﻿Morphological and molecular evidence reveals three new species of Lithocarpus (Fagaceae) from Bidoup-Nui Ba National Park, Vietnam

﻿Abstract Three new species, Lithocarpusbidoupensis Ngoc & Tagane, L.congtroiensis Ngoc & Yahara, and L.hongiaoensis Ngoc & Binh are described from Bidoup-Nui Ba National Park, Central Highland of Vietnam. Morphological analyses and Maximum likelihood tree based on genome-wide SNPs support the distinction of those species from the previously known taxa in the region. The three new species are considered to be endemic to the Bidoup-Nui Ba National Park and the preliminary conservation status for each species is evaluated as Critically Endangered.

Lithocarpus Blume is the largest genus of the family Fagaceae in Vietnam, including 119 species and two varieties, among which 44 species are endemic (Ho 2003;Ban 2005;Linh et al. 2013;Ngoc et al. 2016Ngoc et al. , 2018. The previous taxonomic treatments of Lithocarpus in Vietnam were mostly based on Camus' studies using the specimens collected by French botanists, the results of which were documented in the part of Flore générale de l'Indo-Chine (Hickel and Camus 1930) or Chênes Atlas (Camus 1948). All these early studies relied only on morphological features to identify and construct the keys to species level, so the taxonomic circumscriptions were sometimes inaccurate, leading to continued uncertainty in the taxonomic status and relationship among species.
The phylogenetic approach has become a widespread and efficient way to identify and delimit species, but there is only one study for Lithocarpus in Vietnam (Ngoc et al. 2018). The MIG-seq is a PCR-based method used to identify large numbers of genetic markers throughout the genome (Suyama and Matsuki 2015) that is highly applicable for use in phylogenetic studies (Binh et al. 2018a;Okabe et al. 2021). Recently, a growing number of taxonomic studies of Lithocarpus have yielded new species (Ngoc et al. 2016(Ngoc et al. , 2018, but we often find Lithocarpus material that is difficult to identify to species level. Here we applied MIG-seq for phylogenetic reconstruction to accurately assess the diversity and taxonomy of Lithocarpus species. Bidoup-Nui Ba National Park ( Fig. 1) is in the core zone of UNESCO Langbiang Biosphere Reserve, which is in Lam Dong Province in the central highland of Vietnam. The national park, with the area of 70,038 ha covering almost the entire Langbiang Plateau, harbors 1933 species of vascular plants (Bidoup-Nui Ba National Park 2021) including 62 threatened species (Ban et al. 2007;IUCN 2012), and 42 endemic species (Tagane et al. 2017(Tagane et al. , 2020Bidoup-Nui Ba National Park 2021). For Fagaceae, 25 species of Lithocarpus, nine species of Castanopsis, eleven species of Quercus, and one species of Trigonobalanus have been recorded from Bidoup-Nui Ba National Park (Dung 2005;Ngoc et al. 2016;Binh et al. 2018a).
During our floristic research in Bidoup-Nui Ba National Park from 2015 to present, we found some individuals of the genus Lithocarpus that could not be identified to species level. We here describe them as Lithocarpus bidoupensis Ngoc & Tagane, sp. nov., Lithocarpus hongiaoensis Ngoc & Binh, sp. nov., and Lithocarpus congtroiensis Ngoc & Yahara, sp. nov., based on comparisons of morphology with related species and provide molecular phylogenetic evidence using the MIG-seq method (Suyama and Matsuki 2015).

Taxon sampling
In the present study, we conducted botanical inventories in Bidoup-Nui Ba National Park and the other protected areas in Vietnam and collected a total of 63 samples consisting of 23 species including sixteen samples of three unknown species. Five samples of Lithocarpus balansae (Drake) A.Camus, which is morphologically distinct from the other species of the genus, were included as an outgroup for the phylogenetic analysis. Localities and voucher specimens of these materials are listed in Table 1.

Morphological analysis
We compared morphological traits of three unknown species with those of related species using taxonomic literature (Camus 1931(Camus , 1938(Camus , 1942(Camus , 1943(Camus , 1945(Camus , 1948Huang et al. 1999;Ho 2003;Ban 2005;Phengklai 2008), specimens kept in the herbaria ANDA, BKF, DLU, HN, KAG, KYO, P, and VNM, and digitized plant specimen images available on the web of JSTOR Global Plants (https://plants.jstor.org/) and Chinese Virtual Herbarium (http://www.cvh.org.cn/). The ImageJ software (Schneider et al. 2012) was used to measure the following characters of the new species and related species based on images of type specimens: length, width, aspect ratio and circularity of leaf blade, petiole length, and size of cupules. Aspect ratio and circularity are defined as length/width of leaf blade and 4π × (area/ perimeter squared), respectively. Analysis of variance (ANOVA) and post hoc Tukey's honestly significant difference test (Tukey's HSD) (Tukey 1953) were applied to reveal the mean difference among species. All statistical analyses were performed in R version 4.0.5 (R Core Team 2021) with R-Sutido ver. 1.4.1106 (R-Studio Team 2021).

