Biochemical Systematics and Ecology 69 (2016) 91e100

Contents lists available at ScienceDirect

Biochemical Systematics and Ecology

journal homepage: www.elsevier.com/locate/biochemsyseco

Studies on genetic diversity and phylogenetic relationships of

limpograss (Hemarthria altissima) and related species based
on combined chloroplast DNA intergenic spacer data
Jian Zhang a, **, 1, Xiu Huang b, 1, De-jun Huang a, Yu Zhang c, Lin-kai Huang b, *,
Lu Lu d, Han-dong Yan b
a
Chongqing Academy of Animal Sciences, Herbivorous Livestock Research Institute, Rongchang, Chongqing, 404100, China
b
Department of Grassland Science, Animal Science and Technology College, Sichuan Agricultural University, Chengdu, Sichuan, 611130,
China
c
Institute of Agrifood Research and Technology (IRTA), Centre for Research in Agricultural Genomics (CRAG), CSIC-IRTA-UAB-UB,
Barcelona, 08193, Spain
d
Department of Biochemistry and Molecular Biology, Biosciences Faculty, Universitat Auto
noma de Barcelona, Cerdanyola del Vall

es,
Barcelona, 08193, Spain

a r t i c l e i n f o a b s t r a c t

Article history: Hemarthria R. Br. is a genus which includes important forage grasses. However, there is
Received 25 April 2016 currently a lack of data analysis on the chloroplast DNA (cpDNA) of Hemarthria species.
Received in revised form 27 July 2016 This study is to use three cpDNA intergenic spacers (trnL-F, trnC-ycf6 and psbC-trnS) to
Accepted 30 July 2016
obtain phylogenetic information in 36 Hemarthria samples including four Hemarthria
species: Hemarthria altissima (Poir.) Stapf et C. E. Hubb., Hemarthria compressa (L. f.) R. Br.,
Hemarthria uncinata R. Br., and Hemarthria japonica (Hack.) Roshev. Data analysis revealed
Keywords:
that non-significant genetic diversity existed in our samples, which was implied by
Chloroplast DNA
Hemarthria altissima (Poir.) Stapf et C. E.
nucleotide sequences information and the results of haplotypic and nucleotide diversity.
Hubb. The results of phylogenetic trees based on maximum likelihood (ML) and Bayesian infer-
Hemarthria compressa (L. f.) R. Br. ence (BI) revealed that H. altissima and H. compressa samples were not entirely distinct,
Genetic diversity suggesting that the two species share an intimate genetic relationship. A haplotype
Genetic structure median-joining (MJ) network revealed broadly similar results to those derived from the ML
trnL-F and BI trees and implied that haplotype H3 may represent an ancient haplotype. Analysis
trnC-ycf6 of the population statistic FST revealed little genetic differentiation among the seven
psbC-trnS
populations of H. altissima in Africa.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction

The genus Hemarthria R. Br. in the Poaceae family consists of about 20 species that are geographically widely distributed.
Among these species, the two that are the most important agriculturally and widely studied are Hemarthria altissima (Poir.)

* Corresponding author.
** Corresponding author.
E-mail addresses: [email protected] (J. Zhang), [email protected] (L.-k. Huang).
1
These authors contributed equally to this work.

http://dx.doi.org/10.1016/j.bse.2016.07.024
0305-1978/© 2016 Elsevier Ltd. All rights reserved.
92 J. Zhang et al. / Biochemical Systematics and Ecology 69 (2016) 91e100

Stapf et C. E. Hubb. and Hemarthria compressa (L. f.) R. Br. H. altissima, also known as limpograss, is a warm-season perennial
grass that is native to Africa (Yang et al., 2004; Newman et al., 2011). After being introduced to Florida, limpograss has been
widely used by beef cattle producers as a summer forage plant in the southeastern part of the United States due to its high
forage yield, high quality, and high tolerance of poorly drained soils (Newman et al., 2011). H. compressa (whipgrass), a relative
of limpograss, is another particularly important species of Hemarthria and is mainly found in China. These two species have
been extensively used and commercially grown in subtropical and tropical areas, and made important contributions to agro-
animal husbandry ecosystem development and maintenance (Yang et al., 2004).
Despite their economic and ecological importance, there is a serious lack of genomic information available for H. altissima
and H. compressa. Before the 21th century, studies on H. altissima and H. compressa were mainly focused on naturally
occurring germplasm morphology (Schank et al., 1973), cytology (Quesenberry et al., 1982), stress resistance (Hudson, 1986),
and breeding (Yang et al., 2004). In recent years, molecular studies, including the estimation of genetic diversity, construction
of DNA fingerprints, and association of molecular markers with agronomically important traits, have been conducted on H.
altissima and H. compressa using a variety of molecular markers. However, all of these markers have so far been based on
nuclear DNA (Huang et al., 2008, 2014, 2014a; uang et al., 2014; Huang et al., 2014a, 2014b; Chen et al., 2011). To date, there
have been no molecular research studies based on analysis of chloroplast DNA (cpDNA) in Hemarthria.
Outside of the cell nucleus, the cpDNA of higher plants represents an additional source of genetic information. The
chloroplast genomes of most angiosperms exhibit a simple, circular structure with a low substitution rates and high con-
servation, and they are inherited in a non-Mendelian, uniparental fashion (Clegg et al., 1994). Currently, cpDNA sequence data
is widely used to study the origin, evolution, and genetic diversity of natural populations of the majority of important plant
species (Juszczak et al., 2012). Although cpDNA is conserved in plants, non-coding and coding regions differ greatly in the
evolution rate (Yan et al., 2015). The non-coding regions are more hypervariable and are richer in variations such as sub-
stitutions, translocations, inversions, insertions, and deletions (Katayama et al., 2012). These represent ideal segments for
interspecific phylogenetic studies at low taxonomic levels, such as hybrid cultivar identification, reconstruction of genetic
relationships of plants species, and intraspecific phylogeographic studies (Bakker et al., 1999; Kimura et al., 2003).
Previous molecular research on Hemarthria based on DNAmarkers revealed distinctive genetic differences and extensive
genetic diversity in wild plant resources. These studies also suggested a close genetic relationship between H. altissima and H.
compressa. However, in-depth analysis of the relationship between these two significant Hemarthria species is lacking.
Therefore, in this study, we sought to take advantage of the virtues of cpDNA non-coding regions by assessing three cpDNA
intergenic spacers (trnL-F, trnC-ycf6, and psbC-trnS) in 36 Hemarthria germplasm resources (including 26 H. altissima, 8 H.
compressa, 1 Hemarthria uncinata R. Br., and 1 Hemarthria japonica (Hack.) Roshev.). Our aims were mainly to (1) estimate the
genetic diversity of the 36 Hemarthria samples based on cpDNA sequence variation, (2) explore phylogenetic relationships
between H. altissima and H. compressa, and (3) reveal the genetic structure of African H. altissima. Our results serve to enrich
the existing knowledge on the genetic diversity of H. altissima and H. compressa, providing useful information for the
development of conservation strategies for Hemarthria and for future phylogenetic and phylogeographic studies of
Hemarthria species.

