Phylogenomics and a revised tribal classification of subfamily Dipterocarpoideae (Dipterocarpaceae)

Dipterocarpoideae, the largest subfamily in the Meranti family (Dipterocarpaceae) are an ecologically dominant group of trees throughout much of wet tropical Asia. Increasing anthropogenic pressures on this economically important tree family make it essential to resolve their complex evolutionary relationships and understand the distribution of genetic diversity throughout the family and distribution range. Dipterocarpaceae have been the focal group in a wide range of studies, owing to their economic value, importance in historical biogeography and key role in the evolution of the Asian tropical forest biome. Despite this, persistent taxonomic and evolutionary questions remain, ranging from questions on the geographic origin, sequence of dispersal and the identification of diagnostic characters to circumscribe proper evolutionary groups. Here we present a comprehensive phylogenomic hypothesis for Dipter-ocarpoideae, based on the analyses of plastome and nuclear cistron (NRC) data, and provide an in-depth review on the validity of morphological characters underlying the new tribal classification proposed here for the subfamily. Phylogenomic relationships were inferred using maximum likelihood and Bayesian approaches. Estimates of origin and onset of diversification in major clades and lineages were reconstructed using plastome, nuclear and combined datasets. Results of the separate and combined genomic datasets partly corroborate elements of previous classification systems (with improved support at all levels for major clades) but provide strong support for revising the tribal classification of the subfamily into four main clades: Dipterocarpeae ( Dipterocarpus ), Dryobalanopseae ( Dryobalanops ), Shoreeae ( Hopea , Neobalanocarpus , Parashorea , and all parts of a polyphyletic Shorea ) and Vaterieae (including all other presently accepted Dipterocarpoideae genera). Multi-fossil-dated divergence time estimation suggests Vaterieae first originated in the Late Cretaceous, followed by Dipterocarpeae, with subsequent rise of the Dryobalanopseae and Shoreeae in the Eocene. Diversification of all tribes commenced before the Early Miocene. Our results provide strong support for the position of Neobalano-carpus heimii , Parashorea and (sub-)sections of the genera Anisoptera , Hopea , Shorea and Vatica . Hypotheses on the origin of Neo-balanocarpus heimii by intergeneric hybridisation between Anthoshorea (maternally inherited) and Hopea (paternally inherited) species were corroborated. Finally, our study provides support for future revisionary changes: (1) the elevation to generic rank of sections in Shorea ; and (2) revising the infrageneric classification of Hopea as all (sub-)sections were recovered as not monophyletic.

our understanding of the phylogenetic relationships and the distribution of genetic diversity across the family's range are an essential part of the scientific foundation required to establish conservation and restoration priorities. Dipterocarpaceae have traditionally been classified into three subfamilies: Dipterocarpoideae in Asia, Monotoideae in Africa and South America, and Pakaraimoideae in South America (Maguire & al., 1977;Ashton, 1982;Kostermans, 1985;Londoño & al., 1995;Morton, 1995;Maury-Lechon & Curtet, 1998). Recent molecular phylogenetic studies have suggested that the monospecific genus Pakaraimaea (Pakaraimoideae), previously classified in Dipterocarpaceae, may be more closely related to Cistaceae (e.g., Heckenhauer & al., 2017), and the genus was included in this family in APG IV (2016). Heckenhauer & al. (2017) pointed out that the position of Pakaraimaea among Cistaceae is not supported by its morphology and ecology (placed among Tiliaceae, close to Schoutenia by Kostermans, 1978 andTakhtajan, 1980), and their limited sampling was not sufficient to confirm its position with certainty. The phylogenetic position of Monotoideae is still unclear, and either Sarcolaenaceae (endemic to Madagascar) or Monotoideae have been proposed as sister to Dipterocarpoideae (Takhtajan, 1980(Takhtajan, , 2009APG III, 2009;APG IV, 2016;Heckenhauer & al., 2017).
