Nuclear DNA C-values are correlated with pollen size at tetraploid but not
diploid level and linked to phylogenetic descent in
Royal Botanic Garden Edinburgh, 20A Inverleith Row, Edinburgh EH3 5LR, Scotland, UK
Received 8 May 2017
Received in revised form 7 October 2017
Accepted 23 November 2017
Available online 19 December 2017
Edited by D Honys
Genome size can affect the phenotype of plants by a simple physical effect of the DNA material at the cellular
level. Pollen contains the bare necessities to initiate and sustain pollen tube growth and carries the haploid
genome. This work investigates the extent to which the nuclear DNA content affects pollen size in an evolution-
ary context within Streptocarpus (Gesneriaceae), by correlating genome size with pollen size of 38 samples
representing 36 taxa in a phylogenetic framework. Streptocarpus was found to possess an average genome size
among diploid species of 0.82 pg (1C). Signiﬁcant genome downsizing of up to 44.4% was observed among the
polyploid species which are exclusively found in Madagascar. The pollen size ranged between 11.27 μm and
25.55 μm at the diploid level, but 1C values were not found to drive pollen size. On the other hand, 1C values
in most polyploids showed a strong positive correlation with pollen size, near linear in species of sect.
Parasaintpaulia. In a phylogeneticcontext, polyploidy has evolvedat least twice in the genus, and contrary to pol-
len size, genome size was strongly lineage-speciﬁc rather than adaptive in Streptocarpus. Repeated parallel in-
creases and decreases in genome size (1C, and 1Cx) during the evolution of the genus were inferred. Overall,
in Streptocarpus at least, pollen size is a limited predictor of genome size and only partly reﬂecting ploidy level,
but may be of taxonomic value. The study demonstrates that the relationship between pollen size and genome
size is not straightforward, and their evolutionary trajectories unlinked.
© 2017 SAAB. Published by Elsevier B.V. All rights reserved.
Nuclear DNA amount
Over the last few decades the nuclear DNA C-values have been de-
termined for a large number of plant species (e.g. Bennett and Leitch,
2010). These values can vary by about 2400-fold among angiosperms
(e.g. Greilhuber et al., 2006; Pellicer et al., 2010). Much of the variation
is assumedto be due to repetitive DNA elements (e.g. Kubis et al., 1998;
Gregory, 2001; Meagher and Vassiliadis, 2005), and it was proposed
that the nuclear DNA content can affect the phenotype in two ways,
by its impact on regulatory processes in the genome (Meagher et al.,
2005; Meagher and Vassiliadis, 2005) and by a physical effect of the
nuclear DNA content and volume (e.g. Bennett, 1971). The latter can
inﬂuence the phenotype through effects at the nuclear level such as
developmental rates (Hoffmann et al., 2010), cell cycle time and pollen
maturation (Bennett, 1972, 1987; Smith and Bennett, 1975; Leitch and
Bennett, 2007; Beaulieu et al., 2008; Lomax et al., 2009), guard cell
length and epidermal cell area (Snodgrass et al., 2017) and seeds size
(Beaulieu et al., 2007).
Intriguing features are the variation of nuclear DNA amounts in
homoploid sister species (‘C-value enigma’,Gregory, 2001) and the
decrease of genome sizes
(1Cxvalues) in polyploids (‘genome
downsizing’,Leitch and Bennett, 2004), phenomena not yet fully
understood (Bennett and Leitch, 2005a, 2005b). Polyploidy seems to
have played a signiﬁcant role in plant evolution given the frequency of
its occurrence in speciation events (Stebbins, 1940; Wood et al.,
2009). Two types of polyploids are distinguished: autopolyploids show-
ing duplication of chromosomes within a species, and allopolyploidy
with duplication of chromosomes in interspeciﬁc hybrids. The effect of
the nature of polyploidy on genome downsizing was reviewed by
Eilam et al. (2010), who found a complex situation with typical auto-
polyploids forming multivalents and having some degree of sterility
showed additivity of DNA amounts of the diploid parents, while natural
and synthetic diploidized autopolyploids and allopolyploids showed
South African Journal of Botany 114 (2018) 323–344
E-mail address: firstname.lastname@example.org.
Greilhuber et al. (2005) was followed in terminology and ‘genome size’was used to
indicate the amount of DNA of one non-replicated holoploid genome with the chromo-
some number n(1C), the ‘monoploid genome si ze’for the DNA conte nt of one non-
replicated monoploid genome with chromosome base number x(1Cx), and the ‘nuclear
DNA content’to indicate the amount of DNA in the non-reduced nucleus 2n (2C).
0254-6299/© 2017 SAAB. Published by Elsevier B.V. All rights reserved.
Contents lists available at ScienceDirect
South African Journal of Botany
journal homepage: www.elsevier.com/locate/sajb
genome downsizing. While the ancestor of angiosperms appears to
have had small genomes (Soltis et al., 2003; Leitch et al., 2005), the
genome size was found not to be necessarily a useful marker at
higher taxonomic levels, but may be of value at the species level
(e.g. Greilhuber, 1979; Zonneveld et al., 2005).
Pollen size varies considerably in plants, from about 13 to 130 μm
in diameter (e.g. Knight et al., 2010), by ﬁve orders of magnitude in
volume among angiosperms, and can vary greatly even within genera
(Wodehouse, 1935; Muller, 1979). Pollen has important biological
functions in sexual reproduction including transfer of genetic material
via pollinators, and providing energy for pollen germination and
pollen tube growth. The mode of pollen transport does not appear
to be a main factor in the evolution of pollen size for animal pollinated
angiosperms (Harder, 1998). On the other hand, pollen size per se
has been found to present some taxonomic value in some plant
families such as Malvaceae at the genus level (El Naggar and
Sawady, 2008), or at species level in Salsola (Chenopodiaceae)
(Toderich et al., 2010).
A positive correlation between pollen grain size and nuclear DNA
content may perhaps be most obvious due to its physical effect, be-
cause of the reduction in function of pollen, carrying only the haploid
genome and contains the bare necessities to initiate and sustain
growth of the pollen tube. A direct correlation between nuclear DNA
content and pollen volume was found across grasses (Bennett,
1987). The study by Knight et al. (2010) across the plant kingdom
found a positive relationship between pollen size and C-values for a
split between gymnosperms and angiosperms but only weak correla-
tions for a few larger angiosperm lineages. In comparisons among
congeners they discovered a tendency of larger pollen with increasing
genome size. Although both studies did not account for ploidy level
variation. Very few studies have systematically investigated the
link between pollen size and genome size in relation to phylogenetic
relationships, and none has been carried out at all at the species
level within a genus.
While for about half of the angiosperm families genome size
estimates are available, few studies have involved Gesneriaceae.
In this family, to date species of only half a dozen genera out of
ca. 148 in the family (sensu Weber et al., 2013) have been investi-
gated; the European Haberlea (Zonneveld et al., 2005; Petrova
et al., 2014)andRamonda (Siljak-Yakovlev et al., 2008), the New
World Sinningia (Zaitlin and Pierce, 2010), and the Old World
Primulina (Kang et al., 2014), Saintpaulia (Loureiro et al., 2007) and
Streptocarpus (Hanson et al., 2001). Recent molecular systematic
work on subtribe Streptocarpinae of Gesneriaceae (sensu Weber
et al., 2013) subsumed all taxa from Africa, Madagascar and Comoro
Islands into a single genus Streptocarpus,includingSaintpaulia
and now contains 176 species (Nishii et al., 2015). For this genus
to date, the nuclear DNA content of only two species are published,
for Streptocarpus cyaneus (Hanson et al., 2001), and one for
the hitherto Saintpaulia ionantha,nowStreptocarpus ionanthus
(Loureiro et al., 2007).
Most species in Streptocarpus s.l. are diploid with a basic chromo-
some number of either x=15or16(Möller and Pullan,
2015onwards). Polyploidy, which might affect pollen size most
apparently, is reported for only a few Madagascan species, ranging
from tetraploidy to octoploidy (Milne, 1975; Jong and Möller, 2000;
Briggs, 2004). The origin of the polyploids by auto- or allopolyploidy
has not been addressed so far, but Milne (1975) reported bivalent
formation for the hexaploid Streptocarpus variabilis and the octoploid
Streptocarpus hildebrandtii. One way to detect allopolyploidy in phylo-
genetic studies is to compare phylogenetic trees constructed from
different genomes (e.g. Hegarty and Hiscock, 2004; Lundberg et al.,
The pollen of Streptocarpus species have been studied in detail by
Weigend and Edwards (1996), who reported all 128 species exam-
ined as possessing single pollen grains, except for Streptocarpus
daviesii for which tetrads were reported earlier (Hilliard and Burtt,
1971). The size range found among the species was 7 × 7 to 24
×16μm in polar and equatorial diameters respectively, a roughly
three-fold variation in dimensions across the taxa. The presence of a
range of ploidy levels in Streptocarpus and the publication of a recent
comprehensive phylogeny for the genus (Nishii et al., 2015), make
the genus an ideal object to investigate pollen size evolution in rela-
tion to genome size variation including the effects of polyploidization
in a phylogenetic context.
In the present study the nuclear DNA content for 36 taxa of
Streptocarpus was determined by ﬂow cytometry, and the genome
size (1C values, n) correlated with pollen size to investigate in detail
to what extend it is determined by its DNA content, particularly
where polyploidy is involved. Thus, all naturally occurring polyploid
species in Streptocarpus and an artiﬁcially created allotetraploid of
known recent hybrid origin were included. The artiﬁcial tetraploid
was useful for observations on the effect of polyploidy on pollen size
and as a benchmark for estimating the levels of genome downsizing
of the natural polyploids (Leitch and Bennett, 2004). It was also
attempted to shed light onto the nature of the natural polyploid
Streptocarpus species by comparing nuclear and chloroplast phyloge-
nies. The monoploid genome size (1Cx), genome size (1C) and pollen
size data were analyzed on a dated phylogenetic tree to investigate
the evolutionary trajectories of these characteristics and their inter-
play over evolutionary time. The study may also provide information
on the taxonomic value of pollen size and nuclear DNA levels for
2. Materials and methods
2.1. Plant material
Thirty-eight samples covering 34 species, two subspecies, and one
artiﬁcial hybrid of Streptocarpus were included (Table A1), representing
eight of the 12 sections established by Nishii et al. (2015).Seventeenof
the species were from subgen. Streptocarpus, 21 samples from subgen.
Streptocarpella. Twelve of the species of subgen. Streptocarpus came
from Africa, ﬁve from Madagascar, and 11 of subgen. Streptocarpella
came from Africa, and 10 samples from Madagascar.
Species of subgen. Streptocarpus have a basic number of x=16
chromosomes, while those of subgen. Streptocarpella have x=15
(Möller and Pullan, 2015onwards). Four of the Madagascan species of
Streptocarpus included were polyploids, two in subgen. Streptocarpella
(Streptocarpus beampingaratrensis subsp. antambolorum with 2n =
4x =60,Streptocarpus andohahelensis with 2n =6x = 90) and two
in subgen. Streptocarpus (S. variabilis with 2n =6x =96,
S. hildebrandtii with 2n =8x =128)(Milne, 1975; Jong and Möller,
2000; Briggs, 2004). The Madagascan Streptocarpus perrieri was re-
ported with 2n =4x = 64, for a count from the basal meristem in
the seedling cotyledon (Jong and Möller, 2000). However, the materi-
al used here was determined from root tips as 2n =2x = 32. The
artiﬁcial fertile allotetraploid hybrid (with 2n =4x =60),represented
acrossbetweenStreptocarpus vestitus ×Streptocarpus muscosus,bothof
2.2. Flow cytometry
Sample preparation was based on methods adapted from Costich
et al. (1991). Healthy leaves were collected and kept cold and moist.
