Microsoft Word - 06. BJPT 12-50-Camellia.doc Bangladesh J. Plant Taxon. 19(2): 155-165, 2012 (December) © 2012 Bangladesh Association of Plant Taxonomists FLORAL MORPHOLOGY RESOLVES THE TAXONOMY OF CAMELLIA L. (THEACEAE) SECT. OLEIFERA AND SECT. PARACAMELLIA W. JIANG, M. NITIN1, B. JIANG, Y.P. ZHENG, S.S. HONG AND H.F. LU2 College of Chemistry and Life Science, Zhejiang Normal University, Jinhua, 321004, China Keywords: Camellia; Cladistic; Floral morphology; sect. Oleifera; sect. Paracamellia. Abstract Numerical taxonomy and cladistic analysis of 19 species of Camellia L. were performed using floral morphology containing continuous and discrete units. The current study mostly supports the classifications of 19 species as proposed in previous works. In addition, it also agrees with combining the following species together: C. oleifera and C. vietnamensis; C. sasanqua and C. hiemalis; C. brevistyla and C. puniceiflora; and C. grijsii and C. shensiensis. Further, we propose that C. maliflora be recognized as a variety of C. sasanqua, and C. phaeoclada is best placed in sect. Paracamellia. Moreover, we conclude that these species can be combined: C. tenii and C. miyagii; and C. confusa and C. fluviatilis. Our study indicates that the numerical taxonomy and cladistic analysis based on morphological characters of floral organ is useful in species classification, and this technique appreciated in sect. Oleifera and sect. Paracamellia can be used for identification and classification of other taxa. Introduction The genus Camellia L. (Theaceae) is endemic to southeastern Asia, and 80% species are native to China (Lu et al., 2008). Some species of the plant are used to produce green tea, to cultivate as ornamental plants, and the seeds of others are used to produce edible oils (Lu et al., 2012). More than three million hectares of agricultural land is used to grow Camellia to produce in excess of 164 thousand tons of edible cooking oil (Vijayan et al., 2009). Obviously, the economic value of Camellia is significant. However, the taxonomic relationships between various Camellia species are still unclear. Since Linnaeus (1753) assigned Camellia japonica L. and Thea sinensis L. [=C. sinensis (L.) O. Kuntze] in his first edition of Species Plantarum, there is little consensus upon the combination and species number that should be recognized. Estimates vary from about 82 to 119 or 280 species, depending on the taxonomic authority (Chang, 1998; Ming, 2000). Camellia is regarded as morphologically, anatomically and molecularly heterogeneous genus based on studies of its various sections (Lin et al., 2008; Pi et al., 2009; Pi et al., 2011). Different taxonomic questions in relation to many of its sections remain unresolved, for example, sect. Oleifera and sect. Paracamellia. Sealy (1958) listed six species and two varieties in the sect. Paracamellia. Chang (1998) divided Sealy’s system into sect. Oleifera and sect. Paracamellia, because species of sect. Oleifera have longer styles and androecium, and higher seed oil content (Table 1). There are total 21 species in Chang's taxonomic classification. Ming (2000) concluded that there was no essential difference between sect. Oleifera and sect. Paracamellia, which was largely on the basis of a structural framework proposed by Sealy (1958). By combination of species, Ming (2000) reduced the section to seven species. Since Chang's classification of Camellia provided as part of a comprehensive taxonomic revision, we used it as the primary context for assessing the results of the present study. 1School of Applied Sciences, Health Innovations Research Institute, RMIT University, Melbourne 3000, Victoria, Australia. 2Corresponding author. Email: luhongfei63@yahoo.com.cn 156 JIANG et al. During the last 10 years, efforts to resolve classification issues in sect. Oleifera and sect. Paracamellia have involved data on leaf morphology (Lin et al., 2008), chemical composition analysis (Shen et al., 2008) and molecular marker (Vijayan et al., 2009). However, the taxonomic position of a few species is still ambiguous. It is necessary to seek other information for reassessing the classification of these two sections. The construction of classifications with their positive features depended upon a careful comparison of attributes of the organisms (Stuessy, 2009). Stuessy (2009) reported that although many vegetative characters have also been used to good effect, in a