Phylogeny of the tropical tree family Dipterocarpaceae based on nucleotide sequences of the chloroplast RBCL gene ^†

Three entities are often considered to comprise the tropical tree family Dipterocarpaceae. The original family circumscription, based on the genus Dipterocarpus (Blume, 1825), is confined to Asia and the Seychelles. The Asian dipterocarps include 470 species and dominate the canopy of lowland equatorial forests. Subfamily Monotoideae (Gilg, 1925) comprises ∼30 species in three genera, one of which, Monotes, was earlier associated with Tiliaceae (Heim, 1892). The monotypic subfamily Pakaraimoideae (Maguire and Ashton, 1977) based on Pakaraimaea dipterocarpacea has been described from the Guyana Highlands in South America. The recent discovery of Pseudomonotes tropinbosii from Colombia (Londono et al., 1995; Morton, 1995) extends the range of Monotoideae to the Neotropics. If all three entities are accepted as one family, the Dipterocarpaceae include over 500 species (Table 1). The Asiatic dipterocarps (subfamily Dipterocarpoideae) are further divided into two tribes, Dipterocarpeae and Shoreae (Brandis, 1895), with haploid chromosome numbers of 11 and 7, respectively. The tribe Shoreae is richer in species than Dipterocarpeae due to species richness in the genera Shorea and Hopea. Shorea comprises almost 200 species in 11 sections, while Hopea has over 100 species (Ashton, 1982).

The phylogenetic placement of the Dipterocarpaceae within the angiosperms has long been problematic, with arguments favoring its placement either in the order Malvales or the order Theales (Bessey, 1915; Hutchinson, 1926; Wettstein, 1935; Ashton, 1982). In more recent classifications, Cronquist (1988) placed Dipterocarpaceae in the Theales, while Dahlgren (1983) and Thorne (1992) placed it in the Malvales, again emphasizing the uncertainty of its phylogenetic position. Some of the morphological characters that support the placement of Dipterocarpaceae in the Theales are the possession of persistent imbricate sepals and frequently contorted corolla. Other characters, such as stellate, tufted, or glandular hairs, connectival appendages, mucilage cells, wood rays, floral vascular supply, seed coat, and geniculate petiole, support its placement in the Malvales (Ashton, 1982). Previous analyses of rbcL sequences robustly place Dipterocarpaceae allied to Malvales in a clade with Sarcolaenaceae (Chase et al., 1993; Dayanandan, 1996). A recent comprehensive study of the phylogeny of the Malvales based on rbcL sequence data has provided a more conclusive evidence for the Malvalean affinity of the Dipterocarpaceae (Alverson et. al., 1998).

Maury (1978), after analyzing the morphology of pollen, fruits, embryos, and seedlings of many species in the family Dipterocarpaceae, suggested that the Monotoideae represent a distinct entity and could be treated as a separate family with an alliance to Asiatic dipterocarps. Kostermans (1985) argued that presence of extrafloral glands on the leaf, uniseriate wood rays, androgynophore with numerous stamens, and tricolporate pollen grains in both Monotes and members of Tiliaceae, and the absence of these characters in Asiatic dipterocarps, indicate affinities of the Monotoideae with Tiliaceae. However, he failed to recognize that extrafloral nectaries are widespread among Asiatic dipterocarps. Furthermore, glandular hairs are present in some Asiatic dipterocarps, but absent in Monotes and Tiliaceae. According to Kostermans (1985), the only character shared between Asiatic dipterocarps and Monotes, but not in Tiliaceae, is the imbricate flower sepals, which he did not consider important enough to link Monotes to the dipterocarps. Consequently, Kostermans (1989) created a new family Monotaceae to include both Pakaraimoideae and Monotoideae, recognizing the close affinity between the two subfamilies. He placed Monotaceae close to Tiliaceae and distant from Dipterocarpaceae.

One of the reasons for these disagreements may be due to reversals or parallel evolution of the limited number of morphological characters. It is often difficult to weigh the relative importance of different morphological characters. On the other hand, none of the morphological analyses (Maury, 1978; Ashton, 1982) were done cladistically. An alternative approach is clearly necessary to resolve the relationships of the Dipterocarpoideae, Monotoideae, and Pakaramoideae.

