In Kaeng Krachan National Park (Fig. 1), we established five 100 m × 5 m plots at elevations of 360 m (12°48'11.4"N, 99°26'31.6"E, surveyed on 5 Oct. 2012), 540 m (12°48'18.48"N, 99°25'07.12"E, surveyed on 27 May 2014), 680 m (12°48'25.6"N, 99°24'24.0"E, surveyed on 25 Oct. 2013), 850 m (12°49'03.5"N, 99°22'53.5"E, surveyed on 28 Oct. 2013), and 960 m (12°49'19.7"N, 99°21'57.7"E, surveyed on 21 Oct. 2013). All vascular plants were recorded in each plot. For trees 4 m or taller, we recorded girth and height of trunks. For trees lower than 4 m and herbs, we recorded presence/absence in each of ten 10 m × 5 m sections. For all the species distinguished in the field, we collected voucher specimens and sampled some pieces of leaves for DNA isolation (vouchers were deposited at BKF and FU). Each sample collected for DNA isolation was dried with silica gel in a zipper storage bag. In addition to plants recorded in the five plots, we also collected additional specimens with flowers or fruits from outside the plots.

Figure 1. 

Study area A location of Kaeng Krachan National Park B topographical map of Kaeng Krachan National Park; solid square indicates the type locality (elevation, 960 m) of Cryptocarya kaengkrachanensis M.Z. Zhang, Yahara & Tagane.

We scanned dried herbarium specimens and specimen images were measured for leaf length, leaf width, aspect ratio and circularity using ImageJ (Schneider et al. 2012); aspect ratio and circularity are defined as leaf length/leaf width and 4π × (area/perimeter squared), respectively. To identify species, we examined images of type specimens using JSTOR Global Plants (https://plants.jstor.org/). We also examined specimens kept at the herbaria BKF, BO, FOF, FU, KAG, KYO, RUPP, SAR, SNP and VNM, and reviewed taxonomic literature of Cryptocarya in Southeast Asia (Lecomte 1914, Liou 1934, Kostermans 1988, Kochummen 1989, Hô 1999, Dy Phon 2000, Newman et al. 2007, Li et al. 2008, de Kok 2015, 2016a, Liu et al. 2017).

For DNA extraction, we milled the dried leaf material into fine powder by QIAGEN TissueLyser and the powder was washed three times with 1 ml buffer solution (including 0.1 M HEPES, pH 8.0; 2% Mercaptoethanol; 1% PVP; 0.05 M Ascorbic acid) (Toyama et al. 2015). DNA was then isolated from the washed powder by using the CTAB method (Doyle and Doyle 1987) with a slight modification (Toyama et al. 2015).

We determined partial sequences of the internal transcribed spacer (ITS region) of ribosomal DNA using the following primer sets of Rohwer et al. (2009): ITS18-F (5’-GTCCACTGAACCTTATCATTTAGAGG-3’) and ITS26-R (5’-GCCGTTACTAAGGGAATCCTTGTTAG-3’) and Tks Gflex DNA Polymerase (Takara Bio, Kusatsu, Japan) (Binh et al. 2018). The PCR reaction was carried out following the published protocols of Kress et al. (2009) with some modification by setting PCR cycling conditions as [95 °C 4 min (94 °C 30 sec, 55 °C 1 min, 72 °C 1 min) 25 cycles, 72 °C 10 min]. The PCR products were purified by a diluted mixture of ExoSap-IT (GE Healthcare, Little Chalfont, UK). Using the purified PCR products, forward and reverse sequencing was carried out separately by adding BigDye terminator sequencing mixture (BigDye Terminator v3.1; Applied Biosystems) and setting cycling conditions as [96 °C 1 min (96 °C 30 sec, 50 °C 30 sec, 60 °C 4 min) 25 cycles]. The BigDye reaction products were finally read with ABI 3730 automated sequencer (Applied Biosystems, Foster City, CA, USA).

We determined ITS sequences for 29 samples of Cryptocarya (Table 1) including two samples (numbered as T1883 and T2069) of the Kaeng Krachan taxon and 27 samples of eleven known species. All of these sequences are deposited into the DDBJ database with accession numbers in Table 1. In addition, we downloaded one sequence of a Beilschmiedia sp. collected in Vietnam from the NCBI database (http://www.ncbi.nlm.nih.gov) for outgroup comparison (Table 1).

