Geographical origin of Leucobryum boninense Sull. & Lesq. (Leucobryaceae, Musci) endemic to the Bonin Islands, Japan

Leucobryum boninense is endemic to the Bonin Islands, Japan, and its related species are widely distributed in Asia and the Pacific. We aimed to clarify the phylogenetic relationships among Leucobryum species and infer the origin of L. boninense. We also describe the utility of the chloroplast trnK intron including matK for resolving the phylogenetic relationships among Leucobryum species, as phylogenetic analyses using trnK intron and/or matK have not been performed well in bryophytes to date. Fifty samples containing 15 species of Leucobryum from Asia and the Pacific were examined for six chloroplast DNA regions including rbcL, rps4, partial 5′ trnK intron, matK, partial 3′ trnK intron, and trnL-F intergenic spacer plus one nuclear DNA region including ITS. A molecular phylogenetic tree showed that L. boninense made a clade with L. scabrum from Japan, Taiwan and, Hong Kong; L. javense which is widely distributed in East and Southeast Asia, and L. pachyphyllum and L. seemannii restricted to the Hawaii Islands, as well as with L. scaberulum from the Ryukyus, Japan, Taiwan, and southeastern China. Leucobryum boninense from various islands of the Bonin Islands made a monophylic group that was closely related to L. scabrum and L. javense from Japan. Therefore, L. boninense may have evolved from L. scabrum from Japan, Taiwan, or Hong Kong, or L. javense from Japan. We also described the utility of trnK intron including matK. A percentage of the parsimony-informative characters in trnK intron sequence data (5.8%) was significantly higher than that from other chloroplast regions, rbcL (2.4%) and rps4 (3.2%) sequence data. Nucleotide sequence data of the trnK intron including matK are more informative than other chloroplast DNA regions for identifying the phylogenetic relationships among Leucobryum species.


Introduction
Bryophyte species tend to have broad geographical distribution with a morphological uniformity in comparison with those of seed plants. In the Northern Hemisphere, more than 60% of the flora of the Arctic and boreal regions is made up of the same species (Schofield and Crum 1972). A single sporangium of a bryophyte may contain thousands and sometimes over 50 million spores that have the capacity for long-distance dispersal over thousands of kilometers (Kreulen 1972;van Zanten 1978). Producing abundant air-borne diaspores would appear to guarantee a wide distribution of many bryophyte species (Schofield and Crum 1972). In contrast, extreme geographical isolation such as on oceanic islands affects diversification and speciation, even though bryophyte species have the capability for long-distance dispersal (Oguri et al. 2008). Therefore, oceanic islands may provide models for research on patterns and processes of bryophyte evolution and speciation.
The Bonin (Ogasawara) Islands are oceanic islands located in the northwestern Pacific Ocean, approximately 1000 km south of Tokyo, Japan (Asami 1970). These islands were formed during the Paleocene and rose above sea level before the middle Pleistocene (Kaizuka 1977;Imaizumi and Tamura 1984). Approximately 300 indigenous species of vascular plants are known from these islands, and their percentage of endemism is estimated to be as high as 40-43% (Kobayashi 1978;Ono et al. 1986). A total of 155 species of bryophytes (48 genera and 81 species of mosses, 33 genera and 74 species of liverworts and hornworts) are currently known from the Bonin Islands (Inoue and Iwatsuki 1969, 1970Inoue 1970a,b;Iwatsuki 1985;Furuki et al. 1991). The percentage of bryophyte endemism is approximately 5%, which is much lower than that of vascular plants.
