Two new Russula species (fungi) from dry dipterocarp forest in Thailand suggest niche specialization to this habitat type

Dry dipterocarp forests are among the most common habitat types in Thailand. Russulaceae are known as common ectomycorrhizal symbionts of Dipterocarpaceae trees in this type of habitat. The present study aims to identify collections of Russula subsection Amoeninae Buyck from dry dipterocarp forests in Thailand. A multi-locus phylogenetic analysis placed Thai Amoeninae collections in two novel lineages, and they are described here as R. bellissima sp. nov. and R. luteonana sp. nov. The closest identified relatives of both species were sequestrate species suggesting that they may belong to drought-adapted lineages. An analysis of publicly available ITS sequences in R. subsect. Amoeninae did not confirm evidence of any of the new species occurring in other Asian regions, indicating that dry dipterocarp forests might harbor a novel community of ectomycorrhizal fungi. Macromorphological characters are variable and are not totally reliable for distinguishing the new species from other previously described Asian Amoeninae species. Both new species are defined by a combination of differentiated micromorphological characteristics in spore ornamentation, hymenial cystidia and hyphal terminations in the pileipellis. The new Amoeninae species may correspond to some Russula species collected for consumption in Thailand, and the detailed description of the new species can be used for better identification of edible species and food safety in the region.


Results
Phylogenetic analyses. A total of 21 sequences were newly generated and deposited in GenBank (Supplementary Table 1). The concatenated sequence alignment of the three loci comprised 100 sequences (38 for ITS, 30 for rpb2 and 32 for mtSSU) from 43 collections (Supplementary Table 1). The alignment was 2,004 characters long, including gaps. Multi-locus trees generated from ML and BI analyses showed similar topologies without any supported topological conflict. The multi-locus phylogeny ( Fig. 1) confirmed placement of all Thai collections within the well-supported R. subsect. Amoeninae (ML = 99, BI = 1.0). Five collections from northeastern Thailand and two collections from northern Thailand form two strongly supported clades and are described below as the new species R. bellissima sp. nov. and R. luteonana sp. nov. The new species are not resolved as sister. The first species, R. bellissima, is strongly supported as sister to a clade of Australian sequestrate species that includes R. variispora T. Lebel and an undescribed Russula sp. labeled as Macowanites sp. The Indian species R. intervenosa S. Paloi, A.K. Dutta & K. Acharya is placed as sister to them with bootstrap support of 77. The second species, R. luteonana, is placed with moderate support as sister to the sequestrate European species R. andaluciana T.F. Elliott & Trappe.
The ITS tree (Fig. 2) shows a similar topology and relationships for the studied specimens. In addition, R. intervenosa received good support (ML = 84, BI = 0.99) as sister to the clade of R. bellissima and R. variispora. Five additional ITS sequences that are grouped with strong support within R. bellissima species clade were recovered, three from Thailand, one from Laos, and one from Singapore. We did not recover any other Amoeninae ITS sequences from Thailand. Etymology 'bellus' = latin for beautiful, pretty, lovely; 'bellissima' = the most beautiful. Resembling the species Russula bella which is also belonging to Russula subsection Amoeninae.

Taxonomy. Russula bellissima
Diagnosis Pileus small to medium-sized; cuticle dry, smooth, matt and pruinose, red; stipe white or with a red flush; spore ornamentation of moderately distant to dense amyloid spines or warts, frequently fused into short crests or even long wings; suprahilar spot inamyloid; hymenial cystidia and pileocystidia absent.
Pileus (Fig. 3) small to medium sized, 10-50 mm diam., young hemispherical or convex, becoming plane and depressed at the centre; margin first even, when old distinctly tuberculate-striate up to 10 mm from the margin, often radially cracking; cuticle hardly peeling, radially disrupted into small patches, pruinose when young, later dry, smooth, matt and pruinose in the centre, colour near the margin when young varnish red (9C8), later red to  Note Russula bellissima is a small species with a bright red pileus and pink colour on the stipe. This colour is distinctive and resembles North American R. mariae, Indian R. intervenosa and Asian R. bella. It is very unlikely that the distribution of any European or North American species is overlapping with the Thai species. However, little is known about the distributional ranges and the ecological niches of other Asian Russula species. Therefore discussing the morphological distinguishing characters between Asian species and R. bellissima is more relevant. Russula bellissima is not closely related to R. bella and it differs from this species by larger spores with a more prominent spore ornamentation, absence of hymenial cystidia on lamellae sides, and subterminally short, ellipsoid cells in the suprapellis arranged in unbranched chains of up to four 7 . The Thai species resembles and is closely related to the Indian R. intervenosa, but it has a more prominent spore ornamentation, hymenial cystidia (on lamellae sides) are absent, and hyphal terminations in the pileipellis are wider 22      www.nature.com/scientificreports/ Diagnosis Pileus medium-sized, dry, usually yellow, spores with subreticulate amyloid ornamentation and inamyloid suprahilar spot, hymenial cystidia on lamellae sides large, lamellae edges with combination of subulate, clavate and pyriform marginal cells.
