MOLECULAR ANALYSIS OF THE GENUS CAMELLIA

Xiao Tiaojiang and Clifford Parks, University of North Carolina, U.S.A. Acknowledgements: The authors would like to thank the American Camellia Society for the grants provided to support this work. We also would like to recognize Dr. Jiyin Gao of the Subtropical Forestry Research Institute, Fuyang, China, Mr. Bob Cherry of Paradise Nursery, Kulnura, Australia and Mr. J. C. Rosmann of the Centre for Camellia Studies, Boucau, France for the large number of plants specimens provided for analysis.

SUMMARY

The monographers of the genus Camellia disagree on the boundaries of subgenera, sections and the circumscription of many species. Taxonomic treatments such as these based on herbarium specimens should be verified by objective, non－ morphological studies. Phylogenetic analysis of DNA sequences is an objective approach; and accordingly, 2 segments of the RPB－ 2 gene were sequenced and analyzed for 147 Camellia taxa. The results support the broadly defined genus as treated by all of the monographers. The published subgeneric structures of the genus are not supported by the DNA analysis. The sectional makeup of the genus is partially supported by the phylogenetic analysis.

Introduction

The three most recent classification systems of the genus Camellia were established by J. Robert Sealy (Sealy, 1958), Chang Hungta (Chang, 1981; 1998; Chang and Bartholomew, 1984) and Ming Tianlu (Ming, 2000). In these traditional classifications species that are similar in structure or morphology are grouped together in sections. A large genus such as Camellia is usually divided into many species groups or sections. Sealy divided Camellia into 12 sections, while Chang constructed 20 and Ming 14. Furthermore, in large genera such as Camellia similar sections may be grouped into subgenera. Sealy made no subgeneric divisions in Camellia. Chang divided Camellia into four subgenera and Ming into two.

Sealy, Chang and Ming disagree on the boundaries of subgenera, sections, the circumscription of many Camellia species and the relationships among species. These major disagreements have created confusion among those interested in the genus Camellia. To date, it is uncertain which of these systems provides the most accurate picture of the relationships within Camellia. One difficulty is that the classification systems established by Sealy, Chang and Ming are all based on morphological analysis of a limited number of dry herbarium specimens.

As a traditional method, morphological analysis is an important tool for plant classification, especially for species identification. However, this method has many limitations. First, the traditional morphological analysis employs intuition; thus, the use of morphological data in reconstructing the evolutionary history (phylogeny) may over－ emphasize characters considered by the investigator to be of particular importance. Consequently, strongly conflicting results are obtained when the importance of the selected characters is treated differently by different workers. Secondly, the measurements of the selected characters are often based on a relatively small number of dried herbarium specimens, which may be an inadequate sample of the variation in the whole population. Therefore, the classification system established from these morphological characteristics may be biased and artificial. Traditional taxonomic systems based on morphology must be verified by objective, non－ morphological approaches. In recent years DNA sequence analysis of species relationships has begun to provide this objective evidence.

In the last 20 years, plant molecular systematics (taxonomy) based on DNA sequence analysis has grown rapidly. The small circular segment of DNA in the green plastid or chloroplast (the center for photosynthesis) in the plant cell is known as the chloroplast genome. A relatively small number of genes are carried in the chloroplast. Several of these genes including the rbcL gene, atpB gene, matK gene, ndhF gene, 16s rDNA, atpB－ rbcL intergenic region and other noncoding chloroplast sequences, have been sequenced to determine the relationships between species in many plant groups (Soltis et al., 1998). When the DNA variation among species in a gene is determined by sequencing that gene, the relationships are expressed in a phylogenetic tree. Furthermore, several genes located in the nucleus of the cells (nuclear genes) have similarly been widely explored for their phylogenetic utility. Some nuclear genes used encode ribosomal RNAs (the 18s, 26s, 5.8s, 5s and ITS sequences), phosphoglucose isomerase (PGI), alcohol dehydrogenase (ADH), phytochrome isomerase, and RNA polymerase II subunit (Soltis et al.1998).

Nuclear RNA polymerase II (RPB 2) is found in all multicellular organisms (Eukaryotes). It is a complex enzyme responsible for transcribing pre－ mRNA in nuclei. Here, the functions of the gene are not the concern, but the small differences in gene structure among different species are used to show species relationships. This enzyme has two large subunits and several smaller subunits. The RPB 2 gene has just come into use for studying plant phylogeny, and it shows great potential. In a study of 25 Rhododendron species (Eriaceae family), Denton, et al. (1998) explored the use of the RPB 2 sequences in phylogenetic analysis. They found that RPB 2 contains a considerable number of nucleotide substitutions, especially in the intron region. Genes have sections of DNA that codes life processes in organisms known as exons, and segments of DNA between the exons that do not code cellular activity. These spacer segments are known as introns, which as expected are more variable from species to species than the exons. After her study of 25 Rhododendron species, Denton and her coworkers predicted that the RPB 2 gene could be used to address phylogenetic questions at a number of taxonomic levels (species, section, subgenus and genus).