They were ﬁrst rinsed with distilled water and blotted dry, then
100 mg (for some samples up to 250 mg) of leaf blade placed in a
Petri dish on ice, and 1 ml of solution A added [14.5 ml MgSO
(0.246 g 10 mM MgSO
pH 8.0), 15 mg dithiothreitol (Sigma D-0632), 300 μl propidium iodide
(5 mg/ml, Calbiochem 537,059), 375 μl Triton X-100 (10% w/v)].
To each sample 50 mg leaf material of an internal standard (Pisum
324 M. Möller / South African Journal of Botany 114 (2018) 323–344
sativum ‘Minerva Maple’, 2C = 9.39 pg, Johnston et al., 1999)was
added and the tissues were ﬁnely sliced with a sharp scalpel.
The samples were then ﬁltered through a 30 μmnylonmesh,and
the liquid was centrifuged at 13,000 rpm for 1 min and the superna-
tant was discarded. The pellet was re-suspended in 300 μlofsolution
B[3mlsuspensionA,7.5μl DNA free RNAse (Boehringer Mannheim
Biochemicals 119915)], and the solution incubated for 15 min at
37 °C, then kept on ice until run on a Beckman Coulter Epics
XL·MCL (Beckman Coulter, High Wycombe, UK) ﬂow cytometer,
operated at 488 nm and an output of 15 mW (Fig. A1). Fluorescence
emission was detected at 620 nm using a photomultiplier screened
by a long pass ﬁlter. The samples were run in biological triplicates
on different days and the mean (±SD) nuclear DNA amounts in pg
calculated by dividing the mean ﬂuorescence of the sample
2n peak by the mean ﬂuorescence of the standard 2n peak and
multiplying by the known nuclear DNA amount of the standard
2.3. Pollen measurements
Pollen samples came from ﬂoral material ﬁxed in Farmer's ﬂuid
(3 parts absolute ethanol:1 part glacial acetic acid) and stored in 70%
ethanol. Pollen from dissected anthers were stained on microscopic
slides in 0.5% acetocarmine. Permanent slides were prepared with
Euparal (Agar Scientiﬁc, Stansted, UK) by a vapor exchange method
(Bradley, 1948). Pollen were photographed under a Zeiss Axioskop
(Zeiss, Cambridge, UK) microscope and photographed with a Zeiss
Axiocam MRc5. The diameter of twenty normally formed pollen grains
was measured using Zeiss Axiovision version 4.8, and means (±SD)
determined. In the tetrads-bearing species, the pollen diameter was
measured across one pollen grain.
2.4. Phylogeny, dating and ancestral state reconstruction
The phylogenetic framework used in the present study is based
on previous analyses by Möller and Cronk (2001a, 2001b) and
Nishii et al. (2015). The latter used three gene regions and a 226
sample matrix of concatenated sequences of ITS, trnLF and rpl20–
rps12 to reconstruct a stable and well supported phylogeny.
This matrix was reduced to include 35 ingroup samples while
keeping the three outgroup samples, Haberlea rhodopensis,
Didymocarpus citrinus and Primulina spadiciformis, and adding the
tetraploid S. beampingaratrensis subsp. antambolorum and the dip-
loid S. aff. muscosus (Table A2). The molecular data were acquired
according to the methods of Nishii et al. (2015).Theﬁnal data
matrix contained 40 samples and was subjected to a Partition
Homogeneity (ILD) Test in PAUP* 4.0 (Swofford, 2002)totest
for incongruences between the cpDNA (trnLF: 824 char.; rpl20–
rps12: 846 char.) and ITS (719 char.) data prior to phylogenetic
Bayesian inference (BI) analyses were carried out in MrBayes v3.2.6
(Ronquist et al., 2012) on the cpDNA, ITS and combined data withbest-
ﬁtting models of evolution selected independently as determinedunder
the Akaike information criterion (Akaike, 1974) selected in Mr-
Modeltest v2.3 (Nylander, 2004) and were GTR + I, GTR + G, GTR
+ G and SYM + I for trnLF, rpl20–rps12, the ITS spacers and the 5.8S
gene respectively. One million generations each were run in two inde-
pendent parallel analyses sampled every 1000th generation. The ﬁrst
100 trees (10%) were discarded as burn-in prior to calculating the BI
majority rule consensus trees and posterior probabilities (PP).
Dated metric trees were obtained in BEAST v1.8.3 (Drummond
et al., 2016), using as secondary calibration point the divergence be-
tween subgenera Streptocarpus and Streptocarpella of 17.8 million
years (HPD: 8.56–28.03) of Petrova et al. (2015),andaYuleprior
for the tree topology. A lognormal prior was applied to the second-
ary calibration point (SD = 4). The analysis was run 2 times for
1 million generations under a lognormal relaxed molecular clock
and models of DNA evolution as detailed above and the indepen-
dent Markov chain Monte Carlo (MCMC) chains sampled every
1000th generation, and the convergence checked with TRACER
v1.6 (Rambaut et al., 2014). The effective sample size was checked
for each analysis for asymptotic behavior of the likelihood values
and was N200 for each prior. 10% of the sampled trees were
removed as burn-in in LogCombiner v1.8.3 prior to generating a
majority rule consensus tree in TreeAnnotator v1.8.3 in the
2.5. Character evolution analyses
To test whether pollen size is related to genome size and ploidy
level, a phylogenetic linear regression was used, as implemented in
the phylolm package (Ho and Ané, 2014) in R v3.3.3 (R Core Team,
2017). This controls for the phylogenetic relatedness of species within
a linear modeling framework, where one can compare the ﬁtof
different models with and without speciﬁc explanatory variables.
The response variable for these models was pollen size. The different
models compared were a 1) null model with no explanatory vari-
ables; 2) a model with only genome size as an explanatory variable;
3) a model with whether or not species are polyploid as an explana-
tory variable (yes/no); 4) a model with both genome size and
polyploidy as explanatory variables; and 5) a model with genome
size, polyploidy and their interaction as explanatory variables.
Preliminary analyses indicated that grouping all polyploids
x,8x)together as one level resulted in better model ﬁts,
likely because of the low sample size within individual polyploidy
In order to visualize the distribution of pollen size and genome size
on the phylogeny of Streptocarpus, maximum likelihood ancestral
state reconstruction was performed (Schluter et al., 1997) of these
characters and mapped them onto the BEAST dated phylogenies using
functions in the phytools package (Revell, 2012)inR.
3.1. Nuclear DNA content (2C)
Across the diploid Streptocarpus species the 2C-values ranged
widely, from 1 pg (SD ± 0.1) (S. perrieri) to 2.84 pg (SD ±0.06)
(S. beampingaratrensis subsp. beampingaratrensis), with an average of
1.64 pg (SD ±0.54) (Table 1). The two subgenera had similarly wide-
ranging values at the diploid level, within subgen. Streptocarpus
from 1.00 pg (SD ± 0.10) to 2.58 pg (SD ± 0.08), and from 1.05 pg
(SD ±0.00) to 2.84 pg (SD ± 0.06) for subgen. Streptocarpella, and aver-
ages of 2.16 pg (SD ± 0.90) and 1.80 pg (SD ± 0.10) respectively.
Among the sections, sect. Lignostreptocarpus had the highest values in
subgen. Streptocarpus, while sect. Parasaintpaulia had the highest levels
in subgen. Streptocarpella. Within the sections, the 2C values were rela-
tively homogenous at the diploid level, with sect. Streptocarpus and sect.
Saintpaulia and sect. Hova,sect.Carnosifolii and sect. Caulescentes
possessing similar values respectively (Table 1).
Values for the nuclear DNA content (2C) of polyploid species came
from subgen. Streptocarpus [S. hildebrandtii,8x, 4.96 pg (SD ± 0.09);
S. variabilis,6x, 3.02 pg (SD ±0.02)], and subgen. Streptocarpella
[S. andohahelensis,6x, 4.74 pg (SD ±0.06); S. beampingaratrensis
subsp. antambolorum,4x, 3.87 pg (SD ±0.13)]. The artiﬁcial tetraploid
hybrid, S. muscosus ×S. vestitus, had a 2C value of 2.6 pg (SD ± 0.1),
which was virtually the sum of the somatic amounts determined for
the parents, S. muscosus (1.14 pg, SD ±0.05), and S. vestitus (1.41 pg,
SD ±0.02) (Table 2a).
325M. Möller / South African Journal of Botany 114 (2018) 323–344
3.1.3. DNA amount per monoploid genome
The monoploid genome size (1Cx) across all samples ranged from
0.499 pg (S. perrieri)to1.421pg(S. beampingaratrensis subsp.
beampingaratrensis), an almost threefold range, with an average of
0.827 pg (SD ±0.251). For the diploid species, the 1Cxvalues reﬂected
greatly the patterns of the 2C values, due to the small difference in
basic chromosome number, x=15andx=16(Table 2a).
The 1Cxvalues in polyploids varied greatly, but not necessarily
in line with their differences in ploidy levels and systematic
afﬁliation (Fig. 1). Using the diploid S. beampingaratrensis subsp.
beampingaratrensis for comparison (100%) for the Madagascan
polyploid species of subgen. Streptocarpella sect. Parasaintpaulia, the
genomes of the tetraploid S. beampingaratrensis subsp. antambolorum
were only 68.1% and of the hexaploid S. andohahelensis only 55.6% of
the size of S. beampingaratrensis subsp. beampingaratrensis.Onthe
other hand, the 1Cxvalues for the polyploids of subgen. Streptocar-
pus sect. Plantaginei were very similar between the diploid and
hexaploid species, but the octoploid S. hildebrandtii showed an
over 24% increased 1Cxvalue compared to the diploid S. perrieri
The average 1Cxvalues for the Streptocarpus sections were less
variable than the 2C values and similar values were observed across
subgenus boundaries, with sect. Plantaginei of subgen. Streptocarpus
having a similar value (0.541 pg, SD ± 0.068) compared to sects.
Hova (0.613 pg, SD ± 0.054), Carnosifolii (0.578 pg, SD ±0.042),
and Caulescentes (0.561 pg, SD ± 0.069) of subgen. Streptocarpella
(Tables 1, 2).
3.2. Pollen characteristics
Except for four taxa, all species investigated have single pollen
grains. S. daviesii,S. andohahelensis,S. beampingaratrensis subsp.
antambolorum and S. beampingaratrensis subsp. beampingaratrensis
possessed tetrads (Fig. 2).
Pollen size of the diploid species ranged from 11.27 μm (SD ±0.43–
0.45) (Streptocarpus papangae and Streptocarpus suffruticosus of sect.
Lignostreptocarpus)to25.55μm (SD ±1.51) (S. beampingaratrensis
subsp. beampingaratrensis)(Table 1). Except for the two species of
sect. Lignostreptocarpus with distinctly small pollen, the pollen size
of the diploids was relatively homogeneous (Fig. 3;Table 1). The
Chromatinand pollen size valuesof diploid species of 33 diploid Streptocarpus samples analyzed. Country of origin, somatic chromosome number(2n), nuclear DNAcontent (2C), genome
size in base pairs (MBp), and monoploid genome content (1Cx), pollen diameter (pollen Ø) (means with standard deviation, SD).
