原著論文
RAPD マーカーと葉緑体 DNA 配列によるガクアジサイおよびヤマアジサイの類縁関係の解析
2014 年 83 巻 2 号 p. 163-171
詳細
2014 年 83 巻 2 号 p. 163-171
In Japan, there are many genetic resources for breeding hydrangea cultivars, but it is difficult to utilize them effectively for breeding because of a lack of phylogenetic information. In this study, the phylogenetic relationship of H. macrophylla (Thunb.) Ser. f. normalis (E.H.Wilson) H.Hara and H. serrata (Thunb.) Ser. was evaluated by using RAPD markers and sequences of the plastid genes rbcL and matK. The materials were collected from their wild populations throughout Japan. Both RAPD analysis and chloroplast DNA analysis indicated that the genetic diversity of H. serrata var. serrata was higher than that of H. macrophylla f. normalis or that of H. serrata (Thunb.) Ser. var. yesoensis (Koidz.) H.Ohba. These analyses revealed that H. serrata var. serrata of Japan was separated into two groups; i.e., eastern serrata group and western serrata group. The western serrata group was divided into two or three subgroups by single base substitutions in the matK or rbcL fragment sequences. The results of chloroplast DNA analysis indicated that H. serrata of Shikoku, which was one of the western serrata subgroups, was evolutionarily differentiated from other western serrata subgroups. MatK and rbcL sequences of the eastern serrata group were identical to those of H. macrophylla f. normalis and H. serrata var. yesoensis. The matK sequences of the eastern serrata group, H. macrophylla f. normalis and H. serrata var. yesoensis, contained a duplication of 6 bp (GGTTAT), which was not found in the western serrata group or other Hydrangea species. Analysis of the matK and rbcL sequences revealed that H. serrata var. serrata is paraphyletic and that the eastern serrata group, H. macrophylla f. normalis and H. serrata var. yesoensis, form a monophyletic group. The present study provided useful information for breeding hydrangea cultivars and for the taxonomic treatment of H. macrophylla and H. serrata including the varieties.
Many hydrangea cultivars are descended from Hydrangea macrophylla (Thunb.) Ser. and H. serrata (Thunb.) Ser. (synonym: H. macrophylla subsp. serrata (Thunb.) Makino) introduced from Japan and China to Europe since the eighteenth century. The genetic diversity of cultivated hydrangeas is very low, because most cultivars are derived from the limited numbers of plants that were introduced into Europe. Recently, new attempts using wild populations of H. serrata or H. macrophylla for breeding hydrangea cultivars have been conducted. Wild plants of H. serrata as well as H. macrophylla distributed in Japan are expected to be genetic resources for breeding a wide variety of hydrangea cultivars; however, information on their genetic diversity and phylogenetic relationship has not been accumulated.
H. serrata (Thunb.) Ser. taxonomically consists of two varieties, var. serrata and var. yesoensis (Koidz.) H.Ohba. (H. macrophylla (Thunb.) Ser. var. megacarpa Ohwi is regarded as the synonym). H. serrata var. serrata is distributed widely in mountainous regions of Honshu (southward from Fukushima Pref.), Shikoku, Kyusyu, and the Korean Peninsula. On the other hand, wild plants of H. macrophylla (H. macrophylla (Thunb.) Ser. f. normalis (E.H.Wilson) H.Hara), endemic to Japan, are localized in warm coastal areas including Boso Peninsula, Izu Peninsula, Miura Peninsula, the Izu Islands, and a part of the Bonin Islands (Kita- and Minami-Iwo-To). There are several differences in morphological characteristics between H. macrophylla and H. serrata var. serrata. Leaves of H. macrophylla are larger, thicker, and glossier than those of H. serrata var. serrata. There are many trichomes on the surface of the leaves of H. serrata var. serrata (Sato and Tanaka, 1989). However, in H. macrophylla there are few trichomes on the surface of the leaves. Flower clusters of H. macrophylla are larger than those of H. serrata var. serrata. In H. macrophylla, the inflorescence setting position is generally at the top of primary shoots. In H. serrata, however, there are several inflorescence setting positions, which are the primary shoot type, lateral shoot type, and primary and lateral shoot type (Matsuno et al., 2008).
