0b10 (Swofford 2002) to
assess clade support. The third set, henceforth referred to as the 4-gene backbone analysis, consisted of four loci including the nuclear ribosomal gene regions (5.8S, 18S, and 25S) and the RNA polymerase II (rpb2) region between conserved domains 5 and 7. Positions deemed ambiguous in alignment were pruned from the nexus file before conversion to Phylip format using SeaView 4.2.4 (Gouy et al. 2010). Nexus and Phylip files of the Androgen Receptor Antagonist four-gene region data set can be obtained from http://www.bio.utk.edu/matheny/Site/Alignments_%26_Data_Sets.html. In the final concatenated alignment, rRNA gene regions occupied positions 1–2854; the rpb2 region comprised positions 2855–3995. The four-gene region data set was analyzed using maximum likelihood (ML) in RAxML 7.0.3 (Stamatakis Tubastatin A mouse 2006a) with rapid bootstrapping (Stamatakis et al. 2008) and by Bayesian inference using the parallel version of MrBayes 3.1.2 (Altekar et al. 2004; Huelsenbeck and Ronquist 2001; Ronquist and Huelsenbeck 2003) on the Newton cluster at the University of Tennessee. For both ML and Bayesian analyses, the rRNA gene regions were treated as a single partition following Aime et al. (2006; see Appendix I). First, second, and third codon partitions of rpb2 were partitioned separately. Thus, four partitions were assigned and modeled separately. One thousand rapid bootstraps
and a thorough ML search were conducted in RAxML using four distinct models/partitions CX-6258 with Decitabine solubility dmso joint branch length optimization. All free model parameters were estimated by RAxML and incorporated a GAMMA + P-Invar model of rate heterogeneity, a GTR substitution rate matrix, and empirical base frequencies for the final ML search. Rapid bootstrapping was done using a GTRCAT model (Stamatakis 2006b). Bayesian inference was performed using a mixed models analysis run in parallel for
up to 50 million generations. Four chains were run with trees sampled every 5,000 steps with the heating temperature set to 0.1. Convergence diagnostic features were used to guide burn-in choice. All analyses were rooted with Plicaturopsis crispa (Amylocorticiales; Binder et al. 2010). The fourth data set used a Supermatrix with 1,000 bootstrap replicates (SMBS) to analyze a more comprehensive data set comprising multiple representatives of taxa from various geographic regions, and utilizing all the available ITS, LSU, SSU and RPB2 sequences except those with only ITS sequences. All sequences were from single collections. The four gene partitions used were: rRNA 1–3164, rpb2 1st codon pos 3165–3915/3, rpb2 2nd codon pos 3166–3915/3, rpb2 3rd codon pos 3167–3915/3. In the rRNA partition, SSU comprised pos 1–1754, 5.8S 1755–1956, LSU 1957–3164. A GTRGAMMA model was assigned to each partition. This analysis was restricted to the hygrophoroid clade as delineated by the 4-gene ML analysis above.