Present understanding of phylogenetic relationships of Dipterocarpaceae is based on a few comparative studies (Maury, 1978, 1979; Ashton, 1979). Suggested affinities among dipterocarps have been based on wood anatomy (Gotwald and Parameswaran, 1966; Brazier, 1979; Parameswaran and Gotwald, 1979), secondary metabolites (Ourisson, 1979), and morphological characters (Ashton, 1979, 1982). Within Dipterocarpaceae, Maury (1978, 1979), on the basis of seed, embryo, and seedling characters, recognized two main groups, one with imbricate fruit sepals and the other with valvate fruit sepals. The imbricate group included two monophyletic genera, Hopea and Shorea, while the valvate clade included Dipterocarpus and Vatica. The placement of most genera within these two clades remained unresolved. Ashton (1979) also pointed out the difficulties involved with classification of Hopea, Shorea, and Neobalanocarpus due to their morphological similarities. The taxonomic position of Hopea brevipetiolaris has long been a point of controversy. This species was originally placed in the genus Balanocarpus. Ashton (1963, 1972, 1980) transferred B. brevipetiolaris into Hopea. Kostermans (1992), however, favored its placement in the genus Balanocarpus. The presence of a terminal band of apotracheal parenchyma, silica (SiO₂) in ray cells, and sparsely distributed resin canals in H. brevipetiolaris supports its placement in a separate genus from Hopea, as those characters are absent in other Hopea species (Parameswaran and Gotwald, 1979). However, the type species of Balanocarpus, B. utilis, lacks these characters.

Two alternative hypotheses exist to explain the origin and phytogeography of dipterocarps (Merrill, 1923; Croizat, 1952, 1964; Ashton, 1982). The first suggests that dipterocarps originated on the Eurasian plate, possibly in the Malesian region, and migrated westward towards south Asia and Africa. This is based on the assumption that the high species diversity of dipterocarps found in the Far East is associated with the origin of the family (Merrill, 1923; Prakash, 1972; Meher-Homji, 1979). The second hypothesis suggests that dipterocarps originated in Gondwana (Croizat, 1952, 1964; Ashton, 1982). One clade, comprising the subfamilies Monotoideae and Pakaraimoideae, extended its range westward from Africa to South America; the other clade, the subfamily Dipterocarpoideae, migrated eastward through the Deccan plate (Ashton and Gunatilleke, 1987), entered the Eurasian plate, and diversified in the Far Eastern region (Ashton, 1980). Ashton (1982) pointed out that dipterocarp dispersal must be overland due to limited seed dispersal ability, obligate species-specific mycorrhizal symbiosis, lack of seed dormancy, and salt intolerance of seeds. This hypothesis is consistent with fossil evidence (Ramanujan, 1968; Lakanpal, 1970), tectonic events (Sclater and Harrison, 1971; Audley-Charles, Hurley, and Smith, 1981; Brock, 1981; Klootwijk and Radhakrishnamoorthy, 1981), the existence of the wingless-fruited dipterocarpoid Vateriopsis on the Gondwana fragment of Mahe, Seychelles, and comparative anatomy and morphology (Ashton, 1982). Reconstructing the phylogeny of dipterocarps using an additional set of characters such as molecular data may shed light on the understanding of the historical biogeography of the family Dipterocarpaceae.

In the present study, we reconstructed the phylogeny of Dipterocarpaceae based on rbcL sequence data of selected species. We provide evidence favoring the placement of the Asian dipterocarps in the order Malvales allied to Sarcolaenaceae and show that Monotes and Pakaraimaea are closely related to Asiatic dipterocarps.

MATERIALS AND METHODS

RESULTS

The heuristic search resulted in 12 equally parsimonious trees, each with a length of 359 steps with consistency index (CI) of 0.50 and rescaled consistency index of (RC) of 0.37 (after excluding uninformative characters). The strict consensus tree of 12 equally parsimonious trees is given in Fig. 1. Ten key morphological characters of Dipterocarpaceae were compiled (Gotwald and Parameswaran, 1966; Maury, 1978; Ashton, 1979, 1982; Jong and Kaur, 1979) and mapped on the phylogenetic tree derived from rbcL sequence data. All branches were resolved in the strict consensus tree, except for the branches from Dryobalanops to the remaining Asiatic dipterocarps, and the Anisoptera, Cotylelobium, and Vatica to the Stemonoporus, Upuna, and Vateria clade.