H. serrata (Thunb.) var. yesoensis (Koidz.) H.Ohba is distributed in northern areas of Japan, including Hokkaido and the Tohoku and Hokuriku districts. In general, H. serrata var. yesoensis grows in mountains with heavy snow in winter. Morphological characteristics of H. serrata var. yesoensis are between H. macrophylla and H. serrata var. serrata. In H. serrata var. yesoensis, the leaves are not glossy but have many trichomes on their surface similar to H. serrata var. serrata (Sato and Tanaka, 1989). The sizes of leaves and flower clusters of H. serrata var. yesoensis are large and very similar to those of H. macrophylla.
The relationship between H. macrophylla and H. serrata is still vague. Wilson (1923), Haworth-Booth (1984), and Ohba (2001) separated these taxa at the species level, and differences in nuclear DNA contents between these taxa have supported this taxonomic treatment (Zonneveld, 2004). On the other hand, Makino (1929) and McClintock (1957) united these taxa into the same species, i.e., H. macrophylla ssp. serrata and H. macrophylla ssp. macrophylla. The results of SSR analyses supported their treatment as a single species, as did Makino and McClintock (Reed and Rinehart, 2007; Rinehart et al., 2006).
Most analyses of the phylogenetic relationship between H. macrophylla and H. serrata have been conducted using hydrangea cultivars as representative samples of H. macrophylla. However, it is unclear whether these cultivars are pure H. macrophylla or not because H. serrata was also included in the breeding materials introduced from Japan into Europe. Yamamoto (1979) considered that ‘Rosea’, which was one of the most important breeding materials introduced from Japan, corresponded to a Japanese local variety ‘Himeajisai’. He pointed out that ‘Himeajisai’ was likely to be a natural hybrid between H. macrophylla and H. serrata (var. yesoensis, or var. serrata) having intermediate features of morphology and growth habits between these two.
It is important for the elucidation of the phylogenetic relationship between H. macrophylla and H. serrata that the materials used for analyses are free from crossing between these species. Moreover, samples of H. serrata var. yesoensis should be treated as a distinct entity from both H. macrophylla and H. serrata var. serrata in having intermediate morphological characteristics between H. macrophylla and H. serrata var. serrata.
Random amplified polymorphic DNA (RAPD) has been widely used to study genetic relationships in many plant species because this technique is simple, requires a small amount of DNA, does not require information on the DNA sequence and is economical (Williams et al., 1990). While RAPD markers are frequently used for phylogenetic analysis of samples within the same species, plastid gene sequences are used for phylogenetic analysis between species, genera or families. Sequence data of the plastid genes rbcL and matK has contributed to elucidating the phylogenetic relationships of Hydrangeaceae species (Hufford et al., 2001).
The purpose of this research was to evaluate the phylogenetic relationships among H. macrophylla f. normalis, H. serrata var. serrata and var. yesoensis, and to provide useful information for breeding hydrangeas. In the present study, we analyzed the genetic diversity and relationships of H. macrophylla f. normalis, H. serrata var. serrata and var. yesoensis collected from wild populations throughout Japan by using RAPD markers and sequences of the plastid genes rbcL and matK.
The taxon of each Hydrangea individual used in this study was assigned based on Ohba’s classification system of the genus Hydrangea (Ohba, 2001). Seven H. macrophylla f. normalis, fifteen H. serrata var. serrata, four H. serrata var. yesoensis, and one H. serrata (Thunb.) Ser. var. angustata (Franch. et Sav.) H.Ohba were used for RAPD analysis (Table 1). Figure 1 shows the place where each individual grew as a wild plant. Only individuals whose geographic origins were clear were chosen for this study. One H. hirta (Thunb.) Siebold et Zucc. was used as the outgroup. For analyses of plastid sequences, three H. macrophylla f. normalis, three H. serrata var. serrata, five H. serrata var. yesoensis, one H. luteovenosa Koidz., and one H. petiolaris Siebold et Zucc. were added to the individuals used for RAPD analysis (Table 1; Fig. 1). The geographic origin of the three H. serrata var. serrata was the Republic of Korea.
List of Hydrangea plants evaluated by RAPD analysis and plastid sequence analyses.
The place where the sample grew as a wild plant.