general sense, floral features have been most useful in angiosperm taxonomy. The androecium in flowers was obviously of high taxonomic value within angiosperms, stamens were important in classification of plants, and the anthers occurred in many different sizes and shapes (Hufford and Endress, 1989). Many studies have employed comparisons of several different floral structures (Kocyan and Endress, 2001; Matthews and Endress, 2005). Liston (2003) showed the importance of careful observations of floral morphology to infer homology of ovary position. These all indicate that types of floral characters are more distinguished according to its variability of characters and states. Thus, a detailed analysis of floral morphology can be regarded as a significant method to identify the disputed species among these two sections. The aims of our study were: (1) to explore if numerical taxonomy and cladistic analyses based on morphological characters of floral organ is of value in classification, (2) to investigate the distinction between sect. Oleifera and sect. Paracamellia of the genus Camellia, and (3) to assess the phylogenetic relationships among the 19 species in these two sections. Material and Methods Plant materials Five species from sect. Oleifera and 14 species from sect. Paracamellia were examined (Table 1). As a outgroup, closely related species of sect. Camellia such as C. chekiangoleosa and C. japonica were chosen. The research was based on the investigation of living collections, which were obtained from the International Camellia Species Garden of Jinhua city (ICSG, 29°7´N, 119°35´S, 40 m in altitude). At least three different individual plants per species were selected in this study. Voucher specimens were deposited in the Chemistry and Life Science College of Zhejiang Normal University (ZJNU). Morphological characters For each plant, floral morphological attribute observations were obtained mainly from living plants and partly from the literature. The salient informative morphological variations were selected principally from petal, perule, filament, stamen, style, stigma, ovary, receptacle, pollen and anther (Table 2). A total of five binary characters were coded as 0 and 1, such as petal colour and number of perule. Pollen microscopic observations and measurements were made by using a Zeiss microscope. Other remaining 17 characters were studied by measuring the photographic images (Fig. 1) through the ImageJ software. Seven other variables were added (derived variables, Table 2): petal width-length ratio = petal width / petal length; petal form coefficient = 16×petal perimeter/ (petal area)2; style height-stamen height ratio = height of style / height of stamen; divided style ratio = length of divided style / height of style; ovary diameter-height ratio = diameter of ovary / height of ovary; ovary diameter- receptacle diameter ratio = diameter of ovary/ diameter of receptacle; anther width-length ratio = anther width / anther length. The floral morphological characters we selected in this study were followed by earlier researches (Takahata and Hinta, 1986). TAXONOMY OF CAMELLIA L. 157 Table 1. Species level differences between the classifications of sect. Oleifera and sect. Paracamellia of Camellia L. by Chang (1998) and Ming (2000). Chang (1998) Ming (2000) Framework for classifications Composed 5 species in sect. Oleifera H. T. Chang and 14 species in sect. Paracamellia Sealy Composed 7 species in section sect. Oleifera H.T. Chang Sections with similar members 1. C. gauchowensis H.T. Chang 1. C. gauchowensis H. T. Chang 2. C. oleifera Abel 3. C. vietnamensis T.C. Huang ex Hu 2. C. oleifera Abel 4. C. lanceoleosa H.T. Chang & J.S. Chiu 5. C. fluviatilis Hand.-Mazz. 3. C. fluviatilis Hand.-Mazz. var. fluviatilis var. megalantha 6. C. sasanqua Thunb. 7. C. hiemalis Nakai 8. C. miyagii (Koidz.) Makino & Nemoto 4. C. sasanqua Thunb. 