2. Material and methods

2.1. Plant materials

In this study, 26 H. altissima samples, one H. compressa, and one H. uncinata included, were collected from the USDA
Germplasm Resources Information Network (GRIN) program in October 2012 (Huang et al., 2014). Another seven H. compressa
samples and one H. japonica sample were obtained from the Sichuan Agricultural University (Yaan), Sichuan, China. Detailed
information on these 36 Hemarthria samples is summarized in Table 1. Unfortunately, records of two samples were ambig-
uous. One sample was described as deriving from ‘South Africa’, which could refer to either the Limpopo or KwaZulu-Nata
region. Another source was recorded as simply ‘Japan’, which may refer to the city of Yokohama. All plant samples were
maintained as rhizomes. Fresh and young leaf tissues from each sampled individual clone were harvested and dried at room
temperature in collection bags with color silica gel.

2.2. DNA extraction

Dry and young leaf tissues were first ground by a Tissue Lyser (Grinder, Beijing, China) and were later used to be extracted
genomic DNA using a Plant Genomic DNA Kit (Tiangen, Beijing, China) in accordance with the manufacturer's directions. DNA
quality was analyzed by 0.8% (w/v) agarose gel electrophoresis, and DNA concentration was quantified by NanoDrop 2000
spectrophotometer (Thermo, USA). The DNA stock was diluted to a working concentration of 20 ng/mL and stored at
20  C.

2.3. PCR amplification and sequence determination

A set of universal cpDNA primers were used to amplify the target regions of trnL-F (50 -CGAAATCGGTAGACGCTACG-30 , 50 -
ATTTGAACTGGTGACACGAG-30 ), trnC-ycf6 (50 -CCAGTTCAAATCTGGGTGTC-30 ,50 -CATTAAAGCAGCCCAAGC-30 ) (Demesure et al.,
1995) and psbC-trnS (50 -GGTCGTGACCAAGAAACCAC-30 , 50 -GGTTCGAATCCCTCTCTCTC-30 ) (Murakami et al., 2006). PCR was
 J. Zhang et al. / Biochemical Systematics and Ecology 69 (2016) 91e100 93

Table 1
Sources of the 36 Hemarthria samples.

No. Code Accession Plantid Materials origin Species Types

1 H201 PI 299039 H.A.1 Africa, Zimbabwe Hemarthria altissima Wild materials
2 H228 PI 379613 e Africa, Zimbabwe H. altissima Wild materials
3 H202 PI 299993 Redalta South Africa, Transvaal H. altissima Cultivar
4 H204 PI 299995 Bigalta South Africa, Transvaal H. altissima Cultivar
5 H208 PI 349754 e South Africa, Transvaal H. altissima Wild materials
6 H230 PI 409751 3009 South Africa, Transvaal H. altissima Wild materials
7 H203 PI 299994 Greenalta South Africa, Limpopo H. altissima Cultivar
8 H206 PI 349751 e South Africa, Limpopo H. altissima Wild materials
9 H207 PI 349752 e South Africa, Limpopo H. altissima Wild materials
10 H225 PI 364890 1517 South Africa, Limpopo H. altissima Wild materials
11 H231 PI 410128 1799 South Africa, Limpopo H. altissima Wild materials
12 H232 PI 410129 1806 South Africa, Limpopo H. altissima Wild materials
13 H233 PI 410133 1810 South Africa, Limpopo H. altissima Wild materials
14 H205 PI 349750 e South Africa, KwaZulu-Nata H. altissima Wild materials
15 H211 PI 364863 1036 South Africa, KwaZulu-Nata H. altissima Wild materials
16 H218 PI 364876 1200 South Africa, KwaZulu-Nata H. altissima Wild materials
17 H220 PI 364878 1265 South Africa, KwaZulu-Nata H. altissima Wild materials
18 H221 PI 364881 1284 South Africa, KwaZulu-Nata H. altissima Wild materials
19 H234 PI 410134 1887 South Africa, KwaZulu-Nata H. altissima Wild materials
20 H235 PI 410137 2067 South Africa, KwaZulu-Nata H. altissima Wild materials
21 H236 PI 410138 2068 South Africa, KwaZulu-Nata H. altissima Wild materials
22 H213 PI 364868 1077 South Africa H. altissima Wild materials
23 H227 PI 365145 1697 South Africa, Swaziland H. altissima Wild materials
24 H237 PI 410139 2460 South Africa, Cape Province H. altissima Wild materials
25 H239 PI 413186 e East Africa, Mauritius H. altissima Wild materials
26 H240 PI 508606 385e79 South America, Argentina H. altissima Wild materials
27 H241 PI 404118 Line607 Japan Hemarthria compressa Wild materials
28 H037 e e China, Sichuan, Daxian H. compressa Wild materials
29 H021 e e China, Sichuan, Mianyang H. compressa Wild materials
30 Guangyi e Guangyi China, Sichuan, Guangyi H. compressa Cultivar
31 Yaan e Yaan China, Sichuan, Ya'an H. compressa Cultivar
32 Chonggao e Chonggao China, Chongqing H. compressa Cultivar
33 H046 e e China, Guizhou, Libo H. compressa Wild materials
34 H051 e e China, Yunnan, Qiaojia H. compressa Wild materials
35 H242 PI 400272 e Australia, New South Wales Hemarthria uncinata Wild materials
36 H244 e e China, Heilongjang, meadow of the Songnen Plain Hemarthria japonica Wild materials