Total DNA was extracted using the CTAB method (Doyle and Doyle, 1987). Thirteen random 10-mer primers (A01, A02, A04, A05, A07, A08, A09, A10, A11, A13, A17, A18, A20; Operon Technologies, Inc., Alameda, CA, USA) were used for PCR amplification (Table 2). PCR amplification reactions with primers A01, A02, A04, A05, A07, A08, A10, A20 were carried out in a total reaction volume of 25 μL containing 1 × Taq buffer, 15 ng genomic DNA, 0.2 mM dNTPs, 12.5 pmol primers, and 1.25 units AmpliTaq DNA polymerase Stoffel fragment (PerkinElmer, Waltham, MA, USA). DNA fragments were amplified by repeating 45 cycles of 94°C for 1 min, 50°C for 1 min, and 72°C for 2 min using a GeneAmp PCR system 2400 thermal cycler (PerkinElmer). PCR amplification reactions with primers A09, A11, A13, A17, A18 were carried out in a total reaction volume of 10 μL containing 1 × Taq buffer, 10 ng genomic DNA, 0.25 mM dNTPs, 5 pmol primers, and 0.5 units Blend Taq-Plus DNA polymerase (Toyobo, Osaka, Japan). DNA fragments were amplified by repeating 45 cycles of 94°C for 30 sec, 36°C for 30 sec, and 72°C for 1.5 min using a MJ Mini thermal cycler (Bio-Rad Laboratories, Hercules, CA, USA). Electrophoresis was conducted for 2.5 h at 100 V, using a 2% (w/v) agarose gel in Tris-borate-EDTA (TBE) buffer with 10 μL PCR product from each sample. Staining was performed with ethidium bromide.
Sequences of primers for amplification of RAPD markers and cpDNA fragments.
The presence or absence of bands in the stained gels was used to calculate genetic similarities. Genetic similarity was calculated as Sij = 2 Nij/(Ni + Nj), where Nij is the number of markers shared by individuals i and j, Ni is the number of markers found in individual i, and Nj is the number of markers found in individual j (Dice, 1945). The similarity matrix was subjected to clustering analysis using UPGMA in the program SPSS for Windows, version 15.0 (SPSS, Chicago, IL, USA). Estimates of statistical support for the resulting clusters were obtained from UPGMA bootstrap analysis with 1000 replicates using the program PHYLIP 3.67 (Felsenstein, 2007).
Analysis of rbcL and matK sequencesTotal DNA of additional individuals for analysis of plastid gene sequences was extracted using a MagExtractor TM-Plant Genome-Kit (Toyobo, Osaka, Japan). The primers used to amplify the fragments of rbcL and matK were designed according to highly conserved regions within H. macrophylla (GenBank accession no. AB236030 for matK, L11187 for rbcL), H. anomala (GenBank accession no. AF323202 for rbcL), H. quercifolia (GenBank accession no. AF323203 for rbcL), H. aspera (GenBank accession no. AJ429277 for matK), and H. paniculata (GenBank accession no. AB236029 for matK). The primer sequences are presented in Table 2. PCR amplification reactions were carried out in a total reaction volume 50 μL containing 1 × KOD plus buffer, 20 ng genomic DNA, 0.2 mM dNTPs, 20 pmol primers, and 1.0 unit KOD plus polymerase (Toyobo). DNA fragments were amplified by repeating 35 cycles of 94°C for 15 sec, 58°C for 30 sec, and 68°C for 2.0 min using a MJ Mini thermal cycler (Bio-Rad Laboratories). PCR products were purified using a PCR-M Clean Up System (Viogene, Taipei, Taiwan) in order to remove excess primers and dNTPs after amplification. Sequencing was performed using a BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) in accordance with the manufacturer’s protocol. Sequencing reactions were carried out with the primers shown in Table 2 by repeating 25 cycles of 10 sec at 96°C, 5 sec at 50°C, and 4 min at 60°C. The sequencing reaction products were purified through spin columns and then applied to a 3130xl Genetic Analyzer (Applied Biosystems).
Multiple alignments of the DNA sequences were obtained using the CLUSTAL W program implemented in MEGA5 (Tamura et al., 2011). Two indels were found in matK alignments and they were excluded from the data for constructing the phylogenetic tree. The multiple alignments were used to construct maximum parsimony trees with 500 bootstraps in MEGA5.
