversity in polyploid Aster 5
provide a temporal framework, lineage divergence times were
estimated using the Bayesian relaxed-clock models imple-
mented in BEAST v.1.7 (Drummond et al., 2012). We were
not interested in absolute time estimates but rather in obtain-
ing molecular evidence for testing our hypothesis. Two Markov
Chain Monte Carlos (MCMCs) were run for a minimum of 20
million generations. We used Tracer v.1.6 (Rambaut etal., 2013)
to monitor convergence and EES values (>200) for all parame-
ters, and TreeAnnotator v.1.7 (Rambaut and Drummond, 2013)
to construct a maximum clade credibility tree from the poster-
ior distribution after discarding 20 % of samples as burn-in.
There are no suitable fossils for Aster, so we relied on two
approaches to estimate lineage divergence times. While the two
approaches have their limitations (explained in detail in Mairal
etal., 2015a), they are the most suitable for dating in our sys-
tem. First, we used a standard ‘secondary calibration approach’
in which the more inclusive higher level data set (‘pDNA out-
groups data set’) was used to estimate divergence times within
the ingroup. We used a GTR + G model and a uniform prior for
the ucld.mean within values commonly observed in plant plas-
tid markers (10–6–10–1 substitutions per site Ma–1; Wolfe etal.,
1987) and a default exponential prior for the standard deviation
(s.d.). As calibration points, we used secondary age constraints
drawn from the most comprehensive fossil-rich, meta-calibrated
angiosperm phylogenetic tree reconstruction of Magallón
etal. (2015). Two nodes were calibrated using a normal prior:
the split between Goodeniaceae and the rest of the Asterales
{mean=57.05 Ma [high posterior density (HPD) 50.87–65.69],
s.d.=3}, and the split between Calyceracerae and Asteraceae
[mean= 47.34 Ma (HPD 47.69–53.83), s.d.= 1.5]. The diver-
gences between A.alpinus and A.amellus estimated in the pre-
vious analysis (1.87 Ma, HPD 0.09–6.2) were used to calibrate
the A. alpinus–A. amellus node in the ‘amellus pDNA data set’.
Secondly, because the root and stem nodes of the A.amellus
data set are both constrained with deep-time calibration events,
we used a nested dating approach, similar to that adopted in
previous studies (Pokorny et al., 2011; Mairal et al., 2015a),
in which the higher level data sets calibrated with external evi-
dence are used to constrain the molecular clock rate of the data
set containing population-level data. This approach allowed us
to avoid using ‘all-encompassing’ priors for the mean clock rate,
so that we could assign a branching tree prior for the outgroup
data set and a coalescent constant-size prior for the intraspecic
amellus’ data sets. Choice of model priors was based on Bayes
factor comparisons using the path sampling (PS) and stepping
stone (SS) sampling methods in BEAST (Baele etal., 2012).
Haplotype analyses and demographic history
The relationships among lineages were investigated through
haplotype network analyses, examining separately the plastid
and nuclear genomes. Genealogical relationships among hap-
lotypes were inferred via the statistical parsimony algorithm
(Templeton et al., 1992) implemented in TCS 1.21 (Clement
et al., 2000). The number of mutational steps resulting from
single substitutions among haplotypes was calculated with 95
% condence limits, and gaps were represented as missingdata.
As we detected strong assortative mating between the differ-
ent cytotypes of the A.amellus aggregate (see results here and in
Münzbergová etal., 2013), we performed demographic analyses
for each cytotype separately. We used different approaches to infer
the demographic processes within each cytotype. First of all, to test
for evidence of population expansion, we carried out a neutrality
test – Fu and Li’s tests (Fu and Li, 1993; Fu, 1996) and Tajima’s
D test (Tajima, 1989) – for each cytotype. We used the DNAsp
program, version 5.0 (Librado and Rozas, 2009), and assessed the
signicance level of each test [Fu’s FS (Fu, 1996) and Tajima’s
D (Tajima, 1989) and raggedness]) by generating 10 000 random
samples, using coalescent simulations (Hudson, 1990; Nordborg
etal., 2003) under the innite-site model. Secondly, we plotted the
mismatch distribution for each cytotype using the observed num-
ber of differences between all pairs of sequences with DNAsp. The
goodness of t of the observed mismatch distribution to the the-
oretical distribution under a constant population size model was
tested by generating 10 000 samples by coalescent simulations
between observed and expected mismatch distributions and rag-
gedness index (r) as test statistics (Harpending, 1994).
Ecological niche modelling
To understand whether the two cytotypes differ in their real-
ized niche, we modelled the present distribution of each cyto-
type. Our extensive sampling with ow cytometry allowed us
to use 327 records: 167 for the diploid cytotype and 160 for
the hexaploid cytotype (inset in Fig. 1; Supplementary Data
TableS4), covering the entire distribution range of the A.amel-
lus aggregate in central Europe (see inset in Fig.1), including
adjacent areas. This includes entire species range to the north,
south and west, while, to the east, our sampling only extends
to Moldavia. We combined the available occurrences for the
species within each lineage with a set of bioclimatic variables
available from the WorldClim database (www.worldclim.org;
Hijmans etal., 2005). To avoid the high level of correlation usu-
ally found when using many of these variables, we chose bio-
climatic variables based on prior ecological knowledge of our
species (e.g. Münzbergová et al., 2011) and that were not cor-
related (Pearson correlation coefcient r < 0.75). We chose:
BIO1=annual mean temperature, BIO5=maximum tempera-
ture of the warmest month, BIO6 = minimum temperature of
the coldest month, BIO12=annual precipitation, BIO13=pre-
cipitation of the wettest month and BIO14=precipitation of the
driest month. Pseudoabsences were generated by selecting 5000
random points across the modelled region. We used ensemble
modelling (a procedure integrating the results from multiple
modelling techniques) to generate our models. This procedure
integrates the results from multiple modelling techniques within
an ensemble framework to achieve more robust reconstructions
(Araújo and New, 2007). We used four modelling algorithms that
estimate species distribution using environmental variables and
species occurrences: generalized linear models (GLMs), general-
ized additive models (GAMs), general boosting method (GBM)
and random forests (RFs). These models were run in the R pack-
age ‘biomod2’ (Thuiller et al., 2013) and summarized using
additional R packages (R Core Team, 2014): ‘foreign’, ‘raster’
(Hijmans and van Etten, 2016), ‘SDMTools’ (VanDerWal etal.,
2011), ‘rms’ (Harrell, 2016), ‘gbm’ (Ridgeway, 2015), ‘gam’
(Hastie, 2016), ‘rJava’ (Urbanek, 2010), ‘dismo’ (Hijmans etal.,
2016) and ‘randomForest’ (Liaw and Wiener, 2002) with default
Downloaded from https://academic.oup.com/aob/advance-article-abstract/doi/10.1093/aob/mcy149/5068832
by University of Stellenbosch user
on 18 August 2018
Mairal etal. — Cryptic diversity in polyploid Aster
6
settings. We used repeated split sampling to evaluate the perfor-
mance of the models, successively splitting the data set into 70
% for calibration and 30 % for evaluation by measuring the area
under the curve (AUC). We quantied the performance of the
model using the true skill statistics (TSS; Allouche etal., 2006).
