| Edited by Fumio Tajima. Alfred E. Szmidt: Corresponding author. E-mail: aszmiscb@mbox.nc.kyushu-u.ac.jp |
Dipterocarpaceae is a well-known plant family with approximately 580–680 species (Ashton, 1977; Ashton, 1982; Maury-Lechon and Curtet, 1998). Many members of this family are large forest emergent trees, typically reaching heights of 40–70 m. Their distribution is pan tropical, from northern South America to Africa, Seychelles, Sri Lanka, Philippines, India, China, Thailand, Indonesia and Malaysia with the greatest diversity and abundance in western Malaysia.
The Dipterocarpaceae family is divided into three sub-families: (i) Monotoideae, with three genera and about 30 species, distributed across Africa, Madagascar and South America, (ii) Pakaraimoideae with a single species Pakaraimaea roraimae found in the Guaianan highlands of South America and (iii) Dipterocarpoideae, the largest of the sub-families, with 13 genera and about 470 species (Ashton, 1982), which distribute mainly in South Asian countries such as India, Sri Lanka, Philippines, China, New Guinea, Indonesia, Thailand and Malaysia with the exception of Vateriopsis seychellarum, which is endemic to Seychelles.
The phylogenetic position of the genus Monotes, which is often placed in the sub-family Monotoideae (e.g., Maury, 1978) is still unclear. Initially, it was associated with the family Tiliaceae (Heim, 1892). Later however, it was moved to the sub-family Monotoideae of the family Dipterocarpaceae (Gilg, 1925). On the other hand, based on morphology, Maury (1978) and Kostermans (1989) treated Monotoideae as a separate family.
The sub-family Dipterocarpoideae can be further divided into two tribes: Dipterocarpeae and Shoreae (Brandis, 1895). The genera of the first tribe (Anisoptera, Cotylelobium, Dipterocarpus, Stemonoporus, Upuna, Vateria, Vateriopsis and Vatica) have valvate sepals in fruits, solitary vessels, scattered resin canals and the basic chromosome number n = 11. The genera of the second tribe (Dryobalanops, Neobalanocarpus, Hopea, Parashorea and Shorea) have imbricate sepals in fruits, grouped vessels, resin canals in tangential bands and basic chromosome number n = 7 (Ashton, 1982; Brandis, 1895; Jong, and Kaur, 1979). However, there is still much controversy regarding the number of genera of the Dipterocarpoideae sub-family, especially in the Shoreae tribe, which varies depending on the author between nine and 19 (Ashton, 1977; Ashton, 1982; Kostermans, 1978; Kostermans, 1982; Kostermans, 1984; Kostermans, 1992; Maury, 1978; Maury-Lechon, 1979; Meijer, and Wood, 1964; Meijer, and Wood, 1976). Perhaps the most controversial is classification of the genus Shorea. Based on embryo and leaf epidermal characters Maury (1978) divided this genus Shorea into the following separate genera: Shorea, Anthoshorea, Rubroshorea, Richetia, Doona, and Pentacme. On the other hand, Ashton (1977), Ashton (1980) and Ashton (1982) included them in a single genus Shorea, which was further divided into 11 sections: Shorea, Pentacme, Neohopea, Richetioides, Anthoshorea, Rubella, Brachypterae, Pachycarpae, Mutica, Ovalis and Doona. Yet another classification was proposed by Symington (1943) who divided the genus Shorea into three separate genera: Shorea, Pentacme and Parashorea. Further, based on wood anatomy he divided it into the four following wood groups: Balau, Red Meranti, White Meranti and Yellow Meranti.
The phylogenetic relationships of Dipterocarpaceae have been studied using distribution, fossil and morphological data by Ashton (1982) and the first phylogeny based on molecular data was reported by Tsumura et al. (1996). Since then, several other phylogenetic studies on Dipterocarpaceae were reported based on chloroplast (cp) DNA sequences (Dayanandan et al., 1999; Gamage et al., 2003; Kajita et al., 1998; Kamiya et al., 1998; Morton et al., 1999) and the nuclear gene PgiC (Kamiya et al., 2005). However, previous studies on molecular phylogeny of the Dipterocarpaceae included either limited number of species (Kajita et al., 1998; Morton et al., 1999; Tsumura et al., 1996) or informative sites (Gamage et al., 2003; Kamiya et al., 1998) or both (Dayanandan et al., 1999). The most recent work by Kamiya et al. (2005) has mainly focused on the relationships of Shorea, Hopea, Neobalanocarpus and Parashorea genera and did not include species from the Dipterocarpeae tribe and species of the Doona genus (Kostermans, 1984; Kostermans, 1992; Maury, 1978; Maury-Lechon, 1979). As a result, phylogenetic placement of many species and genera, which belong to sub-family Dipterocarpoideae is still unclear. In particular, little is known about the relationships of the following genera: Vateriopsis, Stemonoporus and Vateria and species from the Doona genus created by Kostermans (1984), Kostermans (1992) and Maury (1978).
