Four new species in the Tremellafibulifera complex (Tremellales, Basidiomycota)

Abstract Samples of species close to Tremellafibulifera from China and Brazil are studied, and T.fibulifera is confirmed as a species complex including nine species. Five known species (T.cheejenii, T.fibulifera s.s., T. “neofibulifera”, T.lloydiae-candidae and T.olens) and four new species (T.australe, T.guangxiensis, T.latispora and T.subfibulifera) in the complex are recognized based on morphological characteristics, molecular evidence, and geographic distribution. Sequences of eight species of the complex were included in the phylogenetic analyses because T.olens lacks molecular data. The phylogenetic analyses were performed by a combined sequence dataset of the internal transcribed spacer (ITS) and the partial nuclear large subunit rDNA (nLSU), and a combined sequence dataset of the ITS, partial nLSU, the small subunit mitochondrial rRNA gene (mtSSU), the translation elongation factor 1-α (TEF1), the largest and second largest subunits of RNA polymerase II (RPB1 and RPB2). The eight species formed eight independent lineages with robust support in phylogenies based on both datasets. Illustrated description of the six species including Tremellafibulifera s.s., T. “neofibulifera” and four new species, and discussions with their related species, are provided. A table of the comparison of the important characteristics of nine species in the T.fibulifera complex and a key to the whitish species in Tremella s.s. are provided.

Although Tremella s.l. was separated into several genera due to its polyphyletism, it is still somewhat confusing because taxonomic positions of some Tremella species are uncertain in Tremellales, especially some species recently described from lichens (Ariyawansa et al. 2015;Malysheva et al. 2015;Millanes et al. 2015;Zamora et al. 2016Zamora et al. , 2018. These lichenicolous species were described as Tremella, but they were not clustered into Tremella s.s. in the phylogeny (Ariyawansa et al. 2015;Malysheva et al. 2015;Millanes et al. 2015;Zamora et al. 2016Zamora et al. , 2018. Recently, Zhao et al. (2019) described four new Tremella species based on the phylogenetic relationship of 19 species in Tremella s.s., and Li et al. (2020) published a new yeast species of Tremella s.s. based on multi-gene analysis.
In this study, samples of species morphologically similar to Tremella fibulifera characterized by cerebriform whitish basidioma and abundant clamp complexes from China and Brazil are studied. Based on morphology, geographic distribution and phylogenetic analyses T. fibulifera is confirmed as a species complex, which was previously mentioned by Bandoni and Oberwinkler (1983) and Malysheva et al. (2015), and nine species are involved in the complex including five known species (T. cheejenii, T. fibulifera s.s., T. "neofibulifera", T. lloydiae-candidae and T. olens) and four new species (T. australe, T. guangxiensis, T. latispora and T. subfibulifera) in the present study. The aim of this paper is to outline the T. fibulifera complex and describe two known species (T. fibulifera s.s., T. "neofibulifera") and the four new species based on our collections.

Sampling and morphological analysis
The studied specimens were collected from Rondônia and Pernambuco states in Brazil, Yunnan, Taiwan, Guangxi, Jilin Provinces in China. They are deposited at the herbaria of Beijing Forestry University (BJFC), Institute of Botânica in São Paulo (SP) and Universidade Federal de Pernambuco, Departamento de Micologia (URM). Macro-morphological illustrations refer to Chen (1998) and Zamora et al. (2017) and microscopic structures refer to Pippola and Kotiranta (2008) and Malysheva et al. (2015). Special color terms followed Petersen (1996). Handmade sections of dried basidioma were examined by a Nikon Eclipse 80i (Japan) microscope (magnification × 1000) after being mounted in 5% KOH for five minutes and treated with 1% Phloxine B (C 20 H 4 Br 4 Cl 2 K 2 O 5 ). Microscopic structures were photographed using a Nikon Digital Sight DS-L3 (Japan) or Leica ICC50 HD (Japan) camera. Microscopic structures were examined and measured in the mix solution of 5% KOH and 1% Phloxine B. To represent variation in the size of spores, 5% of measurements were excluded from each end of the range, and are given in parentheses. Stalks were excluded for basidia measurements and apices were excluded for basidiospores measurements. Length and width of at least 30 basidia and basidiospores from each specimen were measured to micrometers.
Abbreviations as follows: L = mean length (arithmetic average of all basidia or spores length), W = mean width (arithmetic average of all basidia or spores width), Q = L/W ratio for each specimen studied, n (a/b) = number of spores (a) measured from given number of specimens (b).