The nal ensemble model was obtained considering models with
AUC >0.8 and TSS >0.6.
Niche overlap between both cytotypes was quantied by
using the proportional similarity of the distribution using the
metric D (Schoener, 1970; Warren etal., 2010), an index used
in conjunction with SDMs, using ENMtools (Warren et al.,
2010), appropriate for intraspecic lineages that differ in their
geographical distribution (Broennimann etal., 2012). This met-
ric ranges from zero (no overlap) to 1 (complete overlap).
RESULTS
Phylogenetic relationships and moleculardating
The ‘amellus ITS data set’ consisted of 103 sequences of 642
nucleotides (102 individuals + A. alpinus), while the ‘amellus
pDNA data set’ consisted of 101 sequences of 3887 nucleotides
(100 individuals + A. alpinus) (Table 1). Nuclear and plastid
Bayesian phylogenetic reconstructions showed different topolo-
gies (Fig.2). Analyses of each plastid marker separately showed
polytomies and some structured subclades with varying levels of
support (Supplementary Data Fig.S1). The concatenated pDNA
showed a topology where both cytotypes were intermingled
within different clades (Fig.2A). The nuclear phylogeny clearly
separated the hexaploid individuals into one monophyletic clade
and the diploid individuals into another monophyletic clade,
except for one hexaploid individual (no.33 in Supplementary
Data Table S1) from the Vrutice Sad, Czech Republic popu-
lation, which was grouped within the diploid clade. To avoid
mistakes, we checked the unusual position of this individual
by resequencing it (conrming its position) and by sequencing
additional individuals in this populations, which tted within
the clade of hexaploids from other locations. Both cytotypes
of the mixed-ploidy population of Strebersdorf were also sepa-
rated into these two clades. The cytotype segregation was sup-
ported with a moderate clade support in the MrBayes analysis
(PP=88) and a strong support in the BEAST analysis (PP=1,
not included). ML analyses also separated the two cytotypes,
though with lower resolution (Fig.2B). The separation of the
0.6
1
0.93
10.9
1
99
95
99
92
100
100
100
Aster alpinus
A
Aster
alpinus
Aster amellus
hexaploid clade
(6x)
86
87
B
Aster amellus
diploid clade
(2x)
0.88
0.67
0.72
03_PL_2 01_CR_2
03_PL_2
04_SK_2
05_AU_2
08_PL_2
09_PL_2
10_SK_2
11_SK_2
12_SK_2
13_SK_2 14_AU_2
15_AU_2
16_AU_2
17_AU_2
18_AU_2
19_AU_2
20_AU_2
21_AU_2
22_AU_2
23_SN_2
24_GE_2
25_SW_2
26_SW_2
27_GE_2
28_CZ_2
29_CZ_2
31_CZ_2
32_CZ_2
33_CZ_6
34_FR_2
35_GE_2
36_SW_2
37_IT_2
38_CR_2
40_AU_2
57_GE_2
61_AU_2
64_AU_2
66_AU_2
72_CR_2
73_CR_2
84_SK_2
85_SN_2
86_SN_2
87_SN_2
91_CZ_2
92_CZ_2
93_CZ_2
94_CZ_2
95_CZ_2
143_AU_2
152_AU_2
02_PL_6
06_AU_6
30_CZ_6
39_AU_6
41_AU_6
42_AU_6
44_AU_6
45_AU_6
46_AU_6
47_SK_6
48_SK_6
49_SK_6
50_SK_6
51_SK_6
52_SK_6
53_CZ_6
54_PO_6
58_AU_6
59_AU_6
60_AU_6
62_AU_6
63_AU_6
65_AU_6
67_AU_6
68_AU_6
69_AU_6
71_AU_6
74_HU_6
75_HU_6
76_HU_6
77_HU_6
78_HU_6
79_SK_6
80_SK_6
81_SK_6
82_SK_6
83_SK_6
89_CZ_6
96_CZ_6
97_CZ_6
98_CZ_6
99_CZ_6
100_CZ_6
106_AU_6
108_AU_6
118_AU_6
144_AU_6
149_AU_6
151_AU_6
07_PL_2
04_SK_2
06_AU_6
07_PL_2
08_PL_2
09_PL_2
10_SK_2
11_SK_2
13_SK_2
18_AU_2
19_AU_2
21_AU_2
24_GE_2
26_SW_2
27_GE_2
29_CZ_2
31_CZ_2
34_FR_2
36_SW_2
38_CR_2
39_AU_6
40_AU_2
41_AU_6
42_AU_6
44_AU_6
45_AU_6
46_AU_6
48_SK_6
49_SK_6
50_SK_6
52_SK_6
53_CZ_6
54_PO_6
58_AU_6
59_AU_6
60_AU_6
61_AU_2
62_AU_6
63_AU_6
64_AU_2
65_AU_6
66_AU_2
68_AU_6
69_AU_6
71_AU_6
72_CR_2
73_CR_2
75_HU_6
77_HU_6
78_HU_6
79_SK_6
80_SK_6
82_SK_6
83_SK_6
84_SK_2
86_SN_2
87_SN_2
89_CZ_6
108_AU_6
118_AU_6
143_AU_2
144_AU_6
149_AU_6
151_AU_6
152_AU_2
106_AU_6
02_PL_6 16_AU_2
17_AU_2
20_AU_2
25_SW_2
28_CZ_2
30_CZ_6
32_CZ_2
33_CZ_6
47_SK_6
57_GE_2
81_SK_6
85_SN_2
91_CZ_2
92_CZ_2
93_CZ_2
94_CZ_2
96_CZ_6
97_CZ_6
98_CZ_6
99_CZ_6
100_CZ_6
95_CZ_2
01_CR_2
05_AU_2
14_AU_2
15_AU_2
22_AU_2
51_SK_6
12_SK_2
23_SN_2
35_GE_2
37_IT_2
74_HU_6
76_HU_6
F.2. Bayesian majority-rule consensus trees obtained by MrBayes from: (A) the concatenated A.amellus chloroplast data set (atpI–atpH, rps16–trnK, rpl32–
trnL, psbE–petL and petN–psbM); (B) the A.amellus nuclear ribosomal (ITS) data set. Numbers above branches indicate Bayesian credibility values (PP); num-
bers below branches indicate maximum-likelihood bootstrap support values. Depending on their cytotype, each individual is marked in blue (diploids) or green
(hexaploids). Codes for A.amellus populations correspond to those shown in Supplementary Data TableS1.
