Combined data of different loci, whether fully or partially
congruent, have been commonly considered by inferring organismal
phylogeny (Dettman et al. 2003). We therefore performed,
as in previous studies (Kauff & Lutzoni 2002;
Miadlikowska et al. 2006; Muggia et al. 2014), both single locus
and combined datasets. We analysed the single locus datasets
with a Maximum Likelihood (ML) approach (Mason-Gamer &
Kellogg 1996; Reeb et al. 2004) and the combined dataset using
ML and Bayesian approaches. The combined dataset was
treated in partition by genes nuclear 28S and 18S and mitochondrial
16S in both ML and Bayesian approaches. The ML
analyses were performed using the program RAML v. 7.0.3
(Stamatakis et al. 2005). As only a single model of molecular
evolution can be used across gene partitions in RAML, the
ML analyses (for single loci and combined datasets) were performed
with the GTRMIX model and 1000 bootstrap replicates
were run. The Bayesian Markov Chain Monte Carlo (B/MCMC)
analyses were run in MrBayes v. 3.1.2 (Huelsenbeck &
Ronquist 2003; Ronquist et al. 2005). The model of molecular
evolution applied to each gene partition in the Bayesian analysis,
GTR þ I þ G, was estimated in JModeltest v. 2.1.4 (Darriba
et al. 2012) using the Akaike Information Criterion (Posada &
Crandall 1998). The B/MCMC analysis was run with six chains
simultaneously, each initiated with a random tree, for ten