We amplified thousands of short sequences by using the primers of “multiplexed ISSR (inter simple sequence repeats) genotyping by sequencing” (MIG-seq, Suyama and Matsuki 2015) for 24 samples of Cryptocarya, following the protocol of Suyama and Matsuki (2015). Two steps of PCR were performed; for the 1^st PCR step, we amplified ISSR regions from genomic DNA with MIG-seq tailed ISSR primer set-1 and diluted 50 times for each 1^st PCR product with deionized water (Suyama and Matsuki 2015, Binh et al. 2018). The 2^nd PCR step was conducted with common and indexed primers. The 2^nd PCR products were then pooled in equimolar concentrations as a single mixture library. Fragments of size range 350–800 bp were isolated from the purified mixture of 2^nd PCR products by a Pippin Prep DNA size selection system (Sage Science, Beverly, MA, USA) The concentration was measured by quantitative PCR (Library Quantification Kit; Clontech Laboratories, Mountain View, CA, USA) and then sequenced by Illumina MiSeq Sequencer (Illumina, San Diego, CA, USA) with a MiSeq Reagent Kit v3 (150 cycle, Illumina) (Suyama and Matsuki 2015, Binh et al. 2018).

For DNA barcoding analysis, MEGA X (Kumar et al. 2018, http://www.megasoftware.net/) was used to assemble the ITS sequences of 30 samples including 29 of Cryptocarya spp. and an additional sample of Beilschmiedia sp.; MAFFT ver. 7 (http://mafft.cbrc.jp) was used to align ITS sequences. We reconstructed a phylogenetic tree by the maximum likelihood method with Tamura 3-parameter model using MEGA X with a bootstrap test of 1500 replicates. In addition, we drew a TCS haplotype network (Clement et al. 2000) among 29 samples of Cryptocarya using POPART ver.1.7 (Leigh and Bryant 2015).

For MIG-seq, we pretreated the raw data of Cryptocarya samples following the quality control protocol of Suyama and Matsuki (2015) and Binh et al. (2018). We then assembled homologous sequences (designated as loci below) with the de novo map pipelines (ustacks, cstacks, sstacks) using Stacks ver. 1.48 (Catchen et al. 2011). First, we assembled loci by ustacks with the following settings: m = 3, M = 3, N = 2, and maximum gaps = 2 (where “m” is the minimum depth of coverage, “M” is maximum distance allowed between stacks, and “N” is the maximum distance allowed to align secondary reads to primary stacks). We then used cstacks to build a catalogue of consensus loci by assembling loci from ustacks, by setting the parameter of “number of allowed mismatches between sample (n)” as 2. Second, by using the sstacks, we associated all stacks created by ustacks with the catalog produced by cstacks. Third, we got an output vcf file containing genotypes of individuals at each locus. Subsequently, we used the vcf2phylip program (Ortiz 2019) to convert the vcf file to a phylip type file. Finally, we constructed a maximum likelihood tree with RAxML ver. 8.2 (Stamatakis 2014) and examined its reliability by bootstrapping using 1500 replicates.

In the plot at an elevation of 360 m, we recorded three sterile trees of C. pustulata and collected a specimen (voucher specimen number T0524) from one of these trees. In the plot at 540 m, we found no trees of Cryptocarya. However, a sterile specimen of C. pustulata was collected along the roadside at 550 m (T2971). In the plot at 680 m, we recorded two sterile trees of C. pustulata for which we recorded girth × height as 110.7 cm × 25 m and 11.3 cm × 5.5 m, respectively. In addition, we collected a sterile specimen (T2195) from a tree lower than 4 m. Along the roadside at 709 m, we collected a fruiting specimen of C. amygdalina (T3090) on 30 May 2014. In the plot at 850 m, we recorded two sterile trees of Cryptocarya, both of which were lower than 4 m. However, we could not identify these trees and vouchers were not collected. In the plot at 960 m (Fig. 1), we recorded girth × height for two trees of the Kaeng Krachan taxon as 24.3 cm × 5 m and 15.7 × 4.5 m, respectively. In addition, in the vicinity of the plot, a fruiting specimen (T2069) was collected from a tree 12 m tall on 23 Oct. 2013. Young trees of the Kaeng Krachan taxon lower than 4 m were found in all ten sections of 10 m × 5 m in the 100 m × 5 m plot at the elevation of 960 m.

In fruiting specimens, the Kaeng Krachan taxon (T2069; Fig. 2) has relatively shorter and broader leaves than C. amygdalina (T3090; Fig. 3), but the ranges are largely overlapping: the range (and average±SD) of leaf length (cm) is 2.6–10.3 (7.2±2.5) in the Kaeng Krachan taxon (n=17) vs. 9.4–14.1 (11.6±1.4) in C. amygdalina (n=10); the range of leaf width (cm) is 1.5–6.4 (4.2±0.6, n=17) vs. 3.5–5.2 (4.1±1.2, n=10). On the other hand, the two species are distinct in aspect ratio: 1.38–2.28 (1.79±0.25) in the Kaeng Krachan taxon vs. 2.46–3.43 (2.88±0.37) in C. amygdalina. The two taxa were also different in circularity, but with overlapping values: 0.55–0.77 (0.69±0.07) vs. 0.42–0.61 (0.51±0.06). Figure 4 shows that the Kaeng Krachan taxon is distinguishable from C. amygdalina by having a lower aspect ratio and larger circularity.