The RPB 2 gene has 24 intron segments of varying length. In our study, we analyzed the DNA sequences of two different portions of the RPB 2 gene, i.e. connecting exons and introns from intron 11 to 16 and intron 23 from 149 Camellia species that represent 14 sections in Chang’ s treatment of the genus. By comparing the amount of similarity among the DNA sequences, we were able to infer the relationships of species and sections within the genus Camellia. The following discussion is a summary of the results obtained from our DNA sequence analysis.

Monophyly of the genus Camellia

The first, and perhaps the most important question to be addressed is whether the genus Camellia is a natural group of species of common origin (monophyletic) or is it an artificial mixture of small groups of species of divergent origin (polyphyletic). In earlier treatments, the genus Camellia was divided into nine smaller genera that include Camelliastrum, Dankia, Glyptocarpa, Kailosocarpous, Parapiquetia, Piquetia, Thea, Theopsis and Yunnanea. Of these genera, three genera were invalidly published and are thus dropped from consideration. Of these three Parapiquetia and Kailosocarpous were published without descriptions and Glyptocarpa was established by Hu (1965) by transferring a species from Kailosocarpus. For the remaining six genera Dankia was combined into Camellia by Pham (1991) because of common morphology, and the others were merged into Camellia by Sealy (1958).

According to Sealy, the species of the genus Camellia are an inseparable entity because they shared two distinct morphological characters: 1) seeds without wings, and 2) capsules dehiscent from the apex. Sealy felt these two traits distinguish Camellia from other genera in the tea family (Theaceae).

Both Chang (1981) and Ming (2000) are in agreement with Sealy’ s definition of the genus Camellia. The results of our DNA analysis of two segments of the RPB 2 gene also give strong support to the hypothesis that the genus Camellia is a monophyletic group.

Subgenus division

Whether the subgeneric divisions within the genus Camellia are appropriate is still not clear. The monographers of this genus present three different opinions. Sealy (1958) did not recognize any subgenera in the genus Camellia, while Chang (1981) divided the genus into four subgenera. Ming (2000) named two subgenera within the genus Camellia.

In our DNA sequence analysis, three major clades (clusters of related species) can be identified. However, these clades neither fit Chang's four subgenera nor match Ming's two subgenera in any way. At this time, it seems premature to draw any taxonomic conclusion concerning subgenera from the results of the phylogenetic analysis. Additional DNA sequences from other genes may help to clarify this problem.

SECTIONS OF THE GENUS CAMELLIA

A brief comparison of the major sections of the genus Camellia as presented in the morphological treatments by Sealy (1958) with those by Chang and Bartholomew (1984) follows. The treatment by Ming (2000) closely follows the sectional treatment of Sealy. The implications of results from the DNA analysis are compared to the morphological treatments.

Sections Archecamellia and Chrysantha

The definition of section Archecamellia is debated. Both Sealy (1958) and Ming (2000) defined this section by two morphological characters, that is, flowers bearing 1) distinct pedicels (distinct flower stalks), and 2) undifferentiated bracteoles (bracts below the petals). According to Sealy and Ming, the yellow flowered species had obvious pedicels on flowers: and therefore, they should be placed in section Archecamellia with other pedicellate (with long pedicels) species.

Chang (1981, 1996) defined section Archecamellia differently. In terms of Chang’ s definition, species of section Archecamellia are characterized by having 5－ locular ovaries and 5－ parted, free styles. Thus, Chang’ s section Archecamellia only includes three species: C. granthamiana, C. albogigas and C. pleurocarpa. Chang argued that the presence of a floral pedicle was a highly variable character in the genus Camellia, and thus might be of limited value in defining sections. On the other hand, he considered yellow pigment to be a more stable character. For these reasons, Chang (1981) separated the yellow flowered species from Sealy’ s section Archecamellia, and placed them in a new section Chrysantha.

Section Archecamellia as conceived by Sealy is not supported by DNA evidence. The establishment of a separate section Chrysantha from Sealy’ s section Archecamellia by Chang has good support from our DNA sequence analysis. However, the cluster of related yellow－ flowered species (clade) in the DNA based phylogenetic tree includes several non－ yellow species － mostly from section Tuberculata. However, the two species available for analysis from Chang's section Archecamellia, C. granthamiana and C. albigigas, only show moderately close phylogenetic relationship in the DNA analysis, and thus Chang's section Archecamellia is only weakly supported by the DNA sequence analysis. However, these two species are morphologically nearly identical, and in this case the DNA evidence should be reinvestigated.

