Two New Amanita Species in Section Amanita from Thailand

: Based on a survey of macro-fungi in northern and northeastern Thailand, nine samples collected in 2020 are identified as Amanita and introduced here as two new species, Amanita kalasinensis and A. ravicrocina . Typical macro- and microscopical characteristics indicate that both of these two species belong to Amanita section Amanita , but differ from other currently known species. Amanita kalasinensis is characterized by having a greyish yellow pileus covering with a conical to granuliform, yellowish white volval remnant; the presence of clamps; and a broadly ellipsoid to ellipsoid basidiospore. Amanita ravicrocina is characterized by having a brown to greyish orange pileus covering with a patchy, white volval remnant; a collar-like volval remnant on the stipe; and a subglobose to broadly ellipsoid basidiospore. Multi-gene phylogenetic analysis of partial nuclear rDNA internal transcribed spacer region (ITS), partial nuclear rDNA large subunit region (nrLSU), RNA polymerase II second largest subunit ( RPB 2), partial translation elongation factor 1-alpha ( TEF 1- α ), and beta-tubulin gene ( TUB ) also revealed that positions of A. kalasinensis and A. ravicrocina are well-supported within A. section Amanita , but form distinct lineages and do not show any close relationship with any species. The detailed morphological features, line-drawing illustration, and comparison with morphological similar taxa are provided.

During the period of macrofungal investigation in the rainy season of 2020, nine interesting specimens were collected from deciduous forests dominated by Dipterocarpus and Shorea species in northern and northeastern Thailand (Chiang Rai, Kalasin, and Sakon Nakhon Provinces, Thailand). Morphological examination and molecular analyses indicated that these collections herein reported represent two species new to science.

Morphological Study
The following information was recorded at the collecting sites: geographic coordinates, forest type, substrate type, and field photographs. Small pieces of tissue from the cap and/or stipe were taken and dried by silica gel to prepare for the molecular material [13], and the remaining specimens were dried at 35-45 °C for at least twelve hours to prepare for the morphological material and later were deposited at the Herbarium of Biology Department (CMUB) and the Herbarium of Sustainable Development of Biological Resources (SDBR), Faculty of Science, Chiang Mai University, Thailand.
Macroscopic characters were described based on field notes and field images. Color codes and names were recorded according to Kornerup and Wanscher [14]. Marginal striations on the pileus were expressed as a proportion of the ratio of striation length to the radius of the pileus (nR). Microscopic features were observed from dried specimens mounted in distilled water, 5% aqueous KOH (w/v), 1% Congo red (w/v), and Melzer's reagent under a Leica DM500 microscope to depict all tissues [2,3]. Sections of the pileipellis were cut radial-perpendicularly, and halfway between the center and margin of the pileus, and sections of the stipitipellis were taken from the center of the stipe, along the middle part along the longitudinal axis. For the description of basidiospores, the term [n/m/p] represents that n basidiospores were measured from m basidiomata of p collections. Dimensions for basidiospores are given as (a-) b-c (-d), in which 'b-c' represents a minimum of 90% of the measured and extreme values 'a' and 'd' are given in parentheses whenever necessary. Q denotes the ratio of length divided by width of the basidiospore in the side view, Qm denotes the average Q of n measured basidiospores, and SD is their standard deviation. The results are presented as Q = Qm ± SD. Basidiomata size and spores shape are defined according to Bas [15].