A total of 426 RAPD markers were generated with 13 primers in all individuals and 413 markers were polymorphic. Based on the UPGMA dendrogram, the individuals except for H. hirta were separated into two major groups with a similarity value of 0.56 (Fig. 2). One group consisted of H. macrophylla f. normalis and H. serrata var. yesoensis individuals with high bootstrap support (95%). The other group consisted of H. serrata var. serrata individuals and H. serrata var. angustata. H. macrophylla f. normalis was separated into a group of Honshu individuals (M5, M6, M7) and a group of Izu and Bonin Islands individuals (M1, M2, M3, M4) with a similarity value of 0.68. On the other hand, H. serrata var. serrata was divided into groups from the eastern (S12, S13, S15, S16) and western (S1, S2, S3, S4, S5, S6, S7, S8, S9, S10, S11) areas of Japan with a similarity value of 0.57. The H. serrata var. angustata individual (S14) was assigned as a member of the eastern group of H. serrata var. serrata (tentatively called the eastern serrata group). Furthermore, the western group of H. serrata var. serrata (western serrata group) was divided into a subgroup of Honshu individuals (S7, S8, S9, S10, S11) and the subgroup of Shikoku and Kyushu individuals (S1, S2, S3, S4, S5, S6) with a similarity value of 0.59. The Shikoku individuals (S4, S5, S6) were separated from Kyusyu individuals (S1, S2, S3) within the subgroup. UPGMA dendrogram suggested that H. macrophylla and H. serrata var. yesoensis were genetically similar, and that the genetic diversity of H. serrata var. serrata was greater than that of H. macrophylla f. normalis or H. serrata var. yesoensis.
UPGMA dendrogram based on Dice similarities calculated from RAPD data on 27 individuals of H. macrophylla and H. serrata. One individual of H. hirta was used as an outgroup. Bootstrap values (%) from 1000 replicates are indicated above the branches when over 50%.
The matK gene in H. macrophylla is 1521 bp long (AB236030, Setoguchi et al., 2006). Sequences of partial matK fragments (1421–1427 bp) used for alignments in this study covered about 93–94% of the matK gene. The aligned matK matrix using all individuals, including H. hirta, H. luteovenosa, and H. petiolaris, consisted of 1431 bp, of which 32 positions were variable. Two indels were found in the matK alignments. One indel was a 4 bp insertion specific to H. petiolaris. The other indel was a 6 bp (GGTTAT) insertion. Two single base substitutions and a duplication of 6 bp (GGTTAT) were found among individuals of H. macrophylla f. normalis and H. serrata (Table 3). There were no differences in the matK fragment sequences between H. macrophylla f. normalis and H. serrata var. yesoensis. H. serrata var. serrata and H. serrata var. angustata were divided into two groups according to a single base substitution and a duplication of 6 bp (GGTTAT). One serrata group consisted of individuals (S10, S12, S13, S14, S15, S16) from the eastern area of Japan. The matK fragment sequence of this group was identical to those of H. macrophylla f. normalis and H. serrata var. yesoensis and contained a duplication of 6 bp (GGTTAT). The other serrata group consisted of individuals (S1, S2, S3, S4, S5, S6, S7, S8, S9, S11, K1, K2, K3) from the western area of Japan and Korea. Neither this serrata group nor the outgroup including H. hirta, H. luteovenosa, and H. petiolaris contained the 6 bp duplication in their matK sequences. In the maximum parsimony tree based on sequences of matK, H. macrophylla f. normalis, H. serrata var. yesoensis, and the eastern serrata group were clustered together (Fig. 3). The tree indicated that H. serrata var. serrata was paraphyletic. The S10 individual was assigned as a member of the eastern serrata group according to analysis of matK sequences, despite the fact that it belonged to the western serrata group according to RAPD analysis. The H. serrata var. angustata individual (S14) was assigned as a member of the eastern serrata group. The Shikoku group (S4, S5, S6) was discriminated from the western serrata group by the substitution of a nucleotide. The maximum parsimony tree indicated that the genetic diversity of H. serrata var. serrata was greater than that of H. macrophylla f. normalis or H. serrata var. yesoensis.
Substitution and duplication events observed in the aligned matK fragment sequences (1431 bp) of H. macrophylla f. normalis, H. serrata var. yesoensis, H. serrata var. serrata, and H. serrata var. angustata.
The maximum parsimony tree based on matK fragment sequences of 38 individuals of H. macrophylla and H. serrata. H. hirta, H. luteovenosa, and H. petiolaris individuals were used as the outgroup. A tree out of the 10 most parsimonious trees is shown. Tree length = 23, CI = 1.00, RI = 1.00. Bootstrap values (%) from 500 replicates are indicated above the branches when over 50%.