Taxon name Country of origin 2n 2C
SD Pollen Ø
S. papangae Madagascar 32 2.64 0.21 2587 1.322 0.104 11.27 0.45
S.suffruticosus Madagascar 32 2.37 0.12 2313 1.183 0.058 11.30 0.43
Sect. Lignostreptocarpus Mean 2.50 0.20 2450 1.252 0.099 11.28 0.44
S.perrierii Madagascar 32 1.00 0.10 976 0.499 0.051 16.98 0.54
Sect. Plantaginei Mean 1.00 0.10 976 0.499 0.051 16.98 0.54
S.baudertii Africa 32 1.88 0.00 1838 0.940 0.000 17.83 1.07
S.cyaneus Africa 32 1.75 0.03 1711 0.875 0.016 20.53 0.76
S.daviesii Africa 32 1.73 0.09 1689 0.863 0.043 24.07 0.91
S.dunnii Africa 32 2.09 0.15 2043 1.045 0.074 19.65 0.70
S.grandis Africa 32 2.58 0.08 2522 1.289 0.038 21.16 0.61
S.johannis Africa 32 1.73 0.06 1688 0.863 0.032 19.12 1.06
S.kentaniensis Africa 32 1.75 0.00 1714 0.876 0.000 18.41 0.72
S.micranthus Africa 32 1.86 0.03 1818 0.929 0.015 14.32 0.52
S.parvifolius Africa 32 1.83 0.00 1788 0.914 0.000 20.13 0.86
S.primulifolius Africa 32 1.85 0.00 1812 0.926 0.000 22.01 0.78
S.rexii Africa 32 1.90 0.06 1858 0.950 0.028 21.48 0.63
S.wendlandii Africa 32 2.52 0.04 2469 1.262 0.022 18.58 0.88
Sect. Streptocarpus Mean 1.96 0.30 1912 0.978 0.148 19.80 2.52
Subgen. Streptocarpus Mean 2.16 0.90 2115 0.890 0.225 18.44 3.71
S.bea. subsp. beampingaratrensis Madagascar 30 2.84 0.06 2779 1.421 0.029 25.55 1.51
Sect. Parasaintpaulia Mean 2.84 0.06 2779 1.421 0.029 25.55 1.51
S.vestitus Madagascar 30 1.41 0.02 1375 0.703 0.010 17.77 0.72
S.hilsenbergii Madagascar 30 1.26 0.06 1237 0.632 0.031 22.57 0.73
S.muscosus Madagascar 30 1.14 0.05 1115 0.570 0.027 18.63 0.85
S. aff. muscosus Madagascar 30 1.25 0.00 1224 0.626 0.000 18.61 0.35
S.thompsonii Madagascar 30 1.10 0.05 1080 0.552 0.026 19.48 0.69
S.venosus Madagascar 30 1.19 0.00 1162 0.594 0.000 17.15 0.79
Sect. Hova Mean 1.23 0.11 1199 0.613 0.054 19.03 1.88
S.saxorum Africa 30 1.10 0.04 1072 0.548 0.021 22.40 0.72
S.stomandrus Africa 30 1.22 0.02 1189 0.608 0.009 19.82 0.52
Sect. Carnosifolii Mean 1.16 0.08 1130 0.578 0.042 21.11 1.45
S.glandulosissimus Africa 30 1.06 0.02 1032 0.528 0.011 19.09 0.67
S.holstii Africa 30 1.05 0.00 1029 0.526 0.000 20.07 0.55
S.inﬂatus Africa 30 1.33 0.00 1298 0.664 0.000 23.31 1.24
S.pallidiﬂorus Africa 30 1.05 0.02 1029 0.526 0.008 19.78 0.88
Sect. Caulescentes Mean 1.12 0.14 1097 0.561 0.069 20.56 1.85
S.brevipilosus Africa 30 1.94 0.04 1899 0.971 0.022 19.33 0.85
S.shumensis Africa 30 1.88 0.00 1835 0.938 0.000 20.87 0.85
S.ionanthus subsp.grotei-1 Africa 30 1.73 0.03 1691 0.865 0.014 20.64 0.49
S.ionanthus subsp.grotei-2 Africa 30 1.74 0.00 1701 0.870 0.000 19.31 0.64
S. cf. ionanthus subsp. ionanthus Africa 30 1.67 0.02 1632 0.834 0.011 20.22 0.67
Sect. Saintpaulia Mean 1.79 0.11 1752 0.895 0.057 20.07 0.96
Subgen. Streptocarpella Mean 1.80 0.97 1785 0.753 0.301 20.25 2.17
All samples Mean 1.64 0.54 1602 0.819 0.269 19.44 3.09
326 M. Möller / South African Journal of Botany 114 (2018) 323–344
ploidy level (x)
S.vest. x S.musc.
Fig. 1. Relationships between the monoploid genome size, the DNA content of one non-replicated monoploid genome with chromosome base number x(1Cx), and ploidy level (x)for
38 samples of Streptocarpus. Legend indicates sections or hybrid origin (×).
Chromosome number, ploidylevel, genome sizes(a) and pollen characteristics(b) of diploid and polyploidMadagascan species of Streptocarpus analyzed.a. Subgeneric afﬁliation, somatic
chromosome number (2n), basic number, ploidy level (x), nuclear DNA content (2C), genome size in base pairs (Mbp) and monoploid genome content (1Cx), with standard deviations
(SD), withpercentages andexpected values(assuming linearrelationships)based on diploid congeners andpolyploidy series. b. Subgeneric afﬁliation, pollen diameter (Ø), pollen volume
(nL), and expected values (assuming linear relationships) based on diploid congeners and polyploidy series.
327M. Möller / South African Journal of Botany 114 (2018) 323–344
polyploids displayed larger pollen, but not necessarily proportional
with their ploidy levels (Fig. 3;Table 2b). For the polyploid series
in sect. Parasaintpaulia, the pollen sizes increased in line with the
increase in ploidy level, with the tetraploid pollen with 31.78 μm
(SD = 0.92) being 24.4% larger than the diploid pollen (25.55 μm;
SD = 1.51), and the hexaploid pollen (37.23 μm; SD = 1.66)
Fig. 2. Selection of pollen SEM images illustrating size difference s in Streptocarpus.a.S. beampingaratrensis subsp. antambolorum:2n = 4x = 60. b. S. beampingaratrensis subsp.
beampingaratrensis:2n = 2x = 30. c. S. andohahe lensis:2n = 6x = 90. d. S. thompsonii:2n = 2x = 30. e. S. suffruticosus:2n = 2x = 32. f. S. papangae:2n = 2x = 32. SEM images a–d
from Briggs (2004), e & f by Frieda Christie, RBGE.
0.0 0.5 1.0 1.5 2.0 2.5 3.0
pollen diameter (µm)
S.vest. x musc.
Fig. 3. Relationship between pollen diameter (μm) and genome size, the amount of DNA of one non-replicated holoploid genome with the chromosome number n(1C), for diploid and
polyploidsamples of Streptocarpus. Legend indicatessections or hybrid origin(×); continuous linesconnect differentploidy levels within sections: dashed linesindicate theoretical values
for the respective polyploid seriesassuming additivity of genomes(i.e. absence of genome downsizing) and a linearincrease in pollendiameter, and dottedlines point to equivalent ploidy
levels between observed and theoretical values.
328 M. Möller / South African Journal of Botany 114 (2018) 323–344
45.7% larger than the diploid pollen. For the polyploid species of sect.
Plantaginei, the pollen diameter of the hexaploid was 39.2% larger
(23.65 μm; SD = 0.85) than those of the diploid species (16.98 μm;
SD = 0.54). The octoploid species on the other hand, had a pollen size
slightly smaller (21.18 μm; SD = 0.42) than the hexaploid species.
The pollen diameter of the artiﬁcial tetraploid hybrid plant was about
a third larger (27.2% and 33.4%) than those of its two parent species
3.3. Statistical analyses of pollen size
Phylogenetic linear regressions showed that whether or not
species are polyploid is the best explainer of pollen size
(Table 3). Including genome size in the model in addition to
whether or not species are polyploid did not improve model ﬁt
at all. Genome size on its own did perform better in explaining
variation in pollen size than a null model without any explanatory
variables, although this effect was small (delta AIC = −1.6). Thus,
while genome size may appear to be correlated with pollen size
when viewed in isolation, this is likely because it covaries
with whether or not species are polyploid. Once the effect of
polyploidy on pollen size is incorporated into the model, genome
size is no longer a signiﬁcant explainer of variation in pollen size
3.4. Phylogenetic evolution of genome and pollen size
The partition homogeneity (ILD) test indicated no signiﬁcant
levels of incongruence between the cpDNA and ITS datasets (P=
0.12; Fig. A2). The average standard deviation of split frequencies
for the BI analyses on the cpDNA, ITS and combined data
was 0.005201, 0.005698 and 0.003789 respectively indicating
a good convergence of the two independent MCMC runs
Inspection of the individual BI trees indicated that most nodes
received maximum PP values particularly for the sections
(Figs. A3–A5). With view to the polyploid species, no topology
difference was observed in sect. Plantaginei, while in sect.
Parasaintpaulia the cpDNA tree showed a highly supported rela-
tionship between S. beampingaratrensis subsp. antambolorum and
S. andohahelensis (PP = 1; Fig. A3), while the ITS data showed a poorly
supported relationship between S. beampingaratrensis subsp.
beampingaratrensis and S. andohahelensis (PP = 0.63; Fig. A4). The
combined tree showed the former relationship with maximum support
(PP = 1; Fig. A5).
The BEAST chronogram (Fig. 4) was identical with the topology
of the Bayesian inference consensus tree, except for some poorly
supported terminal species relationships in sects. Streptocarpus
and Saintpaulia that had little bearing on the character optimiza-
tion (Fig. A6). The two samples of S. beampingaratrensis, subspecies
beampingaratrensis and antambolorum were not sister, and
the taxonomic implications will be addressed in a forthcoming
Some inferences can be made on the evolution of the genome
and pollen across the genus Streptocarpus, although the ancestral
Fit of different modelsto explain variationin pollen size. The best model (withlowest AIC
score) is listed ﬁrst and with the delta AIC values for alternative models.
Explanatory variables included Delta AIC value
Polyploidy + genome size −2.0
Genome size −2.7
Polyploidy + genome size + interaction −4.0
Null model −4.3
S.vestitus ×S. muscosus
S. aff. muscosus
1Cx value (pg)
S.vestitus ×S. muscosus
S. aff. muscosus
pollen diameter (µm)
S.vestitus ×S. muscosus
S. aff. muscosus
1C value (pg)
Fig. 4. Dated phylogenetic tree of Streptocarpus species analyzed here and ancestral trait
state reconstructions. a. Trait reconstruction for holoploid genome size (1C) and aligned
bar chart of observed values; b. trait reconstruction for monoploid genome size (1Cx)
and aligned bar chart of observed values; c. trait reconstruction for pollen size and
aligned bar charts for observed values. Section abbreviations: Ca = Caulescentes,Cr=
Sa = Saintpaulia,St=Streptocarpus. Species shaded gray from Madagascar, remaining
ones from Africa.
329M. Möller / South African Journal of Botany 114 (2018) 323–344
basic chromosome number could not be determined because the two
main clades each possess one state, x=15orx=16(Fig. 4). Poly-
ploids have evolved at least two times, once in each subgenus, in sect.
Plantaginei of subgen. Streptocarpus, and once in sect. Parasaintpaulia
of subgen. Streptocarpella (Fig. 4). For each polyploidy instance
two levels of polyploidy were attained, hexaploidy and octoploidy
in sect. Plantaginei, and tetraploidy and hexaploidy in sect.
