Genes & Genetic Systems
Online ISSN : 1880-5779
Print ISSN : 1341-7568
ISSN-L : 1341-7568
Full papers
Development of microsatellite markers for the annual andromonoecious herb Commelina communis f. ciliata (Commelinaceae)
Koki R. KatsuharaNaoyuki NakahamaTaketo KomuraMasaya KatoYuko MiyazakiYuji IsagiMotomi ItoAtushi Ushimaru
ジャーナル オープンアクセス HTML

2019 年 94 巻 3 号 p. 133-138


Commelina communis f. ciliata (Commelinaceae), a newly distinguished taxon, is an annual andromonoecious herb exhibiting a mixed mating system, the details of which remain unclear. We developed microsatellite markers for use in exploring the evolution of andromonoecy and mixed mating in the species. Fifteen microsatellite loci were developed using next-generation sequencing. The primer sets were used to evaluate 65 C. communis f. ciliata individuals from three populations in Japan; we found 1–13 alleles per locus and the expected heterozygosity ranged from 0.00 to 0.76. The markers are potentially useful to examine intra- and interspecies genetic structure and the mixed mating strategy of Commelina species via paternity analysis.


Commelina L. is cosmopolitan and is the largest genus (comprising about 170 species) within the family Commelinaceae (Faden, 1998). Commelina communis is an annual herb native to temperate and subtropical regions of eastern and northern Southeast Asia, often growing around rice fields and on roadsides (Satake et al., 1982; Ushimaru et al., 2014). The species has been introduced to non-native regions including Europe and North America. Commelina communis exhibits morphological and chromosomal diversity; several varieties and forms have been described (Fujishima, 2003). Commelina communis f. ciliata Pennell is distinguished from C. communis by the presence of bract hair and the number of chromosomes. Commelina communis f. ciliata and C. communis usually have 2n = 44 or 46 and 2n = 86 or 88, respectively (Fujishima, 2003, 2010, 2017). The two taxa often grow sympatrically but do not hybridize (Fujishima, 2010; Katsuhara and Ushimaru, 2019). We also found significant differences in floral morphologies between the two taxa (Fig. 1; K. R. Katsuhara and A. Ushimaru, unpublished data). These recent findings suggest that C. communis f. ciliata should be a new taxon, although sufficient taxonomic studies have not yet been made.

Fig. 1.

Flower (upper) and short stamen (lower) morphologies of Commelina communis (left) and C. communis f. ciliata (right). Commelina communis f. ciliata is distinguished from C. communis by the presence of bract hair and the shape of short stamens; the stamen shape is illustrated as a silhouette in the circle for each taxon. These pictures are modified from Katsuhara (2017,

Like C. communis, C. communis f. ciliata is andromonoecious and has a relatively high pollen:ovule ratio of ca. 1,300–1,700 for perfect flowers, comparable to ca. 1,000–1,700 in C. communis (Katsuhara and Ushimaru, 2019). Flowers of C. communis f. ciliata are frequently visited by syrphid flies and bees, as are the flowers of C. communis (Ushimaru et al., 2007). Commelina communis f. ciliata can reproduce via both bud pollination and autonomously delayed self-pollination, suggesting that this taxon exhibits a mixed mating system, the details of which remain unclear. Accordingly, studies of the mating systems (e.g., estimates of selfing rates and pollen transfer distances via paternity analysis using microsatellite markers) are necessary to clarify the relative contributions of pollinator-mediated outcrossings to reproductive success and to explore the evolution of andromonoecy. Microsatellite markers have already been developed for the closely related C. communis (Li et al., 2015), which is considered to be tetraploid (Fujishima, 2010). In a preliminary experiment, we found that only four of the twelve markers developed by Li et al. were applicable for analyses of C. communis f. ciliata: one more marker was amplified in more than half of the samples (Supplementary Table S1). Additionally, these previously developed markers may not be suitable for analyzing a large number of samples because they have different annealing temperatures (Li et al., 2015). Thus, markers specifically developed for C. communis f. ciliata are valuable to reveal the mating system of the genus because it is diploid and more suitable to genetic analyses. In addition, work on the population genetics of C. communis f. ciliata and related varieties and forms may afford new insights into the relationships between genetic and morphological diversity and taxonomy of the C. communis complex. Moreover, we are currently studying pollinator-mediated competition between C. communis and C. communis f. ciliata in natural conditions (Katsuhara and Ushimaru, 2019), and therefore genetic analyses using both the existing C. communis markers and our new C. communis f. ciliata markers will reveal the pollen flow dynamics of each taxon under pollinator-mediated competition. Thus, we developed 15 nuclear microsatellite markers for C. communis f. ciliata and explored their applicability to the study of related taxa.