Generic circumscription in Shoreeae has been problematic, and persisting uncertainty in the morphology-based taxonomy of the tribe has been hypothesized to either be the result of a considerable overlap of the morphospaces of the large genera Shorea and Hopea, intergeneric hybridization, or the presence of ancestral polymorphisms (Ashton, 1982; concluded that the current infrageneric classification should be abandoned as most sections and subsections are non-monophyletic, and either support recognizing a single wide circumscription of Shorea (Shorea sensu Ashton) or to recognize Anthoshorea, Doona, Richetia, Rubroshorea and Shorea s.str. at generic level.
Infrageneric relationships in some other groups have also remained problematic, and Maury-Lechon & Curtet (1998) emphasized the mixed taxa of Vatica and Cotylelobium have remained poorly understood. Sunaptea is placed among Vatica, but morphological and anatomical characters in embryos, fruitseeds and seedlings would suggest a close relationship with Cotylelobium.
Hybridization events may have contributed to problematic aspects of the current classification. Nuclear-and plastomebased phylogenies have indicated hard incongruence for the phylogenetic placement of Parashorea within Shorea (Heckenhauer & al., 2017(Heckenhauer & al., , 2018(Heckenhauer & al., , 2019, and a putative hybrid origin of the monotypic genus Neobalanocarpus (Shoreeae) has been hypothesized, based on conflicts between phylogenies derived from the nuclear PgiC gene and those derived from plastome fragments (Kamiya & al., 2005). However, Neobalanocarpus was not included in the recent studies by Heckenhauer & al. (2018Heckenhauer & al. ( , 2019, and the results based on limited DNA data need further corroboration by more extensive phylogenomic analyses. To address some of these enduring conflicts between historically morphology-based classification systems and lacking genomic data, we employed previously released (Cvetković & al., 2017(Cvetković & al., , 2019 and newly sequenced plastome and nuclear ribosomal cistron (NRC) data for 126 species of Dipterocarpaceae. Our main objectives were: (i) to test the monophyly of the proposed two tribes in Asian Dipterocarpaceae (Ashton, 1982): Dipterocarpeae (Anisoptera, Cotylelobium, Dipterocarpus, Stemonoporus, Upuna, Vateria, Vateriopsis, Vatica) and Shoreeae (Dryobalanops, Hopea, Neobalanocarpus, Parashorea, Shorea) and clarify the uncertain phylogenetic position of Dipterocarpus (Heckenhauer & al., 2017); (ii) to test previous hypotheses on the paraphyly of Shorea sensu Ashton and validity of proposed genus-level segregates (Heckenhauer & al., 2019); (iii) to test the monophyly of proposed subsections in Vatica; (iv) to test previous hypotheses of ancient hybridization events in the evolution of Neobalanocarpus (Kamiya & al., 2005) and Parashorea (Heckenhauer & al., 2019) and check for hard phylogenetic incongruence in other groups; (v) to gain insight into molecular divergence age estimates of the main clades and the tempo of diversification of Southeast Asian Dipterocarpaceae.

■ MATERIALS AND METHODS
Sampling. -Leaf material was collected during field work with collected materials frozen in liquid nitrogen or silica gel-dried. Herbarium material in the collections of Naturalis Biodiversity Center (L) were also sampled. Vouchers were deposited in our herbarium (BGT, Brunei Darussalam), Singapore Botanic Gardens herbarium (SING) and Naturalis Biodiversity Center (L, WAG).
DNA extraction, sequencing, and phylogenomic analyses. -Total genomic DNA was extracted from frozen and silica-dried leaf material using the Plant Genomic DNA Kit (Tiangen Biotech, Beijing, China), following Hinsinger & Strijk (2015). The NEBNext Ultra II DNA Library Prep Kit (Ipswich, Massachusetts, U.S.A.) was used for construction of 350-bp paired-end libraries at Novogene (Beijing, China). Sequencing was performed on an Illumina HiSeq2500 platform (San Diego, California, U.S.A.) by Novogene (Beijing, China), with a read length of 2 × 150 bp.