Given the assumptions that went into the maximum likelihood
ancestral state reconstructions, it appears that a relatively small 1C
genome was the ancestral state for Streptocarpus,i.e.largerge-
nomes appeared to have evolved from smaller ones most notably
for the polyploid species, with the reverse, a genome reduction,
observed for three lineages, one representing S. perrieri,another
sect. Hova and a third the lineage leading to sects. Carnosifolii
and Caulescentes (Fig. 4a). Considering the ploidy levels, the ances-
tral condition was suggested to be a medium monoploid genome
size (1Cx)(Fig. 4b). Four separate lineages appear to show a strong
increase in values, one representing sect. Lignostreptocarpus,and
three diploid species, Streptocarpus grandis,Streptocarpus wendlandii and
S. beampingaratrensis subsp. beampingaratrensis. In four separate lineages,
a decrease in monoploid genome size was inferred, for the lineages
representing sects. Plantaginei,Hova,Carnosifolii and Caulescentes
The ancestral Streptocarpus appeared to have possessed a
medium pollen size with two separate lineages, sect. Lignostreptocarpus
and Streptocarpus micranthus evolving signiﬁcantly smaller
pollen, and the polyploid species of sect. Parasaintpaulia evolved
larger pollen linked to their polyploidy. Interestingly, the
polyploid species of sect. Plantaginei did not possess larger pollen,
but the diploid S. perrieri of this section had slightly smaller ones
In light of the topology differences in sect. Parasaintpaulia the
alternative phylogeny topology between the S. beampingaratrensis
taxa and S. andohahelensis was explored (Fig. 4 insets), that sug-
gested that the polyploids arose independently and consequently
their 1C values, rather than be the result of one polyploidization
event (Fig. 4a and inset). This was the same for the pollen size evolu-
tion (Fig. 4c inset). No difference for the 1Cxevolution was evident
between the two scenarios, with the diploid S. beampingaratrensis
subsp. beampingaratrensis only showing an increase in values
4.1. Chromosome numbers and genome size of Streptocarpus in
Streptocarpus varies very little in basic chromosome number with
only two numbers observed, x=15andx= 16, diagnostic for the
two subgenera (Hilliard and Burtt, 1971; Jong and Möller, 2000).
Hidalgo et al. (2010) linked changes in basic chromosome number to
major morphological shifts such as stamen number and ﬂoral symmetry
in Valerianaceae. In Streptocarpus, the two subgenera also show signiﬁ-
cant morphological diversiﬁcations in habit, seed morphology and
other characters across the two subgenera (Hilliard and Burtt, 1971;
Nishii et al., 2015). Apparently, the single change in basic number in
Streptocarpus occurred very early on in the diversiﬁcation of the genus
(Möller et al., 2008), perhaps more than 17 million years ago (Petrova
et al., 2015;ﬁg. B7). This is a signiﬁcant evolutionary period of time suf-
ﬁcient for diverging lineages to acquire morphological and cytological
Lawrence et al. (1939) were not only the ﬁrst to point out the
correlation between basic chromosome number and subgenus as-
signment in Streptocarpus, but also suggested that the chromo-
somes in species of subgen. Streptocarpella are smaller compared
to those of subgen. Streptocarpus.Thishasbeenborneoutinthis
study and Lawrence's proposition was conﬁrmed: on average
species of subgen. Streptocarpella have an about 20% lower DNA
contents than those of subgen. Streptocarpus (Table 1).
Among the few Gesneriaceae genera analyzed to date,
Streptocarpus possess a medium nuclear DNA content at the diploid
level (average 2C = 1.64 pg), which is somewhat lower than that
of the Chinese Primulina (2C = 1.92 pg) (Kang et al., 2014). A 2.8-
fold range in genome size was observed among the diploid species
of Streptocarpus analyzed here. In Primulina, a similar level of varia-
tion was reported (Kang et al., 2014). These are not extreme values.
Studies in other plants reported a 1.7-fold difference across species
of Cirsium (Burešet al., 2004), up to a fourfold difference among
diploid species of Lactuca (Doležalová et al., 2002)andTrifolium
(Vižintin and Bohanec, 2008).
The European Gesneriaceae genera show much higher levels of
nuclear DNA content compared to Streptocarpus with 2C = 2.3 to
2.8 pg. They are regarded as putative paleopolyploids (Zonneveld
et al., 2005; Siljak-Yakovlev et al., 2008; Petrova et al., 2014)
which may explain the higher values. The New World species of
Gesneriaceae on the other hand, show very low values, at least for
the few species investigated (2C = 0.513 to 0.767 pg for wild collec-
tions of Sinningia,Zaitlin and Pierce, 2010). Additionally, all Sinningia
species examined cytologically so far show with 2n =26(Möller and
Pullan, 2015onwards) a lower number of chromosomes compared to
Streptocarpus. Gesneriaceae as a whole do not have greatly varying
chromosome length within genomes (e.g. Jong and Möller, 2000;
Christie et al., 2012) as opposed to many monocot families with bi-
modal karyotypes such as Agavaceae or Hyacinthaceae (e.g. Jong,
1991; Guadalupe et al., 2008) and thus average chromosome
sizes can be compared within Gesneriaceae to some degree; de-
spite the lower number of chromosomes in Sinningia,theaverage
chromosome size in this genus appears to be about half (0.0197
to 0.0295 pg) compared to Streptocarpus (0.053 pg), thus
the New World Gesneriaceae appear to have fewer and smaller
chromosomes compared to those from the Old World. There are
too few data available to date to elaborate on the genome size
evolution at the family level.
4.2. Origin of polyploidy and genome up- and downsizing
Polyploidy has evolved twice independently in Streptocarpus,once
in subgen. Streptocarpella and once in subgen. Streptocarpus (Nishii
et al., 2015). Polyploidy is only found in species from Madagascar
and the Comoro Islands. It has been suggested that island coloniza-
tion favors polyploidy (Stebbins, 1950; Barrier et al., 1999). Its evolu-
tionary advantage may lie in their ﬁxed heterozygous genome, as
shown for arctic plants, that provides a buffer against stochastic or
climate change driven ﬂuctuations in small populations (Brochmann
et al., 2004).
The comparison of topologies of the cpDNA and nuclear (ITS)
phylogenies showed a clear absence of cross-sectional hybridiza-
tion for the genus Streptocarpus which is well in line with
the greatly different vegetative and ﬂoral morphologies (see
Nishii et al., 2015), and geographic discontinuities (Africa and
Madagascar) that precluded opportunities for hybridization. The
absence of topology differences for sect. Plantaginei might suggest
that their genome expanded putatively through autopolyploidy,
the duplication of chromosomes within the species. With the
present data it is impossible to hypothesize whether the ancestor
of the polyploids was already polyploid or whether each species
experienced separate polyploidization events. It is very likely that
intermediate ploidy levels existed and have not been sampled or
have become exist that have led, particularly, to the octoploid
condition of S. hildebrandtii. Recent autopolyploids may show
some level of multivalent formation (Parisod et al., 2010), and
330 M. Möller / South African Journal of Botany 114 (2018) 323–344
their absence in the putatively autopolyploids here (Milne, 1975)
might be explained by their possible old age of around 4.5 million
years that resulted in diploidized genomes (Wolfe, 2001).
The case of the polyploids in sect. Parasaintpaulia
suggests only weakly an allopolyploid origin for the tetraploid
S. beampingaratrensis subsp. antambolorum,fromacrosswiththe
diploid S. beampingaratrensis subsp. beampingaratrensis since
the plastid organelle inheritance in Streptocarpus, and perhaps
Gesneriaceae on the whole, is maternal as detected by empirical
experimental data for Streptocarpus (Möller et al., 2004) and data
from a natural hybrid (Puglisi et al., 2011). All the species of this
section occur in Southeast Madagascar and contact between the
species is theoretically possible. They have similar gross morphol-
ogies in vegetative and ﬂoral form (Hilliard and Burtt, 1971;
Nishii et al., 2015). In this section only S. mandrerensis could
not be included for analysis a species of which its collector
Prof Humbert reported to have observed more robust forms to
exist of this species he suggested to perhaps represent polyploid
forms (Hilliard and Burtt, 1971). Thus, while it seems that
more ﬁeldwork is necessary to fully resolve the situation in this
In Streptocarpus, decreasing monoploid genome sizes (1Cxvalues)
were observed with increasing ploidy levels in the naturally occurring
polyploids (Fig. 1), a phenomenon observed in other genera across
angiosperms (Leitch and Bennett, 2004). A very strong effect of genome
downsizing was found in sect. Parasaintpaulia which formed a perfect
polyploidy series with steadily decreasing 1Cxvalues culminating in a
more than 44% reduction (Table 2a). This is much higher than found
in other studies (e.g. Malus: ~ 2%, Höfer and Meister, 2010;Hepatica:
6–14%, Zonneveld, 2010), though similar ﬁgures were observed for
Turnera (López et al., 2011).
There is some evidence that genome downsizing occurs rapidly
at or in the ﬁrst few generations after the polyploidization event in
natural and synthetic auto- and allopolyploids of Mediterranean
Triticeae (e.g. Eilam et al., 2010). This is somewhat in contrast to
the ﬁnding of a study on Icelandic birch allopolyploids that
showed constancy of 1Cxvalues for three levels of ploidy without
effects of downsizing, which the authors suggest is likely a reﬂec-
tion of the plant's recent colonization and hybridization history
in Iceland after the last glaciation (Anamthawat-Jónsson et al.,
2010). Such constancy in recent polyploids was also reported for
the European Gesneriaceae Ramonda serbica, a species recently
evolved in the Ionian age (0.781–0.126 million years ago)
(Petrova et al., 2015), which shows various levels of polyploidy
but an absence of genome downsizing (Siljak-Yakovlev et al.,
2008). This time aspect may play a role in Streptocarpus as well:
a near perfect additivity of genome size was observed in the
recently artiﬁcially created allotetraploid interspeciﬁchybrid
(Table 2b). Accordingly, the species in sect. Parasaintpaulia could
be regarded as putative paleoalloploids, which is in line with
their inferred age of around 4.6 million years (Fig. 4). Those
putative autopolyploids of sect. Plantaginei have a similar old age.
However, they show no distinct genome downsizing and have
similarly low 1Cxvalues in the section (Table 2a). From the studies
on Triticeae, birch and Streptocarpus here, it appears difﬁcult to
ﬁnd a common pattern of genome evolution in polyploids as
each lineage appears to exhibit its distinct pattern.
On closer inspection it appears that the diploid species in sect.
Parasaintpaulia,S. beampingaratrensis subsp. beampingaratrensis,
possesses a genome 1Cxvalue which is much higher compared
to its related species in the section, and the trait analysis indicates
a sudden genome expansion, irrespective of the phylogenetic
scenario (Fig. 4b and inset). The mechanisms for genome size
variation are not fully understood, but may involve variables such
as heterochromatin, microsatellite and retrotransposon expansion
and removal, coupled with ﬁtness-based selection and adaptation
(Petrov, 2001, 2002; Bennetzen et al., 2005). These mechanisms
may favor small genomes, but larger genomes are not necessarily
impossible to evolve (Oliver et al., 2007). Further studies are need-
ed to fully understand genome evolution in Streptocarpus.