Downloaded from https://academic.oup.com/aob/advance-article-abstract/doi/10.1093/aob/mcy149/5068832
by University of Stellenbosch user
on 18 August 2018
Mairal etal. — Cryptic diversity in polyploid Aster 7
cytotypes based on ITS was partly supported by the Neighbor–
Joining tree constructed using previously obtained microsatel-
lite data. It showed lower genetic distances within diploids and
within hexaploids, although the clades did not precisely sort into
diploids and hexaploids (Supplementary Data Fig.S2).
The ‘standard’ (Supplementary Data Fig. S3) and ‘nested’
(Fig.3) plastid dating approaches were congruent with each other
and with the MrBayes analyses. Additionally, they showed over-
lapping condence intervals: the divergence of Goodeniaceae
and Calyceraceae was dated in the Paleocene [55.62 Ma in the
nested approach (Fig.3) vs. 56.14 Ma in the standard approach
(Supplementary Data Fig. S3)], while the divergence of
Calyceraceae and Asteraceae was dated in the Eocene (49.61 Ma
in nested vs. 51.62 Ma in standard). Aster alpinus and A.amel-
lus diverged in the Late Miocene (1.94 Ma in nested vs. 1.87
Ma in standard) and the rst divergence within the A.amellus
cytotypes was dated in the Pliocene (3.16 Ma, PP=1 in nested
vs. 3.24 Ma, PP=1 in standard). Results for all the dating analy-
ses implemented in BEAST are provided in Supplementary Data
TableS5. PS and SS selected the Yule model and the uncorrelated
log-normal distribution as the tree and clock model priors for all
the analyses (Supplementary Data TableS5).
Haplotype network distribution and demographic analyses
Among the 102 individuals sampled from the 72 populations,
we observed nine plastid DNA haplotypes (Hp1–Hp9 in Fig.4A;
geographically represented in Fig.1) and seven different nuclear
haplotypes (Hn1–Hn7 in Fig. 4B; geographically represented
in Supplementary Data Fig. S4). The plastid network (Fig.4A)
showed that the two dominant haplotypes grouped populations
including both cytotypes (see Hp1 and Hp8 in Fig.4A), and the
haplotypes did not show closer relationships among them, being
separated by a large number of nucleotide changes. In the nuclear
network (Fig.4B), each haplotype was composed exclusively of
either diploids or hexaploids. In addition, the haplotypes of each
cytotype showed closer relationships (fewer changes) among
them than with haplotypes of the other cytotype. Cytotypes
occurring sympatrically in the mixed-ploidy population were also
pooled into different haplotypes in the network. Four haplotypes
dominated and were widespread geographically (Supplementary
Data Fig.S4): two of the diploid cytotype (Hn1 and Hn2) and
two of the hexaploid cytotype (Hn6 andHn7).
Demographic analysis showed positive values of Fu’s and
Tajima’s tests (Table2), with signicant values for the diploid
55.62
49.61
47.72
35.52
12.44
9.5
5.55
18.27
3.45
24.5
14.86
13.04 5.62
3.47
11.47
7.93
9.67
1.94
1
1
1
1
1
0.58
0.99
1
0.55
1
1
1
0.58
0.73
1
3.16
1.35
0.92
0.33
0.31
0.97
0.31
1
0..87
0.57
0.8
1
1
1
AB
Plio-PleistoceneMioceneOligoceneEocenePaleocene
010.020.030.040.050.060.0
02.55.0
Pliocene Pleistocene
2x cytotypes
6x cytoypes
Aster alpinus
Aster amellus
Diaspasis
Lechenaultia
Calendula
Chrysanthemum
Callistephus
Bellis
Solidago
Conyza
Erigeron
Crinitina
Galatella
Tr ipolium
Calycera
Boopis
Nastanthus
Goodenia
Verreauxia
Age (Myr)
Age (Myr)
F.3. Bayesian majority-rule consensus trees obtained by the nested analyses of the linked data sets: (A) ‘pDNA outgroups data set’ and (B) ‘amellus pDNA
data set’. Stars indicate constrained nodes (see text for more details). Horizontal bars show the 95 % HPD condence intervals for the supported nodes. Numbers
above branches indicate mean ages, and numbers below branches indicate Bayesian PP.
Downloaded from https://academic.oup.com/aob/advance-article-abstract/doi/10.1093/aob/mcy149/5068832
by University of Stellenbosch user
on 18 August 2018
Mairal etal. — Cryptic diversity in polyploid Aster
8
individuals. The frequencies of pairwise differences in the mis-
match distribution analyses resulted in bimodal distributions
for the diploid individuals, and unimodal distributions with a
second small peak for the hexaploid individuals (Fig.5). The
bimodal distribution in diploids does not conform to distribu-
tions expected for a population under the sudden expansion
model and it might be interpreted as constant population size
(e.g. Schneider and Excofer, 1999; McMillen-Jackson and
Bert, 2003). The unimodal distribution in the hexaploids was
consistent with a recent demographic expansion, while the
second small peak showed signs of sub-structure (Rogers and
Harpending, 1992). The raggedness statistics derived from the
mismatch distribution were signicant (Table2).
Ecological niche modelling
Species distribution models indicated that both cytotypes
had high suitability values in the current contact zones (Fig.6),
where they grow together. However, overall, there were dif-
ferences in ranges of the cytotype, with the niche of the dip-
loids distributed to the west of the contact zone and the niche
of the hexaploid to the east (Fig. 6). Additionally, the niches
of the two cytotypes were not equivalent (Shoener’s D total
overlap=0.52, P=0.020). However, the relatively high value
of D might indicate some biological overlap of resource use
(Wallace, 1981). It is noteworthy that hexaploids showed a
greater potential to occupy new areas to the east, and partly also
to the west of their current distribution (see Fig.6B). The AUC
values were generally high (with values ranging between 0.80
and 0.96), suggesting that the models are consistent.
DISCUSSION
Can diploids and hexaploids be considered as distinct
evolutionary lineages?
The nuclear and plastid trees showed different phylogenetic
signals, which may indicate distinct evolutionary histories
(Fig.2). Our resolved ITS phylogeny supported an explicit phy-
logenetic hypothesis, separating diploid and hexaploid individ-
uals into two different monophyletic clades (Fig.2B). However,
the inferences obtained from the ITS region could be limited
due to additional difculties such as paralogy, concerted evolu-
tion or directional bias in the homogenization process (Buckler
etal., 1997; Eidesen etal., 2017). We can discard the possibility
of paralogy, because the polymorphism detected when sequenc-
ing the ITS did not show shared sequences between cytotypes
(Feliner and Roselló, 2007). The ITS may be subjected to con-
certed evolution leading to clear ITS differentiation of the two
clades (Fig.2B). Additionally, the phylogram method based on
genetic distances using simple sequence repeats (SSRs) showed
low support values, with diploids and hexaploids not clearly
sorting into distinct clades (Supplementary Data Fig.S2). This
is likely to be because microsatellites are widely distributed in
Hp4
Hp5
Hp6
Hp7
Hp2
Hp3
Hp1
Hp8
Hp
9
A
Hn1 Hn2
Hn4
Hn5 Hn6 Hn7
Hn3
2x cytotypes 6x cytotypes
B
F.4. Statistical parsimony networks inferred using the (A) chloroplast and (B) nuclear sequences by TCS. Black points on connecting lines indicate nucleotide
changes. The circle size of the pie charts is proportional to the frequency of haplotypes. Pie charts are surrounded in blue for the diploid individuals and in green
for the hexaploids. Each haplotype is shown in a different colour, where codes for the (A) chloroplast correspond to populations shown in Fig.1, while codes for
the (B) nucleus correspond to the populations shown in Supplementary Data Fig.S4.