9. C. brevistyla (Hayata) Coh. Stuart 10. C. obtusifolia H.T. Chang 11. C. puniceiflora H.T. Chang 5. C. brevistyla (Hayata) Coh. Stuart var. brevistyla var. microphylla Merr. 12. C. confusa Craib 13. C. kissi Wall. 6. C. kissi Wall. var. kissi var. vonfusa (Craib.) Ming 14. C. grijsii Hance 15. C. yuhsienensis Hu 16. C. shensiensis H.T. Chang 7. C. grijsii Hance var. grijsii var. shensiensis (H.T. Chang) Ming 17. C. maliflora Lindl. hybrid 18. C. phaeoclada H.T. Chang C. saluenensis Stapf ex Bean 19. C. tenii Sealy Sect. Heterogenea Sealy Numerical taxonomy A total of 76 variables for each species were used: 5 binary and 71 continuous (including Max, Ave and Mix). Mix values of diameter of pollen, width of anther, length of anther and anther width-length ratio were ultimately excluded considering dysplasia and abortion of anther and pollen. Gower's (1971) general similarity coefficient (SC) was used to measure pairwise similarities for mixed datasets. Both cluster analysis and principal coordinate analysis were conducted using MVSP software (Version 3.13n, Kovach Computing Services). Cladistic analysis Cladistic analysis was based on a set of 30 characters consisting of five discrete and 25 continuous characters. Discrete characters were scored and entered directly into a data matrix. Continuous characters were converted to discrete characters following Otalora et al. (2008), and modifications were made for our study. Firstly, an ANOVA was analyzed on each character by using SPSS version 17.0 (SPSS Inc., Chicago, IL). If the null hypothesis (H0= mean of each group is equal) was rejected for a given character, a pairwise mean comparison using the Duncans Multiple Range Test (P < 0.05) was performed. Means for groups in homogeneous subsets were obtained. The values with different number were significantly different to convert morphological continuous characters into discrete characters. 158 JIANG et al. Table 2. Salient morphological characters of Camellia species used for the morphometrical analysis. Binary characters are coded as 0 and 1. Morphological characters Morphometrical analysis 1. petal a. colour white (0), pink or red (1) f. area (cm) g. perimeter (cm) h. width (cm) i. length (cm) j. number of petal 2. perule b. number of perule: 6-9 (0), 10-12 (1) k. number of full petaloid perule 3. filament and stamen c. stamen fusion: connate near the base (0), connate in the lower half (1) d. stamen fusion: even (0), uneven (1) e. filament: vertical (0), curved (1) l. number of stamen m. number of stamens layer n. height of stamen 4. style and stigma o. number of stigma p. height of style (mm) q. length of divided style (mm) 5. ovary and receptacle r. diameter of ovary (mm) s. height of ovary (mm) t. diameter of receptacle (mm) 6. pollen and anther u. diameter of pollen (mm) v. width of anther (mm) w. length of anther (mm) 7. derived variables x. petal width-length ration = h/i y. petal form coefficient = 16*g/f2 z. style height- stamen height ratio = p/n ab. divided style ration = q/p ac. ovary diameter-height ratio = r/s ad. ovary diameter- receptacle diameter ratio = r/t ae. anther width-length ration = v/w *See 'Morphological characters' section under Materials and Methods for explanation. The cladistic analysis was performed using Mrbayes 3.1 (Ronquist and Huelsenbeck, 2003) as an alternative for likelihood analysis. A Bayesian Analysis (BA) was performed on standard data set, using a General Time Reversible (GTR) model (nst=6; rates=invgamma) with gamma- distribution rate variation and a proportion of invariant sites, and one million generations of the Markov Chain Monte Carlo (MCMC) chains in two independent runs (Yuan et al., 2010). All other trees sampled were used to calculate a strict consensus tree, thus yielding the Posterior Probability (PP). The tree was rooted with two species of sect. Camellia: C. chekiangoleosa and C. japonica. TAXONOMY OF CAMELLIA L. 159 Results Cluster analysis and principal coordinate analysis The results of cluster analysis and principal coordinate analysis are shown in Figure 2 and Fig. 3, respectively. Fig. 2 shows that the dendrogram divided the 21 species into three clear-cut clusters, Outgroup, Clade 1, and Clade 2. C. chekiangoleosa and C. japonica as outgroup were separated at the top level (SC = 0.715) from the other species of sect. Oleifera and sect. Paracamellia. Clade 1 included 6 species. Except C. hiemalis and C. maliflora; C. gauchowensis, C. vietnamensis, C. oleifera and C. sasanqua belonged to sect. Oleifera. Clade 2 comprised the remaining 12 species of sect. Paracamellia except C. lanceoleosa. On closer inspection, the largest similarity coefficient was recorded within C. tenii and C. miyagii (SC = 0.924), which indicates their close relationship. While C. grijsii and C. shensiensis (SC = 0.885), C. puniceiflora and C. brevistyla (SC = 0.880), C. hiemalis and C. sasanqua (SC = 0.871), C. vietnamensis and C. oleifera (SC = 0.864), C. fluviatilis and C. confusa (SC=0.849), respectively, formed