carried out in a total volume of 50 mL, containing 6 mL DNA template (20 ng/mL), 2 mL each forward and reverse primers
(10 pmol/mL), 25 mL Premix Taq (TakaRa Taq Version 2.0 plus dye; TakaRa Bio Inc., China), and 15 mL distilled water. PCR cycling
conditions observed the following settings: 5 min pre-denaturation at 94  C; 35 cycles of 1 min denaturation at 94  C, 1 min
annealing at 62e65  C, and 1 min extension at 72  C; and a final 10 min extension at 72  C. PCR products were detected on
1.5% agarose gels at 130 V for 30 min and visualized under UV light. Appropriately sized PCR products were then purified by
gel extraction and sent to the Beijing Genomics Institute for bi-directional gene sequencing.

2.4. Data analysis

Obtained sequences of the 25 African H. altissima samples or all samples pertaining to the three cpDNA intergenic spacers
were respectively spliced using DNASTAR SeqMan and then manually adjusted (Swindell and Plasterer, 1997). The multiple
splice sites of each cpDNA intergenic spacer sequence were aligned using the ClustalW algorithm in MEGA 5.0 and improved
by visual inspection (Yan et al., 2015). A BLASTN search (http://blast.ncbi.nlm.nih.gov) was then performed using default
settings in order to identify homologous sequences in other plants. To assess levels of genetic variation, we estimated the
number of haplotypes (h), haplotype diversity (Hd), and nucleotide diversity (p) of each cpDNA intergenic spacer in the 36
Hemarthria samples using DnaSP 5.1 software (Librado and Rozas, 2009).
To determine whether the aligned sequences of each cpDNA intergenic spacer evolved neutrally in the 25 African H.
altissima, we performed neutrality tests, calculating Tajima's D, Fu and Li's F*, and Fu and Li's D* using DnaSP 5.1 following the
method published by Batnini et al. (2014). Next, mismatch distribution was assessed to identify historical processes such as
population equilibrium and recent demographic expansion, which are characterized by a multimodal distribution and
unimodal distribution, respectively (Hu et al., 2011). These were implemented in DnaSP 5.1 and Arlequin 3.1, respectively
(D'Aloia et al., 2015). These two programs were used to draw plots of mismatch distribution based on different theories.
Since a single cpDNA intergenic spacer sequence does not provide sufficient phylogenetic information, we decided to
combine the three cpDNA sequence data sets using SequenceMatrix 1.7 software for subsequent phylogenetic analysis
(Vaidya et al., 2011). Before doing this, we investigated whether the three cpDNA sequence data sets were appropriate for
pooling based on congruence between sequence datasets (Dolphin et al., 2000). The incongruence length difference (ILD) test
94 J. Zhang et al. / Biochemical Systematics and Ecology 69 (2016) 91e100

was implemented by PAUP 4.0 (Swofford, 1998). The likelihood heterogeneity values in this study were not significant at
P ¼ 0.10, indicating no conflicting phylogenetic signals among the three cpDNA intergenic spacers. Hence, we combined the
data from the trnL-F, trnC-ycf6, and psbC-trnS regions.
Combined data were then tested for substitutional saturation in order to detect whether the sequence data were suited for
phylogenetic reconstruction. We applied Xia's test by calculating the saturation parameters Iss and Iss.c using DAMBE 4.5 (Xia
and Lemey, 2009). In order to provide a more intuitive assessment, substitution saturation was also investigated by plotting
the observed divergence against the corrected pairwise divergence calculated by applying the evolutionary model (GTR)
using DAMBE 4.5 (Xia and Lemey, 2009).
Phylogenetic analyses on the combined data of all materials were performed according to the Maximum likelihood (ML)
approach using PhyML 3.1 (Guindon et al., 2009) and the Bayesian inference (BI) approach with MrBayes 3.2 (Huelsenbeck
and Ronquist, 2001). The best-fitting nucleotide substitution models of sequence evolution with the lowest Akaike infor-
mation criterion (AIC) scores were determined by jModelTest 2.1 (Posada, 2008). Clade credibility of the ML phylogram was
assessed with 100 bootstrap replicates. For BI analysis, default priors were applied to each analysis using four Markov chain
Monte Carlo (MCMC) chains that were run for 2,000,000 generations with sampling frequency assigned at every 100 trees.
Node support was evaluated with posterior probabilities of the tree topology, and parameter values were summarized from
20,000 trees (burn-in of 5000 trees).
To evaluate which ML and BI tree topologies were optimal, several testing procedures were performed, including observed
log-likelihood difference (Obs), approximately unbiased (AU), bootstrap probability (BP), approximate Bayesian posterior
probability (PP), Shimodaira-Hasegawa (SH) test, Kishino-Hasegawa (KH) test, weighted Kishino-Hasegawa (WKH), and
weighted Shimodaira-Hasegawa (WSH) test using CONSEL 0.20 with default settings (Shimodaira and Hasegawa, 2001).
CONSEL calculated p-values from the output (the site-wise log-likelihoods for the trees) of the software package TREE-PUZZLE
(Schmidt et al., 2002).
To infer relationships among haplotypes of the 36 Hemarthria samples, the phylogenetic median-joining (MJ) network
algorithm based on parsimony criteria was performed by Network 4.6 software (Forster et al., 2004). The contribution to
genetic variation and parameter of genetic differentiation (FST) of populations of the 25 African H. altissima samples based on
chloroplast sequences were computed in Arlequin 3.5 (Excoffier and Lischer, 2010).