Genomic DNA was extracted from a fresh leaf sample of C. communis f. ciliata collected in Hyogo Prefecture, Japan, using a DNeasy Plant Mini Kit (Qiagen, Germantown, MD, USA). A DNA fragment library was constructed and amplified using an Ion Xpress Plus Fragment Library Kit (Thermo Fisher Scientific, Waltham, MA, USA) and an Ion PGM Hi-Q OT2 Kit, respectively. This was sequenced employing an Ion PGM Hi-Q Sequencing Kit and an Ion 318 Chip v2 BC (both from Thermo Fisher Scientific); 351,799 reads of mean size 237 bp were obtained (DNA Data Bank of Japan Sequence Read Archive accession no.: DRA006993). We screened these reads using PRIMER3 software (Rozen and Skaletsky, 2000) embedded in MSATCOMMANDER version 0.8.2 (Faircloth, 2008) to identify microsatellite loci containing seven or more dinucleotide repeats and eight or more trinucleotide repeats. Based on 128 putatively identified microsatellite loci, we designed a total of 44 primer pairs with 37–71% GC content, annealing temperatures of 57–61 ℃, and fragment sizes of 150–317 bp. These primers were used for DNA amplification of four C. communis f. ciliata individuals. We labeled all forward primers of M13 with one of three fluorescent tag sequences (Boutin-Ganache et al., 2001). All PCR amplifications were performed in a final volume of 6 μl, including approximately 16 ng template DNA, 3 μl 2× Multiplex PCR Master Mix (Qiagen), and primers as follows: 0.05 μM forward primer, 0.1 μM reverse primer and 0.05 μM M13 (fluorescently labeled) primer. One M13 primer (NED) and primer pairs tagged by NED were added to levels five-fold greater than indicated above to adjust the fluorescence level. The PCR thermal profile was as follows: initial denaturation at 95 ℃ for 15 min; 35 cycles of 94 ℃ for 30 s, 57 ℃ for 1.5 min and 72 ℃ for 1 min; and a final extension at 60 ℃ for 30 min. The PCR products were loaded onto an autosequencer (GenomeLab GeXP, Beckman Coulter, Indianapolis, IN, USA) and their sizes calculated using Fragment Analysis ver. 8.0 (Beckman Coulter). Fifteen of the 44 primer pairs yielded amplification products from all four individuals and the fragment lengths were scored uniquely (Table 1). Applicability of these 15 primers was tested by examining three populations of C. communis f. ciliata in Nagano (presumably 2n = 48, Fujishima, 2010), Kyoto and Hyogo Prefectures (presumably 2n = 46, Fujishima, 2010), and three related species, C. communis, C. benghalensis and Murdannia keisak in Hyogo (population localities and specimen data are listed in Supplementary Table S2).

Table 1. Characteristics of microsatellite loci of Commelina communis f. ciliata based on all 65 samples
LocusPrimer sequences (5′-3′)Repeat motifFluorescent labelaSize range (bp)GenBank Accession No.
a  Sequences of the fluorescently labeled primers: 6-FAM = 5′–CACGACGTTGTAAAACGAC–3′, NED = 5′–CTATAGGGCACGCGTGGT–3′, VIC = 5′–TGTGGAATTGTGAGCGG–3′.

To examine genetic structure in three populations of C. communis f. ciliata, we performed Bayesian-based clustering using STRUCTURE ver. 2.3.2 with no prior information on population origins (Pritchard et al., 2000). Ten independent runs each for K = 1 to 10 were performed with a burn-in period of 100,000 steps followed by 100,000 Markov chain Monte Carlo iterations. We determined the appropriate number of clusters (K) based on ∆K (Evanno et al., 2005), calculated by the program STRUCTURE HARVESTER (Earl and vonHoldt, 2012). To infer the genetic differentiation among C. communis and C. communis f. ciliata populations, we conducted principal component analysis (PCA) using genotypic distances between each individual based on presence/absence of a focal allele (R Development Core Team, 2010,


