Three fungus-infected insect specimens were collected from Xishui County (28°29'56.70"N, 106°24'31.04"E) (A1950 and A1952) and Dali, Rongjiang County (26°01'58.70"N, 108°24'48.06"E) (DL1007), Guizhou Province, on 20 July and 1 October 2018, respectively. Isolation of the fungi was done as described by Chen et al. (2019). The surface of the specimens was rinsed with sterile water, followed by surface sterilisation with 75% ethanol for 3–5 sec. A part of the insect body was cut off and inoculated with haemocoel on potato dextrose agar (PDA) and PDA, to which 1% w/v peptone (PDAP) had been added. Fungal colonies emerging from specimens were isolated and cultured at 22 °C for 14 d under 12 h light/12 h dark conditions following protocols described by Zou et al. (2010). Accordingly, strains A19501, A19502, A19521, A19522, DL10071 and DL10072 were obtained. The specimens and the isolated strains were deposited in the Institute of Fungus Resources, Guizhou University (formally Herbarium of Guizhou Agricultural College; code, GZAC), Guiyang City, Guizhou, China.

Macroscopic and microscopic morphological characteristics of the fungi were examined and growth rates determined from PDA cultures incubated at 25 °C for 14 d. Hyphae and conidiogenous structures were mounted in lactophenol cotton blue or 20% lactate solution and observed with an optical microscope (OM, DM4 B, Leica, Germany).

DNA extraction was carried out in accordance with Liang et al. (2011). The extracted DNA was stored at −20 °C. Translation elongation factor 1 alpha (TEF) and RNA polymerase II largest subunit 2 (RPB2) genes were amplified using 983F/2218R and RPB2-5F/RPB2-7Cr primers, according to van den Brink et al. (2012). The RNA polymerase II largest subunit 1 (RPB1) gene was amplified with the primer pair CRPB1 and RPB1-Cr (Castlebury et al. 2004). The internal transcribed spacer (ITS) region was amplified by PCR using ITS4/ITS5, which was described by White et al. (1990). PCR products were purified using the UNIQ-10 column PCR products purification kit (no. SK1141; Sangon Biotech (Shanghai) Co., Shanghai, China) in accordance with the manufacturer’s protocol and sequenced at Sangon Biotech (Shanghai) Co. The resulting sequences were submitted to GenBank.

The DNA sequences, generated in this study, were assembled and edited using Lasergene software (version 6.0 DNASTAR). Generated ITS, RPB1, RPB2 and TEF sequences were aligned with those published by Mongkolsamrit et al. (2018) and others selected on the basis of BLAST algorithm-based searches in GenBank (Table 1). Purpureocillium lilacinum (Thom) Luangsa-ard, Houbraken, Hywel-Jones & Samson (isolates CBS 284.36 and CBS 431.87) and Beauveria bassiana (Bals.-Criv.) Vuill. (ARSEF 1564) were chosen as outgroup taxa for the analysis of Samsoniella in Cordycipitaceae and Samsoniella species and closely-related species, respectively. Multiple datasets of ITS, RPB1, RPB2 and TEF were aligned using MAFFT v7.037b (Katoh and Standley 2013) and alignments were edited with MEGA6 (Tamura et al. 2013). Sequences were concatenated with SequenceMatrix v.1.7.8 (Vaidya et al. 2011). The partition homogeneity test in PAUP4.0b10 (Swofford 2002) was undertaken by using the command ‘hompart’.

Maximum Likelihood (ML) analyses were constructed with RAxMLGUI (Silvestro and Michalak 2012). The GTRGAMMA model was used for all partitions, in accordance with recommendations in the RAxML manual against the use of invariant sites. For Bayesian Inference (BI), a Markov Chain Monte Carlo (MCMC) algorithm was used to generate phylogenetic trees with Bayesian probabilities using MrBayes v.3.2 (Ronquist et al. 2012) for the combined sequence datasets. The selection of the best-fit nucleotide substitution model for each locus was calculated by the Akaike Information Criterion (AIC) with jModelTest 2 (Darriba et al. 2012). The TIM+I+G model was selected for the concatenated ITS+RPB1+RPB2+TEF sequences. The Bayesian analysis resulted in 20,001 trees after 10,000,000 generations. The first 4,000 trees, representing the burn-in phase of the analyses, were discarded, while the remaining 16,001 trees were used for calculating posterior probabilities in the majority rule consensus tree. After the analysis was finished, each run was examined using the programme Tracer v1.5 (Drummond and Rambaut 2007) to determine burn-in and confirm that both runs had converged. The final alignment is available from TreeBASE under submission ID: 24710 (http://www.treebase.org).

The phylogenetic tree of Samsoniella in Cordycipitaceae (Fig. 1) and Samsoniella species and closely related species (Fig. 2) were generated from the ML and BI analysis, based on a combined data set of ITS, RPB1, RPB2 and TEF sequence data. Statistical support (≥ 50%/0.5) is shown at the nodes for ML bootstrap support/BI posterior probabilities (Figs 1, 2). The strain numbers are noted after each species’ name. The concatenated sequences of analysis 1 and analysis 2 included 67 and 17 taxa, and consisted of 2,152 (ITS: 528, RPB1: 488, RPB2: 442 and TEF: 694) and 2,194 (ITS: 477, RPB1: 565, RPB2: 473 and TEF: 679) characters with gaps, respectively.

Figure 1. 

Phylogenetic relationships of the genus Samsoniella in Cordycipitaceae, based on multigene dataset (ITS, RPB1, RPB2 and TEF). Statistical support values (≥ 50%/0.5) are shown at the nodes for ML bootstrap support/BI posterior probabilities.

Analysis 1: Samsoniella in Cordycipitaceae. The RAxML analysis of the combined dataset (ITS+RPB1+RPB2+TEF) yielded a best scoring tree (Fig. 1) with a final ML optimisation likelihood value of –28,809.222105. Parameters for the GTR model of the concatenated dataset was as follows: estimated base frequencies; A = 0.234094, C = 0.301291, G = 0.260521, T = 0.204093; substitution rates AC = 1.111784, AG = 3.130020, AT = 0.930972, CG = 0.886915, CT = 6.300092, GT = 1.000000; gamma distribution shape parameter α = 0.390179. In the phylogenetic tree (Fig. 1), Samsoniella species were clustered in a clade and resolved into two obvious clades. Samsoniella species have a close relationship with Akanthomyces species.

Analysis 2: Samsoniella species and closely-related species. The RAxML analysis of the combined dataset (ITS+RPB1+RPB2+TEF) yielded a best scoring tree (Fig. 2) with a final ML optimisation likelihood value of –9,722.503130. Parameters for the GTR model of the concatenated data set were as follows: estimated base frequencies; A = 0.233473, C = 0.298686, G = 0.261629, T = 0.206212; substitution rates AC = 1.250081, AG = 2.534760, AT = 0.891128, CG = 0.827805, CT = 5.916085, GT = 1.000000; gamma distribution shape parameter α = 0.674468. In the phylogenetic tree (Fig. 2), Samsoniella species were clustered in a clade and easily distinguished with Akanthomyces species. S. coleopterorum and S. lepidopterorum clustered in a clade (Fig. 2) and formed two independent branches. S. hymenopterorum was phylogenetically close to S. inthanonensis and S. aurantia.

Figure 2. 

Phylogenetic relationships between the genus Samsoniella and closely-related species, based on multigene dataset (ITS, RPB1, RPB2 and TEF). Statistical support values (≥ 50%/0.5) are shown at the nodes for ML bootstrap support/BI posterior probabilities.