Among bryophyte taxa growing on the Bonin Islands, members of the genus Leucobryum Hampe (Leucobryaceae, Musci) have been taxonomically well studied by Yamaguchi (1993) and Oguri et al. (2008). This genus is one of the most widely distributed moss genera, containing several widespread species. According to van der Wijk et al. (1964), it includes approximately 180 species. Among members of Leucobryum, L. juniperoideum (Brid.) M€ ull.Hal. is widely distributed in Asia, Europe, Macaronesia, and Madagascar, whereas L. glaucum (Hedw.) Angstr. is widely distributed throughout temperate to cool temperate regions in the Northern Hemisphere (Yamaguchi 1993;Vanderpoorten et al. 2003). In contrast, some endemic species are observed on oceanic islands such as the Hawaiian Islands and the Bonin Islands. Leucobryum pachyphyllum M€ ull.Hal. and L. seemannii Mitt. are endemic to the Hawaii Islands (Bartram 1933;Staples et al. 2004), whereas L. boninense Sull & Lesq. (Oguri et al. 2008) is restricted to the Bonin Islands.
Leucobryum boninense is characterized by its perichaetia terminal on short lateral branches and papillose proration on the abaxial surface of apical parts of leaves (Fig. 1;Yamaguchi 1993). This species seems to be closely related to L. scaberulum Cardot based on morphological characters. In fact, L. scaberulum was treated as a synonym of L. boninense by Yamaguchi (1993).
Molecular phylogenetic analyses of the genus Leucobryum have been performed based on sequence data of internal transcribed spacer (ITS) regions of ribosomal DNA and chloroplast rbcL gene. The results showed that the endemic species, L. boninense, is closely related to L. scaberulum, L. scabrum Sande Lac., and L. javense (Brid.) Mitt. (Oguri et al. 2003(Oguri et al. , 2008. All three species are widely distributed; L. javense is widely distributed in East and Southeast Asia, and L. scabrum and L. scaberulum occur in East Asia (Yamaguchi 1993). Nevertheless, two previous molecular phylogenetic studies did not include plant samples from various parts of the distribution areas and were performed based only on ITS and rbcL DNA sequence regions. Therefore, detailed phylogenetic relationships among L. boninense and its related species remain poorly understood. matK, encoding a splicing-associated maturase in the land plant chloroplast genome, is a very popular region for phylogenetic study and has been extensively applied to reconstruct angiosperm phylogeny (Rev. M€ uller et al. 2006). However, the utility of matK in bryophyte phylogeny is largely unknown. Only one molecular phylogenetic study has been conducted by Long et al. (2000), but it was based on partial matK sequence data.
In this study, we collected L. boninense samples and those of its related taxa from various parts of their distribution and performed molecular phylogenetic studies to clarify the phylogenetic relationships among Leucobryum species and to infer the origin of L. boninense, which is restricted to the Bonin Islands. Phylogenetic trees were constructed based on the combined nucleotide sequences of rbcL, rps4, 5′ trnK intron, matK, 3′ trnK intron, trnL-F intergenic spacer, and ITS regions. Moreover, we verified amplification of the matK region for six moss species, in addition to the Leucobryum species and obtained their sequence data using six primers including four new internal primers designed in this study.

Plant materials
Fifty samples belonging to 15 species of Leucobryum were collected from Asia and the Pacific regions (Table 1). Leucobryum sanctum (Brid.) Hampe was used as an outgroup for the phylogenetic analysis, based on a previous molecular phylogenetic study of the entire genus by Oguri et al. (2003). Six additional moss species of different

DNA extraction, PCR amplification, and sequencing
Total DNA was extracted either from fresh samples or dried herbarium specimens using the phenol-chloroform method of Tsubota et al. (2002) with some modifications. Six cpDNA regions, rbcL, rps4, 5′ trnK intron, matK, 3′ trnK intron, and trnL-F intergenic spacer and one nrDNA region, ITS were amplified by PCR using a thermal cycler (Table 2). Each fragment was amplified with PrimeSTAR Max DNA Polymerase (TaKaRa Bio, Otsu, Shiga, Japan) using 10 ll reactions volumes in a thermal cycle with the following conditions: 98°C for 30 sec followed by 30 cycles of 98°C for 10 sec, 55°C for 5 sec, 72°C for 30 sec and 72°C for 30 sec. After confirming PCR amplification on a 1.0% agarose gel, the amplified products were incubated at 37°C for 30 min and 80°C for 20 min with ExoSAP-IT (usb, Cleveland, OH, USA) to remove any excess primers and nucleotides. Eight primers for rbcL, two primers for rps4, six primers for trnK intron including matK, two primers for trnL-F, and five primers for ITS were used for the cycle sequencing reactions (Table 2) with an ABI PRISM BigDye Terminator Cycle Sequencing Kit v.3.1 (Applied Biosystems, Foster City, CA, USA). The sequencing reaction products were purified, concentrated by ethanol precipitation with sodium acetate and their nucleotide sequences were determined using an automated DNA sequencer (ABI PRISM 3100, Applied Biosystems). The obtained sequences were submitted to the DDBJ database (Table 1 and Appendix S1).