In the three populations of C. communis f. ciliata, the range and mean number of alleles per locus were 1 to 13 and 5.11, respectively (the means were 5.13, 4.53 and 5.67 for the Nagano, Kyoto and Hyogo populations, respectively). The mean observed and expected heterozygosities (HO and HE values, respectively) over loci and populations as calculated by GenAlEx version 6.4 (Peakall and Smouse, 2006) were 0.32 and 0.33, respectively. The HO and HE values for each population are shown in Table 2. The inbreeding coefficients (FIS values) were −0.04, −0.06 and 0.16 in the Nagano, Kyoto and Hyogo populations, respectively. We used FSTAT version 2.9.3 (Goudet, 1995) to explore deviation from Hardy–Weinberg equilibrium (HWE), and linkage disequilibrium (LD) between loci. Significant deviations from HWE were observed in three, three and six loci of the Nagano, Kyoto and Hyogo populations, respectively (Table 2), whereas we found no significant LD. These significant deviations from HWE could be the consequences of autonomous selfing in the species (Katsuhara and Ushimaru, 2019). The combined first- and second-parent non-exclusion probabilities for the 15 markers were 0.0320 and 0.0015, respectively, calculated using Cervus 3.0 software (Marshall et al., 1998; Kalinowski et al., 2007). Analyses of related taxa revealed that 13, 7 and 2 loci were amplified and that 10, 4 and 0 loci were polymorphic in C. communis, C. benghalensis and M. keisak (Table 3).

Table 2. Primer screening for 15 microsatellite loci in three Commelina communis f. ciliata populations
LocusTotal ANagano (N = 22)Kyoto (N = 20)Hyogo (N = 23)

A = number of alleles; FIS = fixation index; HE = expected heterozygosity; HO = observed heterozygosity; N = sample size.

a  Numbers in boldface show deviations from Hardy–Weinberg equilibrium.

Table 3. Ranges of amplicon sizes in cross-amplification testing of three related species for 15 microsatellite loci
LocusC. communis (N = 8)C. benghalensis (N = 6)M. keisak (N = 6)

NG = no signal or nonspecific amplification was detected in PCR amplification.

Our STRUCTURE HARVESTER analysis indicated that ∆K was at a maximum when K = 2 (Supplementary Fig. S1 in supplemental information). Results from STRUCTURE suggested that one genetic cluster was represented by the Nagano population and the other consisted of the Kyoto and Hyogo populations (Fig. 2). According to the PCA result, individuals in the Nagano population formed a group that was independent from those in the Kyoto and Hyogo populations of C. communis f. ciliata (Fig. 3). The genetic differentiation between the Nagano and the other two populations may reflect geographic distribution (distance between populations: Nagano vs. Hyogo, ca. 350 km; Nagano vs. Kyoto, ca. 300 km; Hyogo vs. Kyoto, ca. 60 km) or differences in chromosome number (Fujishima, 2010). The PCA result also revealed a high degree of genetic differentiation between C. communis and C. communis f. ciliata populations in Hyogo, although they were geographically very narrowly distributed (Fig. 3).

Fig. 2.

Bar plots for K = 2 of STRUCTURE analysis results. y- and x-axis labels indicate the probability of a sample being assigned to each cluster represented by two different colors and samples grouped by three populations, respectively. K value was determined from ΔK values based on structure estimation ranging from 1 to 10 (Supplementary Fig. S1).

Fig. 3.

Principal component analysis plot of individuals based on genotypic distances. Blue, red and black indicate the populations of Nagano, Kyoto and Hyogo, respectively. Axis PC1 explains 20.59% of the variance and axis PC2 explains 13.44% of the variance.

We characterized 15 microsatellite loci of C. communis f. ciliata that are useful in terms of paternity analysis and for revealing mixed mating strategies. Although we added 13 new microsatellite loci for the closely related C. communis, only seven out of 15 loci could be applied to C. benghalensis. The results indicate that species-specific markers are necessary for each Commelina species. Our work will contribute to understanding genetic structure and the evolution of andromonoecy in Commelina species.


The authors thank S. Kurata and N. Ishikawa for help with the experiments. We also thank M. Yokogawa and Y. Yaida for help with preparing the voucher specimens and sampling, respectively. This work was supported by Grants-in-Aid for Scientific Research Programs (nos. 16K07517 and 17J01902) from the Japan Society for the Promotion of Science.

© 2019 by The Genetics Society of Japan