4.3. Pollen of Streptocarpus in Gesneriaceae
Streptocarpus pollen was measured to vary greatly in size be-
tween the species by a factor of around 2.5 for diploid (11.27 to
25.55 μm) and about threefold across all species. This has also
been reported previously (Weigend and Edwards, 1996). The
great variation was mainly due to the larger pollen of the poly-
ploid species. In Weigend and Edwards (1996),S. andohahelensis
and S. beampingaratrensis subsp. beampingaratrensis were de-
scribed as possessing single pollen grains, but in studies on living
materials, these and S. beampingaratrensis subsp. antambolorum
were consistently found to possess tetrads (Fig. 2). This seems to
be a consistent characteristic for sect. Parasaintpaulia (Nishii
et al., 2015). The two members of sect. Lignostreptocarpus included
here, S. papangae and S. suffruticosus, had distinctly small pollen
(Figs.2,3;Jong et al., 2012). Other members of this section,
such as Streptocarpus glabrifolius,Streptocarpus macropodus and
Streptocarpus tsaratananensis also have similarly small pollen
(Weigend and Edwards, 1996), a characteristic that appears to be
section speciﬁc for these woody caulescent Madagascan species
that also differ in their non-coherent anthers from other species
in the genus where the two anthers cohere at the tip (Hilliard
and Burtt, 1971). Apart from the large pollen of the polyploids
and small pollen of sect. Lignostreptocarpus, the pollen size across
the genus Streptocarpus appeared to be quite homogeneous
The pollen of the New World Gesneriaceae genus Sinningia is much
larger in diameter (23.1–29.7 × 20.7–27.0 μm, Melhem and Mauro,
1973), than those of Streptocarpus (7–24 × 7–16 μm, Weigend and
Edwards, 1996; average 19.44 μm, Appendix A2) yet the New World
species possess much smaller genomes (2C = 0.513 to 0.767, Zaitlin
and Pierce, 2010) than Streptocarpus (2C = 1.00–2.84 pg, Table 1).
Other neotropical genera have similar sized and even larger pollen as
Sinningia (Melhem and Mauro, 1973; Fritze and Williams, 1988), but
whether this is a consistent pattern requires a larger, more detailed
4.4. Phylogenetic aspects of pollen and genome evolution in Streptocarpus
In a phylogenetic context, it was found that for diploid species
in Streptocarpus large and small genomes (1C) are section-
speciﬁc but occur in several clades in the phylogeny, and the trait
optimization analysis suggested that change was not unidirectional
and included repeated genome expansion and contractions over
evolutionary times (Fig. 4a, b). Someattemptshavebeenmadeto
correlate genome size and ecological factors in a phylogenetic context,
althoughwith conﬂicting results: Grotkopp etal. (2004) reported a neg-
ative correlation between genome size and latitude in Pinus, while in a
recent study on Primulina (Gesneriaceae) a positive relationship with
latitude was described (Kanget al., 2014). In the case of sect. Saintpaulia
here, species with very different ecological requirements were included
(high altitude: Streptocarpus shumensis,Streptocarpus brevipilosus;low
altitude: remaining species), but all have very similarly large ge-
nomes. Furthermore, some of the species of the sister clade to
sect. Saintpaulia,sect.Streptocarpella included here coexist with
species of sect. Saintpaulia but have much smaller genomes. Thus
it seems that ecological factors appear not to be involved in the ge-
nome enlargement of species in Streptocarpus.Variationingenome
size has been linked to retrotransposons (e.g. Hawkins et al., 2006;
Neumann et al., 2006; Vitte and Bennetzen, 2006)thatcanmake
up over 70% of the nuclear genome (SanMiguel and Bennetzen,
331M. Möller / South African Journal of Botany 114 (2018) 323–344
1998), and proliferation of transposable elements can act over very
short periods of time changing the genome size in just a few mil-
lion years (SanMiguel et al., 1996, 1998). Whether this mechanism
is involved in Streptocarpus would require further studies.
It was found that genome size changes affect entire clades rather
than individual species in Streptocarpus (Fig. 4). This strong component
of conservation of genome size for phylogenetic lineages (i.e. shared by
descent within sections) might suggest that in Streptocarpus, descent
overrides adaptation, at least at the diploid level. Similar results of
lineage-speciﬁc genome sizes have been obtained in other studies, for
example in Petunia (Mishiba et al., 2000), and Malus (Höfer and
4.5. Evolutionary dynamics between genome and pollen size
The relationship between pollen and genome size (1C) was found
to be relatively straightforward for diploid species, but more intricate
for the polyploid species in Streptocarpus. Among the diploids several
sections in the genus possessed a unique combination between 1C
value and pollen size (Fig. 3), so much so that the sections could be
characterized and distinguished from each other. For example, species
of sect. Lignostreptocarpus had very small pollen and a relatively high
1C value. Those of sect. Saintpaulia and sect. Streptocarpus had a me-
dium high 1C value and medium large pollen, while those of sects.
Carnosifolii,Streptocarpella and Hova also had medium large pollen
but a low 1C value. While this allows distinguishing sections, it
does not establish a positive link between genome size and pollen
size. Further, since the pollen size of diploids is rather homogeneous,
except for those of sect. Lignostreptocarpus, the pollen size appeared
not to be affected by the shear physical effect of nuclear DNA content
and volume in the pollen (Bennett, 1971). This could only be seen in
The polyploids possessed larger pollen as one would expect from
their larger holoploid genomes (1C), but this was not proportional
to the larger DNA content in the study here. The artiﬁcial allotetra-
ploid hybrid S. vestita ×S. muscosus allows a glimpse into the dynam-
ics of the relationships between genome and pollen over evolutionary
time. It shows that upon duplication of the genome, both the amount
of DNA and the pollen volume are additive between the parent
species, i.e. effectively double (× in Fig. 3). The latter may be
explained by a shear physical effect of the cytoplasm of the parental
gametophytes, and a physical effect of the nuclear DNA mass
Extrapolating this relationship for a range of ploidy levels
(dashed lines in Fig. 3), it can be noted that polyploid species in
the clades representing sects. Parasaintpaulia and Plantaginei,
seem to have responded differently over evolutionary time in
genome and pollen size to polyploidization and are on different
trajectories: the species of sect. Parasaintpaulia plot above the
theoretical line due to their greatly reduced genomes, but their
pollen sizes are in complete conformation with expected values
irrespective of genome downsizing (gray dotted line in Fig. 3;
Table 2b). Thus, while their pollen sizes fully correlate with their
ploidy level, they contain a greatly reduced DNA mass. Here, the
pollen size may be uncoupled from genome size and perhaps
linked to pollen reserves and pistil length (Torres, 2000; López
et al., 2006), since ﬂowers of the polyploid Streptocarpus species
have longer pistils with increasing ploidy level (Hilliard and
Conversely, among the species in sect. Plantaginei the hexaploid
S. variabilis conformed to expectations, but the octoploid
S. hildebrandtii deviates signiﬁcantly and possessed a greatly
reduced pollen size but a high DNA mass (black dotted line in
Fig. 3;Table 2b). Such a reduction in pollen size at this high
polyploidy level may suggest a limit in the correlation, although
this has not been observed in other plants such as grasses up to
octoploidy (e.g. Bennett, 1972). However, pollen downsizing is
not exceptional in Streptocarpus, and is to a degree also noticeable
in species of sect. Lignostreptocarpus (Fig. 3). A post-pollination
pollen reserve/pistil length scenario is unlikely for the polyploids
in sect. Plantaginei, since the octoploid species has the longer pistil
but smaller pollen. However, such post-pollination scenario may
be relevant in the small-ﬂowered S. hildebrandtii and those in
dunnii has an extremely long pistil of 80 mm but does not have a
conspicuously large pollen size (Fig. 4c). Thus, a link between pol-
len size and ﬂower size is not absolute from the available data
here. Further studies are needed here, perhaps along the ﬁndings
of Cruden and Lyon (1985). Their data indicated that stigma
depth, rather than style length is correlated with pollen size.
The study of ploidy, genome and pollen size in a phylogenetic
context allowed a look beyond the mere categorization of numbers.
Overall, it was found that pollen size was not affected strongly by
genome size at the diploid level and was strongly affected by
descent, at least for sect. Lignostreptocarpus where pollen tended
to be very small, and those of sect. Parasaintpaulia that formed tet-
rads. In this respect it was found that 1C values were strongly sec-
tion speciﬁc at the diploid level and in combination with pollen
size, sections in Streptocarpus could be characterized. In a phyloge-
netic context, unrelated parallel increases and decreases in
genome size were inferred to have occurred over evolutionary
times within the genus. Polyploidy has evolved twice in the
genus, in different subgenera but all in Madagascan species. The
putatively allopolyploid species showed a decrease in 1CxDNA
amount in sect. Parasaintpaulia which is in line with previous
studies, but also a case of absence of genome downsizing in
the autopolyploids of sect. Plantaginei which cannot be easily
explained. However, together with published results from other
plants, no common patterns seem to appear in genome downsizing
between allo- and autopolyploids. Polyploidy had a strong effect on pol-
len size and was proportionately reﬂecting ploidy levels and genome
sizes in sect. Parasaintpaulia. Although overall, pollen size varied less
dramatically than genome size. The data allow the postulation that, at
least in species of the genus Streptocarpus,genomesizeismorestrongly
biased by descent, rather than by ecological adaptive factors, and its ef-
fects on pollen size is limited to polyploids.
This work was funded by the Science Division of the Royal Botanic
The author would like to thank the horticulturists at the Royal
Botanic Garden Edinburgh (RBGE), especially S. Barber and A.
Ensoll, for their excellent work in maintaining and accurately
curating the living Gesneriaceae collections. The author is
extremely grateful to Prof. T. Meagher, University of St Andrews,
for his invaluable support and access to the EPICS ﬂow cytometer,
H. Hodge, University of St. Andrews, for his technical expertise and
help in the acquisition of the ﬂow cytometry data, K. Dexter for
support on the trait evolution analysis, and K. Nishii and two anon-
ymous reviewers are acknowledged for their critical and construc-
tive comments on the manuscript. The author is indebted to M.
Briggs, Royal Botanic Gardens, Kew, and F. Christie, for allowing
the use of the excellent pollen SEM images. RBGE is supported
by the Rural and Environment Science and Analytical Services
Division (RESAS) in the Scottish Government.
332 M. Möller / South African Journal of Botany 114 (2018) 323–344
Appendix A. Appendix
Fig. A1. Representative histograms of relative nuclear DNA content for selected species of Streptocarpus with varying ploidy levels, obtained using ﬂow cytometry analysis of propidium
iodide-stained nuclei. Sample (log2n) left peak and Pisum sativum standard (logpea) right peak.
333M. Möller / South African Journal of Botany 114 (2018) 323–344
Fig. A6. Results of a BEAST analysis based on combined 1670 cpDNA and 719 ITS characters. Numbers along the branches are posterior probability values. Scale bar in million years.
341M. Möller / South African Journal of Botany 114 (2018) 323–344
Collection information of the 37 of Streptocarpus and outgroup samples included in the molecular analyses, including current scientiﬁc names, collection origin and details, and GenBank
accession numbers for ITS, rpl20–rps12, trnLF.