Molecular phylogeny
Dry specimens were used to extract DNA after pretreatment using TissuePrep (Jie Ling, China) by CTAB rapid plant genome extraction kit-DN14 (Aidlab Biotechnologies Co., Ltd, Beijing) or directly using the DNA easy Plant Mini Kit (Qiagen, China), according to the manufacturer's instructions with some modifications. The internal transcribed spacer regions (ITS), partial nuclear large subunit rDNA (nLSU), the translation elongation factor 1-α (TEF1), the largest and second largest subunits of RNA polymerase II (RPB1 and RPB2), the small subunit mitochondrial rRNA gene (mtSSU) sequences were amplified with primer pairs listed in the Table 1. All newly generated sequences were submitted to GenBank (Table 2).
Polymerase chain reaction (PCR) cycling schedule for ITS, mtSSU and TEF1 included an initial denaturation at 95 °C for 3 min, followed by 35 cycles at 95 °C for 40 s, 54-56 °C (ITS) and 56-58 °C (mtSSU, TEF1) for 45 s, 72 °C for 1 min, and a final extension at 72 °C for 10 min, for RPB1 and RPB2 included an initial denaturation at 95 °C for 3 min, followed by 9 cycles at 94 °C for 45 s or 1 min , 58 °C for 45 s or 60 °C for 1 min and 72 °C for 1.5 min, then followed by 35 cycles at 95 °C for 1 min, 53 °C or 55 °C for 45 s and 72 °C for 1 min, and a final extension of 72 °C for 10 min, for partial nLSU included an initial denaturation at 94 °C for 1 min, followed by 34 cycles at 94 °C for 30 s, 50-52 °C for 1 min, 72 °C for 1.5 min, and a final extension at 72 °C for 10 min. PCR products were purified at the Beijing Genomics Institute (BGI), China or at the Plataforma Tecnológica de Genômica e Expressão Gênica do Centro de Biociências (UFPE) with the same primers.
Newly generated sequences in this study were aligned with additional related sequences downloaded from GenBank (Table 2) using MAFFT 7.0 online service with the Q-INS-i strategy (Katoh et al. 2019, http://mafft.cbrc.jp/alignment/server/). Prior to phylogenetic analysis, ambiguous sequences at the start and the end were deleted and gaps were manually adjusted to optimize the alignment using the default parameters in BioEdit (Hall 1999). Those positions deemed ambiguous to align were excluded manually. Multi-genes were concatenated as a combined file by Mesquite version 3.2. (Maddison and Maddison 2017). Phylogenetic analyses were applied to the ITS + partial nLSU dataset and the combined ITS+partial nLSU+mtSSU+TEF1+RPB1+RPB2 dataset. Sequences of Cryptococcus depauperatus (Petch) Boekhoutet et al. were used as outgroup, which referred to Malysheva et al. (2015). The final concatenated sequence alignments were deposited in TreeBase (https://treebase.org/treebase-web/home.html; submission ID 28280 for ITS + partial nLSU    (Miller et al. 2012) using tool of RAxML-HPC BlackBox 8.2.6, PAUP on XSEDE (4.a165) and Mrbayes on XSEDE 3.2.6 respectively. All characters were equally weighted, and gaps were treated as missing data. Trees were inferred using heuristic search option with TBR branch swapping and 1000 random sequence additions. MrModeltest 2.3 (Posada and Crandall 1998;Nylander 2004) was used to determine the best-fit evolution model for both datasets for Bayesian analyses using MrBayes3.1.2 (Ronquist and Huelsenbeck 2003). Four Markov chains were run for two runs from random starting trees for 3 million generations (ITS + nLSU) and for 5 million generations (ITS + partial nLSU + mtSSU + TEF1 + RPB1 + RPB2) until the split deviation frequency value < 0.01, and trees were sampled every 100 generation. The first quarter generations were discarded as burn-in. A majority rule consensus tree of all remaining trees was calculated.
Phylogenetic trees were viewed by FigTree v. 1.4.2 (Rambaut 2012) and edited by Adobe Illustrator CS6 (Guide 2012). Branches that received bootstrap support for Maximum parsimony (BP), Maximum likelihood (BS) and Bayesian posterior probabilities (BPP) greater than or equal to 50% (BP/BS) and 0.95 (BPP) were considered as significantly supported, respectively.