Phylogenetic analysis
We obtained rbcL sequence data of 14 samples belonging to 13 taxa and ITS sequence data of 21 samples belonging to 14 taxa of the genus Leucobryum from the DNA database. The obtained rbcL, rps4, 5′ trnK intron, matK, 3′ trnK intron, trnL-F, and ITS sequences were separately aligned using the program MUSCLE (Edgar 2004). We performed the Incongruence Length Difference (ILD) test (Farris et al. 1995) implemented in PAUP* version 4.0 beta (Swofford 2002) before phylogenetic reconstruction to confirm topological congruence between each DNA region. One hundred partition homogeneity replicates were implemented in the test using the heuristic search option with 100 random addition sequences. And then, we performed molecular phylogenetic analyses with combined all six chloroplast DNA plus one nuclear DNA sequences. When these analyses were carried out, identical sequences were pruned to include only one representative from each species. Therefore, a total of 35 operational taxonomic units, including outgroup, were used for the following analyses.
Bayesian inference (BI) analysis was performed using MrBayes version 3.1.2 (Ronquist and Huelsenbeck 2003). The best-fitting model for nucleotide substitution was selected for the combined seven regions based on Akaike information criterion (Akaike 1974) implemented in MrModeltest 2.2 (Nylander 2004), and GTR +I + G model was chosen. The analysis was performed for 1,000,000 generations with four chains, with samples taken every 100 generations.
Maximum likelihood (ML) analysis was conducted with PAUP* 4.0b10 using the best-fitting model GTR + I + G chosen by MrModeltest 2.2. A heuristic search algorithm was engaged with 100 random addition replicates and tree-bisection-reconnection (TBR) branch-swapping, and MulTrees on. The ML bootstrap value were computed in PAUP* 4.0b10 by running 1000 replicates with a full heuristic search using 100 random addition sequences, TBR branch-swapping, and MulTrees off (holding one tree at each step).
Maximum parsimony (MP) analysis was performed using PAUP* 4.0b10. A heuristic search algorithm was engaged with 100 random addition replicates and TBR branch-swapping, and MulTrees on. Parsimony bootstrap values were calculated using PAUP* 4.0b10. The bootstrap analysis used 1,000 bootstrap replicates, the heuristic search algorithm, 100 random addition sequences, TBR branch-swapping, and MulTrees off (holding one tree at each step). Table 3 summarizes the sequence information for all rbcL, rps4, partial 5′ trnK intron, matK, partial 3′ trnK intron, trnL-F, and ITS regions, including the length of each region, numbers of variable and parsimony-informative sites, number of most parsimonious trees, tree length, consistency index (CI), and retention index (RI). The ILD test did not detect incongruence between each pair of DNA data sets tested (combined data of the seven regions: rbcL + rps4 + 5′ trnK intron + matK + 3′ trnK intron + trnL-F + ITS, P = 0.01; other data not shown). Based on these results, we combined all seven DNA sequences into one large data set, and the obtained phylogenetic results based on the combined data are shown ( Table 3). The total aligned length for the combined sequences was 5,240 characters and 531 (10.9%) characters were parsimony informative. Parsimony analysis of all seven data regions resulted in two MP trees (Tree length = 1120, CI = 0.768, RI = 0.862).