It is therefore necessary to examine a larger number of species representing all genera and distribution areas. The main objective of the present work was to provide comprehensive assessment of phylogenetic relationships among Dipterocarpoideae species from Southeastern Asia. Another objective of our study was to ascertain the placement of the genus Vateriopsis (endemic to Seychelles) and classification of many endemic Sri Lankan species from the tribes Dipterocarpeae and Shoreae. In addition, our aim was to investigate the familial affinity of the genus Monotes. Finally, we wanted to determine, which of the two species from outside the Dipterocarpoideae sub-family included in our study (Tilia kuisiana or Monotes madagascariensis) is a better candidate for outgroup species for future studies of Dipterocarpoideae phylogeny.
Among the 79 Dipterocarpaceae species included in our present study, 42 species were from Malaysia, 34 species were from Sri Lanka, one species was from Thailand, one species was from Seychelles and one species was from Madagascar. The species used here represent 14 genera of the family Dipterocarpaceae and thus provide the first comprehensive material for phylogeny reconstruction. To address the issues of outgroup choice and the family placement of the genus Monotes, we have also included one species from the Tiliaceae family: Tilia kiusiana. The chloroplast DNA (cpDNA) used in the present study included the following three regions: trnL-trnF spacer, trnL intron and the partial region of the matK gene, which encodes a splicing-associated maturase (Neuhaus, and Link, 1987).
The total number of Dipterocarpaceae species included in this study was 79. This includes 42 species from Malaysia, 34 species from Sri Lanka and a single species from each of the following regions: Seychelles, Madagascar and Thailand (Table 1). In addition to sequences obtained in the present study (24 sequences for trnL-trnF spacer and trnL intron regions and 65 sequences for matK), we used data reported in the previous studies (Table 1). For the trnL-trnF spacer and trnL intron regions we used seven sequences obtained by Kamiya et al. (1998), 34 sequences obtained by Gamage et al. (2003) and 14 sequences reported by Kajita et al. (1998). For the matK region, we used 14 sequences from Malaysian species reported by Kajita et al. (1998). The remaining matK sequences (65) were obtained in the present study. In addition, we included sequences of the trnL-trnF spacer, trnL intron and the matK gene of Tilia kiusiana, which belongs to the family Tiliaceae.
 View Details | Table 1. List of species used in this study and the database accession numbers of the DNA sequences |
Total DNA was extracted as described by Gamage et al. (2003). The intergenic spacer region between trnL and trnF genes, trnL intron region and a partial region of the matK gene were amplified by polymerase chain reaction (PCR). The primers designed by Taberlet et al. (1991) were used to amplify the trnL-trnF spacer and trnL intron regions. The primers designed by Ooi et al. (1995) were used to amplify the partial matK gene region. Since for most samples the matK gene region could not be amplified, several new primers for both PCR and sequencing were designed. The primers used in this study are listed in Table 2. Amplification was carried out after denaturing the DNA at 94°C for 3 minutes followed by 30 cycles of 1 minute at 94°C, 1 minute at 52–55°C for annealing, 1.3 minutes at 72°C, and ending with 7 minutes at 72°C for extension.
 View Details | Table 2. Primers used in the present study |
PCR products were purified using MiniElute PCR Purification QIAGEN Kit according to the manufacturer’s instructions. Sequencing reactions were carried out using the BigDyeTM Terminator v.3.1 Cycle Sequencing Ready Reaction Kit (Applied Biosystems) following the manufacturer’s instructions. Purified PCR products were directly sequenced using the ABI Prism 3100 Genetic Analyzer (Applied Biosystems). The sequences were determined in both directions. For the trnL-trnF spacer and trnL intron regions sequencing primers were the same as those used for PCR, while sequencing primers designed by Kajita et al. (1998) and newly designed PCR and internal primers were used for sequencing of the matK region (Table 2). Nucleotide sequence data obtained in this study and the sequences used by Kamiya et al. (1998) are diposited in the DDBJ/EMBL/GenBank databases under accession numbers AB246414 through AB246478, AB246479 through AB246543 and AB246544 through AB246608 for the matK, trnL-trnF, and trnL intron regions respectively.