3. Results

3.1. Chloroplast DNA sequence analysis of 36 Hemarthria samples

3.1.1. cpDNA sequence polymorphisms

Three cpDNA intergenic spacers, trnL-F, trnC-ycf6, and psbC-trnS, were amplified and sequenced independently in 36
Hemarthria samples. After visual inspection and adjustment, the total lengths of the aligned trnL-F, trnC-ycf6, and psbC-trnS
sequences were 803 bp, 862 bp, and 539 bp, respectively. Though the psbC-trnS sequences were incomplete, they contain
some parsimony-informative sites, so we decided to add the sequence data in subsequent analyses. We conducted a BLASTN
search with above sequences, afterwards findingthat the trnL-F sequence exhibited strong homology (approximately 99%
identity) to known sequences of other Hemarthria species (Hemarthria pratensis and Hemarthria longiflora). The trnC-ycf6
sequence showed highhomology (97% identity) to sequences of the awn species Miscanthus sinensis, while the incomplete
psbC-trnS sequence similarly showed high homology (99% identity) to the complete psbC-trnS sequences of M. sinensis and
Miscanthus  giganteus. These results suggest that the sequences obtained in our study indeed belong to Hemarthria, and that
among the data in NCBI's GenBank, other Hemarthria sequences (those of H. pratensis and H. longiflora) or those belonging to
Miscanthus are the most similar to our Hemarthria samples. This highlights the fact that the three cpDNA intergenic spacers
are conserved between Hemarthria and related species of the genus Miscanthus.
As summarized in Table 2, we obtained 803 bp of trnL-F and 862 bp trnC-ycf6 sequence, which included 789 and 845
invariable sites, 14 and 17 variable sites, and four and three parsimony-informative sites, respectively. The incomplete, 539 bp
psbC-trnS intergenic spacer consisted of 532 invariable sites, seven variable sites, and two parsimony-informative sites. The
GC contents of the amplified sequences were 0.312 for trnL-F, 0.359 for trnC-ycf6, and 0.441 for psbC-trnS. The trnC-ycf6
sequence exhibited that highest rate of polymorphism (1.97%) with the most haplotypes (11). Although trnL-F exhibited the
second-highest rate of polymorphism (1.74%) with the second most haplotypes (9), it demonstrated the greatest haplotype

Table 2
Chloroplast DNA sequence polymorphisms in the 36 Hemarthria samples.

cpDNA intergenic spacer NS IS VS PIS GCc h Hd p (10
2)

trnL-F 803 789 14 4 0.312 9 0.575 0.151
trnC-ycf6 862 845 17 3 0.359 11 0.522 0.131
psbC-trnS 539 532 7 2 0.441 7 0.354 0.092
trnL-F/trnC-ycf6/psbC-trnS 2204 2166 38 9 0.362 21 0.851 0.129

Note: NS ¼ number of sites; IS ¼ invariable sites; VS ¼ variable sites (no insertions/deletions); PIS ¼ parsimony informative sites; GCc ¼ GC content;
h ¼ number of haplotypes; Hd ¼ haplotype diversity; p ¼ nucleotide diversity.
 J. Zhang et al. / Biochemical Systematics and Ecology 69 (2016) 91e100 95

diversity (0.579) and nucleotide diversity (0.00151). The incomplete psbC-trnS sequence exhibited the lowest rate pf poly-
morphism (1.30%) with the fewest haplotypes (7), the lowest haplotype diversity (0.354), and the lowest nucleotide diversity
(0.00092). All haplotype sequences had been submitted to NCBI's GenBank, the accession numbers were from KX599423 to
KX599454 (Supplementary Table 1).
According to the ILD test, no conflicting phylogenetic signal was discovered among the three cpDNA intergenic spacers, so
we combined the three sequence datasets for follow-up analyses. The total sequence length of the aligned, combined data was
2204 bp, including 2166 invariable sites and 38 variable sites. Among these variable sites, we explore nine parsimony-
informative sites, including eight sites with two variants and one site with five variants. The overall GC content was 0.362,
and a total of 21 haplotypes were presented. The average values of haplotype diversity (Hd) and nucleotide diversity (p) were
0.851 and 0.00129, respectively (Table 2).

3.1.2. Test for substitution saturation

In order to detect whether the sequence data were suited for phylogenetic reconstruction, the combined sequence data
were tested for substitutional saturation applying Xia's test by calculating the saturation parameters Iss and Iss.c using DAMBE
4.5. Once the parameter Iss.c for a set of sequences was obtained, the Iss value would be calculated and compared. With a Iss
vule less than Iss.c value, we deduced that the sequences experiencing little substitution saturation could be used for
phylogenetic reconstruction. In this study, we obtained the average Iss for subsets of 4, 8, 16, and 32 OTUs (Iss ¼ 0.016, 0.019,
0.023, and 0.027, respectively) and found that they were significantly less than the corresponding Iss.c values (Iss.c ¼ 0.819,
0.733, 0.638, and 0.526, respectively; P ¼ 0.000) assuming a symmetrical topology. This means the combined sequence
dataset of 36 Hemarthria samples is useful for phylogenetic analyses. A saturation plot (Supplementary Fig. 1a) showed a basic
linear regression, further demonstrating that there was little substitution saturation.