Downloaded from https://academic.oup.com/aob/advance-article-abstract/doi/10.1093/aob/mcy149/5068832
by University of Stellenbosch user
on 18 August 2018
Mairal etal. — Cryptic diversity in polyploid Aster 9
the euchromatin (Schlötterer and Harr, 2001), and it is thus very
unlikely that a concerted evolution process would have homog-
enized all the alleles. Still, we can, however detect separate
genetic clusters in the microsatellite data due to assortative mat-
ing within cytotypes (Münzbergová et al., 2013). Altogether,
the results of the ITS and SSRs seem to indicate that the current
mating patterns are strongly assortative within cytotypes.
Conversely, the different topology shown by the plastid data
could arise due to processes such as hybridization, chloroplast
capture or deep coalescence. We can discard hybridization since
breeding barriers were detected in A. amellus and the inter-
mediate forms found previously are inviable (Mandáková and
Münzbergová, 2006; Castro et al., 2011, 2012). In terms of
chloroplast capture, we detected some identical sequences in
cytotypes growing sympatrically in the only mixed-ploidy popu-
lation detected in nature. Because individuals with intermediate
ploidy levels have rarely been found in the populations (but were
never fertile), one cannot completely dismiss the possibility of
punctual hybridization and backcrossing in sympatric locations,
though it seems unlikely. Thus, the discrepancy between the ITS
and chloroplast trees seems to be due to shared ancestral relation-
ships (deep coalescence events) or incomplete lineage sorting
(ILS) among the chlorotypes (Maddison and Knowles, 2006).
Chloroplast DNA is smaller and more conserved compared
with the nuclear genome, usually reecting deep evolution-
ary events (Zurawski, 1987; Patwardhan et al., 2014). In this
case, the deep pDNA evolutionary events are further supported
by the deep divergence found among the plastid cytotypes
(Fig. 3; Supplementary Data Fig S3). While ILS may occur
during divergences, gene ow decreases over time and even-
tually disappears (Rogers and Gibbs, 2014). These theoretical
expectations agree with our ndings: while pDNA patterns
seem to reect the ancient history of colonization of the seeds,
the nuclear genome does not show admixture, conrming the
absence of gene ow (pollen transport) between the cytotypes
(Münzbergová etal., 2013).
Morphological and cytological differences between the cyto-
types do not all appear at once, leading to conicts between the
different species concepts in the early stages of autopolyploid evo-
lution (De Queiroz, 2007). For example, in the case of A.amellus,
the phylogenetic species concept is debatable because of phenom-
ena such as concerted evolution and ILS. However, the cytotypes
seem to behave as separate species according to the biological spe-
cies concept as the cytotypes are reproductively isolated (Castro
et al., 2011). Regarding the ecological species concept (different
habitat), our evidence demonstrates that habitats of diploid and
hexaploid populations differ, with cytotypes showing some signs
of local adaptation (Mandáková and Münzbergová, 2006; Raabová
etal., 2008; Münzbergová etal., 2011)). However, the magnitude
of this differentiation was low, indicating that the two cytotypes
may occupy the same habitats in the contact zone, with broader
niche differentiation across all the geographical range (Fig.6). In
contrast, A.amellus does not satisfy the morphological species con-
cept, since both cytotypes are morphologically indistinguishable
T2. Results from the DNA polymorphism of plastid and nuclear haplotypes, neutrality test and mismatch raggedness for the
Aster amellus aggregate
DNA polymorphism Fu’s FS test Tajima’s D test Mismatch distribution
π θH (d) nFSDDistribution Raggedness
2x Cytotype 0.00377 0.00225 0.773 52 2.55** 2.12* Multimodal 0.016*
6x Cytotype 0.00223 0.00196 0.471 49 1.533 0.421 Unimodal 0.091*
The mismatch distribution analyses are shown in Fig.5.
Asterisks denote signicant differences: *P<0.05, **P<0.001.
010 20 30
Pairwise differences
0
0.05
0.1
0.15
0102
03
0
Pairwise differences
0
0.05
0.1
0.15
0.2
Exp
Obs
Exp
Obs
AB
F.5. Demographic analyses for the diploid (A) and hexaploid (B) cytotypes of Aster amellus showing the mismatch distribution. The observed (Obs) and
expected (Exp) mismatch distributions (left) show the frequencies of pairwise differences.
Downloaded from https://academic.oup.com/aob/advance-article-abstract/doi/10.1093/aob/mcy149/5068832
by University of Stellenbosch user
on 18 August 2018
Mairal etal. — Cryptic diversity in polyploid Aster
10
(Mandáková and Münzbergová, 2008). However, this is an intrinsic
feature of autopolyploids, which tend to be morphologically simi-
lar to their diploid progenitors (Soltis et al., 2007). This may be
because species divergence and phenotypic diversication are often
highly temporally detached from the WGD event (Robertson etal.,
2017). Overall, the reproductive isolation between the A.amellus
cytotypes seems to indicate an ongoing separation of the two line-
ages. These lineages can be viewed as two cryptic species (Soltis
etal., 2007) yet carrying the signatures of ancient relationships.
Multiple-origin vs. single-originmodel
A previous study of A. amellus using microsatellites
(Münzbergová etal., 2013) postulated that A.amellus cytotypes
probably had a single origin. However, our new evidence shows
the need to include chloroplast markers to unravel the origin of
cytotypes. The ancient divergences and topological relation-
ships reected by the pDNA suggest that the hexaploid cytotype
arose and was established several times from the diploid cytotype
(Figs2A and 3B). On the other hand, the parapatric mosaic of
A.amellus cytotypes makes it difcult to interpret the geographical
patterns clearly, and a single-origin model could also be expected
if the secondary contact occurred after Pleistocence range expan-
sion (Mandáková and Münzbergová, 2006). Differentiation
between these alternatives requires an accurate temporal frame-
work, which is not a straightforward task due to the difculties that
exist when dating polyploids (Doyle and Egan, 2010). However,
it is remarkable that our different dating approaches were con-
gruent and resulted in very similar age estimates (Supplementary
Data TableS5). Weneed, however, to emphasize that our dating
identied the point at which gene trees coalescent, which does not
necessarily coincide with the polyploidization event (Doyle and
Egan, 2010). Overall, dating estimates agree well with an ancient
origin of the genetic diversication in A. amellus, suggesting a
multiple-origin model for the hexaploid cytotypes. This is further
supported by the large number of undetected mutation events
separating the chlorotypes (gaps in Fig. 4A), which probably
corresponds to extinct ancestral haplotypes (Mairal etal., 2015b),
showing older relationships in this marker.