Sequences for the trnL-trnF spacer, trnL intron and matK regions were aligned individually and as combined data set using the ClustalX program (Thompson et al., 1997). The aligned sequences were corrected manually using the BioEdit program (Hall, 1999).
Kimura’s two-parameter distance (Kimura, 1980) was used to calculate the genetic distances for all pairs of sequences. Neighbor-joining (NJ) trees (Saitou, and Nei, 1987) for both individual and combined data sets were constructed excluding and including alignment gaps using the ClustalX program (Thompson et al., 1997). Pair-wise deletion option was used when gaps were included in the distance calculations. Phylogenetic tree using Maximum Likelihood (ML) method (Felsenstein, 1981) was obtained based on the combined data using the SEQBOOT, DNAML and CONSENSE programs from the PHYLIP v. 3.6 package (Felsenstein, 2004). In this analysis, empirical nucleotide frequencies were used and the transition/transversion ratio was set to 0.5 as estimated from the combined data set. The statistical support for the nodes of the trees was determined using bootstrap (BT) method (Felsenstein, 1985) based on 1000 replicates.
In the present study, we determined sequences of the trnL-trnF spacer and trnL intron regions for additional 24 species, which were not included in our previous study (Gamage et al. 2003). For the trnL-trnF spacer region the total number of sites, including gaps was 408 after alignment of which 101 sites were variable and 49 were parsimony informative. For the trnL intron region the total number of sites after alignment, including gaps was 560 of which 165 sites were variable and 39 were parsimony informative. There were 14 indels ranging from 1 bp to 89 bp in the trnL-trnF spacer while 30 indels ranging from 1 bp to 101 bp were found in the trnL intron. A long indel of 89 bp was found in the trnL-trnF spacer region of Hopea latifolia and H. helferi. The longest indel, 101 bp was found in the trnL intron of all Stemonoporus species. Sri Lankan endemic species belonging to the Doona genus established by Kostermans (1984) and Maury (1978) had a common 33 bp indel in the trnL intron. All species from the genus Vatica had one 6 bp indel in both the trnL-trnF spacer and the trnL intron regions. Monotes madagascariensis had two indels of 1 bp and 4 bp in the trnL-trnF spacer and one indel of 7 bp in the trnL intron. Tilia kiusiana had eight indels (<4bp) in the trnL-trnF spacer and three indels (two of 1 bp and one of 38 bp) in the trnL intron.
In the present study, we determined 65 sequences of the partial matK gene region. The aligned matrix of the matK region comprised 972 bp. There were 235 polymorphic sites, of which 106 sites were parsimony informative. One 6 bp long indel was found in the T. kiusiana sequence.
Topologies of the NJ trees obtained separately for the trnL-trnF spacer, trnL intron, and the partial matK gene region were generally congruent, with small differences in the resolution of some genera such as Cotylelobium, Dipterocarpus, Hopea and Vateriopsis (data not shown). The topology of the NJ tree based on the matK sequence was identical with that of the tree based on the combined data set including all three regions used in the present study (data not shown). There were no considerable topological differences between the trees constructed including and excluding alignment gaps. The topology and BT support of the ML tree based on the combined data set were very similar to those of the corresponding NJ tree (data not shown). Therefore, only the NJ tree constructed using the combined data set including alignment gaps is presented (Fig. 1).
 View Details | Fig. 1. Neighbor joining tree constructed using the combined data set for the trnL-trnF spacer, trnL intron and matK regions, based on Kimura’s two parameter distance (Kimura, 1980). Bootstrap values (BT) in percent from 1000 replicates are indicated above the nodes. The BT values, which were < 50% are not shown. The tree is unrooted and branch lengths are proportional to the scale given in nucleotide substitution per site. |
Monotes madagascariensis was grouped together with Tilia kiusiana. Next to this group was Vateriopsis seychellarum, which occupied a single branch sister to the clade containing species from the Dipterocarpeae tribe.