In addition to the likelihood analyses, we performed a coalescent method (ASTRAL) to analyse the rate variation and signal in the plastid protein-coding genes. Coding genes were extracted from assembled plastomes with Geneious R11 v.11.0.4 (http://www.geneious.com) (Kearse & al., 2012). Extracted genes were aligned individually using MAFFT v.7.475 with the FFT-NS-i algorithm and the "--adjustdirectionaccurately" option (suppl. Appendix S4). A maximum likelihood (ML) tree was then built for each gene with IQ-TREE v.1.6.12 (Chernomor & al., 2016), including 1000 replicates for both ultrafast bootstrap and SH-aLRT. Substitution model for each gene was automatically chosen by IQ-TREE, then used for tree building. Resulting ML trees for individual genes were then concatenated and analyzed using ASTRAL v.5.7.4 (Mirarab & al., 2014) with default parameters.
Fossil calibration and molecular divergence time estimation. -Molecular divergence age estimation was performed using four calibrations (Fig. 1)

S. confusa
Here, we have applied the same settings from Cvetković & al. (2021), with the following modifications: two independent MCMC (Drummond & al., 2002) runs were conducted for 515 million generations for the plastome dataset, 200 million generations for the NRC dataset and 470 million generations for the combined dataset, each with the first 80% of tree samples discarded as burn-in. Effective sample size values (ESS), assessed in Tracer v.1.7.1 (Rambaut & al., 2018), were over 150 for plastome, NRC and combined datasets.
The The RAxML tree constructed using our combined dataset was largely congruent with the topology recovered using the plastome data (suppl. Figs. S5, S6).
Moreover, the four main clades and the placement of genera in Dipterocarpoideae in the ML analyses were additionally supported by Bayesian analysis (for details about the posterior probability values, see Figs. 1 and 2, suppl. Fig. S7).
Tribe Vaterieae first originated in the Late Upper Cretaceous (ca. 62/76/52 Ma, plastome, NRC and combined datasets, respectively), followed by tribe Dipterocarpeae (ca. 56/ 71/47 Ma, as before), and tribes Dryobalanopseae and Shoreeae in the Eocene (ca. 48/62/41 Ma, as before). The longest period of relative stasis between origin and onset of diversification occurred in tribe Dryobalanopseae (ca. 30 Myr; plastome data), and tribe Dipterocarpeae (ca. 40-50 Myr; NRC and combined data). The shortest intervals were detected in tribe Shoreeae (ca. 4-7 Myr; all datasets). Plastome and NRC data show divergence of all generic precursors occurring as early as the Early Miocene. Onset of species diversification of all tribes was initiated before the Late Miocene (combined dataset, Fig. 1) or Early Miocene (plastome and NRC data, Fig. 2, suppl. Fig. S7).

■ DISCUSSION
A revised tribal classification for Dipterocarpoideae and phylogenetic affinities of Dipterocarpus. -Our analyses of plastome data resolved the backbone of the Dipterocarpoideae phylogenetic placement and clarified the previously unresolved positions of Dipterocarpus and Dryobalanops (Heckenhauer & al., 2017). The current tribal classifications of Dipterocarpoideae recognize either the two tribes Dipterocarpeae and Shoreeae (Ashton, 1982;Cao & al., 2006), or the four tribes Dipterocarpeae, Dryobalanopseae, Parashoreeae, and Shoreeae (Takhtajan, 2009), which is not supported by our results. Tribe circumscription here is in partial agreement with that outlined by Brandis (1895), Ashton (1982), Kamiya & al. (2005), Takhtajan (2009), and Heckenhauer & al. (2018. Our study recovers the following four strongly supported tribes, and we propose to restructure the tribal classification of Dipterocarpoideae accordingly: Tribe Dipterocarpeae Rchb.,  1837 -Type: Dipterocarpus C.F.Gaertn. Emergent or canopy trees, columnar but hardly buttressed with untidy globose crowns; prominently lenticellate orangebrown massively flaky bark; large leaf buds; amplexicaul bud scales; furnished stipules with diverse species-defining indumenta; plicate venation resulting in corrugation of their coriaceous leaves; thickly geniculate and often long petioles with often complex rings of vascular bundles and resin canals; variously thickened calyx ornamentations (tubercules, simple or folded wings); large flowers bearing a tubular calyx united at base into a smooth, angled, tuberculate or flanged tube enclosing but free from the ovary; two aliform, valvate sepals all along their development; stamens (15-40) are elongate orange anthers and stout tapering connectival appendages; dispersed resin canals in the wood and the largest stamens are the unique characteristic for Dipterocarpus; chromosome number n = 11 (summarized in Heckenhauer & al., 2017). From India and Sri Lanka to SE Asia.