Sequence characteristics
We also tested the utility of matK for resolving phylogenetic relationships among Leucobryum species. The total aligned length for trnK intron including matK was 1,969 characters and 115 (5.8%) characters were parsimony informative. A percentage of parsimony-informative characters of trnK intron sequence data was significantly higher than that of other chloroplast sequence data (rbcL: 34 characters, 2.4%; rps4: 15 characters, 3.2%), except for trnL-F sequence data (26 characters, 6.1%).
We sequenced the chloroplast trnK intron including matK from six additional moss species of other genera including Tetraphis pellucida (Tetraphidaceae), Brothera leana (Dicranaceae), Dicranodontium denudatum (Dicranaceae), Hypnum plumaeforme (Hypnaceae), Isopterygium propaguliferum (Hypnaceae), and Rhytidium rugosum (Hylocomiaceae) (Appendix S1). The region was not amplified for the Hepaticae and Anthocerotae plant materials when we used PCR primers for exon 1 and exon 2 of the trnK intron (Demesure et al. 1995; see also Table 2). Figure 2 shows a majority rule consensus tree generated by BI analysis. The major five clades recognized in the analyses are indicated with Roman numerals (I-V). These clades were supported by high statistical values. Clade I contained L. bowringii Mitt. and L. sumatranum (Brid.) Hampe. ex M.Fleisch., and clade II contained L. albidum (P.Beauv.) Lindb., L. glaucum, and L. juniperoideum. Clades I and II were supported by high statistical support (Bayesian posterior probabilities/ML bootstrap/MP bootstrap = 1.00/ 100/100). Clade III contained only one species, L. chlorophyllosum M€ ull.Hal., from the Philippines and Indonesia. Clade IV contained L. candidum (Brid. ex Beauv.) and L. aduncum Dozy & Molk.. All three species contained in Clades III and IV are distributed in southeastern Asia and the south Pacific region. Clade V contained six species: L. boninense restricted to the Bonin Islands, L. javense,  matK, 3′ trnK intron and trnL-F were the same between them. Three species, L. scabrum, L. scaberulum, and L. javense showed similar sequences to that of L. boninense. Our phylogenetic results showed that the plant samples of L. scabrum and L. scaberulum were monophylic, in contrast that those of L. javense were polyphyletic. Leucobryum scaberulum contained two different groups: the Ryukyus group consisting of plant materials from the Ryukyus and Taiwan and the China group consisting of those from Hong Kong and Guangdong. Leucobryum javense was divided into four clades, samples #1 and 2 from Japan were closely related to L. boninense and L. scabrum, sample #4 from Malaysia was sister to the Hawaiian endemic species, L. pachyphyllum and L. seemannii, samples #3, 5, and 6 were sister to L. scaberulum, and sample #7 from Malaysia formed an independent clade.

Discussion
Origin of Leucobryum boninense, endemic to the Bonin Islands, Japan In this study, the endemic species L. boninense formed a robust clade with five related species including L. scabrum, L. javense, L. scaberulum, L. pachyphyllum, and L. seemannii, as suggested by Oguri et al. (2003Oguri et al. ( , 2008 (Fig. 2; clade V), and was closely related to L. scabrum from Japan, Taiwan, and Hong Kong and L. javense from Japan. No differences in the rps4 sequences were observed between the L. boninense samples and those of L. scabrum, in contrast, only 1-bp difference was observed in the rps4 sequences between the L. boninense samples and those of L. javense from Japan. In the rbcL sequences, 1-bp difference was observed between the L. boninense samples from the Ogasawara Islands (Chichijima Island, Hahajima Island, and Anijima Island) and those of L. boninense from Kita-iwo Island, as well as between those of L. boninense from the Ogasawara Islands and those of L. scabrum. Leucobryum boninense samples from the Ogasawara Islands and L. javense from Japan had the same rbcL sequences. In morphological characters, Yamaguchi (1993) mentioned that L. boninense is morphologically similar in the absence of a central strand in the stem and perichaetia terminal on short lateral branches to L. scabrum and L. javense. However, this species is clearly distinguishable from L. scabrum based on leaves being papillose-prorate on the abaxial surface, and is also clearly distinguishable from L. javense based on its small plant size (Yamaguchi 1993). Therefore, this molecular phylogenetic result suggests that L. boninense, which is restricted to the Bonin Islands, originated from Japan, Taiwan, or Hong Kong. The bryophyte flora of the Bonin Islands is generally regarded as similar to that of East and Southeast Asia (Iwatsuki 1985). However, this is still the first demonstration that molecular phylogenetic data directly support an East Asian origin of a moss species endemic to the Bonin Islands.