DNA extraction and sequencing
Leaf pieces were dried using silica-gel in the field, and DNA was isolated with the CTAB method (Doyle and Doyle 1987) with minor modifications described in Toyama et al. (2015). The extracted DNA was diluted to 10 ng/µl and used as templates to amplify thousands of short sequences (loci) from a wide variety of genomes with a standard PCR protocol according to Suyama and Matsuki (2015). MIG-seq library was constructed as described in Suyama and Matsuki (2015) with a minor update by using dual-indexed primers (Suyama et al. 2021). The 1 st PCR, multiple non-repetitive regions from various inter-simple-sequence repeats (ISSRs) were amplified from genomic DNA by multiplexed PCR with tailed ISSR forward and reverse primers sets. The first PCR products were diluted and used as the templates for the 2 nd PCR with dual indexed primers sets. Then, 3 µl of each 2 nd PCR product was pooled in equimolar concentrations as single mixture library. The mixture was then purified and the size range of 350-800 bp were isolated by a Pippin Prep DNA size selection system (Sage Science, Beverly, MA, USA). Quantitative PCR was performed to measure final concentration of size-selected library with approximately 10 pM and then used for sequencing on an Illumina MiSeq Sequencer (Illumina, San Diego, CA, USA), using a MiSeq Reagent Kit v3 (150 cycle, Illumina).

Phylogenetic analysis
A total of 50 samples of 22 species of Lithocarpus including samples of unknown species were sequenced (except NAF122, NAF123, NAF185, NAF192, V3205, V9470, V9492, V9555), of which five samples of L. balansae were used as an outgroup. The low-quality reads and primer sequences were eliminated from raw data by using the trimmomatic software version 0.40 (Bolger et al. 2014). The quality-filtered sequence data were demultiplexed and filtered through the software Stacks v1.46 (Catchen et al. 2011;Catchen et al. 2013) following the parameters set as described by Takata et al. (2019) with minor modifications: in the U-stacks, the option settings of 'maximum distance allowed between stacks (M)' = 4,'maximum distance allowed to align secondary reads to primary stacks (N)' = 4; in the population program, the minimum percentage of individuals required to process a locus across all data (r) was set at 10% and the minimum minor allele frequency required to process a nucleotide site at a locus (min_maf ) = 0.005, the maximum observed heterozygosity required to process a nucleotide site at a locus (max_obs_het) = 0.6. Phylogenetic analyses were conducted using maximum likelihood method on SNPs data set. The model of sequence evolution was set to GTR+G as selected by jMrModeltest 2.1.10 (Darriba et al. 2012). Maximum Likelihood analyses were implemented using the RAxML ver. 8.2 (Stamatakis 2014). The topological reliability of the maximum likelihood tree was evaluated with 1000 bootstrap replicates.

Phylogeny inference
The Maximum likelihood tree based on data set of 4962 genome-wide SNPs strongly supports two sister clades, clade 1 and 2, with 100% bootstrap value (Fig. 2). Clade 1 is divided into two subclades, 1a and 1b, each with 100% bootstrap value. Clade 2 is divided into four subclades of 2a, 2b, 2c, and 2d, each with 100% bootstrap value. The three candidates for new species were included in the clade 2 and supported the monophyly of L. congtroiensis and L. bidoupensis with 100% bootstrap value. The specimen here described as Lithocapus hongiaoensis is sister to two specimens of L. vuquangensis. Lithocarpus bidoupensis (clade 2a) was clearly separated from the morphologically similar species of L. licentii (Clade 1a) and L. blaoensis (Clade 1b) and is sister to many different species in clade 2b. Lithocarpus congtroiensisis is included in a clade with L. encleisocarpus, L. lemeeanus and L. pseudomagneinii (clade 2.b2) with 97% bootstrap support, but it is well supported as monophyletic The morphologically similar L. honbaensis is in a different clade (clade 2.b1).
Lithocarpus hongiaoensis (clade 2.c1) was clearly separated from the morphologically similar species of L. vinhensis (clade 2.b3) but showed a sister relationship with L. vuquangensis in the clade 2.c1 with a strongly bootstrap value (100%). Those three species share the character of solitary cupules but the distribution of L. hongiaoensis is narrowly restricted and apart from the two species: L. vinhensis and L. vuquangensis are distributed in Nghe An and Ha Tinh Provinces, which are located in the north of the Central Coast of Viet Nam, whereas L. hongiaoensis was found only in Bidoup-Nui Ba National Park, Lam Dong Province about 1000 km further south.  Note: diff. = mean difference; Bold font indicates statistically significant differences, (-) not available.