a group distinct from the other species. Principle coordinate analysis (PCO) is displayed in a two- dimensional plot using the first two principal coordinates (Fig. 3). It permitted a visualization of the degree of affinity among these species. Figure 3 also shows that all the species formed three groups, which were consistent with results obtained from cluster analysis. Fig. 1. Petal shape variability of Camellia species. Petals from A) C. fluviatilis; B) C. kissi; C) C. confusa; D) C. miyagii; E) C. tenii; F) C. puniceiflora ; G) C. obtusifolia; H) C. phaeoclada; I) C. brevistyla; J) C. lanceoleosa; K) C. oleifera; L) C. gauchowensis; M) C. vietnamensis; N) C. yuhsienensis; O) C. shensiensis; P) C. grijsii; Q) C. maliflora; R) C. hiemalis; and S) C. sasanqua. Cladistic analysis The phylogenetic relationship deduced from the morphology characters using Bayesian Analysis (BA) (Fig. 4) was largely consistent with cluster analysis (Fig. 2). The species included 160 JIANG et al. in Clade 1 and Clade 2 was exactly the same in both trees. The monophyly of Clade 1 was supported with posterior probabilities (PP = 0.66). The difference being that the BA majority rule tree was somewhat worse resolved than the cluster tree: 1) C. hiemalis, C. sasanqua and C. maliflora diverged at the same time in the BA tree, while C. hiemalis and C. sasanqua were more closely related to each other than to C. maliflora in the cluster tree; and 2) in the BA tree, C. gauchowensis was separated as the most distant from two subclades comprising C. oleifera and C. vietnamensis, and C. hiemalis, C. sasanqua and C. maliflora. While C. oleifera and C. vietnamensis were more closely related to each other than C. gauchowensis. The Clade 1 consisted of these three species, and together with another subclade including C. hiemalis, C. sasanqua and C. maliflora. In Clade 2, BA tree and cluster tree were consistent with one another in two places: 1) C. puniceiflora and C. brevistyla were clustered in subclade at the same level (PP = 0.65), and C. lanceoleosa was closely related to these two species (PP = 0.67); and 2) C. phaeoclada was separated as a distant from two subclades comprising C. obtusifolia, and C. tenii and C. miyagii (PP = 0.80). However, the distinct interrelationships among Clade 2 remained: 1) C. shensiensis, C. grijsii and C. yuhsienensis diverged earlier on in Clade 2 in BA tree, while C. shensiensis and C. grijsii were closer to each other than C. yuhsienensis in cluster tree; and 2) C. kissi had distinct systematic position in both trees. Fig. 2. UPGMA dendrogram based on Gower's general similarity coefficient for the sect. Oleifera and sect. Paracamellia species of Camellia. Discussion Relationship between sect. Oleifera and sect. Paracamellia The present study is the first attempt to reconstruct the phylogenetic relationships between sect. Oleifera and sect. Paracamellia based on floral organ morphology. A comparison between topologies obtained by cluster tree and BA majority rule tree indicated that both methods provided highly similar estimates of phylogeny (Figs 2-4). Based on the present taxon samples, the relationships among the species in sect. Oleifera and sect. Paracamellia are closer. Our results suggest that sect. Oleifera is closely related to sect. Paracamellia. Sealy (1958) and Ming (2000) TAXONOMY OF CAMELLIA L. 161 combined the species into one section because they are according to their common morphological characteristics, like the colour of flowers, bloom and perules drop time. However, Chang (1998) divided them into sect. Oleifera and sect. Paracamellia because he found plants of sect. Oleifera have longer styles and androecium and higher seed oil content than of sect. Paracamellia. Shen et al. (2008) also demonstrated that the mergence of the two sections is quite unnatural based on macro-morphology, micro-morphology and chemical characteristics. In this study, we selected various morphological characters of floral organs combined numerical taxonomy to investigate the distinction between sect. Oleifera and sect. Paracamellia, our results agree on Chang’s system and Shen et al. (2008). Thus, two clusters are supported with low PP value. In addition, there were several morphological traits that demonstrated the discrimination within sect. Oleifera and sect. Paracamellia. The most prominent ones are that sect. Oleifera species had bigger floral organs, more stamen