Multilocus studies

Although the ITS locus is quite variable and sufficient to distinguish between different species, this region provides too few information when looking deeper into closely related species or species complexes. In such cases further loci are needed, to produce reliable trees. In some publications regarding smut fungi a combination of ITS and LSU is used (e. g. McTaggart et al. 2016, Savchenko et al. 2014a, see chapter 1.5.1). As mentioned in chapter 1.5.1, the LSU is mostly sufficient for the discrimination of suprageneric splits and the ITS for the differentiation between different species.

Protein-coding loci show varying results and successes. Liu & Hall (2004) demonstrated that protein-coding genes resolve fungal lineages that could not be distinguished based on rDNA loci. Although some different protein-coding loci for smut fungi are available, most of them are not regularly used. Most commonly used in different studies are atp6, tef-1α, ß-tubulin, rpb1, rpb2, and gadph (e.g Albu et al. 2015, Begerow et al. 2006, Lotze-Engelhard 2010, McTaggart et al. 2012, Wang et al. 2006).

Sequences of rpb1 and rpb2 were mostly derived from culture material of smut fungi, without any other DNA present, e. g. from other fungi or plants. However, the available primers result in no or unsatisfactory amplification results for several smut fungi. In some cases, the amplified sequences of the different protein-coding genes showed a patchy distribution among the samples used (e. g. Begerow et al. 2006, McTaggart et al. 2012). It is still not fully resolved, which impact these incomplete datasets have on the calculation of phylogenetic trees and controversial opinions exist (e.g. Lemmon et al. 2009). Roure et al. (2013) pointed out, that the outcome strongly depends on the calculation method used. Furthermore, their analysis showed a decrease of the resolving power of the phylogenetic branches. They also highlighted, that there are problems with phylogenetic artefacts, if too many incomplete data (samples) are used. But in many cases it is difficult to produce an entire dataset. For smut fungi, most of the multilocus trees published in different journals rely on sequencing of two loci of rDNA, ITS and LSU (Begerow et al. 2002, Piątek et al. 2012,

2013, Stoll et al. 2005, Vánky & Lutz 2007, 2010) and these datasets are mostly complete. Rarely multilocus studies exist with more than two loci used, but always as incomplete data sets. In general, ribosomal loci were combined with protein coding loci to produce reliable trees (e. g. Begerow et al. 2006, McTaggart et al. 2012, Wang et al. 2015).

Lotze-Engelhard (2010) based his analysis of the genus Urocystis on three different loci (LSU, ITS, tef-1α), Begerow et al. (2006), McTaggart et al. (2012) and Kruse et al. (2018b) used four different loci for their analysis. Piątek et al. (2016) used five different loci and Wang et al. (2015) analysed seven different loci for their phylogenetic analysis. More than 80% of the species analysed by Wang et al. (2015) were collection samples available as pure cultures, simplifying the amplification of different loci due to the lack of other contaminant DNA which is present in herbarium material. For the first time, nine different loci were used by Kruse et al. (2018a) to produce phylogenetic trees for the Ustilaginomycotina (Ustilago striiformis).

Lotze-Engelhard (2010) analysed the rpb1/2-locus of Urocystis, which was used by Wang et al. (2015), but they could not amplify most of the samples. Our preliminary lab work yielded unsatisfactory results (weak bands, dual bands, regularly no amplification), too. As Lotze-Engelhard (2010) was working with herbarium samples, and our preliminary primer tests had problems in amplifying these regions for herbarium samples, it is hypothesized that cultures are more readily amplified than non-cultured fungi. It seems that this primer is not specific enough to amplify the target DNA in a mixture of plant and other fungal DNA. Thus, it only worked for pure cultures, where only the target organism is available.

The atp6 locus and the ß-tubulin locus used in Begerow et al. (2006) were not used again in any other smut phylogeny to date. Beside the difficulties to guess which primer pair they used (this information is missing in the paper as well as in the supplementary material), we assume that they used the atp6 primers published in Kretzer & Brunns (1999) and the ß-tubulin primers of van der Mewe et al. (2007).

Lotze-Engelhard (2010) tested these loci for the genus Urocystis and did not get any amplification result. This is in line with our initial tests, which also achieved only negative amplification result (data not shown), and is assumed to be one reason for the rare usage of these two gene loci (less than 30 sequences each deposited in GenBank, https://www.ncbi.nlm.nih.gov/). Furthermore, the ß-tubulin primers were designed for the amplification of the Boletales, and the atp6 primers for the

amplification of rust fungi. It is likely that they were too specific to amplify smut fungi.

All these preliminary tests and available publications regarding the usefulness of the above mentioned primers resulted in the attempt of designing new and more specific primers for multilocus approaches to reveal species boundaries and the phylogeny of smut fungi.

Most of the loci presented in Kruse et al. (2017c) were not yet used for phylogenetic analysis within the smut fungi. Only atp (ATP-Synthase) and ß-tubulin (beta-Tubulin) were used by Begerow et al. (2006), but with other primers and/or regions (atp2 vs. atp6). Since some genomes are published for several smut fungi (e.

g. Malassezia globosa (Xu et al. 2007), Melanopsichium pennsylvanicum (Sharma et al. 2014), Moesziomyces [Pseudozyma] aphidis (Lorenz et al. 2014), Sporisorium reilianum (Schirawski et al. 2010), Ustilago hordei (Laurie et al. 2012) and U. maydis (Kämper et al. 2006)), there was an opportunity to design new primers instead of trying to improve already available ones. The analysis in Kruse et al. (2018a) shows, that these newly designed primers are suitable for distinguishing closely related species of Ustilago. This was a result of design based on smut genomes of the Ustilaginomycetes, published previously. Most of the lineages which were only poorly resolved in Savchenko et al. (2014a) are given here with medium to high support values and showing different lineages based on host plant genus or species.

For the Exobasidiomycetes, Kruse et al. (2018b) presented new primers for the gene regions shown in Kruse et al. (2017c). For the map and the ssc1-locus Kruse et al. (2018b) presented new primer combinations. The atp2 locus is a promising locus for amplifying a wide range of smut fungi (Ustilaginomycetes and Exobasidiomycetes).

Own preliminary work has shown, that the newly designed primers are not working for every species of a genus, due to sequence differences between these species. For example, the atp2 primer gives good amplification results for Entyloma species, except of Entyloma on Apiaceae. Within this family only a small set of different species worked (data not shown). This aspect shows the diversity of smut fungi and that it is impossible to get all atp2 sequences with only one primer set. The primers given in Kruse et al.

(2017c, 2018b) give the chance to further improve these primers and optimize them for specific species complexes and/or species. With the opportunity to use now nine different gene loci for smut phylogenies the work in Kruse et al. (2018a) allow to discriminate also closely related species and species complexes, which could not be solved previously (e. g. Entyloma on Asteraceae, Begerow et a. 2006).