Dipterocarpaceae and Sarcolaenaceae formed a strongly supported monophyletic clade sister to Thymelaeaceae. Within the dipterocarp clade, Monotes occupied a basal position and Pakaraimaea occupied the next position basal to Asiatic dipterocarps. However, the branches among Sarcolaena, Monotes, and Pakaraimaea were weakly supported with bootstrap values of <50%.

The monophyly of the subfamily Dipterocarpoideae is well supported. Members of Dipterocarpoideae formed four clades (Dryobalanops, Dipterocarpus, and remaining species of the tribes Shoreae and Dipterocarpeae) and relationships among these clades remain unresolved (Fig. 1). Among the remaining Dipterocarpeae taxa, Vateria, Stemonoporus, Upuna, Anisoptera, Vatica, and Cotylelobium formed a monophyletic group with a close affinity among Vateria, Stemonoporus, and Upuna.

The monophyly of the tribe Shoreae is supported except for the position of Dryobalanops (Fig. 1). Species of Shorea section Shorea (S. dyeri, S. lissophylla, S. robusta) formed a clade sister to the clade of S. curtisii (section Mutica) and S. ovalis (section Ovales). Shorea species of section Anthoshorea and the species of the genus Hopea formed a monophyletic clade sister to Shorea section Doona. Neobalanocarpus was sister to the Anthoshorea, Hopea, and Doona clade. Within the Dipterocarpoideae clade, bootstrap results are in agreement with the strict consensus tree obtained from the heuristic search, with few exceptions. All branches showed a bootstrap consensus value of >50%, except for branches supporting the monophyly of Shorea, Hopea, and Neobalanocarpus; Neobalanocarpus, Doona, Hopea, and Anthoshorea; Doona, Hopea, and Anthoshorea; and Upuna, Vateria, and Stemonoporus.

Mapping of morphological characters on the phylogenetic tree reconstructed from rbcL sequence data (Fig. 1; Tables 3 and 4) showed five character state changes (wood parenchyma, resin canals, flower petals, pollen aperture, and pericarp dehiscence) on the branch supporting the monophyly of the Dipterocarpoideae. Within Dipterocarpoideae, monophyly of the tribe Shoreae is supported by five characters (grouped xylem vessels, resin canal grouped into series, chromosome number of seven, imbricate fruit sepal aestivation, and thickened fruit sepal base). The monophyly of the tribe Dipterocarpeae is supported by three characters (solitary resin canals, chromosome number of 11, and free flower petals). Two morphological character state changes (solitary resin canals and chromosome number of 11) were observed on the branch with the genus Dipterocarpus, and three character state changes (resin canals grouped into series, chromosome number of seven, and thickened fruit sepal base) were observed on the branch with Dryobalanops.

DISCUSSION

Analysis of rbcL sequence data suggests that the family Dipterocarpaceae, including Monotes and Pakaraimaea, is monophyletic and is sister to Sarcolaenaceae. Although the Sarcolaenaceae have been considered as a member of Theales (Cronquist, 1988), recent studies showed a malvalean affinity (Dahlgren, 1983, 1989; Thorne, 1992; Conti, Litt, and Sytsma, 1996; Alverson et al., 1998). Close affinity of Dipterocarpaceae to Sarcolaenaceae had earlier been suggested based on wood anatomical and other features (Ashton, 1982). A common origin of Dipterocarpaceae and Sarcolaenaceae from an ancient tiliaceous stock has also been suggested (Maguire et al., 1977). The rbcL data support earlier authors who argued a malvalean affinity for Dipterocarpaceae (Lindley, 1846; Heim, 1892; Hallier, 1912; Corner, 1946; Ashton, 1982; Thorne, 1992) but disputes the thealean affinity (Endlicher, 1840; Lindley, 1846; De Candolle, 1868; Cronquist, 1981). According to Corner (1976), palisade cells in the seed coat of Dipterocarpus and Vatica show similarity to Durio and Cullenia of Bombacaceae (Malvales). Analysis of rbcL sequences also supports a close affinity of that Dipterocarpaceae with Cistaceae and Malvales s. str. (Bombacaceae, Malvaceae, Sterculiaceae, and Tiliaceae; Chase et al., 1993; Alverson et al., 1998).