After the emergence of polyploid lineages, their subsequent
success requires reproductive isolation (Soltis etal., 2007). The
establishment of a new cytotype will only be possible when
intracytotype mating increases (Rieseberg and Willis, 2007;
Paun etal., 2009), or the new entity has an advantage (such as
increased asexual reproduction or selng) or disperses to other
localities to avoid minority cytotype exclusion. In A.amellus,
the hexaploid establishment has probably been supported by
reproductive isolation caused by strong intercytotype gametic
barriers together with some level of selng that enabled off-
spring production in the hexaploids (Castro et al., 2011) and
resulted in functional isolation between the ploidy levels
(Münzbergová etal., 2013).
Changes in population size and niche differentiation of the
cytotypes
The evolutionary advantages intrinsic to autopolyploidy may
provide higher reproductive success over the diploid entities
(Lewis, 1980; Levin, 1983). If this hypothesis is correct, the poly-
ploid cytotype should be more persistent, increasing the prob-
ability of detecting its demographic expansion. In this study, we
found signicant differences in the trajectories of the two A.amel-
lus cytotypes. On the one hand Fu’s and Tajima’s tests (Table2)
for the diploid individuals suggest either balancing selection or
a recent population decrease. On the other hand, our results for
hexaploids allowed us to accept the null hypothesis of recent
population expansion (Rogers and Harpending, 1992) (Table2;
Fig.5), supporting the idea that, after their origin, the hexaploids
settled in a growing number of habitats. This is further suggested
by the greater potential of the hexaploids to colonize new areas,
especially to the east [compare distribution (inset in Fig.1) with
Fig.6]. To the east, hexaploids clearly found open niches beyond
the limits of its diploid progenitor where they could easily estab-
lish (Levin, 1975; Mandáková and Münzbergová, 2006).
AB
510152025305 10 15 20 25 30
0.
8
0.
6
0.
4
0.
2
45
50
55
Diploid occurrences Hexaploid occurrences
F.6. Geographic projections of the climatic niche model of the (A) diploid and (B) hexaploid cytotypes of Aster amellus for current climate. The color scale
indicates habitat suitability.The area delimited geometrically in black corresponds to the contact area of the cytotypes. The x-axis shows the longitude co-
ordinates, while the y-axis shows the latitude.
Downloaded from https://academic.oup.com/aob/advance-article-abstract/doi/10.1093/aob/mcy149/5068832
by University of Stellenbosch user
on 18 August 2018
Mairal etal. — Cryptic diversity in polyploid Aster 11
Despite the niche differentiation detected across the whole
distribution range, the niche of both cytotypes seems to be
overlapping in the contact zone (Fig.6). However, a previous
study that performed a reciprocal transplant experiment among
neighbouring populations in the eld showed niche differen-
tiation between the two cytotypes and local adaptation within
each cytotype (Raabová et al., 2008). This may contribute to
the maintenance of single-ploidy populations of A. amellus
along the contactzone.
The only mixed-ploidy population detected (Strebersdorf)
could provide new insights into the trajectories of mixed ploidy
populations. This population was detected at a contact zone of
secondary origin, where the hexaploids show higher seed set and
seedling emergence (Castro etal., 2012). The hexaploids are thus
expected to displace the diploids by means of minority cytotype
exclusion, possibly also linked with direct competition. In line
with this expectation, reduction in the proportion of diploids in
this population has been conrmed in its resampling in 2017 (J.
Raabová, pers. comm.). Additionally, recent eld observations
have detected a second mixed-ploidy population where, interest-
ingly, cytotypes show clear niche differentiation: the hexaploids
have been detected growing in grasslands, while the diploids are
relegated to sub-optimal zones inside the forest (J. Raabová, pers.
comm.). This further supports the expansion of the hexaploids,
and subsequent displacement of diploid individuals.
Additionally, the accumulated evidence detected in A.amel-
lus further supports an increasing colonization potential of
hexaploids. For example: (1) hexaploids show wider ecological
amplitude (Mandáková and Münzbergová, 2006); (2) hexaploids
grow equally well with and without arbuscular mycorrhizal fun-
gal (AMF) symbiosis, while diploids grow signicantly better
only with AMF (Sudova et al., 2014); (3) hexaploids occur in
both low and high productive habitats, while diploids are conned
only to low productive habitats (Mandáková and Münzbergová,
2006; Münzbergová, 2007); and (4) the damage to seeds by her-
bivory decreases with habitat isolation in hexaploids, whereas no
such trend can be found for diploids (Münzbergová, 2006).
The greater genomic exibility acquired by polysomic inher-
itance provides polyploids with new evolutionary advantages
for colonization over their diploid relatives (Ramsey et al.,
2008; Parisod et al., 2010; Alix et al., 2017). While evolu-
tionary advantages of polyploids have been extensively docu-
mented for allopolyploids (Lewis, 1980; Soltis etal., 2014), in
autopolyploids the evolutionary consequences of these advan-
tages remain largely unknown (Parisod et al., 2010; Ramsey,
2011). Although our demographic analyses seem to point in
this direction, our data set is not powerful enough to test this
hypothesis, and further studies are necessary to test the possible
evolutionary advantages of autopolyploidy.
Conclusions
Autopolyploid evolution has been overlooked, underestimat-
ing its evolutionary implications (Soltis et al., 2007; Barker
et al., 2016). Most studies have focused on diploid–tetra-
ploid allopolyploids, where the effects of genome duplication
with interspecic hybridization may obscure the signature of
ancient relationships. In addition, identifying autopolyploid
lineages has been limited due to the need for a high sampling,
an insufcient differentiation in the markers among cytotypes
and a lack of taxonomic recognition (Kolář etal., 2017). The
analytical approach performed here provides clues about an
ongoing diversication process in a unique diploid–autohexa-
ploid aggregate, in which cytotypes are morphologically indis-
tinguishable. Here we demonstrate that the two cytotypes have
common evolutionary history and probably diverged due to
multiple polyploidization events. This knowledge is essential
to understand the cryptic diversity of morphologically identical
autopolyploids, and shows the importance of performing add-
itional studies on autopolyploid cytotypes, potentially recog-
nizable as different lineages. We suggest that the great cryptic
diversity masked by the autopolyploidy and its role as a source
of evolutionary advantages should be further evaluated.
SUPPLEMENTARYDATA
Supplementary data are available online at https://academic.
oup.com/aob and consist of the following. TableS1: voucher
information and GenBank accession numbers for all samples
included in this study. TableS2: primers used for PCR ampli-
cation and sequencing. TableS3: evolutionary model selected
with Jmodeltest. Table S4: geographical co-ordinates for the
diploid and hexaploid individuals used in the ecological niche
modelling of Aster amellus. TableS5: results for the analyses
implemented in BEAST to test accuracy of dating estimates.