series, longer androecium, longer styles, bigger ovary, and even bigger anthers and pollens than sect. Paracamellia. The analysis of leaf anatomy revealed that sect. Oleifera may differ from sect. Paracamellia. The sect. Oleifera shared the same pattern of anticlinal cells and the same size between adaxial and abaxial epidermal cells, and long ovate stomatal shape (Lin et al., 2008). Shen et al. (2008) reported that sect. Oleifera was distinguishable from sect. Paracamellia in FTIR (Fourier Transform Infrared) fingerprint-like spectra, and they concluded that the merger of the two sections was quite surprising. The formation of sect. Oleifera by Chang (1998) was also confirmed by molecular phylogeny (Vijayan et al., 2009). Differences among the classifications based on leaf anatomy (Lin et al., 2008) and leaf FTIR (Lu et al., 2008), and our results are highlighted in Figure 5. The problems with classification of C. hiemalis, C. maliflora, C. lanceoleosa and other species are discussed below separately. Fig. 3. Scatter plots of two principal coordinate axes. OTUs and characters used are the same as for Figure 2. A square represents a species. Numbers in the figure correspond to Camellia species numbers in Table 1. Interspecies relationships in sect. Oleifera and sect. Paracamellia Classification of C. sasanqua, C. hiemalis and C. maliflora Floral morphology data has certainly improved the comprehension of evolutionary relationships between sect. Oleifera and sect. Paracamellia, but some questions remain. Sealy 162 JIANG et al. (1958) noted that there was a little doubt that C. hiemalis was a form of C. sasanqua. Parks et al. (1981) later reported that C. hiemalis was well known as a group of C. sasanqua. However, Chang and Bartholomew (1984) considered C. hiemalis a distinct species based on short and thick androecium and styles, and free petals at the base, and placed it in sect. Paracamellia. Recently, Lin et al. (2008) showed that characteristics of the leaf anatomy in C. sasanqua and C. hiemalis, such as stomata, size of adaxial and abaxial epidermal cells, and the thickness ratio of palisade parenchyma, were different. Nonetheless their close relationships were demonstrated in the dendrogram of FTIR data, which showed Ming’s (2000) combination of the two species is reasonable (Shen et al., 2008). In the trees we constructed, C. hiemalis had closest relationship with C. sasanqua. Thus, the merge of C. hiemalis and C. sasanqua seems to be natural. Sealy (1958) considered C. maliflora a hybrid of unknown origin and placed it in sect. Theopsis. Chang (1998) also believed that C. maliflora was a hybrid according to the floral characteristics, but leaved its status as a distinct species in sect. Paracamellia. Subsequently, Ming (2000) revealed the C. maliflora was a cultivar and should not be recognized at the species level. Leaf micro-morphology characteristics displayed C. maliflora had closer relationship with sect. Oleifera which convincing evidenced that it was a cultivar of sect. Oleifera species. Our data suggested that it is quite reasonable to place it in sect. Oleifera. It had closest relationships with a clade of C. hiemalis and C. sasanqua. From a morphological point of view, these three species had similar petal colour and shape (Fig. 1Q-S), stamen fusion connate near the base, longer androecium and styles than sect. Paracamellia species. Therefore, based on these above evidence, we cannot dismiss the correctness of results and should do more work to demonstrate that C. maliflora may be regarded as a hybrid of C. sasanqua. Fig. 4. Phylogram from Bayesian phylogenetic analysis of 30 morphological characters of 21 Camellia species. Number behind branches indicate Bayesian clade-credibility values (posterior probability). Classification of C. sasanqua, C. oleifera and C. vietnamensis Chang and Bartholomew (1984) believed that C. vietnamensis was closely related to C. sasanqua and C. oleifera. Ming (2000) merged it into C. oleifera, although there were significant differences between C. oleifera and C. vietnamensis. The combination of C. oleifera and C. vietnamensis by Ming (2000) also had good support from our results, and cluster analysis based on leaf anatomy characters. Although, from our observation, C. vietnamensis had larger petals, they TAXONOMY OF CAMELLIA L. 163 had similar petal form coefficient, height of style, number of stamen, and diameter of ovary. Further, they had