The evidence presented here suggests that Monotes and Pakaraimaea are more closely related to Asiatic dipterocarps than to Tiliaceae. This is in agreement with other recent interpretations (Maguire et al., 1977; Maury, 1978). Wood anatomical characters also support the close alliance between Monotes and Asiatic dipterocarps (Bancroft, 1935; Gotwald and Parameswaran, 1966). However, this is contrary to Kostermans (1989) who considered that Monotes is more closely related to Tiliaceae than to Dipterocarpoideae. Kostermans (1989) also raised Monotoideae to a family status, including both Monotes and Pakaraimaea. In phylogenetic trees based on rbcL sequence data, Monotes and Pakaraimaea form two distinct, though weakly supported clades, paraphyletic to Asian dipterocarps, and therefore formation of a new family, Monotaceae, cannot be justified. However, all three groups, Monotoideae, Pakaraimoideae, and Dipterocarpoideae, together form a monophyletic group and therefore treating the whole clade as one family is justified. However, a detailed study involving members of Cistaceae and Elaeocarpceae, particularly Muntingia, would be necessary for a comprehensive understanding of the relationships between Sarcolaenaceae and subfamilies of the Dipterocarpaceae.

Phylogenetic relationships within Dipterocarpaceae based on rbcL data are mostly in agreement with present taxonomic treatments (Ashton, 1980, 1982) as well as phylogenetic speculations (Maury, 1978, 1979; Ashton, 1982). The inferred phylogenetic tree topology is in agreement with the division of Dipterocarpaceae into two tribes, mainly based on chromosome numbers (Jong and Kaur, 1979). The tribes Dipterocarpeae and Shoreae form monophyletic groups with the exception of Dipterocarpus and Dryobalanops, for which phylogenetic positions remain unresolved. However, presence of solitary resin canals and the base chromosome number of 11 support the placement of Dipterocarpus within the clade of the tribe Dipterocarpeae. Similarly, grouping of resin canals into groups, chromosome number of seven, and the thickened fruit sepal base support the placement of Dryobalanops within tribe Shoreae.

Meijer (1979) postulated that Dipterocarpus may represent the basal clade of Dipterocarpoideae. Maury (1978) recognized Dipterocarpus as a primitive group among the members of the tribe Dipterocarpeae, and she placed Dipterocarpus along with other taxa of Dipterocarpeae as sister to the group with species of the tribe Shoreae. Dipterocarpus has a number of unique characters, including the winged free calyx tube and large flowers associated with pollination by macrolepidoptera. Dipterocarpus may be a clade with early divergence, but most of the primitive morphological characters appear to have been lost in Dipterocarpus over the course of evolution.

The monophyly of the remaining taxa of the tribe Dipterocarpeae (Vateria, Stemonoporus, Vatica, Cotylelobium, Anisoptera, and Upuna) is in agreement with previous taxonomic and phylogenetic treatments of the dipterocarps (Maury, 1978, 1979; Ashton, 1982). The separate status of Cotylelobium and Vatica and the unity of the latter genus, from which we sampled species from each of its two sections, are upheld. This is contrary to Kostermans’ (1987) decision to unite Cotylelobium with Vatica section Sunaptea and separate both from the type section of Vatica. Morphological data suggest that Vateriopsis also belongs to this group, and further molecular studies will be necessary to infer the phylogenetic position of Vateriopsis.

The phylogenetic placement of the genus Dryobalanops remains unresolved. Ashton (1979) considered Dryobalanops as a member of the tribe Shoreae, while Maury (1978) placed Dryobalanops in the tribe Dipterocarpeae. The presence of solitary vessels suggests the affinity of Dryobalanops to the tribe Dipterocarpeae (Gotwald and Parameswaran, 1966), while chromosome number indicates an affinity to the tribe Shoreae (Jong and Kaur, 1979). Among other morphological characters, valvate fruit sepals (Maury, 1978) support the placement of Dryobalanops in Dipterocarpeae, while connate petals (Ashton, 1982) and grouped resin canals support the placement of Dryobalanops in the tribe Shoreae. A detailed analysis using several species of Dryobalanops would be valuable in further resolving the phylogenetic position of this genus.