Current name, country of origin and locality, collection details, ITS, rpl20–rps12, trnLF
Didymocarpus citrinus Ridl., Malaysia: Perlis, Kedah Peak, Davis P.H. 69437 (E), DQ912669, KR703821, AJ492293. Haberlea rhodopensis Friv., ex cult (Bulgaria, Greece), s.n. (E),
AF316898, KR703820, AJ492296. Primulina spadiciformis (W.T.Wang) Mich.Möller. & A.Weber, China: unknown locality, ex Smithsonian Institution 94-087 (E), AF316900,
KR703822, AJ492291. Streptocarpus andohahelensis Humbert, Madagascar: Tulear, Ranomafana, Moeller M. & Rafanantsoa G. MM9717B (E), KR704086, KR703849,
KR703944. Streptocarpus baudertii L.L.Britten, South Africa: KwaZulu-Natal, Otterspoort, Hughes M. & al. MH1067 (E), HQ719049, HQ719166, KR703945. Streptocarpus
beampingaratrensis subsp. antambolorum Humbert, Madagascar: Tulear, Col de Beampingaratra, Moeller M. & Rafanantsoa G. MM9719 (E), this study, this study, this study.
Streptocarpus beampingaratrensis subsp. beampingaratrensis Humbert, Madagascar: Tulear, Ranomafana, Moeller M. & Rafanantsoa G. MM9715 (E), KR704089, KR703851,
FJ501448. Streptocarpus brevipilosus (B.L.Burtt) Mich.Möller & Haston, Tanzania: Kanga forest, Mt. Kanga (Nguru Mts), Pócs T. s.n. (E), AF316924, KR703832, KR703926.
Streptocarpus cyaneus S.Moore, South Africa: Limpopo, Mariepskop, Hughes M. & al. MH1329 (E), HQ719024, HQ719141, HQ718943. Streptocarpus daviesii N.E.Brown ex
C.B.Clarke, South Africa: Kwa-Zulu Natal, Laager Farm, Wartburg, Bellstedt D.U. DUB955 (STE), KR704099, HE861713, HE956759. Streptocarpus dunnii Hook.f., South Africa:
Mpumalanga, Uitvlugt Farm, Hughes M. & al. MH1268 (E), HQ718988, HQ719105, HQ718912. Streptocarpus glandulosissimus Engl., ex cult. (Congo, Rwanda, Rurundi,
Uganda, Tanzania, Kenya), Hilliard O.M. s.n. (E), AF316918, KR703872, KR703972. Streptocarpus grandis N.E.Brown, South Africa: KwaZulu-Natal, NW of Inanda Mt., Styles
D.G.A. 3007_01 (STE), HQ718993, HQ719110, HQ718917. Streptocarpus hildebrandtii Vatke, ex cult. Madagascar: Parc Botanique et Zoologique de Tsimbazaza, Moeller M. &
Rafanantsoa G. MM9725 (E), AF316930, KR703874, MacMaster & al. (2005). Streptocarpus hilsenbergii R.Brown, Madagascar: Asaranitra, Andringitra, Bellstedt D.U. DUB1274
(STE, TAN), KR704105, KR703875, KR703974. Streptocarpus holstii Engl., ex cult. (Tanzania, E. Usambara Mts), ex Cornell Univ. (Bail. Hort) 88 (E), AF316917, KR703880,
AJ492304. Streptocarpus inﬂatus B.L.Burtt, Tanzania: Udzungwa Mts, Luke Q. (MMOG-112) (E), KR704111, KR703882, KR703980. Streptocarpus ionanthus subsp. grotei
(Engl.) Christenh.-1, ex cult. (Tanzania), ex RBG Kew 1983-8183 (E), KR704052, –, KR703935. Streptocarpus ionanthus subsp. grotei (Engl.) Christenh.-2, Tanzania: Tanga
Region, Bogner J. BNR (E), KR704045, KR703833, KR703927. Streptocarpus cf. ionanthus (H.Wendl.) Christenh., Tanzania: Tanga, Sigi River, Moors D.R. s.n. (–), AF316923, –, KR703934.
Streptocarpus johannis L.L.Britten, South Africa: Eastern Cape, Embotyi, Bellstedt D.U. DUB0840 (STE), HQ719054, HQ719170, HQ718965. Streptocarpus kentaniensis L.L.Britten & Story,
South Africa: Eastern Cape, Kentani, Joannou J. 6 (STE), HQ719014, HQ719131, HQ718934. Streptocarpus micranthus C.B.Clarke, South Africa: Mpumalanga, Shia Longubu dam,
Hughes M. & al. MH1375 (E), KR704121, –, KR703990. Streptocarpus muscosus C.B.Clarke, Madagascar: Tulear, Col de Tanatana, Moeller M. & Rafanantsoa G., MM9703 (E), KR704128,
KR703895, KR703994. Streptocarpus aff. muscosus C.B.Clarke, Madagascar: Tulear, Col de Beampingaratra, Moeller M. & Rafanantsoa G. MM9731 (E), this study, this study, this study.
Streptocarpus pallidiﬂorus C.B.Clarke, Tanzania: Arusha region, Masai distr., Longido Mts, Longido Stream, Carmichael Rev. W. s.n. (E), AF316921, KR703900, KR704000. Streptocarpus
papangae Humbert, Madagascar: Tulear, Col de Beampingaratra, Moeller M. & Rafanantsoa G. MM9718 (E), HQ718980, HQ719097, HQ718905. Streptocarpus parviﬂorus Hook.f.,
South Africa: Limpopo, Magoebaskloof Hotel, Hughes M. & al. MH1292 (E), HQ719021, HQ719138, KR704004. Streptocarpus perrieri Humbert, Madagascar: Antananarivo, Angavo near
Ankazobe, Moeller M., Rafanantsoa G. & Irapanarivo S. MM9726 (E), AF316931, KR703903, MacMaster & al. (2005). Streptocarpus primulifolius Gand., South Africa: Eastern Cape, Port
Information of the 38 samples of Streptocarpus analyzed, including their previous scientiﬁc names and taxonomic placements.
Current name Old name Subgenus Section Country Sample no.
Streptocarpus papangae Streptocarpus papangae Streptocarpus Lignostreptocarpus Madagascar 1997 2886 A1 2011-40
Streptocarpus suffruticosus Streptocarpus suffruticosus Streptocarpus Lignostreptocarpus Madagascar 1999 0122 D 2011-82
Streptocarpus hildebrandtii Streptocarpus hildebrandtii Streptocarpus Plantaginei Madagascar 2009 0476 2009 0476
Streptocarpus variabilis Streptocarpus variabilis Streptocarpus Plantaginei Madagascar 2002 0722 2011-34
Streptocarpus perrierii Streptocarpus perrierii Streptocarpus Plantaginei Madagascar 2009 1009 2011 0018
Streptocarpus baudertii Streptocarpus baudertii Streptocarpus Streptocarpus Africa 2003 0808 I 2008-60
Streptocarpus cyaneus Streptocarpus cyaneus Streptocarpus Streptocarpus Africa 2004 0867 J 20040867N
Streptocarpus daviesii Streptocarpus daviesii Streptocarpus Streptocarpus Africa 2002 0576 D 2002 0576 D
Streptocarpus dunnii Streptocarpus dunnii Streptocarpus Streptocarpus Africa 2004 0917 B 1994 1754
Streptocarpus grandis Streptocarpus grandis Streptocarpus Streptocarpus Africa 2008 0662 B 2011-85
Streptocarpus johannis Streptocarpus johannis Streptocarpus Streptocarpus Africa 1999 0271 O 1999 0271 O
Streptocarpus kentaniensis Streptocarpus kentaniensis Streptocarpus Streptocarpus Africa 2006 0942 B 2006 0942 B
Streptocarpus micranthus Streptocarpus micranthus Streptocarpus Streptocarpus Africa 2009 075x A 2009-197
Streptocarpus parvifolius Streptocarpus parvifolius Streptocarpus Streptocarpus Africa 2004 0865 C 2008-149
Streptocarpus primulifolius Streptocarpus primulifolius Streptocarpus Streptocarpus Africa 1966 0432 F 19660431
Streptocarpus rexii Streptocarpus rexii Streptocarpus Streptocarpus Africa 2009 0477 A 20080299
Streptocarpus wendlandii Streptocarpus wendlandii Streptocarpus Streptocarpus Africa 2009 0481 A 2011-37
Streptocarpus andohahelensis Streptocarpus andohahelensis Streptocarpella Parasaintpaulia Madagascar 1997 2885 2011-88
Streptocarpus bea. subsp. antambolorum Streptocarpus bea antambolorum Streptocarpella Parasaintpaulia Madagascar 1997 2887 C 2011-35
Streptocarpus bea.subsp.beampingaratrensis Streptocarpus bea beampingaratrensis Streptocarpella Parasaintpaulia Madagascar 2007 1148 2011-87
Streptocarpus vestitus Hovanella vestita Streptocarpella Hova Madagascar 2001 0230 2011-27
Streptocarpus hilsenbergii Streptocarpus hilsenbergii Streptocarpella Hova Madagascar 1999 0212 A 1999 0212 A
Streptocarpus muscosus Streptocarpus muscosus Streptocarpella Hova Madagascar 2008 0693 2008-48
Streptocarpus aff. muscosus Streptocarpus aff. muscosus Streptocarpella Hova Madagascar 2007 1152 2011-36
Streptocarpus thompsonii Streptocarpus thompsonii Streptocarpella Hova Madagascar 2009 0474 2009 0474
Streptocarpus venosus Streptocarpus venosus Streptocarpella Hova Madagascar 1998 2247 U 1998 2247 U
Streptocarpus saxorum Streptocarpus saxorum Streptocarpella Carnosifolii Africa 2005 1680 2011-41
Streptocarpus stomandrus Streptocarpus stomandrus Streptocarpella Carnosifolii Africa 2005 1683 2008-42
Streptocarpus glandulosissimus Streptocarpus glandulosissimus Streptocarpella Caulescentes Africa 1965 2118 B 1965 2118 B
Streptocarpus holstii Streptocarpus holstii Streptocarpella Caulescentes Africa 1959 2272 1959 2272
Streptocarpus inﬂatus Streptocarpus inﬂatus Streptocarpella Caulescentes Africa 2007 1151 2011-92
Streptocarpus pallidiﬂorus Streptocarpus pallidiﬂorus Streptocarpella Caulescentes Africa 2006 0040 2011-32
Streptocarpus brevipilosus Saintpaulia brevipilosa Streptocarpella Saintpaulia Africa 1970 0909 2008-53
Streptocarpus shumensis Saintpaulia shumensis Streptocarpella Saintpaulia Africa 2004 2093 2011-89
Streptocarpus ionanthus subsp. grotei-1 Saintpaulia magungensis Streptocarpella Saintpaulia Africa 1992 3185 2011-31
Streptocarpus ionanthus subsp. grotei-2 Saintpaulia difﬁcilis Streptocarpella Saintpaulia Africa 1987 2169 2011-42
Streptocarpus cf. ionanthus subsp. ionanthus Saintpaulia ionantha Streptocarpella Saintpaulia Africa 1996 1886 A 2011-29
Streptocarpus vestitus ×muscosus Hovanella vestita ×S.muscosus Streptocarpella Artiﬁcial hybrid Madagascar 2008 0567 2008-49
342 M. Möller / South African Journal of Botany 114 (2018) 323–344
Table A2 (continued)
Current name, country of origin and locality, collection details, ITS, rpl20–rps12, trnLF
Shepstone,Rooivaal, Hughes M. & al. MH1088 (–), HQ719082, HQ719199, KR704009. Streptocarpus rexii (Bowie ex Hook.) Lindl., South Africa: Eastern Cape, Kologha center, Hughes M.
& al. MH1149 (E), HQ719088, HQ719205, KR704015. Streptocarpus saxorum Engl., Tanzania: Usambara, Tanga region, Mather S. 1330 (E), AF316914, KR703906, KR704019.