Figure S1: Bayesian majority-rule consensus trees for each
chloroplast marker individually. FigureS2: Neighbor–Joining
tree with microsatellites. Figure S3: Bayesian majority-rule
consensus trees obtained by the standard dating analyses of the
linked data sets. FigureS4: haplotype nuclear distribution for
the 72 populations sampled for Aster amellus in central Europe.
ACKNOWLEDGEMENTS
We thank Andrea Jarošová for her help in the laboratory, and
Judith Fehrer for the help provided in the laboratory in the
initial stages of the project. We are grateful to the Population
Ecology Seminar Group of Průhonice (Czech Academy of
Science) for a critical review of the manuscript. We also thank
Isabel Marques, Myriam Heuertz, Pilar Catalán, Tomáš Fér and
two anonymous reviewers for comments to the manuscript,
and J.Raabová for providing the unpublished information on
the mixed-ploidy populations. This work was supported by
Grant Agency Czech Republic [grant no. P505-13-32048S];
Institutional research projects Academy of Science, Czech
Republic [grant no. 67985939]; Ministry of education, youth
and sports and Programa Operacional Potencial Humano –
Fundo Social Europeu/Portuguese Foundation for Science and
Technology [grant no. IF/01267/2013 to S.C.].
LITERATURECITED
Alix K, Gérard PR, Schwarzacher T, Heslop-Harrison JS. 2017. Polyploidy
and interspecic hybridization: partners for adaptation, speciation and
evolution in plants. Annals of Botany 120: 183–194.
Allouche O, Tsoar A, Kadmon R. 2006. Assessing the accuracy of species
distribution models: prevalence, kappa and the true skill statistic (TSS).
Journal of Applied Ecology 43: 1223–1232.
Araújo MB, New M. 2007. Ensemble forecasting of species distributions.
Trends in Ecology and Evolution 22: 42–47.
Downloaded from https://academic.oup.com/aob/advance-article-abstract/doi/10.1093/aob/mcy149/5068832
by University of Stellenbosch user
on 18 August 2018
Mairal etal. — Cryptic diversity in polyploid Aster
12
Arrigo N, Barker MS. 2012. Rarely successful polyploids and their legacy in
plant genomes. Current Opinion in Plant Biology 15: 140–146.
Baele G, Lemey P, Bedford T, Rambaut A, Suchard MA, Alekseyenko AV.
2012. Improving the accuracy of demographic and molecular clock model
comparison while accommodating phylogenetic uncertainty. Molecular
Biology and Evolution 29: 2157–2167.
Barker MS, Arrigo N, Baniaga AE, Li Z, Levin DA. 2016. On the relative
abundance of autopolyploids and allopolyploids. New Phytologist 210:
391–398.
Broennimann O, Fitzpatrick MC, Pearman PB, et al. 2012. Measuring
ecological niche overlap from occurrence and spatial environmental data.
Global Ecology and Biogeography 21: 481–497.
Buckler ES, Ippolito A, Holtsford TP. 1997. The evolution of ribosomal DNA
divergent paralogues and phylogenetic implications. Genetics 145: 821–832.
Castro S, Münzbergová Z, Raabová J, Loureiro J. 2011. Breeding barriers at
a diploid–hexaploid contact zone in Aster amellus. Evolutionary Ecology
25: 795–814.
Castro S, Loureiro J, Procházka T, Münzbergová Z. 2012. Cytotype distri-
bution at a diploid–hexaploid contact zone in Aster amellus (Asteraceae).
Annals of Botany 110: 1047–1055.
Catalán P, Segarra-Moragues JG, Palop-Esteban M, Moreno C, González-
Candelas F. 2006. A Bayesian approach for discriminating among alter-
native inheritance hypotheses in plant polyploids: the allotetraploid origin
of genus Borderea (Dioscoreaceae). Genetics 172: 1939–1953.
Catalán P, Müller J, Hasterok R, etal. 2012. Evolution and taxonomic split
of the model grass Brachypodium distachyon. Annals of Botany 109:
385–405.
Clement M, Posada D, Crandall KA. 2000. TCS: a computer program to
estimate gene genealogies. Molecular Ecology 9: 1657–1659.
Cuadrado Á, de Bustos A, Jouve N. 2017. On the allopolyploid origin and genome
structure of the closely related species Hordeum secalinum and Hordeum cap-
ense inferred by molecular karyotyping. Annals of Botany 120: 245–255.
De Queiroz K. 2007. Species concepts and species delimitation. Systematic
Biology 56: 879–886.
Dong W, Liu J, Yu J , Wang L, Zhou S. 2012. Highly variable chloroplast
markers for evaluating plant phylogeny at low taxonomic levels and for
DNA barcoding. PLoS One 7: e35071.
Doyle JJ, Egan AN. 2010. Dating the origins of polyploidy events. New
Phytologist 186: 73–85.
Drummond AJ, Suchard MA, Xie D, Rambaut A. 2012. Bayesian phyloge-
netics with BEAUti and the BEAST 1.7. Molecular Biology and Evolution
29: 1969–1973.
Eidesen PB, Alsos IG, Popp M, Stensrud Ø, Suda J, Brochmann C. 2017.
Nuclear vs. plastid data: complex Pleistocene history of a circumpolar key
species. Molecular Ecology 16: 3902–3925.
Feldman M, Levy AA. 2005. Allopolyploidy – a shaping force in the evolu-
tion of wheat genomes. Cytogenetic and Genome Research 109: 250–258.
Feliner GN, Rosselló JA. 2007. Better the devil you know? Guidelines for
insightful utilization of nrDNA ITS in species-level evolutionary studies
in plants. Molecular Phylogenetics and Evolution 44: 911–919.
Fu YX. 1996. New statistical tests of neutrality for DNA samples from a popu-
lation. Genetics 143: 557–570.
Fu YX, Li WH. 1993. Statistical tests of neutrality of mutations. Genetics 133:
693–709.
Grant V. 1981. Plant speciation. New York: Columbia University Press.
Halverson K, Heard SB, Nason JD, Stireman JO. 2008. Origins, distribu-
tion, and local co-occurrence of polyploid cytotypes in Solidago altissima
(Asteraceae). American Journal of Botany 95: 50–58.
Harrell JFE. 2016. rms: regression modeling strategies. R package version
3.4-0. https://cran.r-project.org/web/packages/ rms/rms.pdf
Harpending H. 1994. Signature of ancient population growth in a low-resolution
mitochondrial DNA mismatch distribution. Human Biology 66: 591–600.
Hastie T. 2016. gam: generalized additive models. R package version, 1.
https://cran.r-project.org/web/packages/gam/ gam.pdf.
Hijmans RJ, van Etten J. 2016. raster: geographic data analysis and mod-
eling. R package version, 2–15. https://cran.r-project.org/web/packages/
raster/ras- ter.pdf
Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A. 2005. Very high
resolution interpolated climate surfaces for global land areas. International
Journal of Climatology 25: 1965–1978.