Phylogeny
The ITS + partial nLSU dataset included 50 fungal specimens representing 27 species. The dataset has an aligned length of 2282 total characters including gaps, of which 1777 characters are constant, 128 variable characters are parsimony-uninformative, and 377 are parsimony-informative. MP analysis yielded four equally parsimonious trees (TL = 1394, CI = 0.529, RI = 0.792, RC = 0.419, HI = 0.471). The best model The samples used in this study are in bold.
for the ITS + partial nLSU dataset estimated and applied in the BI analysis was GTR. BI and ML analyses generated similar topologies as MP analysis, with an average standard deviation of split frequencies = 0.002648 (BI). The best tree obtained from the ML analysis with bootstrap values for BP, BS and BPP is shown in Fig. 1. The phylogeny shows that eight species are clustered into the T. fibulifera complex and four new species form four independent lineages with robust support. The combined dataset of ITS + partial nLSU + mtSSU + TEF1 + RPB1 + RPB2 has an aligned length of 5113 total characters including gaps, of which 3332 characters are constant, 383 variable characters are parsimony-uninformative, and 1398 are parsimony-informative. MP analysis yielded two equally parsimonious trees (TL = 4519, CI = 0.607, RI = 0.730, RC = 0.443, HI = 0.393). The best model for the combined dataset estimated and applied in the BI analysis was GTR+I+G. BI analysis generated similar topology to MP and ML analysis, with an average standard deviation of split frequencies = 0.008566 (BI). The best tree obtained from the ML analysis with bootstrap values for BP, BS and BPP is shown in Fig. 2. The phylogeny results in similar topology to the phylogeny based on the ITS + partial nLSU sequences, which supports four new species separated from T. fibulifera s.s. and T. "neofibulifera". Basidioma. Sessile, when fresh gelatinous, pale whitish, lobed to irregularly cerebriform, becoming pale yellowish when dry, 0.5-2.5 cm in diameter, broadly attached to substratum.
Notes. Tremella guangxiensis is closely related T. "neofibulifera" in our phylogenies (Figs 1, 2). The most distinctive characteristic of the species is branched hyphidia and umbelliform conidiophores, but T. "neofibulifera" has parallel hyphidia and lacks of conidia. In addition, T. guangxiensis are different from T. australe and T. "neofibulifera" by 6.35% and 5.09% sequence differences in the ITS sequences and 3.39% and 1.97% in the partial nLSU sequences respectively. Etymology. Refers to the species having wide basidiospores. Basidioma. Sessile, when fresh soft gelatinous, creamy-white to lvory, translucent, pustulate to irregularly cerebriform, with thick and undulate lobes, up to 4.0 cm long, 2.0 cm broad and 1.0 cm high from base, distinctly shrinking into a film and becoming whitish to pale yellow when dry, broadly attached to substratum.
Notes. Phylogenetically, Tremella latispora formed a distinct lineage closely related to T. cheejenii (Figs 1, 2). Morphologically, the species has significantly larger basidia and basidiospores than T. cheejenii or other similar species (Table 3), and it has globose to subglobose basidiospores rather than more or less ellipsoid basidiospores in other species. And T. latispora are different from T. cheejenii and T. fibulifera s.s. by 4.63% and 5.09% sequence differences in the ITS sequences and 3.39% and 2.95% in the partial nLSU sequences respectively.

Discussion
Tremella fibulifera was originally described from Blumenau of Brazil (Möller 1895); later two similar species, T. olens and T. neofibulifera, were respectively described from Tasmania of Australia and Simotuke of Japan (Hooker 1860;Kobayasi 1939). The Russian Far East specimen LE303445 was identified as T. fibulifera by Malysheva et al. (2015). Our results demonstrated the Northeastern Chinese specimens and Russian Far East specimen formed an independent lineage, and this lineage is distantly related to the lineage formed by two Brazilian specimens, SP 211759 and Alvarenga 471 (Figs 1, 2). The location of SP 211759 is near to the type locality of T. fibulifera. So, we treat SP 211759 and Alvarenga 471 as representatives of T. fibulifera s.s. Molecular data are not available from type or type locality specimens of T. neofibulifera. Neither is its type re-examined, but the Northeastern Chinese specimens have more or less similar morphology as the description of T. neofibulifera, so we temporarily treat Northeast Chinese specimens and Russian Far East specimen as T. "neofibulifera".
The Southern Chinese specimen GX20172028 was also identified as Tremella fibulifera by Zhao et al. (2019), but it clustered with another Southern Chinese specimen Wu 3 into a distinct lineage which is closely related to T. "neofibulifera" (Figs 1, 2). T. guangxiensis is different from T. "neofibulifera" by 5.09% sequence differences in the ITS sequences and 1.97% in the partial nLSU sequences respectively. In addition, the Southern Chinese specimens have translucent basidioma, branched hyphidia and umbelliform conidiophores, and they are readily distinguished from T. "neofibulifera". So, these two specimens are identified as a new species T. guangxiensis.
Seven species, Tremella fibulifera, T. olens, T. "neofibulifera", T. guangxiensis, T. australe, T. latispora and T. subfibulifera have cerebriform whitish basidioma and abundant clamp complexes, and they nested in the same clade. So, we treat these seven species as members of the T. fibulifera complex.
Tremella lloydiae-candidae Wojewoda and T. cheejenii Xin Zhan Liu & F.Y. Bai also have whitish basidioma and similar micro-morphology with T. fibulifera, but clamp complexes were not observed (Malysheva et al. 2015;Zhao et a. 2019), and we did not examine their types. Because these two species are nested in the same clade as other species of the T. fibulifera complex with robust support in our phylogenies (Figs 1, 2), we treat them as members of the T. fibulifera complex, too.
Currently, more than 30 morphological characteristics are applied for identification species of Tremella s.s. (Chen 1998;Zhao et al. 2019), and some features including basidioma color and basidia shape are variable at different stages. The shape and size of basidiospores are relatively stable characteristics for each species, but they are very similar among some species in the T. fibulifera complex; that is why several taxa were previously treated as T. fibulifera s.l. (Malysheva et al. 2015;Zhao et al. 2019). Consequently, combined morphology and molecular evidence are essential to distinguish species within the complex, and ITS + partial nLSU dataset are selected for species delimitation.
Key to the whitish species in Tremella s. s.