The aligned rbcL matrix consisted of 1257 bp, of which 24 positions were variable. Four events of single base substitutions were found in the rbcL fragment sequences among individuals of H. serrata var. serrata (Table 4). On the other hand, there were no differences in the rbcL fragment sequences among individuals of H. macrophylla f. normalis and H. serrata var. yesoensis. The topology of the maximum parsimony tree based on sequences of rbcL was almost identical to that of matK, except for a few minor differences in the grouping of H. serrata var. serrata individuals (Fig. 4). The tree indicated that H. serrata var. serrata was paraphyletic. H. serrata var. serrata was divided into the eastern Japan group, Shikoku group, and western Japan and Korea group, which was subdivided into two subgroups. H. macrophylla f. normalis, H. serrata var. yesoensis and the eastern serrata group were united into a single cluster. The S10 individual was assigned as a member of the eastern serrata group, which was in accordance with the results of matK sequence analysis. The H. serrata var. angustata individual (S14) was assigned as a member of the eastern serrata group.
Substitution events observed in the aligned rbcL fragment sequences (1257 bp) of H. macrophylla f. normalis, H. serrata var. yesoensis, H. serrata var. serrata, and H. serrata var. angustata.
The maximum parsimony tree based on rbcL fragment sequences of 38 individuals of H. macrophylla and H. serrata. H. hirta, H. luteovenosa, and H. petiolaris individuals were used as the outgroup. The most parsimonious was obtained and is shown. Tree length = 26, CI = 0.91, RI = 0.97. Bootstrap values (%) from 500 replicates are indicated above the branches when over 50%.
H. serrata var. serrata is geographically distributed widely in Japan and also in the Korean Peninsula, whereas H. macrophylla f. normalis and H. serrata var. yesoensis are endemic to Japan. In this study, we conducted phylogenetic analysis of H. macrophylla f. normalis, H. serrata var. serrata, and H. serrata var. yesoensis using individuals derived from wild populations in Japan for the purpose of providing useful information for breeding hydrangeas.
Both RAPD analysis and chloroplast DNA analysis indicated that the genetic diversity of H. serrata var. serrata was higher than that of H. macrophylla f. normalis or that of H. serrata var. yesoensis. The high genetic diversity of H. serrata var. serrata can be attributed to its wide-ranging geographical distribution, including Japan and the Korea Peninsula. Results of RAPD analysis indicated that H. serrata var. serrata of Japan was separated into two groups; i.e., eastern and western serrata groups (Fig. 2). This result was supported by phylogenetic analysis using chloroplast DNA sequences (Figs. 3 and 4). The grouping of H. serrata var. serrata in this study was consistent with the geographical distribution of flower color. White flowers are predominant on the Pacific Ocean side of the eastern area of Japan, including Tokai district and Mt. Fuji. Plants S12, S14, S15, assigned as members of the eastern serrata group in this study, have white flowers. On the other hand, flowers with an anthocyan color are found in the area west of the Kinki district. All serrata individuals assigned as members of the western group have flowers with an anthocyan color.
In this study, the serrata individuals derived from the area west of the Suzuka Mountains were assigned as members of the western group, whereas serrata individuals derived from the area east of the Suzuka Mountains were assigned as members of the eastern group. Furthermore, individual S10, which was sampled from a wild population in Suzuka Mountains, was assigned as a member of the western serrata group by RAPD analysis, whereas it was included in the eastern serrata group in the analysis using chloroplast DNA sequences (Figs. 2, 3, and 4). These results indicate that the Suzuka Mountains are one of the borders between the eastern and western serrata groups.
Results of chloroplast DNA analysis indicated that there was more genetic diversity within the western serrata group than within the eastern serrata group (Figs. 3 and 4). No differences were found in DNA sequences of either matK fragments or rbcL fragments within the eastern serrata group, including the variety Angustata individual (S14). On the other hand, the western serrata group was divided into two or three subgroups by single base substitutions in the matK or rbcL fragment sequences. Individuals S4, S5, S6, whose geographical origins were Shikoku Island, were distinguished from other members of the western serrata group on analysis of both matK and rbcL. This result suggests that H. serrata var. serrata of Shikoku is evolutionarily differentiated from other western serrata. The matK sequences of the western serrata group except for individuals of Shikoku were the same as those of Korean serrata. Furthermore, a single base substitution in rbcL sequences within the western serrata group, except for the Shikoku individuals, was found within the Korean serrata group (Table 4). These results indicate that the Korean serrata group and the western serrata group, except for the Shikoku samples, form a monophyletic group and that the single base substitution in the rbcL sequence occurred in a common ancestor.
