A taxonomic reassessment of the genus Balsamia from China

Abstract Molecular analysis of the genus Balsamia was conducted with ITS and 28S sequences available, including newly gained sequences from Chinese specimens. Combined with the morphological examinations, a new hypogeous species, Balsamia lishanensis was described and illustrated from North China, which is morphologically characterized by reddish brown ascomata covered with fine warts, the whitish gleba with numerous small chambers, 3–5 layers peridium with reddish brown polygonal cells and the smooth and regular ellipsoid ascospores with one large oil drop. Two species previously described as Barssia were transferred to Balsamia. Balsamia platyspora was confirmed to be in existence in China based on newly collected specimen. A key to the Chinese Balsamia species was provided.

Recently, two new species of the genus Barssia have been described from China (Xu et al. 2018), their taxonomic position, however, needs to be reassessed because Hansen et al. (2019) synonymized Barssia under Balsamia based on their phylogenetic analysis from three loci (28S, RPB2,  and morphological studies. More recently, an un-described Balsamia species is recognized when we check the specimens newly collected from north China. In this paper, both the molecular analyses and morphological examinations are conducted for the Chinese samples,, and our aims are:1) to illustrate the position of Chinese Balsamia species based on ITS and 28S sequences newly obtained from Chinese Balsamia collections with distinct features in this study as well as recently published and used sequences ; 2) to give a detailed characterization of a new species based on morphological features and phylogenetic evidences.

Morphological studies
Collections were obtained and photographed in the field from Shanxi regions in China, and they were dried and deposited in BJTC (Herbarium, Biology Department, Capital Normal University). One specimen was studied from HMAS (Herbarium Mycologicum Academiae Sinicae, Institute of Microbiology, Chinese Academy of Sciences). Macroscopic characters were recorded from fresh specimens. Microscopic characters were observed in thin sections of dry specimens mounted in 3% KOH, Melzer's reagent (Dring 1971) or 0.1% (w/v) cotton blue in lactic acid. Thirty mature ascospores were measured, and the symbol Q is used to indicate length/width ratios of ascospores in side view.

DNA extraction, PCR amplification and DNA sequencing
Herbarium specimens were crushed by shaking for 30 s at 30 Hz 2-4 times (Mixer Mill MM 301, Retsch, Haan, Germany) in a 1.5 ml tube together with one 3 mm diam. tungsten carbide ball, and total genomic DNA was extracted using the modified CTAB method (Gardes and Bruns 1993). The internal transcribed spacer (ITS) region of nuclear ribosomal DNA (nrDNA) was amplified using primers ITS1f/ITS4 (White et al. 1990;Gardes and Bruns 1993). The 28S large subunit nrDNA (nrLSU) region was amplified using primers LR0R/LR5 (Vilgalys and Hester 1990). PCRs were performed in a volume of 50 μl consisted of 4 μl of DNA template; 2 μl of (10 μM) per primer; 25 μl 2× Master Mix (Tiangen Biotech Co., Beijing). The procedure for PCR reaction was: an initial denaturation at 94 °C for 3 min; followed by 35 cycles at 94 °C for 30 s, 55 °C for 45 s, 72 °C for 1 min; and a final extension at 72 °C for 10 min. The PCR products were sent to Beijing Zhongkexilin Biotechnology Co. Ltd. (Beijing, China) for purifying, sequencing and editing. Validated sequences are stored in the NCBI database (http://www.ncbi.nlm.nih.gov/) under the accession numbers provided (Table1). The other sequences used in the molecular phylogenetic analysis were downloaded from the NCBI database (Suppl. material 1).