Pileus (Fig. 7) medium-sized, 28-53 mm diam., plano-convex with depressed centre, infundibuliform when mature; margin striated and radially cracking in dry condition; cuticle dry, peeling to almost ½ of radius, smooth to minutely wrinkled, dull in dry condition, color very variable, some collections pale cream and with darker pale brownish-yellow centre, other yellow brownish and with darker orange-brown centre, sometimes also bright red-brown and with discolored centre, always with rusty-brown spots especially when near the centre. Lamellae: 3-5 mm deep, moderately distant, intervenose, forking near the stipe, white to cream, edges even, concolorous. Stipe: 26-40 × 6-9 mm, cylindrical or narrowed at the base, surface dry, longitudinally wrinkled, white, turning brown when bruised. Context: 2-4 mm in at the half pileus radius, soft, solid, becoming partially hollow when mature, white, unchanging when cut. Taste mild; odour rather strong, fishy. Spore print: not observed.
Spores  Note Russula luteonana is extremely variable but based on the pileus colour we can eliminate some species with purple or green tints. If we combine this with the white or nearly so stipe, it can only be confused with either R. bella (if it has redder colouration) or R. orientipurpurea. The unique character of R. luteonana is the large cystidia which range 14-21 μm in width and are often also obtuse. While R. bella has many microscopic characters that distinguish it from this proposed species (e.g. smaller spores, narrower hymenial cystidia), R. orientipurpurea resembles, in many aspects, the Thai species (i.e. relatively large spores, obtuse and relatively wide hymenial cystidia on the lamellae sides, and usually only one unbranched short cell below the terminal cell of hyphae in pileipellis). Distinguishing features of these two species are the more prominent spore ornamentations and the often acute hymenial cystidia of R. luteonana 7 .

Discussion
Members of Russula subsect. Amoeninae were identified in the field based on the typical dryness of the pileus, the velvety-granulose aspect of pileus cuticle, and white lamellae. Later each collection was checked under the microscope for the absence of cystidia in the pileipellis. In the field, both new species are distinct from many other species, because of the pileus with red, yellow or brown tints and completely without purple or greenish colours. The stipes of both species are white or only partly flushed by pink, and never with purple or deep red on more than half of their surface. There is a high degree of infraspecific variability of pileus colours in Amoeninae and therefore field observations need sufficient sampling or further verifications under the microscope to identify them accurately to species 7,23 .
Microscopically, all Amoeninae representatives have a distinctive spore ornamentation composed of relatively prominent ridges and crests connected also by line connections to form a subreticulate to reticulate    www.nature.com/scientificreports/ terminations in the pileipellis. While hymenial cystidia on lamellae sides are completely absent in R. bellissima, they are very large and often base-inflated (lageniform or subulate) in R. luteonana. Thus these microscopic structures distinguish both species from all known Asian Amoeninae. Hyphal terminations in the pileipellis are very different between the two newly described species and they prove importance of specific characters used already for European species, e.g. size of subterminal cells and their number 7,23 . Russula bellissima has relatively short terminal cells measuring up to 50 µm and usually 2-3 short, subglobose subterminal cells that make it similar to R. intervenosa from India 24 . Russula luteonana has terminal cells usually longer than 50 µm and zero to one cylindrical subterminal cells which make it similar to R. orientipurpurea 7 .