Both phylogenetic trees based on RAPD markers and chloroplast DNA sequences suggest that H. macrophylla f. normalis as well as H. serrata var. yesoensis is monophyletic. On the other hand, phylogenetic trees based on the matK and rbcL sequences indicate that H. serrata var. serrata is paraphyletic. Although RAPD analysis demonstrated that H. serrata var. serrata was monophyletic, the bootstrap value was low (27.5%). MatK and rbcL sequences of the eastern serrata group were identical to those of H. macrophylla f. normalis and H. serrata var. yesoensis. Both phylogenetic trees based on matK and rbcL sequences indicated that H. macrophylla f. normalis, H. serrata var. yesoensis, and the eastern serrata group form a monophyletic group, and that these taxa are descended from a common ancestor. This is supported by the geographic distribution of these taxa and the 6 bp duplication in the matK sequence. H. serrata var. serrata is distributed more widely than H. macrophylla f. normalis and H. serrata var. yesoensis, and the distribution areas of H. macrophylla f. normalis, H. serrata var. yesoensis, and the eastern serrata groups neighbor each other. This pattern of distribution supports the hypothesis that H. macrophylla f. normalis, H. serrata var. yesoensis, and the eastern serrata group were derived from a common ancestor, which was H. serrata var. serrata. Moreover, this hypothesis is supported by the matK sequences specific to these taxa. The matK sequences of these taxa contained a duplication of 6 bp (GGTTAT) (Table 3). On the other hand, neither the serrata individuals, except for the eastern serrata group, nor the outgroup including H. hirta, H. luteovenosa and H. petiolaris contained the 6 bp duplication in their matK sequences. In addition, other Hydrangea species in the databases (H. heteromalla, GU217275; H. paniculata, GU217276; H. quercifolia, GU217277; H. arborescens, GU217285; H. involucrata, GU217290; H. sikokiana, GU217291; H. longipes, GU217292; H. sargentiana, GU217293; H. glabripes, GU217294; H. aspera, GU217295; H. serratifolia, GU217300; H. integrifolia, GU217302; H. seemannii, GU217303; H. anomala, GU217304; H. indochinensi, GU217312; H. scandens, GU217328; H. lobbii, GU217330; H. chungii, GU217332; H. angustipetala, GU217336) did not contain the 6 bp duplication in their matK sequences. The 6 bp duplication seems to be specific to H. macrophylla f. normalis, H. serrata var. yesoensis, and the eastern serrata group, so the duplication event probably occurred in their common ancestor.
There are two distinct interpretations of the taxonomic positions of H. macrophylla and H. serrata. One interpretation is that these taxa should be treated as different species (Haworth-Booth, 1984; Ohba, 2001; Wilson, 1923; Zonneveld, 2004). The other is that H. serrata var. serrata should be placed as a subspecies of macrophylla (Makino, 1929; McClintock, 1957; Reed and Rinehart, 2007; Rinehart et al., 2006). The results of this study revealed that H. serrata var. serrata consists of two groups; one group is a near relative of H. macrophylla and the other group is obviously different from H. macrophylla. Therefore, the interpretations of the taxonomic relationship between the H. macrophylla and H. serrata are possibly affected by the serrata group used for analysis if the serrata samples are restricted to only one serrata group. Furthermore, it is important for the elucidation of the phylogenetic relationships between H. macrophylla and H. serrata that hydrangea cultivars are excluded from macrophylla samples, because it is unclear for most hydrangea cultivars whether they are absolutely free from crossing with H. serrata.
The present study provided valuable information for the taxonomic treatment of H. macrophylla and H. serrata, including the varieties. Analysis of the matK and rbcL sequences revealed that H. serrata var. serrata was paraphyletic and that the eastern serrata group, H. macrophylla f. normalis, and H. serrata var. yesoensis formed a monophyletic group. However, these taxa have clearly distinct morphological and physiological characteristics and specific distribution areas, so it is reasonable that H. macrophylla and H. serrata should be treated as different species. Two concepts for the present H. serrata var. yesoensis are reliable; to treat it as a distinct species or as an infrataxa of H. macrophylla or H. serrata. Until further evidence for elucidation is obtained, its present treatment as a variety of H. serrata should be tentatively maintained because of their morphological similarities. This study indicates that H. serrata var. serrata is genetically polymorphic; however, we have not yet found morphological features except for flower color corresponding to a genetic separation. Further morphological and geographical studies as well as analysis of sequence data from more informative regions of the genome are needed for taxonomic evaluation of its infraspecific separation.
There are abundant genetic resources for hydrangea breeding in Japan. Effective utilization of these genetic resources for breeding hydrangeas could produce novel cultivars with excellent properties (e.g., ornamental value, resistance to environmental stress), and the results obtained in this study will provide useful information for breeding hydrangeas.
