Departmental coffee hours are supposed to be cordial affairs, a chance to catch up with colleagues and exchange news and ideas. So why was this graduate student I’d just met getting up in my face about using the “wrong” locus for my work characterizing fungal communities in Hawaii? I’ve been using the D1/D2 domain of 26S rRNA and finding an amazing amount of diversity in plant substrates Hawaiian drosophilids use for reproduction, oviposition, and larval development. The grad student seemed to think if I wasn’t pyrosequencing using ITS, I was just wasting my time. But I’m not pyrosequencing, although I am accepting contributions from benevolent patrons to do so. Instead, I’m focusing on one locus that will be informative over wide phylogenetic space without becoming saturated with mutations. In a recent paper comparing the utility of D1/D2 and ITS for yeast species delimitation, Groenewald et al (2011) found 3 to 4 times greater nucleotide diversity in ITS. That worked out great for their work, in which they combined the sequences from both loci with morphology, mating studies, and physiology to propose three new yeast species in the genus Candida. In another study, Zimmerman and Vitousek used only ITS to identify over 4200 fungal OTUs across 13 sites on Mauna Loa. BLAST searches showed these OTUs spread over a dozen or so fungal classes, but most were either Dothideomycetes (36%) or Not assigned/environmental sample (40%). The problem I see with this is that they identified 4200+ OTUs BLASTing to distant classes using 157 bp of sequence. The experiments were rigorously carried out, but saturation of variable sites, where the same base location is hit with the same mutation in multiple lineages, must be an issue. That can cause errors in defining OTUs, making less related taxa appear to be more related. So in fact Zimmerman and Vitousek may have picked up less diversity than was actually present in their samples.
Getting back to the grad student’s criticism, is the D1/D2 domain obsolete? I did a quick search on Web of Science to find out. I searched for citations of the paper in which the NL1 and NL4 primers, which amplify the domain, were first described (Kurtzman 1997), and on the paper (Gardes & Bruns 1993) in which one of the primers, ITS1-F, used by Zimmerman & Vitousek (2012) was presented. These searches gave the following hits, broken down by year.
Kurtman’s D1/D2 domain primers have seen pretty steady use over the last decade, with a slight uptick in the late 2000’s. Use of the ITS primer has steadily increased over the same period. Why? I think it’s a combination of one locus becoming accepted as the “standard”, and the fact that there is more phylogenetic information (variation) contained within the ITS locus, which is probably why it became the standard. In addition, the sequence length works quite well with high-throughput pyrosequencing, allowing the generation of very large data sets, as in Zimmerman & Vitousek (2012).
So, which locus would you use? I still think it makes sense to choose a less variable locus when you’re searching across wide phylogenetic space. But to make more refined identifications to the level of species, I’d use ITS, and I’d like to make the switch to pyrosequencing to do it.
Brian Ort
Kurtman’s D1/D2 domain primers have seen pretty steady use over the last decade, with a slight uptick in the late 2000’s. Use of the ITS primer has steadily increased over the same period. Why? I think it’s a combination of one locus becoming accepted as the “standard”, and the fact that there is more phylogenetic information (variation) contained within the ITS locus, which is probably why it became the standard. In addition, the sequence length works quite well with high-throughput pyrosequencing, allowing the generation of very large data sets, as in Zimmerman & Vitousek (2012).
So, which locus would you use? I still think it makes sense to choose a less variable locus when you’re searching across wide phylogenetic space. But to make more refined identifications to the level of species, I’d use ITS, and I’d like to make the switch to pyrosequencing to do it.
Brian Ort
RSS Feed