3.1.3. Phylogenetic tree inference and topology comparison

To clarify the genetic relationships among the 36 Hemarthria samples, the combined sequences of the three cpDNA
intergenic spacers were used to construct ML and BI phylogenetic trees. The log likelihood scores of 88 substitution models
varied from 3318.4581 to 3417.8324, and the jModelTest program suggested that the best-fit model under AIC was TIM1þI
(-lnL ¼ 3318.8688). The obtained ML and BI trees were presented in Figs. 1 and 2, respectively. Both trees revealed the ex-
istence of three main groups of Hemarthria samples. Each of the two groups contained only one single sample each: H.
uncinata sample H242 (Australia, New South Wales) and H. japonica sample H244 (China, Heilongjang, Songnen Plain
meadow). The third group covered the remaining 34 samples belonging to H. altissima and H. compressa. From these results,
we submitted an inference that hierarchical classification structure was mainly independent of species and geographic origin
of the samples.
The ML and BI trees exhibited similar patterns, though they converged on different topologies. To further evaluate the
optimal tree topology, p-values for the two trees were calculated using CONSEL. As seen in Supplementary Table 2, most p-
values from the ML tree (AU, BP, PP, KH, SH, WKH) were larger than those of the BI tree, indicating that for our experimental
data, the ML method was more appropriate than the BI method.

3.1.4. Median-joining (MJ) network analysis

The relationships among haplotypes in the 36 Hemarthria samples were further assessed by constructing a phylogenetic
median-joining (MJ) network with the combined sequences of the three cpDNA intergenic spacers (trnL-F, trnC-ycf6, and
psbC-trnS) (Fig. 3). According to the network tree, 21 haplotypes were analyzed with a distribution to the four species.
Haplotype 3 (H3), containing 10 H. altissima (H201, H211, H213, H218, H220, H221, H228, H233, H234, and H236) and 4 H.
compressa (‘Chonggao’, H046, H051, and H241) samples, was found to be the most ancestral haplotype. This revealed the close
genetic relationship between H. altissima and H. compressa. The remaining 20 haplotypes were derived directly from the H3
haplotype. The two independent lineages on the ML and BI trees e H. uncinata sample H242 (haplotype H20) and H. japonica
sample H244 (H21) e were segregated in the network from H. altissima and H. compressa by numerous mutations and were
connected to H3 by median vector 2. Haplotypes H10 (H. altissima sample H206) and H12 (H. altissima sample H208) were also
segregated and connected to H3 by MV1. In addition, haplotypes H6 (H202), H9 (H205), H11 (H207), and H16 (H235) were
jointly connected to haplotype H14 (H227, H230, and H232). H13 (H225) was connected to H18 (H239), which was subse-
quently linked to H19 (H240). These results illustrated a broadly similar distribution of samples to those suggested by the ML
and BI trees.

3.2. Genetic structure of African H. altissima

3.2.1. Genetic diversity and neutrality test

Twenty-five H. altissima coming from seven populations were collected from Africa: five populations from South Africa,
one from Zimbabwe, and one from East Africa. We performed saturation analyses of the combined sequence dataset in the 25
H. altissima samples, obtaining an Iss value of 0.0241, which was much less than the Iss.c (0.7951 assuming a symmetrical
topology and 0.5796 assuming an asymmetrical topology). This indicated that the sequences experienced little substitution
saturation, which was further confirmed by a linear regression in the saturation plot (Supplementary Fig. 1b). Next, we
obtained the genetic diversity and performed neutrality tests for the cpDNA sequences of the 25 African samples
96 J. Zhang et al. / Biochemical Systematics and Ecology 69 (2016) 91e100

Fig. 1. Maximum likelihood (ML) tree based on the combined sequences of three cpDNA intergenic spacers (trnL-F, trnC-ycf6, and psbC-trnS) in 36 Hemarthria
samples. Numbers above branches are bootstrap values, which were computed based upon 100 replicates. A: Africa; Au: Australia; SA: South Africa; EA: East
Africa; C: China; SAM: South America; J: Japan.

(Supplementary Table 3). Statistical neutrality tests (Tajima's D, Fu and Li's F*, and Fu and Li's D*) of trnC-ycf6 and the
combined sequence data yielded highly significant (P < 0.05) negative values, but only the observed mismatch distribution of
the combined sequence data was unimodal (Supplementary Fig. 2), closely fitting the expected distribution under a model of
sudden expansion. For trnL-F and psbC-trnS, no distinct signal of population equilibrium or expansion was observed, as the
neutrality tests were not significant and the mismatch distribution did not reflect multimodal or unimodal distributions
(Supplementary Table 3, Supplementary Fig. 2).

3.2.2. Population genetic structure analysis

We calculated the FST value of the seven African populations of H. altissima using Arlequin. As described in Table 3, the FST
was extremely low (0.0298, P ¼ 0.343), indicating little genetic differentiation between these populations. In other words, the
genetic structure between subpopulations was weak (Balloux and Lugon-Moulin, 2002). Moreover, a high percentage of total
variance (97.02%) rooted in within-population differences, while only 2.98% of the variance resulted from differences between
populations. This might be due to the strong capacity of Hemarthria to propagate asexually and its ecological adaptation to
 J. Zhang et al. / Biochemical Systematics and Ecology 69 (2016) 91e100 97

Fig. 2. Phylogenetic tree constructed from the combined data sequences of three cpDNA intergenic spacers (trnL-F, trnC-ycf6, and psbC-trnS) in 36 Hemarthria
samples using Bayesian inference (BI). Numbers above branches are Bayesian posterior probability values. A: Africa; Au: Australia; SA: South Africa; EA: East
Africa; C: China; SAM: South America; J: Japan.

disperse, which could lead to high gene flow between populations, generating genetic relatedness between samples with
further geographic distant.