Discussion
The morphological comparison and phylogenetic analysis provided evidence of the validity of three new species. Lithocarpus bidoupensis is most similar to L. blaoensis that occurred in the same locality with L. bidoupensis, and also similar to L. licentii that was collected in Kon Tum Province, which is the type locality of L. licentii. However, the new species is clearly different from both in many morphological traits (Table 2 and Table  3) as well as shown in the phylogenetic results (Fig. 2). The molecular phylogenetic tree strongly supports this disjunction in that the monophyly of L. bidoupensis was supported by 100% bootstrap value, while L. blaoensis and L. licentii were placed in another clade. Lithocarpus congtroiensis is placed in the same clade with L. encleisocarpus and L. dahuoaiensis (2.b2), but the morphology was clearly distinct. Lithocarpus congtroiensis is distinguished from L. encleisocarpus by its greater number of secondary vein (13-15 pairs in L. congtroiensis vs. 8-10 pairs in L. encleisocarpus), cupules with tiny imbricate scales, enclosing 1/3-1/2 of the nut (vs. the scales forming 5-7 dimly concentric flanges, cupules completely enclosing the nut). Especially, the cupule of L. congtroiensis usually clustered of three, while the cupules of L. encleisocarpus is solitary. The morphological distinctness between L. congtroiensis and L. dahuoaiensis is clearly shown in Tables 2 and 3.
Although we could not collect any specimens of L. honbaensis with mature fruits, the morphological analysis of leaf and cupule characters provided enough evidence to distinguish species from L. congtroiensis. In addition, the molecular phylogenetic tree showed that L. honbaensis has a close genetic relationship with L. lemeeanus and L. pseudomagneinii than L. congtroiensis (clade 2b).
Lithocarpus hongiaoensis is most similar to L. vinhensis and L. vuquangensis, of which the latter showed the sister relationship to L. hongiaoensis in the molecular phylogenetic tree (Fig. 2, clade 2.c1). However, L. hongiaoensis is narrowly endemic to the Hon Giao area of Bidoup-Nui Ba National Park, Lam Dong Province, in the southern part of Vietnam while L. vuquangensis is endemic to Vu Quang National Park of Ha Tinh Province, north-central coast of Vietnam. From 2015 to 2017 we conducted three field trips at Bach Ma National Park, Ba Na Nature Reserve and Ngoc Linh National Park. These protected areas are located between Lam Dong and Ha Tinh Province, but we did not find any individual of Lithocarpus similar to L. hongiaoensis or L. vuquangensis. Also, morphological differences are distinct enough to distinguish them as different species (Table 2 and 3). While L. hongiaoensis was collected in Lam Dong Province, L. vinhensis occurred in Nghe An Province, the province located in the north central coast of Vietnam. The genetic differences between L. hongiaoensis and L. vinhensis were presented in the phylogenetic tree, L. hongiaoensis and L. vuquangensis a sister to each other (2.c1), while the two samples of L. vinhensis formed a clade not closely related to these two species (2.b3).
Conservation status. Critically Endangered (CR). From our intensive field survey in Bidoup-Nui Ba Naitonal Park and its vicinities from 2015 to present (Tagane et al. 2017, Binh et al 2018b, Lithocarpus bidoupensis was found only in a narrow area within the protected areas of Bidoup-Nui Ba National Park, and its adjascent area of Son Thai Commune at the elevation range between 1400 and 1669 m. In the area,  Conservation status. Critically Endangered (CR). We found around ten individuals of L. congtroiensis along the road and inside the permanent plot at Cong Troi area, and three individuals in Mt. Langbiang, both located inside the protected area of Bidoup-Nui Ba National Park. Based on criterion D of the IUCN Red List criteria (IUCN 2012), this species is qualified as CR. The new species is endemic to Bidoup-Nui Ba National Park, Lam Dong Province. Diagnosis. Lithocarpus hongiaoensis is similar to L. vinhensis but differs in having much longer petioles, fewer secondary veins, longer infructescences, bigger cupules, and bigger nuts. It is also similar to L. vuquangensis but differs in having much longer petioles, more secondary veins, longer infructescences, and bigger cupules (Table 2).