Streptocarpus shumensis (B.L.Burtt) Christenh., ex cult. (Tanzania: W. Usambara Mts, Mt. Shume), ex Clements T. s.n. (E), KR704056, –, KR703940. Streptocarpus stomandrus B.L.Burtt,
Tanzania: Nguru Mts, Mabberley D.J. 687C (E), AF316915, KR703910, KR704023. Streptocarpus suffruticosus Humbert, Madagascar: Antsiranana, Marojezy RN12, Moeller M. &
Andriantiana J. MM9877A (E), MacMaster & al. (2005), KR703912,MacMaster & al. (2005). Streptocarpus thompsonii R.Brown, Madagascar: Antananarivo, Angavoke, Moeller M. &
Rafanantsoa G. MM9851 (E), KR704144, KR703914, KR704028. Streptocarpus variabilis Humbert, Madagascar: Road to Beolamana, Moeller M. MMOG32 (E), KR704148, KR703918,
KR704033. Streptocarpus venosus B.L.Burtt, Madagascar: Tulear, Andranohela River, Moeller M.& Rafanantsoa G. MM9711 (E), KR704149, KR703919, KR704034. Streptocarpus vestitus
(Baker) (Baker) Christenh., Madagascar: Antananarivo, Maromiza, Moeller M. & Andriantiana J. MMO0115 (E), KR704042, KR703828, KR703923. Streptocarpus wendlandii Spreng.,
South Africa: KwaZulu-Natal: Ngoye Forest, Mtunzini, Bellstedt D.U. DUB1336/1345 (STE), KR704150, HE861729, HE956773.
Akaike, H., 1974. A new look at the statistical model identiﬁcation. IEEE Transactions on
Automatic Control 19, 716–723.
Anamthawat-Jónsson,K., Thórsson, Æ.T.,Temsch, E.M., Greilhuber, J., 2010. Icelandicbirch
polyploids —the case of a perfect ﬁt in genome size. Journal of Botany 2010, 347254.
Barrier, M., Baldwin,B.G., Robichaux, R.H., Purugganan, M.D., 1999. Interspeciﬁchybridan-
cestry of a plant adaptive rad iation: allopolyploidy of the Hawaiian silversword alliance
(Asteraceae) inferred from ﬂoral homeotic gene duplications. Molecular Biology and
Evolution 16, 1105–1113.
Beaulieu,J.M., Moles, A.T.,Leitch, I.J., Bennett, M.D., Dickie, J.B.,Knight, C.A., 2007.Correlat-
ed evolution of genome size and seed mass. New Phytologist 173, 422–437.
Beaulieu, J.M., Leitch, L.J., Patel, S., Pendharkar, A., Knight, C.A., 2008. Genome size is a
strong predictor of cell size and stomatal density in angiosperms. New Phytologist
Bennett, M.D., 1971. The duration of meiosis. Proceedings of the Royal Society of London.
Series B 178, 277–299.
Bennett, M.D., 1972. Nuclear DNA content and minimum generation time in herbaceous
plants. Proceedings of the Royal Society of London. Series B 181, 109–135.
Bennett, M.D., 1987. Variation in genomic form in plants and its ecological implications.
New Phytologist 106, 177–200 (Suppl.).
Bennett, M.D., Leitch, I.J., 2005a. Plant genome size research: a ﬁeld in focus. Annals of
Botany 95, 1–6.
Bennett, M.D., Leitch, I.J., 2005b. Nuclear DNA amounts in angiosperms: progress, prob-
lems and prospects. Annals of Botany 95, 45–90.
Bennett, M.D., Leitch, I.J., 2010. Angiosperm DNA C-values Database (Release 7.0, Dec.
Bennetzen, J.L., Ma, J.X., Devos, K.M.,2005. Mechanisms of recent genome size variation in
ﬂowering plants. Annals of Botany 95, 127–132.
Bradley, M.V., 1948. A method for making aceto-carmine squashes permanent without
removal of coverslip. Stain Technology 23, 41–42.
Briggs, M., 2004. The Taxonomy of Madagascan Saintpaulia-like Streptocarpus. (MSc The-
sis). The University of Edinburgh, Edinburgh, UK.
Brochmann, C., Brysting, A.K., Alsos, I.G., Borgen, L., Grundt, H.H., Scheen, A.C., Elven, R.,
2004. Polyploidy in arctic plants. Biological Journal of the Linnean Society 82,
Bureš, P., Wang, Y.F., Horová, L., Suda, J., 2004. Genome size variation in Central
Europeanspecies of Cirsium (Compositae) andtheir natural hybrids. Annals of Botany
Christie, F., Bar ber, S., Möller, M. , 2012. New chromosome counts in Old World
Gesneriaceae: numbers for species hitherto regarded as Chirita, and their systematic
and evolutionary signiﬁcance. Edinburgh Journal of Botany 69, 323–345.
Costich,D.E., Meagher, T.R.,Yurkow, E.J., 1991. A rapid means of sex identiﬁcationin Silene
latifolia by use of ﬂow cytometry. Plant Molecular Biology Reporter 91, 359–370.
Cruden, R.W., Lyon, D.L., 1985. Correlations among stigma depth, style length, and pollen
grain size: do they reﬂect function or phylogeny? Botanical Gazette 146, 143–149.
Doležalová, I., Lebeda, A., Janeček, J., Číhalíková, J., Křístková, E., Vránová, O., 2002. Varia-
tion in chromoso me numbers and nuclear DNA contents in genetic resources of
Lactuca L. species (Asteraceae). Genetic Resources and Crop Evolution 49, 385–397.
Drummond, A.J., Rambaut, A., Suchard, M.A., 2016. BEAST v1.8.3: Bayesian evolutionary
analysis by sampling trees. Available from. http://tree.bio.ed.ac.uk/software/beast/.
Eilam, T., Anikster, Y., Millet, E., Manisterski, J., Feldman, M., 2010. Genome size in dip-
loids, allopolyploids and autopolyploids of Mediterranean Triticeae. Journal of Botany
El Naggar, S.M., Sawady, N., 2008. Pollen morphology of Malvaceae and its taxonomic
signiﬁcance in Yemen. Flora Mediterranea 18, 431–439.
Fritze, K.J., Williams, N.H., 1988. The taxonomic signiﬁcance of pollen morphology in the
Columnea alliance (Gesneriaceae). Annals of th e Missouri Botanical Garden 75,
Gregory, T.R., 2001. Coincidence, coevolution, or causation? DNA content, cell size, and
the C-value enigma. Biological Reviews 76, 65–101.
Greilhuber, J., 1979. Evolutionary changes of DNA and heterochromatin amounts in the Scilla
bifolia group (Liliaceae). Plant Systematics and Evolution 263–280.
Greilhuber, J., Doležel, J., Lysák, M.A., Bennett, M.D., 2005. The origin, evolution and pro-
posed stabiliza tion of the terms ‘gen ome size’and ‘C-value’to describ e nuclear
DNA contents. Annals of Botany 95, 255–260.
Greilhuber, J., Borsch, T., Müller, K., Worberg, A., Porembski, S., Barthlott, W., 2006 .
Smallest angiosperm genomes found in Lentibulariaceae, with chromosomes of bac-
terial size. Plant Biology 8, 770–777.
Grotkopp, E., Rejmanek, M., Sanderson, M.J., Rost, T.L., 2004. Evolution of genome size in
pines (Pinus) and its life-history correlates: supertree analyses. Evolutio n 58,
Guadalupe, P., M artínez, J., Méndez, I., 2008. Karyo type studies in cultivars of Agave
tequilana Weber. Caryologia 61, 144–153.
Hanson, L., McMahon, K.A., Johnson, M.A.T., Bennett, M.D., 2001. First nuclear DNA C-
values for 25 angiosperm families. Annals of Botany 87, 251–258.
Harder, L.D., 1998. Pollen-size comparisons among animal pollinated angiosperms with
different pollination characteristics. Biological Journal of the Linnean Society 64,
Hawkins, J.S., Kim, H.R., Nason, J.D., Wing, R.A., Wendel, J.F., 2006. Differential lineage-
speciﬁcampliﬁcation of transposable elements is responsible for genome size varia-
tion in Gossypium. Genome Research 16, 1252–1261.
Hegarty, H.J., Hiscock, S.J., 2004. Hybrid speciation in plants: new insights from molecular
studies. New Phytologist 165, 411–423.
Hidalgo,O., Mathez, J., Garcia, S., Garnatje, T., Pellicer, J., Vallès, J., 2010.Genome size study
in the Valerianaceae: ﬁrst results and new hypothes es. Journal of Botany 2010,
Hilliard, O.M., Burtt, B.L., 1971. Streptocarpus: An African Plant Study. University of Natal
Ho, L.S.T., Ané, C., 2014. A linear-time algorithm for Gaussian and non-Gaussian trait evo-
lution models. Systematic Biology 63, 397–408.
Höfer, M., Meister, A., 2010. Genome size variation in Malus species. Journal of Botany
Hoffmann,M.H., Schmuths, H., Koch, C., Meister, A., Fritsch,R.M., 2010. Comparative anal-
ysis of growth, genome size, chromosome numbers and phylogeny of Arabidopsis
thaliana and three cooccurring species of the Brassicaceae from Uzbekistan. Journal
of Botany 2010, 504613.
Johnston, J.S., Bennett, M.D., Rayburn, A.L., Galbraith, D.W., Price, H.J., 1999. Reference
standards for determination of DNA content of plant nuclei. American Journal of Bot-
any 86, 609–613.
Jong, K., 1991. Cytological observations in Albuca, including a survey of polymorphic var-
iation in the Sat-chromosome pair. Edinburgh Journal of Botany 48, 107–128.
Jong, K., Möller, M., 2000. New chromosome counts in Streptocarpus (Gesneriaceae) from
Madagascar and the Comoro Islands and theirtaxonomic signiﬁcance. Plant System-
atics and Ev olution 2 24, 173–182.
Jong, K., Christie, F., Paik, J.H., Scott, S.M., Möller, M., 2012. Unusual morphological andan-
atomical features of two woody Madagascan endemics, Streptocarpus papangae and
S. suffruticosus (Gesneriaceae), and their potential taxonomic value. South African
Journal of Botany 80, 44–56.
Kang, M., Tao, J.J., Wang, J., Ren, C., Qi, Q.W., Xiang, Q.Y., Huang, H.W., 2014. Adaptive and
nonadaptive genome size evolution in Karst endemic ﬂora of China. New Phytologist
Knight, C.A., Clancy, R.B., Götzenberger, L., Dann, L., Beaulieu, J.M., 2010. On the relation-
ship between pollen size and genome size. Journal of Botany 2010, 612017.
Kubis, S., Schmidt, T., Heslop-Harrison, J.S., 1998. Repetitive DNA elements as a major
component of plant genomes. Annals of Botany 82, 45–55.
Lawrence, W.J.C., Scott-Moncrieff, R., Sturgess, V.C., 1939. Studies in Streptocarpus I. Ge-
netics and chemistry of ﬂower colour in the garden strains. Journal of Genetics 38,
Leitch, I.J., Bennett, M.D., 2004. Genome downsizingin polyploid plants. Biological Journal
of the Linnean Society 82, 651–663.
Leitch, I.J., Bennett, M.D., 2007. Genome size and its uses: the impact of ﬂow cytometry.
In: Doležel, J., Greilhuber, J., Suda, J. (Eds.), Flow Cytometry With Plant Cells: Analysis
of Genes, Chromosomes and Genomes. Wiley-VCH, Weinheim, pp. 153–176.
Leitch, I.J., Soltis, D.E., Soltis, P.S., Bennett, M.D., 2005. Evolution of DNA amounts across
land plants (Embryophyta). Annals of Botany 95, 207–217.