Origin of the Hawaiian endemic species of Leucobryum
In the case of Hawaiian mosses, their geographical origins remain unclear, although it is known that that Hawaiian moss flora, especially of cosmopolitan taxa, shows almost no connection with those of the American continents (Bartram 1933). Leucobryum pachyphyllum and L. seemannii are endemic to the Hawaii, and the two species are morphologically characterized by medium-sized plants and abaxially rough leaves (Bartram 1933;Staples et al. 2004). Our phylogenetic tree showed that the two species formed a monophyletic group, and were closely related to L. javense from Malaysia (Fig. 2). Leucobryum albidum, which is restricted in North America, formed a clade with L. glaucum from Japan and L. juniperoideum from Japan, and is genetically distinct from the Hawaiian Leucobryum ( Fig. 2; clade II). This species is clearly distinguished from the Hawaiian endemic species by smooth abaxial leaf surface and terminal perichaetia on stems (Bartram 1933). Molecular phylogenetic results suggested that the two Hawaiian endemic species may be originated from a southeastern Asia, not from the America.
Utility of the chloroplast matK gene for resolving phylogenetic relationships among Leucobryum species Bryophyte phylogeny and biogeography have been studied using nucleotide sequence information of nuclear and plastid DNAs such as those of nuclear ITS regions, chloroplast rbcL, rps4, trnG and trnL-F, for resolving origin and species delimitation (e.g. Huttunen et al. 2008;Oguri et al. 2008;Shaw et al. 2008;Preußing et al. 2010;Villarreal et al. 2010). However, phylogenetic analyses using chloroplast matK have not been well performed yet in bryophytes, although this gene is a powerful source for angiosperm phylogenetic analyses (Rev. M€ uller et al. 2006). A molecular phylogenetic study of Asterella (Aytoniaceae, Marchantiopsida), inferred from partial matK sequences (aligned length = 759 bp) by Long et al. (2000), is the only study to date. Their phylogenetic analysis strongly supported monophyly of Aytoniaceae; therefore, they concluded that the matK region is a useful source of phylogenetic signals in Asterrella and related marchantioid liverworts. In the present study, we compared useful sequence information among each sequence data for 50 samples containing 15 species of Leucobryum (Table 3) mative characters in the trnK intron (5.8%) was significantly higher than other chloroplast DNA regions, rbcL and rps4, although its percentage in the ITS (25.6%) was the highest among the seven regions. Maximum parsimony trees based on the trnK intron sequence data (CI = 0.824, RI = 0.916) were relatively robust than those based on the rbcL (CI = 0.783, RI = 0.885), ITS (CI = 0.778, RI = 0.894), and the combined seven sequence data (CI = 0.768, RI = 0.862). Therefore, the sequence data of trnK intron region including matK provided more informative signals for phylogenetic reconstruction among Leucobryum species.
In the present study, we also sequenced the chloroplast trnK intron region including matK of six moss species from various taxonomic groups (Appendix S1). Among these six moss species, Brothera leana and Dicranodontium denudatum were mostly closely related to Leucobryum species, whereas the remaining four species had largely different rbcL sequences from Leucobryum species, according to the results of a previous molecular phylogenetic study by Tsubota et al. (2004). Therefore, six primers (four primers of the six were newly designed in the present study, Table 2) for the trnK intron and matK are expected to be useful for molecular phylogenetic analyses in various moss taxa.