unique curved filament, while the others had vertical filament. Based on the above features we support the Ming’s (2000) combination of C. oleifera and C. vietnamensis. Classification of C. grijsii, C. shensiensis and C. yuhsienensis When reconsidering C. grijsii, C. shensiensis and C. yuhsienensis, Ming (2000) treated C. shensiensis and C. yuhsienensis as a variety of C. grijsii. According to our floral morphological observations, C. grijsii, C. shensiensis and C. yuhsienensis had the most similar petal shape - broadly obcordate (Fig. 1), whereas other species had obcordate petals. However, C. yuhsienensis had larger flowers than C. grijsii and C. shensiensis. Our results support the combination of C. grijsii and C. shensiensis (Fig. 2). Lin et al. (2008) showed area of adaxial and abaxial epidermal cell of C. grijsii and C. shensiensis nearly match, but C. yuhsienensis was larger. Pollen exine sculpture characteristics (Ao et al., 2001) and molecular taxonomy from nrITS sequence (Vijayan et al., 2009) also support the combination of C. shensiensis into C. grijsii. Close relations of these two species was also demonstrated in the dendrogram of FTIR data (Shen et al., 2008). Therefore, combining of C. grijsii and C. shensiensis was considered here to be a reasonable one, further investigation may show if C. yuhsienensis should be treated as a variety of C. grijsii. Fig. 5. Differences among species classifications based on leaf anatomy, leaf FTIR and our results from sect. Oleifera and sect. Paracamellia of Camellia species. Classification of C. puniceiflora, C. brevistyla and C. obtusifolia Camellia puniceiflora and C. obtusifolia were designated as the varieties of C. brevistyla by Ming (2000). In our study, C. puniceiflora and C. brevistyla formed a clade with PP = 0.65 and SC = 0.88, which is in accordance with the results of studies on leaf anatomy (Lin et al., 2008) and FTIR (Shen et al., 2008). However, the position of C. obtusifolia was unclear, because it was considered to belong to C. brevistyla by leaf anatomy (Lin et al., 2008), but not supported by FTIR (Shen et al., 2008) and our results. These results indicate that C. puniceiflora should be treated as a variety of C. brevistyla, but their relationship with C. obtusifolia requires more evidence. 164 JIANG et al. Classification of C. phaeoclada, C. tenii, C. miyagii Ming (2000) combined C. phaeoclada into C. saluenensis (belong to sect. Camellia), and placed C. tenii in sect. Heterogenea. Our analyses, combined with FTIR (Shen et al., 2008), leaf anatomy (Lin et al., 2008), strongly suggest that C. phaeoclada is the best placed in sect. Paracamellia. This species is morphologically distinct from other C. saluenensis as well as the remainder of sect. Camellia. We also propose combining C. miyagii and C. tenii, which is congruent with results of leaf FTIR (Shen et al., 2008), but disagrees with results of leaf anatomy (Lin et al., 2008). On the basis of present study in conjunction with other studies based on leaf anatomy (Lin et al., 2008), FTIR (Shen et al., 2008; Lu et al., 2008) and molecular data (Vijayan et al., 2009), we propose a bifurcation of sect. Paracamellia supporting Chang’s (1998) creation of sect. Oleifera from sect. Paracamellia. This proves that there is consistency of the information from different organs of a plant. It also highlights that every new technical development offers promise for improving the description of relationships among species. The present study shows that sect. Oleifera included six species: C. oleifera, C. vietnamensis, C. gauchowensis, C. sasanqua, C. hiemalis and C. maliflora. The sect. Paracamellia comprised the remaining 13 species. Moreover, we extended this observation for interspecies relationship confirming Ming's (2000) combination of C. oleifera - C. vietnamensis, C. sasanqua - C. hiemalis, C. brevistyla - C. puniceiflora, and C. grijsii - C. shensiensis. Further, we suggest combining C. tenii - C. miyagii, and C. confusa - C. fluviatilis. Moreover, we recognize C. maliflora as a variety of C. sasanqua, and have assigned C. phaeoclada to sect. Paracamellia. Numerical and cladistic analyses based on the floral morphometric data employed in this study had enough discriminating power to classify a group of species at section level. Additional floral information is needed to classify individuals at the species level. This technique appears to have taxonomic value and can be