Lomax, B.H., Woodward,F.I., Leitch, I.L.,Knight, C.A., Lake, J.A., 2009. Genome size as a pre-
dictor of guard cell length in Arabidopsis thaliana is independent of environmental
conditions. New Phytologist 181, 311–314.
López, H.A., Anton, A.M., Galetto, L., 2006. Pollen-pistil size correlation and pollen size-
number trade-off in species of Argentinian Nyctaginaceae with different pollen re-
serves. Plant Systematics and Evolution 256, 69–73.
343M. Möller / South African Journal of Botany 114 (2018) 323–344
López, A., Panseri, A.F., Poggio, L., Fernández, A., 2011. Nuclear DNA content in the poly-
ploid complex Turnera ulmifolia (Turnera L., Passiﬂoraceae). Plant Systematics and
Evolution 296, 225–230.
Loureiro, J., Rodriguez, E., Doležel, J., Santos, C., 2007. Two new nuclear isolation buffers
for plant DNA ﬂow cytometry: a test with 37 species. Annalsof Botany 100, 875–888.
Lundberg, M., Töpel, M., Eriksen, B., Nylander, J.A.A., Eriksson, T., 2009. Allopolyploidy in
Fragariinae (Rosaceae): comparing four DNA sequence regions, with comments on
classiﬁcation. Molecular Phylogenetics and Evolution 51, 269–280.
Meagher, T.R., Vassiliadis, C., 2005. Phenotypic impacts of repetitive DNA in ﬂowering
plants. New Phytologist 168, 71–80.
Meagher, T.R., G illies, A.C.M., Costich, D.E., 2005. Genome size, quantitative genetics
and the genomic basis for ﬂower size evolution in Silene latifolia. Annals of Botany
Melhem, T.S.A., Mauro, C., 1973. Pollen morphological studies in Gesneriaceae. Hoehnea
Milne, C., 1975. Chromosome numbers in the Gesneriaceae: V. Notes from the Royal
Botanic Garden Edinburgh 33, 523–525.
Mishiba, K.I., Ando, T., Mii, M., Watanabe, H., Kokubun, H., Hashimoto, G., Marchesi, E.,
2000. Nuclear DNA content as an indexcharacter discriminating taxain the genus Pe-
tunia sensu Jussieu (Solanaceae). Annals of Botany 85, 665–673.
Möller, M., Cronk, Q.C.B., 2001a. Evolution of morphological novelty: a phylogenetic anal-
ysis of growth patterns in Streptocarpus (Gesneriaceae). Evolution 55, 918–929.
Möller, M., Cronk, Q.C.B., 2001b. Phylogenetic studies in Streptocarpus (Gesneriaceae):
reconstruction of biogeographic history and distribution patterns. Systematics and Ge-
ography of Plants 71, 545–555.
Möller, M., Pullan, M., 2015. RBGE WebCyte2 —an updated Gesneriaceae cytology data-
base. http://elmer.rbge.org.uk/Webcyte/webcyteintro.php (onwards).
Möller, M., Brooks, K.J., Hughes, M., 2004. Plastid inheritance in Streptocarpus
(Gesneriaceae) and an inferred hybrid origin for a population of S. aff. primulifolius
from Igoda River, South Africa. Edinburgh Journal of Botany 60, 389–408.
Möller, M., Kokubugata, G., Jong, K., 2008. Aspects of genome evolution in Gesneriaceae:
patterns of 45S-nrDNA site evolution based on evidence from ﬂuorescent in-situ hy-
bridization (FISH). International Journal of Plant Sciences 169, 667–676.
Muller, J.,1979. Form and function in angiosperm pollen. Annals of the Missouri Botanical
Garden 66, 593–632.
Neumann, P., Koblizkova, A., Navratilova, A., Macas, J., 2006. Signiﬁcant expansion of Vicia
pannonica genome size mediated by ampliﬁcation of a single type of giant
retroelement. Genetics 173, 1047–1056.
Nishii, K.,Hughes, M., Briggs, M.,Haston, E., Christie, F., DeVilliers, M.J., Hanekom, T., Roos,
W.G., Bellstedt, D.U., Möller, M., 2015. Streptocarpus redeﬁned to include all Afro-
Malagasy Gesneriaceae: mo lecular phylogenies prove congruent with geography
and cytology and uncovers remarkable morphological hom oplasies. Taxon 64 ,
Nylander, J.A.A., 2004. MrModeltest v2. Program Distributed by the Author. Evolutionary
Biology Centre, Uppsala University, Sweden.
Oliver, M.J., Petrov, D., Ackerly, D., Falkowski, P., Schoﬁeld, O.M., 2007. The mode and
tempo of genome size evolution in eukaryotes. Genome Research 17, 594–601.
Parisod, C., Holderegger, R., Brochmann, C., 2010. Evolutionary consequences ofautopoly-
ploidy. New Phytologist 186, 5–17.
Pellicer,J., Fay, F.M., Leitch, I.J., 2010. Thelargest eukaryotic genome of themall? Botanical
Journal of the Linnean Society 164, 10–15.
Petrov, D.A., 2001. Evolution of genome size: new approaches to an old problem. Trends
in Genetics 17, 23–28.
Petrov, D.A., 2002. Mutational equilibrium model of genome size evolution. Theoretical Pop-
ulation Biology 61, 531–544.
Petrova, G., Dzhambazova, T., Moyankova, D., Georgieva, D., Michova, A., Djilianov, D.,
Möller, M., 2014. Morphological variation, genetic diversity and genome size of crit-
ically endangered Haberlea (Gesneriaceae) populations in Bulgaria do not support
the recognition of two different species. Plant Systematics and Evolution 300, 29–41.
Petrova, G., Moyankova, D., Nishii, K., Forrest, L., Tsiripidis, I., Drouzas, A.D., Djilianov, D.,
Möller, M., 2015. The European paleoendemic Haberlea rhodopensis (Gesneriaceae)
has an Oligocene origin and a Pleis tocene diversiﬁcation and occurs in a long-
persisting refugialarea in southeasternEurope. International Journalof Plant Sciences
Puglisi, C., Wei, Y.G., Nishii, K., Möller, M., 2011. Oreocharis × heterandra (Gesneriaceae):
a natural hybrid from the Shengtangshan Mountains, Guangxi, China. Phytotaxa 38,
R Core Team, 2017. A Language and Environment for Statistical Computing. R Foundation
for Statistical Computing, Vienna, Austria Available at. https://www.R-project.org.
Revell, L.J.,2012. phytools: an R packagefor phylogenetic comparative biology (and other
things). Methods in Ecology and Evolution 3, 217–223.
Ronquist, F., Teslenko,M., Van der Mark, P., Ayres, D., Darling, A., Höhna, S., Larget, B., Liu,
L., Suchard, M.A., Huelsenbeck, J.P., 2012. MrBayes 3.2: Efﬁcient Bayesian phylogenet-
ic inference and model choice across a large model space. Systematic Biology 61:
SanMiguel, P., Bennetzen, J.L., 1998. Evidence that a recent increase in maize genome size
was caused by the massive ampliﬁcation of intergene retrotransposons. Annals of
Botany 82, 37–44.
SanMiguel, P., Tikhonov, A., Jin, Y.K., Motchoulskaia, N., Zakharov, D., Melake-Berhan, A.,
Springer, P.S., E dwards, K.J., Lee, M., Avramova, Z., Bennetzen, J.L., 19 96. Nested
retrotranspos ons in the intergenic regions of the maize genome. Science 274,
SanMiguel, P., Gaut, B.S., Tikhonov, A., Nakajima, Y., Bennetzen,J.L., 1998. The paleontolo-
gy of intergene retrotransposons of maize. Nature Genetics 20, 43–45.
Schluter, D., Price, T., Mooers, A.Ø., Ludwig, D., 1997. Likelihoodof ancestor states in adap-
tive radiation. Evolution 51, 1699–1711.
Siljak-Yakovlev, S., Stevanovic, V., Tomasevic, M., Brown, S.C., Stevanovic, B., 2008. Ge-
nome size variation and polyploidy in the resurrection plant genus Ramonda:
cytogeography of livingfossils. Environmental andExperimental Botany 62, 101–112.
Smith, J.B., Bennett, M.D., 1975. DNA variation in the genus Ranunculus. Heredity 35,
Snodgrass, S.J., Jareczek, J., Wendel, J.F., 2017. An examination of nucleotypic effects in
diploid and polyploid cotton. AoB Plants 9, plw082. https://doi.org/10.1093/aobpla/
Soltis, D.E., Soltis, P.S., Bennett, M.D., Leitch, I.J., 2003. Evolution of genome size in the
angiosperms. American Journal of Botany 90, 1596–1603.
Stebbins, G.L., 1940. The signiﬁcance of polyploidy in plant evolution. American Naturalist
Stebbins, G.L., 1950. Variation and Evolution in Plants. Columbia University Press, New
Swofford, D.L., 2002. PAUP*. Phylogenetic Analysis Using Parsimo ny (*and Other
Methods). Version 4. Sinauer Associates, Sunderland, Massachusetts.
Toderich, K.N., Shuyskaya, E.V., Ozturk, M., Juylova, A., Gismatulina, L., 2010. Pollen mor-
phology of some Asiatic species of genus Salsola (Chenopodiaceae) andits taxonomic
relationships. Pakistan Journal of Botany 42, 155–174.
Torres, C.,2000. Pollen size evolution correlation betweenpollen volume and pistillength
in Asteraceae. Sexual Plant Reproduction 12, 365–370.
Vitte, C., Bennetzen, J.L., 2006. Analysis of retrotransposon structural diversity uncovers
properties and propensities in angiosperm genome evolution.Proceedings of the Na-
tional Academy of Sciences of the United States of America 103, 17638–17643.
Vižintin,L., Bohanec, B., 2008. Measurement of n uclear DNA content of t he genus Trifolium
L. as a measure of genebankaccession identity. Genetic Resources andCrop Evolution
Weber, A., Clark , J.L., Möller, M., 2013. A new formal classiﬁcation of Gesneriaceae.
Selbyana 31, 68–94.
Weigend, M., Edwards, T.J., 1996. The palynology of Streptocarpus and the other African
and Malagasy Gesneriaceae and its systematical implications. Botanische Jahrbücher
für Systematik 1 18, 59–80.
Wodehouse, R.P., 1935. Pollen Grains. McGraw-Hill, New York.
Wolfe, K.H., 2001. Yesterday's polyploids and the mystery of polyploidization. Nature Re-
views Genetics 2, 333–341.
Wood, T.E., Takebayashi, N., Barker, M.S., Mayrose, I., Greenspoon, P.B., Rieseberg, L.H.,
2009. The frequency of polyploid speciation in vascularplants. Proceedingsof the Na-
tional Academy of Sciences of the United States of America 106, 13875–13879.
Zaitlin,D., Pierce, A.J., 2010. Nuclear DNA content in Sinningia (Gesneriaceae); intraspecif-
ic genome size variation and genome characterization in S. speciosa. Genome 53,
Zonneveld, B.J.M., 2010. Genomes in Hepatica Mill: (Ranunculaceae) show a relati ve loss
of DNA, not a gain, in polyploids. Journal of Botany 2010, 758260.
Zonneveld, B.J.M., Leitch, I.J.,Bennett, M.D., 2005. First nuclear DNA amounts in more than
300 angiosperms. Annals of Botany 96, 229–244.
344 M. Möller / South African Journal of Botany 114 (2018) 323–344