4. Discussion

This is the first report analyzing the genetic diversity of Hemarthria species using cpDNA molecular markers. Although
cpDNA non-coding regions are generally hypervariable and rich in variation, the 36 Hemarthria samples exhibited relatively
low genetic diversity of the cpDNA intergenic spacers trnL-F, trnC-ycf6, and psbC-trnS (Hd ¼ 0.575, 0.522, and 0.354,
respectively; p ¼ 0.00151, 0.00131, and 0.00092, respectively). The combined sequence data also exhibited low haplotype
diversity (Hd ¼ 0.851) and nucleotide diversity (p ¼ 0.129). This low level of genetic diversity resulted in low efficiency of the
cpDNA markers for resolving Hemarthria genotypes or, in other words, poor detection of variable sites (the highest proportion
of variable sites was only 2.0%). The proportion of variable sites is a major index that reflects the level of genetic diversity.
Similar findings were reported previously for trnC-ycf6 (proportion of variable sites ¼ 0.23%) in Anemoclema glaucifolium
(Guan et al., 2013) and for psbC-trnS (0.13%) in pear germplasm resources (Chang et al., 2014) but stood in contrast to results
98 J. Zhang et al. / Biochemical Systematics and Ecology 69 (2016) 91e100

Fig. 3. Median-joining haplotype network based on the combined sequences of three cpDNA intergenic spacers (trnL-F, trnC-ycf6, and psbC-trnS) in 36 Hemarthria
samples. H and mv denote haplotype and median vector, respectively.

Table 3
Molecular variance and FST in populations of H. altissima based on combined sequence data.

Source of population Source of variation d.f. Sum of squares Variance components Percentage of variation FST
All population Among populations 7 12.357 0.04994 2.98% FST ¼ 0.02980
Within populations 17 27.643 1.62605 97.02% (P ¼ 0.34311)
Total 24 40 1.67599

for trnL-F (44%) in field vittarioid gametophytes (Chen et al., 2013). Moreover, such low genetic diversity in our cpDNA analysis
contradicts previous studies based on nuclear DNA molecular markers (SCoT and EST-SSR markers) with higher levels of
genetic diversity in Hemarthria samples (Huang et al., 2014). A similar situation e low diversity in cpDNA and higher diversity
in nuclear markers e was also found in the Miscanthus sinensis (Yan et al., 2015). Lower genetic diversity in cpDNA non-coding
spacers compared to nuclear genes may be due to the influences of random genetic drift and genetic bottlenecks (Gong et al.,
2011).
Two methods of phylogenetic inference (ML and BI) were applied to group samples to elucidate relationships between
samples. Although the ML and BI trees displayed similar patterns, the hierarchical classification structures were not directly
associated with either the species labels or the geographic distributions of the samples. These results largely conflict with
results from nuclear EST-SSR and SCoT markers (Huang et al., 2014). Samples of the two species H. altissima and H. compressa
could not be clearly distinguished, indicating that they are closely related. This could be explained by the fact that the selected
markers did not cover the entire chloroplast genome, or it could reflect the strong conservation of the trnL-F, trnC-ycf6, and
psbC-trnS regions across plant species. The phylogenetic inferences were found to be consistent with a median-joining
network analysis, which showed that 10 H. altissima samples and 4 H. compressa samples were included together in
Haplotype 3. In addition, the results of the ML and BI analyses showed that some Hemarthria samples that were genetically
divergent were geographically close in proximity, which agrees with the mismatch distribution analysis of the combined data,
the population genetic structure analysis, and previous studies using cpDNA sequences (Batnini et al., 2014). This could be
explained by 1) the strong capacity of Hemarthria individuals to undergo asexual propagation and ecological adaptation,
which may be susceptible to natural factors (e.g., river flow), livestock activity (e.g., foraging), and human activities (e.g.,
domestication and cultivation) resulting in widespread dispersal of Hemarthria clones (Yang et al., 2004); 2) the likely
occurrence of natural hybridization between Hemarthria clones, resulting in the production of new genetic combinations that
may increase the complexity of the genetic and geographic distribution (Wu and Du, 2000); and 3) the probable occurrence of
gene mutations during plant growth.
 J. Zhang et al. / Biochemical Systematics and Ecology 69 (2016) 91e100 99

There is a common belief that branch-support values from each partition of a tree can be interpreted as the probability that
the tree is correct or as an indicator of phylogenetic accuracy (Huelsenbeck et al., 2002). However, some believe that this is
misleading and branch-support values should not be interpreted as probabilities that clades are correctly resolved, as the
validity of ML or BI methods depend on the validity of the likelihood model and subsets of data (Simmons et al., 2004).
Although the bootstrap support values of the ML tree and the Bayesian posterior probability values of the BI tree were quite
weak for most of the clades in this study, we used normative processes to simulate the phylogenetic trees. Our low branch-
support values may be due to the restricted genetic background of the tested samples, as branch-support values are poor
when taxon sampling increases without an increase in genetic diversity (Simmons et al., 2004).
The ML and BI methods have been shown to be accurate in numerous simulation studies (Gill and Fast, 2006) and are now
broadly recognized as better approaches than the neighbor-joining (NJ) or maximum parsimony (MP) methods (Guindon
et al., 2010). In the case that an explicit evolutionary model was fitted for ML or BI, Huang (2012) held the ML method
was more consistent with evolutionary fact. By testing a variety of conditions, Hall (2005) reported that the BI method was the
more accurate. Negrisolo et al. (2004) believed that the ML method was very flexible due to its plasticity, and was theoretically
sound and statistically consistent. For this study, we employed the CONSEL program, which calculates probability values (i.e.,
p-values), to assess the tree topologies of the ML and BI trees in order to reveal which method was better suited to our data.
The results demonstrated that the ML tree exhibited a better topological structure, which may indicate that the ML method is
better suited at illustrating genetic relationships between samples with a narrow genetic background than the BI method.
FST is the most commonly reported statistic for the assessment of genetic differentiation (Balloux and Lugon-Moulin,
2002). The FST value reflects the amount of genetic differentiation between populations. An FST value between 0 and 0.05
indicates little genetic differentiation, from 0.05 to 0.15 suggests moderate differentiation, from 0.15 to 0.25 indicates strong
differentiation, and over 0.25 reflects very strong genetic differentiation (Curnw and Wright, 1979). Our FST of 0.0298
calculated for the seven African populations of H. altissima is generally considered to imply low genetic differentiation;
however, such a low FST value may actually indicate important genetic differentiation that is by no means negligible (Curnw
and Wright, 1979).