Section Paracamellia

Sealy (1958) placed C. sasanqua, C. oleifera and eight other related species into section Paracamellia because these species had short styles and minimal fusion of floral parts. Chang (1981) did not agree with Sealy on the definition of section Paracamellia. He divided section Paracamellia into two sections, Oleifera and Paracamellia. The new section Oleifera includes four species: C. sasanqua, C. oleifera, C. vietnamensis and C. gauchowensis. Chang argued that these four species differed from other species in section Paracamellia by having more stamen series and relatively longer styles. However, the wild forms of C. miyagii (Chang’ s section Paracamellia) and C. sasanqua (Chang's section Oleifera) are virtually identical and considered by most workers to be the same species. Cultivated C. sasanqua with its brightly colored flowers is much coarser with larger parts, but it is introgressed with C. japonica (Parks, et al, 1981). In addition, considering the demonstrated similarity between C. sasanqua and C oleifera, there is not basis remaining for the separation of section Oleifera out of section Paracamellia.

In our DNA sequence analysis, all species from Chang's sections, Paracamellia and Oleifera, except C. grijsii, C. odorata and C. yusienensis, fall in a strongly supported group. The three very closely related species, C. grijsii, C. odorata and C. yuhsienensis, fall into the clade of section Camellia species from western China. This DNA evidence generally reinforces Sealy's definition of section Paracamellia, but there is no obvious explanation for the divergent position of C. yuhsienensis and the two species closely allied to it.

Section Thea

The definition of section Thea is not disputed. Sealy, Chang and Ming all view this section as a monophyletic group.

In our DNA sequence analysis, all of the available species of section Thea fall in one tight species cluster or clade, strongly supporting the concept that section Thea is a natural unit. The species of section Thea fall within the major clade (grouping) of red flowered species of section Camellia from southern and southwestern China. This is evidence that species of section Thea may have the same ancestor as species of section Camellia distributed in southern and southwestern China.

Section Camellia

Section Camellia is treated as a single unit by all the monographers, but it forms four distinct groups in our DNA sequence analysis. The species of section Camellia in Japan form the first clade and are a sister group to section Paracamellia. The species of section Camellia in eastern China form the second clade and are also closely related to section Paracamellia. Thus, C. japonica and the species of section Camellia in eastern China are separated by section Paracamellia. Camellia amplexicaulis, a red－ flowered species from Vietnam, is in the eastern China grouping despite its different morphology. The species of section Camellia from southern and southwestern China, including the highly variable species C. reticulata, C. saluenensis and their relatives, compose the third clade and have a close relationship with section Thea. The species C. albogigas, C. granthamiana, C. grijsii, C. odorata, and C. yusienensis are in the clade of section Camellia from western China. In this case further DNA research is necessary to resolve the correct position of these morphologically different species. Camellia hongkongensis stands alone, since it falls in a clade formed by the majority of species of section Furfuracea. Compatibility data and morphology also place C. hongkongensis further away from section camellia. This result indicates that section Camellia is a polyphyletic assemblage (not originated from one ancestral group). The species from Japan, Eastern China and South and southwestern China may have had different origins. The sectional position of C. hongkongensis must be reconsidered.

Sections Heterogenea and Furfuracea

When Sealy established section Heterogenea, he wrote that,“ As its name implies, this section is a collection of diverse species, and since they did not fit easily into any of the other groups of species, it is convenient to put them together in a separate section" (Sealy 1958). Apparently, Sealy considered section Heterogenea an artificial group. For this reason, Chang completely dissolved Sealy’ s section Heterogenea, and placed its species into seven different sections.

Ming (2000) agreed with Sealy on the definition of section Heterogenea. However, he believed that section Heterogenea was a natural group based on his observation that all species of section Heterogenea had free styles.

Since free styles are widespread in the genus Camellia, including species in sections Chrysantha and Thea, it is doubtful that the characteristic of free styles is important in defining sections in the genus Camellia. Thus, section Heterogenea as conceived by Ming includes species from Chang's sections Archecamellia, Furfuracea, Pseudocamellia and stereocarpus.