Note: The name Dryobalanopseae was incorrect and superfluous when published, since it included Dipterocarpus, but it was validly published with the correct name on page 213 and the description on 210, and is available for a tribe containing Dryobalanops and not Dipterocarpus; see Art. -Based on Parashorea Kurz. Emergent or canopy, understorey trees; resin canals in tangential bands; thickened sepal base; fruit sepals imbricate at the incrassate-cupped base of the ripe fruit; 3 strata in pollen exine: absent tilioid structure of exine; T and Y columellae shapetype; always grouped vessels with cellular divisions; radial canal formation; 2 or 3 incrassate bases of sepals (and accrescent sepals) in fruits; free bases of fruit sepals; frigid pericarp tissue; circular fruit equatorial section; embryo cotyledons "coveringpiled"; inferior or median-inferior hypocotyl; bilobed seedling 92 Version of Record cotyledons; 4 root-xylem poles; uni-to tri-lacunar cotyledonary vascular bundles; stomatal types in first leaves paracytic, or para-cyclocytic, or anomo-cyclocytic; elongate stomata, sunken in the epiderm; chromosome number n = 7 (summarized in Appanah & Turnbull, 1998). From India and Sri Lanka to Malesia.
Shoreeae has also been spelled as 'Shoreae'. Given this name has a non-Latin base (named after Sir John Shore), but has been Latinized, where the stem is the single syllable Shore, it seems proper to follow the original spelling of Miquel, who also used 'Shoreeae'. In case we consider this name to have alternative possible genitives, Art. 18.1 (via Art. 17.1) even obliges us to do so. In botanical Latin, double ee's are often avoided, but many recent tribal names based on generic names ending on -ea have also regularly been formed including the double ee. Since Parashorea is derived from Shorea, a tribe named after it should be spelled with the double ee as well, but so far this name lacks a formal description and is not in alignment with our classification superfluous. Anisoptera Korth. Emergent or canopy trees; pollen grains tricolpate and lack endexine; universal presence of intercellular resin canals; valvate base of sepals in calyx of ripe fruit (imbricate at first, after only retain some traces of imbrication); solitary vessels, scattered resin canals; pericarp thickenings; 2 strata in pollen exine; tilioid structure of exine; columellae shape-type V and U; solitary vessels with cellular divisions of canal formation oblique; 0 or 5 incrassate bases of sepals (and accrescent sepals) in fruits; fused bases of fruit sepals; rigid to soft pericarp tissue; circular to 3-symmetric fruit equatorial section; cotyledons in embryo neither covering nor piled; hypocotyl apical or median; seedling cotyledons entire; 6, 8 or 10 rootxylem poles; 3-to multilacunar cotyledonary vascular bundles; stomatal types in first leaves anomocytic or anisocytic; elongate stomata; and sunken in the epiderm (imbricate) or round and raised above the epiderm (valvate); chromosome number n = 11 (summarized in Appanah & Turnbull, 1998). From the Seychelles through India and Sri Lanka to SE Asia.