The present study shows that Neobalanocarpus, Hopea, and Shorea sections Anthoshorea and Doona form a clade sister to the remaining species of Shorea. The close affinity between Neobalanocarpus, Hopea, and section Doona, is supported by wood anatomical characters (Parameswaran and Gotwald, 1979). Although the close alliance between sections Doona and Anthoshorea has been recognized by Maury (1978), she considered Hopea as a separate clade sister to all species of Shorea.

Neobalanocarpus occupies a weakly supported basal position in the clade comprising section Doona, section Anthoshorea, and Hopea. Floral biology, especially the diurnal anthesis and the stamen structure of Neobalanocarpus, shows similarity to section Doona (Dayanandan et al., 1990). Presence of medium-sized vessels and storied rays and absence of silica in Neobalanocarpus led Parameswaran and Gotwald (1979) to place it in a special position allied to Hopea. Anthocyanin development (Bate-Smith and Whitmore, 1959) and bark morphology (Whitmore, 1962) in Neobalanocarpus are similar to Hopea. Jong and Kaur (1979) observed abnormalities in meiosis in Neobalanocarpus: failure in pairing of chromosomes at metaphase I, irregular disjunction of chromosomes, and tendency to associate in pairs at metaphase II and anaphase II. Therefore, they speculated that Neobalanocarpus may be a result of hybridization between Hopea and Shorea.

Although Maury (1978) placed the genus Hopea and section Anthoshorea in two different but closely related groups, rbcL data support the monophyly of section Anthoshorea and the genus Hopea. The floral morphologies of the genus Hopea and section Anthoshorea are similar, both having an urceolate corolla and stamens with an acicular connectival appendage. The positioning of Hopea and Neobalanocarpus in a clade with Shorea disrupts the monophyly of the genus Shorea. Since the objective of a natural system of classification is to have monophyletic groups as taxonomic units, this relationship needs further evaluation using representative taxa from all sections of both Hopea and Shorea for subsequent consideration in the classification of the genera Shorea and Hopea. According to rbcL data, Hopea brevipetiolaris forms a clade with other Hopea species, and therefore it is reasonable to include it in Hopea rather than in a separate genus.

The monophyly of the three species of Shorea of section Shorea studied is in agreement with taxonomical grouping (Ashton, 1982). Despite the disjunct distribution pattern of Shorea robusta in India and S. dyeri and S. lissophylla in Sri Lanka, these three species form a clade sister to species of Shorea of section Mutica, which are distributed in Malesia. Section Ovales is closely related to Mutica and is also distributed in Malesian forests. Although S. ovalis is a polyploid (Ashton, 1982), the wood anatomy of S. ovalis is similar to that of section Mutica. Therefore, S. ovalis (section Ovales) and section Mutica may be a result of recent divergence from a common ancestor.

The distribution of the taxa of the basal clades, Sarcolaenaceae and Monotes, in Madagascar and Africa where Sarcolaenaceae are now extinct and the absence of such taxa in the Asian region are in agreement with the Gondwanan origin hypothesis of dipterocarps. Presence of fossils of Dipterocarpoxylon in Africa (Chiarugi, 1933) and fossil records of Dryobalanoxylon in India (Ramanujan, 1968) together with the basal position of Dipterocarpus and Dryobalanops in the present cladogram are also consistent with the hypothesis that dipterocarps originated in Gondwana and migrated eastwards through India to the Malesian region. However, the disjunct distribution of closely related taxa both in Sri Lanka and Malesia suggests that dipterocarps must have already diverged to generic or even infrageneric sections before they entered the Laurasian plate from the Deccan plate. For instance, Hopea jucunda and Shorea stipularis, both of section Anthoshorea, are present in Sri Lanka (Ashton, 1980), while their respective sister taxa, Hopea dryobalanoides and Shorea assamica, are present in Malesia (Ashton, 1982).

In summary, this study shows that taxa of Dipterocarpaceae, including Monotes and Pakaraimaea, form a monophyletic group closely related to the family Sarcolaenaceae. The present taxonomic grouping of Dipterocarpaceae is mostly in agreement with the phylogeny with the exception of the genus Hopea, which forms a clade with Shorea section Anthoshorea. Phylogenetic placement of Dipterocarpus and Dryobalanops remains unresolved. Further studies involving representative taxa from both Hopea and Shorea as well as Dipterocarpus and Dryobalanops are needed for a comprehensive understanding of the phylogeny within the Dipterocarpaceae.