Hijmans RJ, Phillips S, Leathwick J, Elith J. 2016. dismo: species distribu-
tion modeling. R package version 0.8–17. https://cran.r-project.org/web/
packages/dismo/dismo.pdf
Hudson RR. 1990. Gene genealogies and the coalescent process. Oxford
Surveys in Evolutionary Biology 7: 44.
Husband BC, Baldwin SJ, Suda J. 2013. The incidence of polyploidy in
natural plant populations: major patterns and evolutionary processes. In:
Greilhuber J, Dolezel J, Wendel J, eds. Plant genome diversity, Vol. 2.
Vienna: Springer, 255–276.
Jiao Y, Wickett NJ, Ayyampalayam S, etal. 2011. Ancestral polyploidy in
seed plants and angiosperms. Nature 473: 97–100.
Kelchner SA. 2000. The evolution of non-coding chloroplast DNA and its
application in plant systematics. Annals of the Missouri Botanical Garden
87: 482–498.
Kellogg EA. 2016. Has the connection between polyploidy and diversication
actually been tested? Current Opinion in Plant Biology 30: 25–32.
Kolář F, Čertner M, Suda J, Schönswetter P, Husband BC. 2017. Mixed-
ploidy species: progress and opportunities in polyploid research. Trends in
Plant Science 22: 1041–1055.
Levin DA. 1975. Minority cytotype exclusion in local plant populations. Taxon
24: 35–43.
Levin DA. 1983. Polyploidy and novelty in owering plants. American
Naturalist 122: 1–25.
Lewis WH. 1980. Polyploidy in species populations. In: Lewis WH, ed.
Polyploidy: biological relevance. New York: Plenum Press, 103–144.
Lexer C, van Loo M. 2006. Contact zones: natural labs for studying evolution-
ary transitions. Current Biology 16: 407–409.
Li WP, Yang FS, Jivkova T, Yin GS. 2012. Phylogenetic relationships and
generic delimitation of Eurasian Aster (Asteraceae: Astereae) inferred
from ITS, ETS and trnL-F sequence data. Annals of Botany 109:
1341–1357.
Liaw A, Wiener M. 2002. Classication and regression by randomForest. R
News 2: 18–22.
Librado P, Rozas J. 2009. DnaSP v5: a software for comprehensive analysis
of DNA polymorphism data. Bioinformatics 25: 1451–1452.
Maddison WP, Knowles LL. 2006. Inferring phylogeny despite incomplete
lineage sorting. Systematic Biology 55: 21–30.
Magallón S, Gómez–Acevedo S, Sánchez–Reyes LL, Hernández–
Hernández T. 2015. A metacalibrated time‐tree documents the early
rise of owering plant phylogenetic diversity. New Phytologist 207:
437–453.
Mairal M, Pokorny L, Aldasoro JJ, Alarcón M, Sanmartín I. 2015a.
Ancient vicariance and climate-driven extinction explain continental-
wide disjunctions in Africa: the case of the Rand Flora genus Canarina
(Campanulaceae). MolecularEcology 24: 1335–1354.
Mairal M, Sanmartín I, Aldasoro JJ, Culshaw V, Manolopoulou I, Alarcón
M. 2015b. Palaeo-islands as refugia and sources of genetic diversity within
volcanic archipelagos: the case of the widespread endemic Canarina
canariensis (Campanulaceae). Molecular Ecology 24: 3944–3963.
Mandáková T, Münzbergová Z. 2006. Distribution and ecology of cytotypes
of the Aster amellus aggregates in the Czech Republic. Annals of Botany
98: 845–856.
Mandáková T, Münzbergová Z. 2008. Morphometric and genetic differen-
tiation of diploid and hexaploid populations of Aster amellus agg. in a
contact zone. Plant Systematics and Evolution 274: 155–170.
Masterson J. 1994. Stomatal size in fossil plants: evidence for polyploidy in
majority of angiosperms. Science 264: 421–423.
McMillen-Jackson, AL, Bert TM. 2003. Disparate patterns of population
genetic structure and population history in two sympatric penaeid shrimp
species (Farfantepenaeus aztecus and Litopenaeus setiferus) in the eastern
United States. Molecular Ecology 12: 2895–2905.
Müntzing A. 1932. Cytogenetic investigations on synthetic Galeopsis tetrahit.
Hereditas 16: 105–154.
Münzbergová Z. 2006. Ploidy level interacts with population size and habitat
conditions to determine the degree of herbivory damage in plant popula-
tions. Oikos 115: 443–452.
Münzbergová Z. 2007. Population dynamics of diploid and hexaploid popula-
tions of a perennial herb. Annals of Botany 100: 1259–1270.
Münzbergová Z, Raabová J, Castro S, Pánková H. 2011. Biological ora
of central Europe: Aster amellus L. (Asteraceae). Perspectives in Plant
Ecology, Evolution and Systematics 13: 151–162.
Münzbergová Z, Šurinová M, Castro S. 2013. Absence of gene ow between
diploids and hexaploids of Aster amellus at multiple spatial scales.
Heredity 110: 123–130.
Nesom GL. 1994. Subtribal classication of the Astereae (Asteraceae).
Phytologia 76: 193–274.
Downloaded from https://academic.oup.com/aob/advance-article-abstract/doi/10.1093/aob/mcy149/5068832
by University of Stellenbosch user
on 18 August 2018
Mairal etal. — Cryptic diversity in polyploid Aster 13
Nordborg M. 2003. Coalescent theory. In: Balding D, Bishop M, Cannings C,
eds. Handbook of statistical genetics, 2nd edn. New York: John Wiley and
Sons, 602–635.
Noyes RD, Rieseberg LH. 1999. ITS sequence data support a single origin for
North American Astereae (Asteraceae) and reect deep geographic divi-
sions in Aster sl. American Journal of Botany 86: 398–412.
Parisod C, Holderegger R, Brochmann C. 2010. Evolutionary consequences
of autopolyploidy. New Phytologist 186: 5–17.
Patwardhan A, Ray S, Roy A. 2014. Molecular markers in phylogenetic stud-
ies – a review. Journal of Phylogenetics & Evolutionary Biology 2: 131.
doi: 10.4172/2329-9002-2-131
Paun O, Forest F, Fay MF, Chase MW. 2009. Hybrid speciation in angio-
sperms: parental divergence drives ploidy. New Phytologist 182: 507–518.
Pokorny L, Oliván G, Shaw A. 2011. Phylogeographic patterns in two south-
ern hemisphere species of Calyptrochaeta (Daltoniaceae, Bryophyta).
Systematic Botany 36: 542–553.
Posada D. 2008. jModelTest: phylogenetic model averaging. Molecular
Biology and Evolution 25: 1253–1256.
Raabová J, Fischer M, Münzbergová Z. 2008. Niche differentiation between
diploid and hexaploid Aster amellus. Oecologia 158: 463–472.
Rambaut A, Drummond AJ. 2013. TreeAnnotator v1. 7.0.