The genera: Stemonoporus, Anisoptera, Vateria, Upuna, Cotylelobium and Vatica formed distinct clade (designated as A on Fig. 1) supported by high bootstrap (BT) probability (99%). It contained two clades: the Stemonoporus clade, and the clade with the following genera: Anisoptera, Vateria, Upuna, Cotylelobium and Vatica. Within the latter clade Anisoptera, Cotylelobium and Vatica were monophyletic, while Upuna and Vateria formed sister clades. Except for the Upuna, Vateria and Vatica clades, other generic clades had high BT support (80%–99%).
Most of the remaining genera included in our study formed separate clades on our tree. Here, the genera Dipterocarpus and Dryobalanops formed two separate monophyletic clades each with 100% BT support. Dipterocarpus clade was divided into two groups, one (BT = 77%) with Malaysian species (D. kerrii, D. baudii and D. cornutus) and the other (BT = 71%) with three Sri Lankan species (D. zeylanicus, D. insignis, and D. glandulosus) and D. alatus from Thailand.
The remaining genera formed two main clades (designated as B and C on Fig. 1). Clade B contained species belonging to the genus Richetia (BT = 100%) (Maury, 1978) or Yellow Meranti (Symington, 1943) wood group, Shorea (Maury, 1978) or Balau (Symington, 1943) wood group, Parashorea (BT = 65%), Rubroshorea (BT = 81%) (Maury, 1978) or Red Meranti wood group (Symington, 1943). The clade C contained the following genera: Doona (BT = 99%) (Kostermans, 1984; Kostermans, 1992; Maury, 1978), Anthoshorea (BT = 99%) (Maury, 1978) or White Meranti wood group (Symington, 1943), Neobalanocarpus heimii (BT = 97%) and Hopea (BT = 97%). The clades B and C had 93% BT support. Species belonging to the genera Richetia and Shorea formed separate clades. Parashorea and Rubroshorea were grouped separately with the former genus as a sister clade. Richetia clade was sister to the Shorea clade.
We obtained two NJ trees (with and without alignment gaps) for the combined data set including trnL-trnF spacer, trnL intron, and the partial region of the matK gene. Except for the differences in the bootstrap (BT) support for some nodes, the topologies of these trees were similar. Thus, we think that alignment gaps had little effect on the topology of our phylogenetic trees. Furthermore, topologies of our NJ and ML trees were nearly identical. Hence, only the NJ tree based on combined data set including gaps is discussed.
The generic relationships revealed by our NJ tree are mostly in agreement with previous molecular phylogenies based on cpDNA (Dayanandan et al., 1999; Gamage et al., 2003; Kajita et al., 1998; Kamiya et al., 1998). However, our current results provide new evidence regarding the relationships of additional genera such as Monotes, Vateriopsis and Stemonoporus, which were not included or discussed well (except Monotes) in most previous studies (Dayanandan et al., 1999; Kajita et al., 1998; Kamiya et al., 1998). Our study also gives new information about the validity of the previous classifications of the genus Shorea.
To date, the genus Monotes was included only in three previous studies of Dipterocarpaceae phylogeny (Dayanandan et al., 1999; Gamage et al., 2003; Morton et al., 1999). Moreover, no effort employing molecular data was made to directly test its placement in the family Tiliaceae suggested by Heim (1892) based on flowers and fruit features. Morphological similarity of Monotes to Tiliaceae was also suggested by Kostermans (1985). Other classifications however, placed it in a separate family Monotaceae, which includes two sub-families Monotoideae and Pakaraimoideae (Kostermans, 1989). Based on the rbcL sequences, Dayanandan et al. (1999) suggested that Monotes is more related to the Asian Dipterocarpaceae than to Tiliaceae. Our present results showed that Monotes madagascariensis was placed together with Tilia kiusiana. However, the M. madagascariensis branch was much shorter (0.0243) than the branch with T. kiusiana (0.1159). Furthermore, the internal branch of the clade containing these two species was relatively short (0.0080). This result supports suggestion by Dayanandan et al. (1999) to place the genus Monotes within the Dipterocarpaceae family. Furthermore, it also indicates that M. madagascariensis appears to be better candidate for outgroup species than T. kiusiana, which was used for this purpose by Kajita et al. (1998).