Figure 2. 

Cryptocarya kaengkrachanensis M.Z. Zhang, Yahara & Tagane A branch with immature fruit B lower leaf surface C holotype: Tagane et al. T2069 (KYO) D young branchlet E part of an infructescence with immature fruits F longitudinal sections of an immature fruit.

Figure 3. 

Cryptocarya amygdalina A leafy branchlet B lower leaf surface C specimen Tagane et al. T3090 (KYO) D lower leaf surface (dry) E fresh immature fruit F dried fruit.

The abaxial leaf blade surface of the Kaeng Krachan taxon is sparsely covered with minute hairs that are almost invisible to the naked eye or hand lens (10 ×), but visible under a microscope (25 ×). Similarly, the lower leaf surface of C. amygdalina (T3090) is sparsely covered with minute hairs that are visible only under a microscope (25 ×). Both C. amygdalina and the Kaeng Krachan taxon have scalariform to scalariform-reticulate tertiary veins and it is difficult to distinguish between the two species by their venation.

The specimen T2069 of the Kaeng Krachan taxon had smaller fruits than the specimen T3090 of C. amygdalina: the range (and average±SD) of fruit length (mm) is 9.88–13.82 (11.49±1.28, n=14) vs. 22.07–28.1 (25.67±2.3, n=6), and the range (average±SD) of fruit width (mm) is 5.81–9.67 (7.53±2.36, n=14) vs. 11.27–12.63 (12.03±0.49, n=6). However, the fruits of T2069 (Fig. 2E) and T3090 (Fig. 3E) were green and not fully matured when collected. While the fruits of the Kaeng Krachan taxon are ovoid (Fig. 2E, F), the fruits of C. amygdalina are ellipsoid (Fig. 3E, F) but fruits of both specimens are still immature. Fruit stalks of C. amygdalina (T3090) are smooth and not or only slightly swollen (Fig. 3E, F) as in the lectotype (Wallich Cat. 2585, K001116509) and isolectotype (Francis 990, E00393147) of C. amygdalina. On the other hand, fruit stalks of the Kaeng Krachan taxon (T2069) are rough and swollen (Fig. 2E).

Figure 4. 

Scatter plot of leaf aspect ratio (horizontal axis) and circularity (vertical axis). Solid circles: Cryptocarya amygdalina, solid triangles: C. kaengkrachanensis.

A phylogenetic tree constructed from ITS sequences with length of about 670 bp (Fig. 5) showed that C. amygdalina is close to C. albiramea Kosterm. (T3902) and C. pustulata, and the bootstrap support for the monophyly of the clade including these species was 84%. For the ITS sequence, two samples initially identified as C. amygdalina (T3090 collected from Kaeng Krachan, Thailand and MY479 collected from Myanmar) were identical in the ITS sequences determined. Also, two samples of C. pustulata (T2195 of Kaeng Krachan and T1545 collected from Kao Soi Dao, Chanthaburi, Thailand, the type locality) were identical but another ITS sequence of C. pustulata (T2971) differed from T2195 in one base pair (Fig. 6). On the other hand, C. amygdalina is one species in a well supported clade with the plants of the Kaeng Krachan taxon sister to this clade, differing by 10 base pairs in the ITS sequences (Fig. 6) and are distinct from each other in the ITS haplotype network (Fig. 6).

Figure 5. 

Maximum likelihood tree of Cryptocarya species from Thailand reconstructed from ITS sequences. Numbers: bootstrap values; Scale bar: mean number of nucleotide substitutions per site; solid triangle and square bracket: clade of 95% bootstrap value.

Figure 6. 

A haplotype network of Cryptocarya species from Thailand constructed from the ITS sequences.

For 24 of the samples that belonged to a clade supported by 95% bootstrap value in the ITS tree (Fig. 5), we constructed a MIG-seq tree in which conspecific clusters of C. amygdalina (T3090, MY0479), the Kaeng Krachan taxon (T1883, T2069) and C. pustulata (T2195, T2971, T1545) were supported by 100% bootstrap values (Fig. 7). While C. amygdalina and C. pustulata were sister to each other, the Kaeng Krachan taxon was not sister to either of these species, but instead to C. albiramea (Fig. 7), with 99% bootstrap (BS).

Based on the evidence of morphology and phylogenetic analysis presented above, the two samples (T1883 and T2069) collected from Kaeng Krachan national park clearly represent a distinct and new species, which is named as Cryptocarya kaengkrachanensis.

Figure 7. 

A maximum likelihood tree of Cryptocarya species from Thailand. reconstructed using MIG-seq data. Numbers: bootstrap values; Scale bar: mean number of nucleotide substitutions per site;