Note: Miquel (1859) seems to have taken up the name from Korthals (1839), who had described the unranked name Vaterieae for a group containing Vateria and Retinodendron Korth. Subsequently, Blume (1852) used Vaterieae, but as a subfamily name ("subord.") that included both Vateria and Vatica. Miquel was the first to explicitly use the term at tribal level, and provided a new description. Since in this work he only had to deal with some species of Vatica (Vateria does not occur in Indonesia), it is not immediately apparent here that it is based on Vateria instead of Vatica. Since Vaterieae is only the correct tribe name when it is based on the generic name Vateria, and it likely is a classification following previous works of Korthals and Blume, we consider Vateria the type of the tribe, and hence Vaterieae the correct spelling.
Resolving power of genomic data and morphological traits, and taxonomic areas requiring additional investigation. -In this revised setup, the chromosome number of x = 11 is considered a synapomorphy of Dipterocarpeae, and imbricate flowers as a synapomorphy of Shoreeae . Unique morphological characters of Dipterocarpus were summarized in Heckenhauer & al. (2017Heckenhauer & al. ( , 2018. Stemonoporus, Upuna, Vateria, and Vateriopsis were not included in this study and will need to be included in an expanded survey. Monophyly of Stemonoporus and Vateriopsis was confirmed by both morphology (Ashton, 1982) and molecular phylogenetics Gamage & al., 2003Gamage & al., , 2006Heckenhauer & al., 2017). However, phylogenetic placement of Upuna and Vateria remains unresolved (Heckenhauer & al., 2017), and additional genomic data is needed for resolving their placement.
New insights on a hybrid origin of Neobalanocarpus. -Various affinities of Neobalanocarpus heimii have been suggested using morphological and anatomical data: N. heimii was hypothesized to be closely related to Hopea sect. Hopea on the basis of the inflorescence, fruit embryo and germination mode (Ashton, 1982;Yulita & al., 2005) and Doona based on wood anatomy (Parameswaran & Gotwald, 1979). Neobalanocarpus heimii shares morphological characters with both Anthoshorea and Hopea (urceolate corolla and stamens with an acicular connective appendage , and a linear anther in the flower and sub-equal short woody fruit sepals [Kamiya & al., 2005]).
The placement of Neobalanocarpus heimii as sister to Hopea (plastome data, Fig. 2A, suppl. Figs. S3A, S4A, S7A, S8) is in concordance with Gamage & al. (2006) and Tsumura & al. (2011); in NRC-derived phylogenetic trees, Neobalanocarpus is nested within the Hopea clade (Fig. 2B, suppl. Figs. S3B, S4B, S7B), which is not only incongruent with the results of the plastome data, but also with the results from phylogenetic inference based on the nuclear PigC gene that indicates that the genus is nested within Anthoshorea (White Meranti; Kamiya & al., 2005). Kamiya & al. (2005) hypothesized this incongruence is a likely indicator of an ancient hybridization event involving ancestors of Anthoshorea as paternal progenitor and ancestors of the Hopea crown group as maternal progenitor. This hypothesis for a hybrid origin is also corroborated by an irregular behaviour during meiosis Version of Record 93 in Neobalanocarpus (Jong & Lethbridge, 1967;Kamiya & al., 2005). The strong support for inclusion of Neobalanocarpus in Hopea (Fig. 2B, suppl. Figs. S3B, S4B, S7B) was unexpected and may indicate an additional level of complexity not previously recovered. As the NRC reads in our study are derived from the same read pool as the plastome reads, the latter indicating the expected relationship as sister to Hopea, lab artefacts such as sample mix-ups are unlikely. NRC copies can homogenize to either maternal or paternal parent after hybridization (concerted evolution; see Álvarez & Wendel, 2003;Nieto Feliner & Rosselló, 2007). The phylogenetic signal presented by the NRC data may be the result of an additional hybridization event with a species in Hopea. This would have occurred after the hybridization event between the Anthoshorea crown group species and the ancestor of the Hopea crown group that gave rise to Neobalanocarpus. Additional nuclear data is clearly required to further disentangle this complex pattern of reticulation.