The topology of the present phylogenetic tree, was to a certain extent, consistent with the current division of the sub-family Dipterocarpoideae into two tribes: Dipterocarpeae (n = 11) and Shoreae (n = 7) (Ashton, 1982; Brandis, 1895; Jong, and Kaur, 1979; Maury-Lechon, 1979). The two tribes formed two monophyletic clades (BT = 98%) on our tree except for the genus Dipterocarpus from the Dipterocarpeae tribe, which was placed as a sister clade to species from the Shoreae tribe.
Similar to result reported by Gamage et al. (2003) monotypic species Vateriopsis seychellarum endemic to Seychelles Island was placed on a separate branch sister to the Dipterocarpeae clade (Fig. 1). On our tree, the relationship of this species with other species of the tribe Dipterocarpeae, which have the same chromosome number (n = 11) is relatively well supported (BT = 86%). The origin of this species is still unclear and the possibilities of both plate tectonic movements and the human transportation should be considered (Kostermans, 1992). Embryological evidence suggested that it is related to Dipterocarpus, Hopea, Shorea and Vateria (Oginuma et al., 1999). However, it more resembles Dipterocarpus than the other three genera in having the micropyle formed by both the inner and outer integument and a conspicuously enlarged chalaza (basal part of the ovule opposite the micropyle, where integument and nucellus are joined) with ample vascular tissues (Oginuma et al., 1999). The placement of Vateriopsis seychellarum on our tree did not support its relationship with Dipterocarpus, which occupied a separate clade, sister to the clade containing species from the Shoreae tribe. Based on our results it appears that it rather represents a relatively diverged member of the Dipterocarpeae or Shoreae tribe.
The genus Stemonoporus, which is endemic to Sri Lanka formed a distinct and well supported monophyletic clade (BT = 98%). This is in agreement with the phylogenetic analyses based on rbcL data and noncoding cpDNA (Dayanandan et al., 1999; Gamage et al., 2003). Based on comparative morphology Stemonoporus was considered as one of the most archaic genera of the Asian sub-family Dipterocarpoideae (Ashton, and Gunatilleke, 1987). We found a long 101 bp indel in the trnL intron in all Stemonoporus species included in the present study. Divergent status of Stemonoporus revealed in the present and other studies is also consistent with its unique morphological features such as peculiar anthers with apical dehiscence and apical leaf traces, which separate from the central vascular cylinder well before the node (Ashton, 1982; Kostermans, 1992).
Our present results showed that (except for Upuna and Vateria) Anisoptera, Vatica and Cotylelobium clades are monophyletic, although only the Cotylelobium clade had high BT support (93%). Vatica also showed monophyly in the study by Dayanandan et al. (1999). However, its relationships with Anisoptera and Cotylelobium were not elucidated. On our tree, Upuna and Vateria were grouped together but with low BT support. On the other hand, the rbcL analysis placed Upuna and Vateria on two separate branches sister to Stemonoporus (Dayanandan et al., 1999). Kostermans (1992) suggested that Vateria is closely related with Vatica and that there is no consensus whether these two genera should be fused or kept separate. Further studies are necessary to elucidate the relationship of these two genera.
On our NJ tree, the genus Dipterocarpus formed a distinct, highly supported monophyletic clade (BT = 100%, Fig. 1). Similar results were reported in the previous molecular phylogenies (Gamage et al., 2003; Kajita et al., 1998; Kamiya et al., 1998). Morphological evidence also supports highly divergent character of this genus. Dipterocarpus has many unique characters, including the winged free calyx tube and large flowers (Dayanandan et al., 1999). Some studies suggested that Dipterocarpus might represent the basal clade of Dipterocarpoideae sub-family (Meijer, 1979). On the other hand, others placed Dipterocarpus (together with other members of the Dipterocarpeae tribe) as a sister to the group with species of the tribe Shoreae (Maury, 1978). Our present results also indicate that the genus Dipterocarpus was among the most diverged genera of the sub-family Dipterocarpoideae. Finally, Sri Lankan Dipterocarpus species (D. glandulosus, D. hispidus, D. insignis and D. zeylanicus) formed a separate clade but they were not much diverged from the other species of the Dipterocarpus clade. Thus, our results suggest early divergence of this genus from other species of the Dipterocarpoideae and independent evolution of Sri Lankan species.