The first and only phylogenetically confirmed records of Amoeninae members in Thailand were published by Wisitrassameewong et al. 7 , and they were represented by two samples with GenBank accession numbers AB459514 and AB854679. In our ITS analysis both these samples cluster within the R. bellisima clade together with three other samples from Thailand, Laos and Singapore. Our ITS tree covers nearly all previously published species of Amoeninae in Asia and all available sequence data of this group from public databases. The only Asian species without available DNA that are not included in our analyses are R. mukteswarica K. Das, S.L. Mill., J.R. Sharma & R.P. Bhatt and R. punicea W.F. Chiu. Both of these species are morphologically very different from species described here. Russula mukteswarica has a purple and green coloured pileus and low spore ornamentation (only 0.75 μm), and the R. punicea has small spores (up to 7 μm) and small hymenial cystidia 7 . In conclusion, our study does not confirm any species of Amoeninae previously described or recorded from India, Korea, Japan or China occurring in Thailand. We were unable to locate any sequence record for R. luteonana in public sequence databases, but R. bellisima is represented by two additional sequences from Thailand and two more from Laos and Singapore. The Thailand sequences originated from a dry dipterocarp forest (GenBank accession number AB854679) and from an evergreen dipterocarp forest with annual precipitation of approximately 1030 mms (GenBank accession number AB459514) 27,28 . The Laos record (UNITE accession number UDB033872) is from an urban area of Vientiane city on the northeastern Thai border and the Singapore record (GenBank accession number MZ519838) is also from Singapore Botanical Garden collected with Shorea leprosula. Because all collections of our new species are located in Mainland Southeast Asia, this result suggests that this area's endemic Russula diversity developed under specific climate, geomorphology and available ECM host range, such as already suggested for southwestern Himalayas 29 . In this respect it is worth to mention that sister to both our new species collected in dry dipterocarp forests are sequestrate species (Figs. 1, 2) 30 and both species can be members of two independent seasonal drought tolerant lineages within Amoeninae. Despite our effort, we did not collect any sequestrate species of Russula from dry dipterocarp forests, but during our field excursions, members of our expedition collected sequestrate Lactarius and Entoloma 31,32 .
The fungal flora of dipterocarp forests is still very poorly known, but there is a multitude of evidence that roots of Dipterocarpaceae trees are colonized by ectomycorrhizal fungi 33 . Based on sequencing of root tips, Dell et al. 4 estimated that the fungal richness of dry dipterocarp forests in Thailand is comparable to other tropical rain forest sites, but the phylogenetic community structure has elements of both tropical and temperate ecosystems 4 . They confirmed insufficient knowledge of fungal species diversity of this habitat, only 9 of the 69 species matched with sequences from public databases at the 97% sequence similarity cut-off and only four of these taxa were identified to species based on the available reference sequences and identifications. In terms of species richness, Russulaceae lineage was the richest and a Russula cf. pectinata was the most frequent molecular operational taxonomic unit retrieved from all dataset. Phosri et al. 3 reported that the diversity of ECM fungi in dipterocarp forests in Northern Thailand is the second most abundant after Fagaceae forest and the Russula-Lactarius lineage was among the dominant fungal groups. In lowland dry dipterocarp forest of Malaysia, Russula is reported as one the most abundant genera found from bulk soil and root tip samples 34  Amoeninae species are not the first Russula members described from dry dipterocarp forests, also R. aurantiopectinata was collected and is reported with Dipterocarpus tuberculatus from this type of habitat 12 . However, our study is the first discussing possibly unique ectomycorrhizal communities and species adapted to this kind of ecological and climatic conditions.
To understand distributional ranges of Russula species collected in dry dipterocarp forests, we need to understand their niche limits in term of habitat specificity, ecological adaptations to certain climate and soil conditions, and their life strategy in general. Based on sequencing of Dipterocarpaceae root samples, Sato et al. 43 suggested that specific lineages of closely related dipterocarp taxa are associated with some specific ECM forming Basidiomycota OTUs and have uniquely characteristic community structure. However, seedling experiments of Dipterocarpaceae species with contrasting soil specializations proved that there is little host specificity and soil environment was the primary determinant of ectomycorrhizal diversity 44 . This result is in contrast with the study of Essene et al. 34 , reporting that majority of the taxa detected in root tip samples had not only restricted preference of soil type (either sand or clay soil), but also were associated with only one species of Shorea. Two different Shorea tree species on the same soil type were colonized by different ECM fungal communities. These results suggested that these biotic and abiotic factors might have played a role in structuring ECM fungal communities in Dipterocarpaceae forests.
In Thailand, wild edible mushrooms are one of the three main non-timber forest products collected during the rainy season 45 . Mushroom foraging is considered to be a recreational activity, seasonal food source, or an option for additional household income. Several species of Russula are considered choice for consumption in rural regions, particularly in northern and northeastern Thailand and they are a commonly eaten mushrooms in the region due to their high yields in Thai forests. Various morphological types of Russula are frequently  47 . Therefore, it is likely that the two species described in this study are edible. However, the correct identification of well documented edible species is important because a number of Russula species are reported as poisonous 48,49 or can cause gastrointestinal irritations, and the latter ones include Russula collections identified as R. cf. rosacea 50 .