Author contributions

Lin-kai Huang and Jian Zhang conceived and designed the experiments; Xiu Huang and Lu Lu performed the experiments;
Xiu Huang, De-jun Huang, Lu Lu, and Han-dong Yan analyzed the data; Jian Zhang, Xiu Huang, De-jun Huang, Yu Zhang, Lin-
kai Huang, Lu Lu, and Han-dong Yan contributed reagents/materials/analysis tools; Jian Zhang and Xiu Huang wrote the
paper.

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgments

The research was supported by Strategic Planning and Management Innovation of Science and Technology Support
Demonstration Project of Chongqing Modern Cattle and Sheep Industry (cstc2014zktjccxbx0021), and Modern Agro-industry
Technology Research System (CARS-35-05). We thank Dr. Fang-luan Gao (Fujian Agriculture and Forestry University) for
comments and suggestions that improved the manuscript.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.bse.2016.07.024.

References

Bakker, F.T., Culham, A., Daugherty, L.C., Gibby, M., 1999. A trnL-F based phylogeny for species of Pelargonium (Geraniaceae) with small chromosomes. Plant
Syst. Evol. 216, 309e324.
Balloux, F., Lugon-Moulin, N., 2002. The estimation of population differentiation with microsatellite markers. Mol. Ecol. 11, 155e165.
Batnini, M.A., Bourguiba, H., Trifi-Farah, N., Krichen, L., 2014. Chloroplast DNA sequence data provides new insights into genetic diversity and phylogenetic
relationships of Tunisian apricot germplasm. Sci. Hortic. 178, 241e247.
Chang, Y.J., Cao, Y.F., Zhang, J.M., Tian, L.M., Dong, X.G., Zhang, Y., Qi, D., Zheng, Y.C., 2014. Studies on genetic diversity of pear germplasm resources in
Liaoning province of China based on chloroplast DNA analysis. Acta Hortic. Sin. 41, 1307e1316.
Chen, C.W., Huang, Y.M., Kuo, L.Y., Nguyen, Q.D., Luu, H.T., Callado, J.R., Farrar, D.R., Chiou, W.L., 2013. trnL-F is a powerful marker for DNA identification of
field vittarioid gametophytes (Pteridaceae). Ann. Bot. 111, 663e673.
Chen, Y.X., Zhang, X.Q., Ma, X., Xie, W.G., 2011. Association analysis between SSR molecular markers and agronomic characters of Hemarthria compressa.
Hubei Agric. Sci. 7, 059.
Clegg, M.T., Gaut, B.S., Learn, G.H., Morton, B.R., 1994. Rates and patterns of chloroplast DNA evolution. Proc. Natl. Acad. Sci. 91, 6795e6801.
Curnw, R.N., Wright, S., 1979. Evolution and the Genetics of Populations, Volume 4: Variability within and among Natural Populations. Biometrics 35, 359.
Demesure, B., Sodzi, N., Petit, R.J., 1995. A set of universal primers for amplification of polymorphic non-coding regions of mitochondrial and chloroplast
DNA in plants. Mol. Ecol. 4, 129e134.
100 J. Zhang et al. / Biochemical Systematics and Ecology 69 (2016) 91e100