Rambaut A, Drummond AJ, Suchard M. 2013. Tracer v1. 6.
Ramsey J. 2011. Polyploidy and ecological adaptation in wild yarrow.
Proceedings of the National Academy of Sciences, USA 108: 7096–7101.
Ramsey J, Robertson A, Husband B. 2008. Rapid adaptive divergence in New
World Achillea, an autopolyploid complex of ecological races. Evolution
62: 639–653.
R Core Team. 2014. R: a language and environment for statistical comput-
ing. R Foundation for Statistical Computing, Vienna, Austria. URL http://
www.R-project.org/
Rieseberg LH, Willis JH. 2007. Plant Dpeciation. Science 317: 910–914.
Ridgeway G. 2015. Generalized boosted regression models. Documentation
on the R Package ‘gbm’, version 1? 5, 7. https://cran.r-project.org/web/
packages/gbm/gbm.pdf
Robertson FM, Gundappa MK, Grammes F, etal. 2017. Lineage-specic
rediploidization is a mechanism to explain time-lags between genome
duplication and evolutionary diversication. Genome Biology 18: 111.
doi: 10.1186/s13059-017-1241-z.
Rogers AR, Harpending H. 1992. Population growth makes waves in the dis-
tribution of pairwise genetic differences. Molecular Biology and Evolution
9: 552–569.
Rogers J, Gibbs RA. 2014. Comparative primate genomics: emerging patterns
of genome content and dynamics. Nature Reviews Genetics 15: 347.
Ronquist F, Teslenko M, van der Mark P, etal. 2012. MrBayes 3.2: efcient
Bayesian phylogenetic inference and model choice across a large model
space. Systematic Biology 61: 539–542.
Rosato M, Castro M, Rosselló. 2008. Relationships of the woody Medicago
species (section Dendrotelis) assessed by molecular cytogenetic analyses.
Annals of Botany 102: 15–22.
Ryan WB, Carbotte SM, Coplan JO, et al. 2009. Global multi-resolution
topography synthesis. Geochemistry, Geophysics, Geosystems 10:
Q03014. doi.org/10.1029/2008GC002332.
Schlötterer C, Harr B. 2001. Microsatellite instability. eLS. doi.org/10.1038/
npg.els.0000840.
Schneider S, Excofer L. 1999. Estimation of past demographic parameters from
the distribution of pairwise differences when the mutation rates vary among
sites: application to human mitochondrial DNA. Genetics 152: 1079–1089.
Schoener TW. 1970. Nonsynchronous spatial overlap of lizards in patchy habi-
tats. Ecology 51: 408–418.
Soltis DE, Soltis PS, Schemske DW, et al. 2007. Autopolyploidy in angio-
sperms: have we grossly underestimated the number of species? Taxon
56: 13–30.
Soltis DE, Buggs RJ, Doyle JJ, Soltis PS. 2010. What we still don’t know
about polyploidy. Taxon 59: 1387–1403.
Soltis DE, Visger CJ, Marchant DB, Soltis PS. 2016. Polyploidy: pitfalls and
paths to a paradigm. American Journal of Botany 103: 1146–1166.
Soltis PS, Soltis DE. 2000. The role of genetic and genomic attributes in the
success of polyploids. Proceedings of the National Academy of Sciences,
USA 97: 7051–7057.
Soltis PS, Liu X, Marchant DB, Visger CJ, Soltis DE. 2014. Polyploidy and
novelty: Gottlieb’s legacy. Philosophical Transactions of the Royal Society
B: Biological Sciences 369: 20130351. doi:10.1098/rstb.2013.0351.
Stamatakis A, Hoover P, Rougemont J. 2008. A rapid bootstrap algorithm for
the RAxML web servers. Systematic Biology 57: 758–771.
Stebbins GL. 1947. Types of polyploids: their classication and signicance.
Advances in Genetics 1: 403–429.
Stebbins GL. 1970. Variation and evolution in plants: progress during the
past twenty years. In: Hecht MK, Steere WC, eds. Essays in evolu-
tion and genetics in honor of Theodosius Dobzhansky: a supplement to
Evolutionary Biology. New York: Appleton-Century-Crofts, 173–208.
Sudová R, Pánková H, Rydlová J, Münzbergová Z, Suda J. 2014.
Intraspecic ploidy variation: a hidden, minor player in plant–soil–mycor-
rhizal fungi interactions. American Journal of Botany 101: 26–33.
Tajima F. 1989. Statistical method for testing the neutral mutation hypothesis
pDNA polymorphism. Genetics 123: 585–595.
Templeton AR, Crandall KA, Sing CF. 1992. A cladistic analysis of pheno-
typic associations with haplotypes inferred from restriction endonuclease
mapping and DNA sequence data. III. Cladogram estimation. Genetics
132: 619–633.
Thuiller W, Georges D, Engler R. 2013. biomod2: ensemble platform for
species distribution modeling. R package version, 2, r560. http://www2.
uaem.mx/r-mirror/web/ packages/biomod2/biomod2.pdf
Urbanek S. 2010. rJava: Low-level R to Java interface. https://cran.r-project.
org/web/packages/rJava/rJava.pdf
Van de Peer Y, Mizrachi E, Marchal K. 2017. The evolutionary signicance
of polyploidy. Nature Reviews. Genetics 18: 411–424.
Wallace RK. 1981. An assessment of diet-overlap indexes. Transactions of the
American Fisheries Society 110: 72–76.
VanDerWal J, Falconi L, Januchowski S, Shoo L, Storlie C. 2011. SDMTools:
species distribution modelling tools: tools for processing data associated
with species distribution modelling exercises. R package version, 1.
Warren DL, Glor RE, Turelli M. 2010. ENMTools: a toolbox for comparative
studies of environmental niche models. Ecography 33: 607–611.
Wendel JF. 2000. Genome evolution in polyploids. Plant Molecular Biology
42: 22–249.
Wendel JF. 2015. The wondrous cycles of polyploidy in plants. American
Journal of Botany 102: 1753–1756.
Wolfe KH, Li W-H, Sharp PM. 1987. Rates of nucleotide substitution vary
greatly among plant mitochondrial, chloroplast, and nuclear DNAs.
Proceedings of the National Academy of Sciences, USA 84: 9054–9058.
Wood TE, Takebayashi N, Barker MS, Mayrose I, Greenspoon PB,
Rieseberg LH. 2009. The frequency of polyploid speciation in vascu-
lar plants. Proceedings of the National Academy of Sciences, USA 106:
13875–13879.
Zurawski G, Clegg MT. 1987. Evolution of higher plant chloroplast DNA-
encoded genes: Implications for structure–function and phylogenetic stud-
ies. Annual Review of Plant Physiology and Plant Molecular Biology 38:
391–418.
Downloaded from https://academic.oup.com/aob/advance-article-abstract/doi/10.1093/aob/mcy149/5068832
by University of Stellenbosch user
on 18 August 2018