Materials and methods
Studied collections and morphological study. Our  To describe all microscopic features, we used the description template and terminology of Adamčík et al. 6 . Microscopic characters were studied on dried material mainly in Congo Red 53 , except for spore morphology, which was observed in Melzer reagent. Chemical tests were applied with Cresyl Blue 54 , sulfovanillin 55 , and carbolfuchsin 56 to observe colour changes, incrustations, and cystidia contents. Line drawings and measurements of all microscopic characters were done using Nikon Eclipse Ni (Nikon Instrument Inc., Japan), with the aid of the software NIS-element BR 5.02.03 at a projection scale of 2000 ×. Scanning electron microscopy (SEM) images were done using Zeiss Auriga crossbeam microscope at a 6330 × magnification and FESEM Hitachi SU5000 at a 5000 × magnification. Ranges of measurements were estimated as average plus/minus two times of standard deviation; in parenthesis are minimum and maximum values. Q value corresponds to length/width ratio of spores. The central value in italics represents an average. The measurements of all microscopic characters of a species were observed for three collections if available with 30 measurements per collection.
DNA extraction, PCR amplification and sequencing. Genomic DNA was extracted from dried materials using E.Z.N.A. Forensic DNA kit (Omega Bio-Tek). The amplification of the internal transcribed spacer region (ITS) of the nuclear ribosomal DNA was done using primers ITS1-F 57 and ITS4 58 . The primers RPB2-6F and fRPB2-7cr 59 and the primers MS1 and MS2 58 were used to amplify the second largest subunit of RNA polymerase II (rpb2) and mitochondrial small subunit ribosomal DNA region (mtSSU), respectively. DNA sequencing of the successful PCR products using the same primers was performed by an ABI 3500 Genetic analyzer (Thermo Fisher Scientific) at the National Biobank of Thailand. Obtained sequences were checked and edited using FinchTV 1.4 (Geospiza, Inc.) and then assembled in MEGA X 60 .
Alignment and phylogenetic analyses. Phylogenetic analysis was based on three DNA regions: ITS nrDNA, mtSSU and rpb2. The DNA sequences of Amoeninae members used for the multi-loci analysis were downloaded from GenBank (https:// www. ncbi. nlm. nih. gov/ genba nk/) based on the previous study of Wisitrassameewong et al. 7 (Supplementary Table 1). Three other members of R. subg. Heterophyllidiae, R. aff. crustosa, R. grisea, and R. vesca, were chosen as an outgroup for this multi-locus analysis. For better estimation of overall diversity of Amoeninae, we downloaded available ITS Amoeninae sequences with special emphasis to Asian taxa. Our selection of ITS sequences covers species recognized by Wisitrassameewong et al. 7 and additionally also all Asian members with at least 90% BLAST similarity to any described Amoeninae member. We also searched the UNITE database (https:// unite. ut. ee/) for Asian samples matching Amoeninae with a 3% identity threshold. Based on previous phylogenetic studies, members of the other subsections of R. subg. Heterophyllidia were included in the ITS dataset and members of R. subsect. Virescentinae Singer was selected as the outgroup. All datasets were aligned using MAFFT 61 . We inferred phylogenies for each single-gene dataset using Maximum Likelihood (ML). ML analyses were performed using RAxML 8.2.12 62 . The parameters for RAxML analyses were the GTRGAMMA model and the rapid bootstrapping algorithm for 1,000 replicates. We compared the tree topology between different single-gene trees and examined for conflict at nodes with bootstrap support value (BS) above 70% and posterior probability (PP) higher than 0.90. A conflict can be considered as significant if two different relationships for the same set of taxa were observed among different single-gene phylograms. A concatenated dataset of ITS + rpb2 + mtSSU was constructed using BioEdit 7.2.5 63 . jModelTest 2.1.6 was used to estimate the best-fit model of substitution for each partition 64 . The models based on Bayesian information criterion (BIC) were selected as follows: K80 + G for rpb2pos1, rpb2pos2 and rpb2pos3, TIM1 + G for mtSSU, K80 + G for ITS1, TPM1 for 5.8 s rDNA and TrNef + G for ITS2 regions. Bayesian Inference (BI) analysis was executed using MrBayes 3.2.7a 65 . The analysis was executed for four independent runs with four chains each that was run for 10 million generations and sampled every 100th tree until the standard deviation of split frequency was less than 0.01. Before finalizing a consensus tree, the convergence and ESS values were checked using Tracer 1.6 66 . The burn-in of 10,000 for each run was used. All phylogenetic analyses were done using the CIPRES Science Gateway 67 .