Phylogenetic analyses
Two datasets, ITS and 28S, were compiled to identify Balsamia species and investigate relationships among species. The taxa Tuber anniae and T. bellisporum were selected as outgroups. The ITS and 28S sequences were aligned using the MAFFT v.7.110 online program under default parameters (Katoh and Standley 2013), and manually adjusted to allow maximum sequence similarity in Se-Al version.2.03a. (Rambaut 2000). Ambiguously aligned regions and gaps in alignment were excluded by Se-Al version.2.03a. (Rambaut 2000) before the phylogenetic analysis. Alignments were submitted to Tree-BASE under accession number S25937. We conducted maximum likelihood (ML), most parsimonious (MP) and Bayesian inference (BI) analyses on the two datasets.
Maximum likelihood (ML) analysis of the dataset was carried out using RAxML 8.0.14 (Stamatakis 2014) and the GTRGAMMA substitution model with parameters unlinked. The ML bootstrap replicates (1000) were computed in RAxML using a rapid bootstrap analysis and search for the best-scoring ML tree. The ML trees were viewed with TreeView32 (Page 2001). Clades with bootstrap support (MLBS) ≥ 70% were considered as significant-supported (Hillis and Bull 1993).
A most parsimonious (MP) analysis was constructed with PAUP* 4.0b10. (Swofford 2002). The bootstrap values were generated using the following settings: 1000 replicate searches on all parsimoniously informative characters using 100 random sequence addition replications and TBR (tree-bisection reconnection) branch swapping algorithms in PAUP*. Tree statistics (TL), consistency index (CI), retention index (RI) and homoplasy index (HI) were also calculated. Tree was viewed with TreeView32 (Page 2001). Clades with bootstrap support (MPBS) ≥ 70% were considered to be significant (Hillis and Bull 1993). Bayesian inference (BI) analyses was performed with MrBayes v3.1.2 (Ronquist and Huelsenbeck 2003) based on the best substitution models determined by MrModeltest 2.3 (Nylander 2004), which were GTR+I+G for the ITS dataset and SYM+I+G for the 28S dataset. Two independent runs of four chains were conducted for 4 000 000 for ITS and 2 000 000 for 28S datasets Markov Chain Monte Carlo generations using the default settings and sampled every 100 generations. The temperature value was lowered to 0.20, burn-in was set to 0.25, and the run was automatically stopped as soon as the average standard deviation of split frequencies reached below 0.01. A 50% majority-rule consensus tree was constructed and visualized with TreeView32 (Page 2001). Clades with Bayesian posterior probability (BPP) ≥ 0.95 were considered as significantly supported (Alfaro et al. 2003).

Phylogenetic analysis
For ITS dataset, we comprehensively collected the ITS sequences of Balsamia and the fungi previously described as Barssia, and sequences that are high similarity to Balsamia. For 28S dataset, we collected all sequences of Balsamia and the fungi previously described as Barssia, and representative sequence of other genera of Helvellaceae. Sequences of each locus were aligned and analyzed separately.
The 28S dataset contained 72 sequences (9 were newly gained in this study), and 4 from the outgroup Tuber anniae and T. bellisporum. The dataset had an aligned length of 886 characters, of which 578 were constant, 308 were variable, and 278 of these variable sites were informative. The maximum parsimony analysis resulted in one most parsimonious tree with a length (TL) of 842 steps, consistency index (CI) of 0.570, retention index (RI) of 0.896, homoplasy index (HI) of 0.430. MP, ML and BI analyses yielded similar tree topologies, and only the tree inferred from the MP analysis is shown (Fig.  1). The 28S sequences of Balsamia were grouped into a distinct clade with high supports (MPBS = 100%, MLBS = 100%, BPP = 1.00). The Chinese materials were well clustered in the Balsamia clade (Fig. 1), including the sequences of the fungi previously described as Barrsia guozigouensis L. Fan & Y.Y. Xu and Barssia luyashanensis L. Fan & Y.Y. Xu (Xu et al. 2018). Three distinct branches with strong supports can be recognized from Chinese collections, respectively representing Balsamia guozigouensis, Balsamia luyashanensis, and a new species Balsamia lishanensis proposed in this study. In addition, the Chinese sequence from BJTC FAN557 grouped together with a reliably identified sequence (MK100252) of B. platyspora (Hansen et al. 2019) with strong support value (MPBS = 99%, MLBS = 99%, BPP = 1.00), and they shared 99.83% 28S sequences similarity, indicating the Chinese specimen BJTC FAN557 was B. platyspora.
The ITS dataset contained 108 sequences (9 were newly gained in this study), and 4 from the outgroup T. anniae and T. bellisporum. The dataset had an aligned length of 1056 characters, of which 310 were constant, 745 were variable, and 622 of these variable sites were informative. The maximum parsimony analysis resulted in one most parsimonious tree with a length (TL) of 2220 steps, consistency index (CI) of 0.580, retention index (RI) of 0.900, homoplasy index (HI) of 0.420. MP, ML and BI analyses yielded similar tree topologies, and only the tree inferred from the Bayesian analysis is shown (Fig. 2). The ITS sequences of Balsamia were grouped into a distinct clade with high supports (MPBS = 100%, MLBS = 100%, BPP = 1.00), and the sequences from the Chinese collection unambiguously clustered in the Balsamia clade, including the sequences of the fungi previously described as Barrsia guozigouensis and Barssia luyashanensis (Xu et al. 2018) (Fig. 2). The sequences of all Chinese collections excepting specimen BJTC FAN557 were grouped into three independent clades with strong supports (Fig. 2), respectively representing Balsamia guozigouensis, Balsamia luyashanensis and a new species Balsamia lishanensis proposed in this study. The sequence of BJTC FAN557, which was identified as B. platyspora by morphology and 28S phylogeny (Fig. 1) in this study, formed a strong support clade together with nine European sequences isolated from ascomata of Balsamia platyspora or ectomycorrhizal root tips of Balsamia. These ten samples showed high sequences similarity so the clade was considered as representing B. platyspora.