widely used for identification and classification of other taxa when the species are closely related. Acknowledgments This study was supported by Science and Technology Project of Jinhua City (No. 2009-2- 020), and Excellent Master Thesis Cultivation Project of Zhejiang Normal University. The authors thank The International Camellia Species Garden of Jinhua City for providing grant to support this work. Thanks are also due to Ms. Wei Zhang and Ms. Jia Lou for their help during specimens collection. References Ao, C.Q., Chen, G.X., Xi, J.B., Zheng, Z.W. and Zhang, H.T. 2001. Study on the pollen morphology of section Paracamellia in genus Camellia. J. South China Agr. U. 22: 66-68. Chang, H.T. 1998. Theaceae. In: Reip. Pop. Sin. Editorial Committee (ed.), Vol. 49. Reip. Pop. Sin. Beijing, Science Press. Chang, H.T. and Bartholomew, B. 1984. Camellias. Oregon, Timber Press. Gower, J. 1971. A general coefficient of similarity and some of its properties. Biometrics 27: 857-871. Hufford, L.D. and Endress, P.K. 1989. The diversity of anther structures and dehiscence patterns among Hamamelididae. Bot. J. Linn. Soc. 99: 301-346. Kocyan, A. and Endress, P.K. 2001. Floral structure and development of Apostasia and Neuwiedia (Apostasioideae) and their relationships to other Orchidaceae. Int. J. Plant Sci. 162: 847-867. Lin, X.Y., Peng, Q.F., Lu, H.F., Du, Y.Q. and Tang, B.Y. 2008. Leaf anatomy of Camellia sect. Oleifera and sect. Paracamellia (Theaceae) with reference to their taxonomic significance. J. Syst. Evol. 46: 183-193. Linnaeus, C. 1753. Species Plantarum. Vol. 1. Stockholm: Impensis Laurentii Salvii. TAXONOMY OF CAMELLIA L. 165 Liston, A. 2003. A new interpretation of floral morphology in Garrya (Garryaceae). Taxon 52: 271-276. Lu, H.F., Jiang, B., Shen, Z.G., Peng, Q.F. and Cheng, C.G. 2008. Comparative leaf anatomy, FTIR discrimination and biogeographical analysis of Camellia section Tuberculata (Theaceae) with a discussion of its taxonomic treatments. Plant Syst. Evol. 274: 223-235. Lu, H., Jiang, W., Ghiassi, M., Lee, S. and Nitin, M. 2012. Classification of Camellia (Theaceae) species using leaf architecture variations and pattern recognition techniques. PLoS ONE 7(1): e29704. Matthews, M.L. and Endress, P.K. 2005. Comparative floral structure and systematics in Celastrales (Celastraceae, Parnassiaceae, Lepidobotryaceae). Bot. J. Linn. Soc. 149: 129-194. Ming, T.L. 2000. Monograph of the Genus Camellia. Kunming, Yunnan Science and Technology Press. Otalora, M.A.G., Martinez, I., Molina, C.M., Aragon, G. and Lutzoni, F. 2008. Phylogenetic relationships and taxonomy of the Leptogium lichenoides group (Collemataceae, Ascomycota) in Europe. Taxon 57: 907-921. Parks, C.R., Kondo, K. and Swain, T. 1981. Phytochemical evidence for the genetic contamination of Camellia sasanqua Thunberg. Japan Journal Breed 31: 168-182. Pi, E.X., Lu, H.F., Jiang, B., Huang, J., Peng, Q.F. and Lin, X.Y. 2011. Precise plant classification within genus level based on simulated annealing aided cloud classifier. Expert Syst. Appl. 38: 3009-3014. Pi, E.X., Peng, Q.F., Lu, H.F., Shen, J.B., Du, Y.Q., Huang, F.L. and Hu, H. 2009. Leaf morphology and anatomy of section Camellia (Theaceae). Bot. J. Linn. Soc. 3: 456-476. Ronquist, F. and Huelsenbeck, J.P. 2003. MrBayes 3: bayesian phylogenetic inference under mixed models. Bioinform. 19: 1572-1574. Sealy, J.R. 1958. A Revision of the Genus Camellia. The Royal Horticultural Society, London. Shen, J.B., Lu, H.F., Peng, Q.F., Zheng, J.F. and Tian, Y.M. 2008. FTIR spectra of Camellia sect. Oleifera, sect. Paracamellia, and sect. Camellia (Theaceae) with reference to their taxonomic significance. J. Syst. Evol. 46: 194-204. Stuessy, T.F. 2009. Plant Taxonomy: The Systematic Evaluation of Comparative Data, 2nd edn. New York, Columbia University Press. Takahata, Y. and Hinata, K. 1986. A consideration of the species relationships in subtribe Brassicinae (Cruciferae) in view of cluster analysis of morphological characters. Plant Spec. Biol. 1: 79-88. Vijayan, K., Zhang, W.J., Tsou, C.H. 2009. Molecular taxonomy of Camellia (Theaceae) inferred from nrITS sequences. Am. J. Bot. 96: 1348-1360. Yuan, W., Zhang W.R., Han Y.J., Dong M.F. and Shang, F.D. 2010. Molecular phylogeny of Osmanthus (Oleaceae) based on non-coding chloroplast and nuclear ribosomal internal transcribed spacer regions. J. Syst. Evol. 48: 482-489. (Manuscript received on 20 June 2012; revised on 9 October 2012)