Phylogenetic Analyses
Detailed information about the sequences retrieved from GenBank and the sequences newly generated in this study was analyzed ( Table 1). Sequences of five gene regions were aligned with MAFFT v.7 [22] using the G-INS-i iterative refinement algorithm, and then checked visually and manually optimized using BioEdit v.7.0.9 [23]. Gblocks v. 0.91b [24] was used to check and exclude the ambiguously aligned regions for ITS, based on two options "Allow smaller final blocks" and "Allow gap positions within the final blocks". Maximum likelihood analyses were carried out for each single gene dataset using the same settings used for concatenated analysis to test potential conflicts among the five genes. Sequence Matrix v.100.0 was applied to combine the five gene fragments for further phylogenetic analysis and the concatenated dataset was deposited in TreeBASE under the number of 29155. Phylogenetic tree inference was performed using both Bayesian inference (BI) and maximum likelihood (ML), as detailed in Dissanayake et al. [25]. The best-fit model of nucleotide substitution was determined for each single gene dataset using MrModeltest v. 2.3 [26] following the default parameters.  Newly generated sequences in this study are in black bold, while holotypes are marked with "T".
The ML analysis was performed at the CIPRES web portal [27] using RAxML v.8.2.12 as part of the "RAxML-HPC BlackBox" tool [28] with default settings, except the "Estimate proportion of invariable sites (GTRGAMMA+I)" was set to be "yes" for both single-gene and combined gene analyses. Phylogenetic inference was first performed on each single-gene alignment, and as there was no evident conflict (with ML bootstrap support ≥75%), then multiple-gene alignments and trees were built. The Bayesian analysis was performed using MrBayes v.3.1.2 [29]. Posterior probabilities [30] were determined by Markov chain Monte Carlo sampling (MCMC) [31] in MrBayes v.3.1.2. Six simultaneous Markov chains were run from random trees for 1 million generations and trees were sampled every 100th generation (the critical value for the topological convergence diagnostic is 0.01). The first 25% of trees were discarded and the remaining trees were used for calculating posterior probabilities in the majority rule consensus tree. The phylogenetic tree was visualized with FigTree v.1.4.4 [32].

Phylogenetic Analyses
The best-fit models for the five genes were as follows: general time reversible + proportion of invariable sites + gamma distribution (GTR + I + G) for ITS, nrLSU, and TEF1-α; Hasegawa-Kishino-Yano (HKY) + I + G for RPB2; and Kimura 2-parameter (K80) + G for TUB. The concatenated dataset was partitioned into five parts by sequence region. The model HKY + I + G and K80 + G could not be implemented in RAxML, thus the GTR + I + G model, which included all parameters of the selected model, was used instead.
The multi-gene dataset comprised 212 sequences, including 33 newly generated and 179 retrieved from GenBank. Bayesian and RAxML analysis of the combined dataset resulted in phylogenetic reconstructions with largely similar topologies, thus the result of maximum likelihood (RAxML) tree is shown in Figure 1. In our phylogenetic results, five collections representing Amanita kalasinensis and four collections representing A. ravicrocina, respectively, formed a monophyletic lineage from other extant species with credible support values, which could be recognized as two new species. RAxML tree based on a combined of ITS + nrLSU + RPB2 + TEF1-α + TUB dataset. Bootstrap values (BS) for ML ≥50% and posterior probabilities (PP) for BI ≥0.95 are placed above or below the branches, respectively. Newly generated sequences are indicated in red and sequences from type material are marked with (T). The tree is rooted with Amanita yuaniana (HKAS58807 and HKAS68662) and A. caesareoides (HKAS92009 and HKAS92017).

Discussion
As mentioned above, thirteen species in section Amanita have been described in Thailand. Although most of these species are based on both of morphologic and phylogenetic evidences, four species, namely, A. aff. mira, A. obsita, A. siamensis, and A. subglobosa, have only morphologic data [6], which need to be supplemented with more molecular data to confirm their taxonomic status. Our study, along with other previous studies, allowed to discover the high diversity of Amanita species in Thailand [6][7][8][9][10][11][12], indicating that more taxa remain to be discovered and documented.
In our multi-gene phylogenetic analyses, both Amanita kalasinensis and A. ravicrocina formed a well-supported (BS = 100%, PP = 1.0) monotypic clade. However, the position of A. ravicrocina is not stable. Despite that many potential causes could lead to these unstable topologies, the primary reason might be that A. ravicrocina has significant differences from any other existing Amanita species in each single gene (the highest similarity and its query cover of initial BLAST searches results in GenBank for ITS and nrLSU of A. ravicrocina samples were from following: A. frostiana-RET 588-6 (KP313583) with 89.23% similarity and 99% query cover for ITS, and A. altipes-BZ2013-42 (MH716040) with 97.58% similarity and 100% query cover for nrLSU).
The erratic positions of Amanita ravicrocina and its distinct sequences provide indirect evidence that there are probably some new taxa related to A. ravicrocina that remain to be discovered. Therefore, further exploration of Amanita diversity is critical, which could reveal more members in this section.