Similar to Dipterocarpus, the genus Dryobalanops also formed a distinct, highly supported monophyletic clade on our tree (Fig. 1, BT = 100%). Ashton (1979) placed Dryobalanops in the tribe Shoreae due to the presence of connate petals. Such placement was also suggested by the presence of solitary vessels (Gotwald, and Parameswaran, 1966) and the chromosome number (n = 7) (Jong, and Kaur, 1979). However, Maury-Lechon (1979) placed it in the tribe Dipterocarpeae based on the presence of valvate fruit sepals. Our results showed that Dryobalanops was placed as a sister clade to the cluster containing species from the Shoreae tribe, which supports classification proposed by Ashton (1979).
The topology of our tree lends some support to the classification of Shorea species proposed by Maury (1978) and Maury-Lechon (1979). On our tree, the genera created by this author (Richetia, Shorea, Rubroshorea, Doona and Anthoshorea) are resolved as separate groups, although the clade containing Shorea members had weak BT support (<50%). Some of these genera (Richetia, Shorea and Rubroshorea) also formed separate groups on the tree reported by Kamiya et al. (2005) although their study did not include Doona and Pentacme species. Our present study also did not include species from the genus Pentacme recognized by Maury (1978) and Maury-Lechon (1979). Therefore, it is important to include them in future phylogenies for obtaining further support for the classification of Shorea species proposed by this author.
The placement of Shorea species on our tree is also in agreement with classification proposed by Symington (1943). White Meranti, Red Meranti, Balau and Yellow Meranti were all monophyletic. Similar result, except monophyly of the Red Meranti, was reported by Gamage et al. (2003) and Kamiya et al. (1998). On our tree, White Meranti, Red Meranti, and Yellow Meranti had high BT support (>81%). However, the Yellow Meranti-Balau clade was sister to the clade with Parashorea and Red Meranti. Thus, the topology of our tree is not consistent with that of the tree obtained by Kamiya et al. (2005), where Parashorea was placed on a long separate branch sister to Yellow Meranti, Balau and Red Meranti. On the other hand, the placement of Parashorea (within the clade containing Balau, Red Meranti and White Meranti) revealed in our present study is similar to that reported in the previous cpDNA based phylogenies (Gamage et al., 2003; Kamiya et al., 1998). Actually, the wood groups recognized by Symington (1943), well correspond with the generic classification proposed by Maury (1978) and Maury-Lechon (1979). That is, Yellow Meranti with Richetia, Balau with Shorea, Red Meranti with Rubroshorea and White Meranti with Anthoshorea.
Taking into account many distinctive morphological differences between Shorea and Doona species, several studies suggested that Doona should be regarded as a separate genus (Kostermans, 1984; Kostermans, 1992; Maury, 1978; Maury-Lechon, 1979). In our present study, species placed by these authors in the genus Doona (S. megistophyla, S. ovalifolia, S. worthingtonii, S. gardneri, S. trapezifolia, S. zeylanica, S. disticha, S. cordifolia, S. congestiflora and S. affinis) formed particularly distinct, monophyletic clade with 99% BT support. Based on morphology these species were placed by Ashton (1972), Ashton (1977) and Ashton (1982) in a separate section Doona. The common 33 bp indel in the trnL intron present in all Doona species provides further evidence for distinct character of this group. Our present study also resolved the position of additional two Doona species (S. disticha and S. ovalifolia), which were not included in our previous study (Gamage et al., 2003). Therefore, we could determine phylogenetic position of almost all the Doona species present in Sri Lanka.
Parameswaran and Gotwald (1979) reported that the genus Neobalanocarpus has close affinity with Doona based on wood anatomy. Floral characters such as diurnal anthesis and stamen structure of Neobalanocarpus also show similarity to Doona (Dayanandan et al., 1999). However, on our tree Doona had the sister relationship to the Neobalanocarpus heimii branch, which in turn was sister to the Hopea clade. This agrees with results of other cpDNA phylogenies (Gamage et al., 2003) but is incongruent with phylogeny based on nuclear (n) DNA, which placed Neobalanocarpus together with Anthoshorea species in the most basal and first diverged clade sister to clades containing Shorea, Parashorea and Hopea (Kamiya et al., 2005). The different placement of Neobalanocarpus in cpDNA and nDNA based phylogenies together with the morphological characters shared by Neobalanocarpus, Anthoshorea and Hopea, and the irregular behavior of Neobalanocarpus during meiosis (Jong, and Lethbridge, 1967) lead Kamiya et al. (2005) to suggest that it may be a hybrid between Anthoshorea and Hopea. Our results also showed that Neobalanocarpus has an intermediate position between Hopea and Anthoshorea (BT = 97%). If this placement is associated with the hybrid nature of Neobalanocarpus our present result would also imply the occurrence of recombination in the cpDNA. This is surprising because it is believed that due to its uniparental inheritance in most plants cpDNA does not undergo recombination (Chiu and Sears, 1985). Therefore, our result suggests that cpDNA in some Dipterocarpaceae species is inherited biparentally and undergoes recombination. Further investigation regarding this mater is necessary.