Chang constructed section Furfuracea in 1981. According to Chang, the defining features of species of this section were spongy and furfuraceous pericarps. Ming (2000) did not support Chang's establishment of section Furfuracea. He placed all Chang's species from section Furfuracea into section Heterogenea. Ming argued that the character of spongy and furfuraceous pericarp was not unique to species from Chang's section Furfuracea. He noted that this character could also be found in section Tuberculata (i.e., C. tuberculata). For reasons that are not obvious the species in section Tuberculata group with the yellow－ flowered species in section Chrysantha in the DNA analysis.

Our DNA sequence analysis produced a complicated result. On one hand, the majority of species from section Furfuracea form a single clade, indicating section Furfuracea is a distinct group, which favors Chang's definition of section Furfuracea.

On the other hand, the clade formed by Furfuracea species fell within a large clade that included C. yunnanensis, C. hongkongensis and other free－ styled species as the sister taxa to section Furfuracea. In fact, species of section Furfuracea also have free styles. Therefore, it is apparent that this large clade is formed by free－ styled species. These facts, to some degree, support the definition of section Heterogenea made by Sealy and Ming.

Sections Theopsis and Eriandria

These two sections are poorly resolved in our DNA sequence analysis. Species from sections Theopsis and Eriandria fall into two distinct clades. The majority of species from these two sections cluster together in a clade with species of section Camellia from Japan, while the remaining species fall within a second clade along with the yellow flowered species. In these two clades, species from section Eriandria are not distinguishable from the species from section Theopsis.

Sections Tuberculata, PseudoCamellia, Longipedicellata

These three sections are poorly resolved in our DNA sequence analysis. There is no indication that these sections are distinct groups. The reason for such a poor resolution is unknown. However, the number of species from these sections available for analysis is very limited. Sequencing more taxa from these sections or sequencing the same taxa with different genes may help to improve the resolution.

C.vidalii

C.vidalii is a new species published by Rosmann (2000). Its sectional position remains to be determined. In the most parsimonious, phylogenetic trees, this species does not cluster with any other Camellia species. This suggests that either C. vidalii may not belong to the genus Camellia or it is distantly related to all Camellia species. It seems more research is necessary in order to verify the results presented here.

A CONCLUDING COMMENT: An overview of the results shows an interesting correlation. The sections containing species with similar morphology hold together in the DNA analysis. Sections Chrysantha, Paracamellia and Thea hold together well in morphology, DNA sequence similarity and in other respects as well. The two sections Theopsis and Eriandria merge in the DNA analysis, but the similar species of the two sections cluster together. The species of the three sections with morphologically diverse species including Heterogenea, Camellia and Archecamellia do not cluster together in the DNA analysis. Other sections of the genus were not well enough represented in the DNA analysis to make definite statements concerning relationships. As one would anticipate, clear morphological similarity is strongly correlated with DNA sequence similarity.

Literature cited

Chang, H.T. 1981. A taxonomy of the genus Camellia. Acta Scientarum Naturalium Universitatis. Sunyatseni, monographic series 1 : 1－ 180.

1996. Diagnosis on the systematic development of Theaceae I. A review on the Sect. Chrysantha and Archecamellia of the genus Camellia. Acta Scientarum Naturalium Universitalis Sunyatseni, 35 (1):77－ 83.

1998. Revision of the 1981 taxonomy of the genus camellia (in Chinese).Chang, H.T. and B. Bartholomew. 1984. Camellias. Timber Press, Portland, Oregon. ( A translation of the 1981 revision by H. T. Chang).

Denton, A.L., B. L. McConaughy, and B. D. Hall. 1998. Usefulness of RNA polymerase II coding sequences for estimation of green plant phylogeny. Molecular Biology and Evolution 15(8): 1082－ 1085.

Hu, H. H. 1965. Glyptocarpa, a new genus of Theaceae. Acta Phytotaxonomica Sinica 10 (1):25－ 26.

Ming, Tian－ lu. 2000. Monograph of the genus Camellia. Yunnan Science and Technology Press, Kunming, P. R. China.

Parks, C. R, K. Kondo, and T. Swain. Phytochemical evidence for the genetic contamination of Camellia sasanqua Thunberg. Japanese Journal of Breeding 31(2):168－ 182.

Pham－ Hoang, Ho.1991. Theaceae. In: Cayco Vietnam: An illustrated flora of Vietnam.

Rosmann, J. C. 2000. A new camellia species of Vietnam. International Camellia Journal 32:66－ 68.

Soltis, D. E, P. S. Soltis, M. E. Mort, M. W. Chase, V. Salvolainen, S. B. Hoot, and C. M. Morton. 1998. Inferring complex phylogenies using parsimony: an empirical approach using three large DNA data sets for Angiosperms. Systematic Biology 47:32－ 42.

Sealy, J. R. 1958. A revision of the genus Camellia. The Royal orticultural Society, London. 239pp.