Hopea is consistently retrieved as sister to Anthoshorea (Heckenhauer & al., 2018(Heckenhauer & al., , 2019, but there are some inconsistencies in the placement of Parashorea. Previous analyses of plastome data recovered Parashorea as sister to a Shorea s.str. + Rubroshorea clade, while RADseq-derived SNP data indicated a sister relationship to Richetia (Heckenhauer & al., 2019). Our results, with extended taxon sampling in the generic segregates of Shorea, corroborate the phylogenetic position in the plastome phylogenetic analysis; the backbone of tribe Shoreeae in the NRC data-derived phylogenetic tree was moderately supported. Heckenhauer & al. (2019) hypothesized that the incongruent placement in the plastome and nDNA phylogenetic analyses may indicate ancient hybridization, and this hypothesis remains plausible given the signals from the extended plastome and NRC data.
Monophyly of the genera Anisoptera, Cotylelobium and Vatica among Vaterieae was confirmed by our study (as in Gamage & al., 2003Gamage & al., , 2006. Two monophyletic sections in Vatica (Cao & al., 2006) were not retrieved here. In addition, two entries of V. sect. Sunaptea were placed among sect. Vatica in our study, resolving the previously doubtful position of this group (Maury-Lechon & Curtet, 1998).
Dating analyses.
-Here we focused on species in subfamily Dipterocarpoideae and present in detail the origin and divergence of outgroups used elsewhere (Cvetković & al. 2021). Results obtained with our combined dataset (51.78 [45.91-55.77] Ma) partly confirm results from Heckenhauer & al. (2017: 54.9 Ma [39.3-71.6 Ma]) but provide improved phylogenetic resolution. Major clades in Heckenhauer & al. (2017) showed wider range age estimates. A key difference is the position of Dipterocarpus forming a monophyletic clade with the rest of species belonging to Vaterieae, in contrast to all our analyses that recovered four tribes in Dipterocarpoideae, including a well-supported Dipterocarpeae. The placement of Neobalanocarpus in their study is compatible with our plastome dataset; however, our NRC-based results present an additional previously undetected hybridization event with species in Hopea. We agree with Heckenhauer & al. (2017) that calibration remains difficult in the group, despite the large numbers of reported fossils for the family (see discussion further below).

■ CONCLUSIONS
Our plastome and NRC datasets confirm some results from previous studies, but also provide novel insights into the tribal classification of Dipterocarpoideae and present strong support for a new tribal classification for the group. Our data resolves the poorly understood phylogenetic relationships of Dipterocarpus, establishes non-monophyly of sections in Hopea, Shorea and Vatica, and re-assesses hybridization of Neobalanocarpus and Parashorea, revealing a previously undetected event. In our study we have focused on the use of a single (extended) nuclear region but in order to fully corroborate this signal, more extensive data, especially from the nuclear genome, is needed. This will aid in ruling out other potential explanations such as incomplete lineage sorting, chloroplast capture and other organism-level processes that can cause phylogenomic discordance (Spooner & al., 2020).
Our molecular results are consistent with previous hypotheses that several Shorea sections including S. sect. Anthoshorea, sect. Doona, sect. Richetia, sect. Rubroshorea and sect. Shorea are distinct and could be elevated to the genus rank (Maury, 1978;Maury-Lechon, 1979a,b;Heckenhauer & al., 2018Heckenhauer & al., , 2019; this could also resolve the paraphyly of Shorea sensu Ashton (1982) in which the genera Hopea, Neobalanocarpus and Parashorea are nested. Most Shorea subsections were shown to be non-monophyletic, indicating that the infrageneric classification at this rank needs to be revised.

■ ACKNOWLEDGEMENTS
We thank the horticultural staff of Singapore Botanic Gardens, Xishuangbanna Tropical Botanical Garden and Naturalis Biodiversity Center (L, WAG) for their kind assistance in our sampling efforts. This work was supported by the China Scholarship Council (2016GXZS80 to TC), grants from Guangxi University and funding through the Bagui Scholarship team funding (C33600992001 to JSS) and China Postdoctoral Science Foundation Grants (2015M582481 and2016T90822 to DDH).