An unresolved feature in the previous classifications of the Shoreae tribe was that the well recognized genus Hopea was placed within the clade containing other species of that tribe. Our results showed that Hopea group was monophyletic within the clade containing Anthoshorea and Neobalanocarpus and had high BT support (97%). The topologies of the PgiC (Kamiya et al., 2005) and rbcL trees (Dayanandan et al., 1999) also showed the monophyly of Hopea and placed it within the clade containing other species of the Shoreae tribe. The floral morphology of the genera Hopea and Anthoshorea are similar, both having an urceolate corolla and stamens with an acicular connective appendage (Dayanandan et al., 1999). There are also some unique morphological characters shared by Hopea, some Shorea species and Neobalanocarpus (Kamiya et al., 2005). Therefore, it is possible that these genera have yet not reached the generic level of divergence at molecular level, even though they have already evolved some different morphological characters. Further evaluation using morphological and molecular data is important for detailed classification of these genera.
There are about 58 species of Dipterocarpaceae in Sri Lanka (Kostermans, 1992). They belong to the genera: Dipterocarpus, Shorea, Doona, Hopea, Stemonoporus, Cotylelobium, Vatica, and Vateria. Ninety eight percent of the species are endemic. Our present phylogeny revealed the monophyly of Sri Lankan endemic genus Stemonoporus and Doona species while other species formed separate clades. However, the present phylogeny did not reveal much divergence between Sri Lankan and other Dipterocarpideae species. The isolated position of Sri Lankan species on our tree may be due to their independent evolution caused by the geographic isolation. The placement of Sri Lankan Shorea stipularis, which belongs to the Anthoshorea section (Ashton, 1980; Ashton, 1982) or genus (Maury, 1978) with other Anthoshorea species (S. bracteolata and S. assamica) from Malaysia is in agreement with such taxonomical grouping. Geographical distribution of S. stipularis in Sri Lanka and its morphological similarity to Malaysian Shorea species suggest that Dipterocarpaceae must have already diverged to generic or infrageneric sections before they entered the Laurasian plate from the Deccan plate according to the Gondwanan origin of Asian Dipterocarpaceae (Dayanandan et al., 1999). The other Sri Lankan Shorea species (S. lissophylla, S. dyeri and S. pallescens), which belong to section (Ashton, 1982) or genus (Maury, 1978) Shorea were monophyletic and had close relationship with other species from this group. To obtain more refined phylogeny and further insights into the evolutionary history of Sri Lankan species, additional sequence data are necessary. They should also include species from India. Sri Lanka was intermittently connected to mainland India and this could have enabled biotic interchange with southern India during the Pleistocene ice ages (Bossuyt et al., 2004). Thus, Sri Lankan and Indian Dipterocarpaceae species may be closely related. The most likely possibility is that Dipterocarpaceae spread to Sri Lanka through India.
We would like to thank H. Ishiyama from Kyushu University, Japan and A. Yoshida from the Research Institute of Evolutionary Biology, Japan and two anonymous reviewers for help and comments in the course of this work, K. Kamiya also from Kyushu University, Japan for providing unpublished sequence data for some Shorea species, M. Jayaweera from the Plant Genetic Resources Center, Sri Lanka for identifying Dipterocarpaceae species collected from Sri Lanka, the Officers of the Forest Department and Forest Conservation Department, Sri Lanka for permission to collect samples. This work was financially supported by the grant (No.1535 SRI (SF)) to the Department of Botany, University of Ruhuna, Sri Lanka from the Asian Development Bank and the grants 13490022 and 13575002 to AES and 072898 to TY from the Ministry of Education, Culture, Sports, Science and Technology, Japan and the Sasakawa Scientific Research grant 16–268 to TGD from the Japan Science Society.
|
Index