2.7　Phylogenetic analysis

A phylogenetic tree was inferred by the maximum likelihood(ML)method using the nucleotides of 14 concatenated PCGs(atp6，atp8，atp9，cox1～3，nad1～6，nad4L，cob)in mt genome.The 20 fungal mt genomes were downloaded from NCBI.Multiple sequence alignment was performed using the MAFFT program(Katoh and Standley，2013).Poorly aligned positions and gap positions were removed with Gblocks(Castresana，2000).Three species belonging to Glomerellales(Colletotrichum acutatum，Colletotrichum lupini)and Eurotiales(Penicillium polonicum)were used as outgroup taxa in the phylogenetic analysis.The best model used in the ML phylogenetic tree was determined by using“find best DNA/protein models(ML)”in MEGA 7.0(Kumar et al.，2016).The“GTR+G”model produced the lowest values for both the Bayesian Information Criterion and the corrected Akaike information criterion，therefore it was chosen for phylogenetic analysis.The phylogenetic tree was constructed using RAx ML 8.0.19 with 500 bootstrap replicates(Stamatakis，2006).Bayesian inference(BI)analyses were processed with 1，000，000 generations and four chains(one cold and three hot chains)，with sampling every 500 generations and a burn-in of 25%.The confidence values of the BI tree were shown as Bayesian posterior probabilities in percentages.The neighbor-joining(NJ)phylogenetic tree based on 14 PCGs or rps3 was constructed by MEGA 7.0 and the evolutionary distances were computed using the Tajima-Nei method.(https://www.daowen.com)

For rps3 analysis，the outgroup P.polonicum was changed to P.nordicum to avoid a large number of gaps and to get more genetic information.Because rps3 gene in P.polonicum contained a premature termination codon，resulting a shorter aa sequence(Kang et al.，2016).The nonsynonymous(d N)，synonymous substitution(dS)and the ratio(d N/dS)of each rps3 sequence with the reference of P.nordicum were calculated using CODEML(RateAncestor=1)in PAML v4.7a(Yang，1997；Yang and Nielsen，2000).Sequence alignment was performed for rps3 genes in Hypocreales(length from 398 to 544 amino acids，aa)using Clustal W version 2.0.12(Sievers et al.，2011).The positive selection pressure on rps3 was detected using CODEML implemented in PAML v4.8(Yang and Nielsen，2002；Yang etal.，2005；Yang，2007；Lin et al.，2015).Excluding the ambiguous sequences，the rps3 gene sequences were represented by 322 consensus aa sequences.The models of M0(one-ratio)，M1a(neutral)，M2a(selection)，M7(beta)，and M8(beta&ω)were used to calculate the dN/dS values.