Dolphin, K., Belshaw, R., Orme, C.D.L., Quicke, D.L.J., 2000. Noise and incongruence: interpreting results of the incongruence length difference test. Mol.
Phylogenet. Evol. 17, 401e406.
D'Aloia, C.C., Azodi, C.B., Sheldon, S.P., Trombulak, S.C., Ardren, W.R., 2015. Genetic models reveal historical patterns of sea lamprey population fluctuations
within Lake Champlain. PeerJ 3, e1369.
Excoffier, L., Lischer, H.E.L., 2010. Arlequin suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows. Mol.
Ecol. Resour. 10 (3), 564e567.
€hl, A., 2004. Network 4.2. 0.1. Fluxus Technology Ltda. Software available free at www. fluxus-engineering. com.
Forster, P., Bandelt, H.J., Ro
Gill, E.E., Fast, N.M., 2006. Assessing the microsporidia-fungi relationship: combined phylogenetic analysis of eight genes. Gene 375, 103e109.
Gong, X., Luan, S.S., Hung, K.H., Hwang, C.C., Lin, C.J., Chiang, Y.C., Chiang, T.Y., 2011. Population structure of Nouelia insignis (Asteraceae), an endangered
species in southwestern China, based on chloroplast DNA sequences: recent demographic shrinking. J. Plant Res. 124, 221e230.
Guan, M.M., Ma, R., Gong, X., 2013. Conservation genetics of and endemic plant, Anemoclema glaucifolium, in the Jinsha river valley. Plant Divers. Resour. 35,
555e562.
Guindon, S., Delsuc, F., Dufayard, J.F., Gascuel, O., 2009. Estimating Maximum Likelihood Phylogenies with PhyML. Bioinformatics for DNA Sequence
Analysis. Springer, pp. 113e137.
Guindon, S., Dufayard, J.F., Lefort, V., Anisimova, M., Hordijk, W., Gascuel, O., 2010. New algorithms and methods to estimate maximum-likelihood phy-
logenies: assessing the performance of PhyML 3.0. Syst. Biol. 59, 307e321.
Hall, B.G., 2005. Comparison of the accuracies of several phylogenetic methods using protein and DNA sequences. Mol. Biol. Evol. 22, 792e802.
Hu, Z.M., Uwai, S., Yu, S.H., Komatsu, T., Ajisaka, T., Duan, D.L., 2011. Phylogeographic heterogeneity of the brown macroalga Sargassum horneri (Fucaceae) in
the northwestern Pacific in relation to late Pleistocene glaciation and tectonic configurations. Mol. Ecol. 20, 3894e3909.
Huang, L.K., Huang, X., Yan, H.D., Yin, G.H., Zhang, X.Q., Tian, Y., Zhang, Y., Jiang, X.M., Yan, Y.H., Ma, X., 2014a. Constructing DNA fingerprinting of Hemarthria
cultivars using EST-SSR and SCoT markers. Genet. Resour. crop Evol. 61, 1047e1055.
Huang, L.K., Zhang, X.Q., Ma, X., Liu, W., Li, F., Zeng, B., 2008. Genetic differentiation among Hemarthria compressa populations in south China and its genetic
relationship with H. japonica. Hereditas 145, 84e91.
Huang, L.K., Zhang, Y., Zhang, J., Zhang, X.Q., Xie, W.G., Jiang, X.M., Peng, F., Yan, Y.H., Ma, X., Liu, W., et al., 2014b. Genetic stability and DNA fingerprinting of
the Hemarthria compressa cultivar “Guangyi”. Biochem. Syst. Ecol. 55, 310e316.
Huang, X., Zhang, X.Q., Huang, L.K., Ma, Y.M., Yin, G.H., Lee, S., Zeng, J., Liu, H., 2014. Genetic diversity of Hemarthria altissima and its related species by EST-
SSR and SCoT markers. Biochem. Syst. Ecol. 57, 338e344.
Huang, Y., 2012. Molecular Phylogenetics. Science Press, Beijing.
Hudson, W.G., 1986. Mole cricket (Orthoptera: Gryllotalpidae) damage to Hemarthria altissima: resistance or nonpreference? J. Econ. Entomol. 79, 961e963.
Huelsenbeck, J.P., Larget, B., Miller, R.E., Ronquist, F., 2002. Potential applications and pitfalls of Bayesian inference of phylogeny. Syst. Biol. 51, 673e688.
Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics 17, 754e755.
Juszczak, I., Rudnik, R., Pietzenuk, B., Baier, M., 2012. Natural genetic variation in the expression regulation of the chloroplast antioxidant system among
Arabidopsis thaliana accessions. Physiol. Plant. 146, 53e70.
Katayama, H., Tachibana, M., Iketani, H., Zhang, S.L., Uematsu, C., 2012. Phylogenetic utility of structural alterations found in the chloroplast genome of pear:
hypervariable regions in a highly conserved genome. Tree Genet. Genomes 8, 313e326.
Kimura, T., Iketani, H., Kotobuki, K., Matsuta, N., Ban, Y., Hayashi, T., Yamamoto, T., 2003. Genetic characterization of pear varieties revealed by chloroplast
DNA sequences. J. Hortic. Sci. Biotechnol. 78, 241e247.
Librado, P., Rozas, J., 2009. DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25, 1451e1452.
Murakami, A., Darby, P., Javornik, B., Pais, M.S.S., Seigner, E., Lutz, A., Svoboda, P., 2006. Molecular phylogeny of wild hops, Humulus lupulus L. Heredity 97,
66e74.
Negrisolo, E., Minelli, A., Valle, G., 2004. The mitochondrial genome of the house centipede Scutigera and the monophyly versus paraphyly of myriapods.
Mol. Biol. Evol. 21, 770e780.
Newman, Y., Vendramini, J., Sollenberger, L.E., Quesenberry, K., 2011. Limpograss (Hemarthria Altissima): Overview and Management. EDIS Publication. SS-
AGR-320. http://edis.ifas.ufl.edu.
Posada, D., 2008. jModelTest: phylogenetic model averaging. Mol. Biol. Evol. 25, 1253e1256.
Quesenberry, K.H., Oakes, A.J., Jessop, D.S., 1982. Cytological and geographical characterizations of Hemarthria. Euphytica 31, 409e416.
Schank, S.C., Klock, M.A., Moore, J.E., 1973. Laboratory evaluation of quality in subtropical grasses: II. Genetic variation among Hemarthrias in in vitro
digestion and stem morphology. Agron. J. 65, 256e258.
Schmidt, H.A., Strimmer, K., Vingron, M., von Haeseler, A., 2002. TREE-PUZZLE: maximum likelihood phylogenetic analysis using quartets and parallel
computing. Bioinformatics 18, 502e504.
Shimodaira, H., Hasegawa, M., 2001. CONSEL: for assessing the confidence of phylogenetic tree selection. Bioinformatics 17, 1246e1247.
Simmons, M.P., Pickett, K.M., Miya, M., 2004. How meaningful are Bayesian support values? Mol. Biol. Evol. 21, 188e199.
Swindell, S.R., Plasterer, T.N., 1997. Seqman. Methods in Molecular Biology, pp. 75e89.
Swofford, D., 1998. PAUP 4.0: Phylogenetic Analysis Using Parsimony. Encyclopedia of Genetics, Genomics, Proteomics and Informatics, p. 1455.
Vaidya, G., Lohman, D.J., Meier, R., 2011. SequenceMatrix: concatenation software for the fast assembly of multi-gene datasets with character set and codon
information. Cladistics 27, 171e180.
Wu, Y.Q., Du, Y., 2000. Some reproductive characteristics of Hemarthria compressa. J. Sichuan Agric. Univ. 18, 262e264.
Xia, X.H., Lemey, P., 2009. Assessing Substitution Saturation with DAMBE. The Phylogenetic Handbook: a Practical Approach to DNA and Protein Phylogeny,
2, pp. 615e630.
Yan, H.D., Zhang, X.Q., Fu, C., Huang, L.K., Yin, G.H., Nie, G., Xu, W.Z., Liu, X.C., Chen, T., Lee, S., et al., 2015. Chloroplast DNA variation and genetic structure of
Miscanthus sinensis in southwest China. Biochem. Syst. Ecol. 58, 132e138.
Yang, C.H., Zhang, X.Q., Li, X.L., Du, Y., Wu, Y.Q., 2004. Hemarthria germplasm resources and breeding. Acta Prataculturae Sin. 13, 7e12.