Discovering new species of porcini with citizen science

This is the second installment of the expanding porcini ITS tree thanks to specimens provided by citizen scientists that were sent in response to our solicitation on MushroomObserver (MO). For this dataset, I had my students in Mycology (BIOL 5425) at the University of Utah extract DNA, PCR amplify, and Sanger sequence the ITS region from 23 specimens from MO (plus a few others). The reference matrix is a curated dataset of single representative sequences for each species of porcini in existence that I am aware of (including a handful of undescribed species). A number of these sequences/specimens are still unpublished. I used MAFFT L-INS-i to align them. The tree is the best ML tree using a partitioned analysis in IQ-tree and ultrafast bootstraps (sign. at 95%). I rooted with a sequence of Buchwaldoboletus hemichrysus. The MO sequences are highlighted in red and are labeled with the MO number, followed by the UT-M accession number, followed by the putative identification.



There are a few exciting results to highlight:

  1. There are undescribed species in North America that were unknown to me (i.e. the second B. separans clade, the highly divergent MO279831, and possibly the B. variipes-like MO299848)
  2. Two species of B. separans — not totally surprising (the name B. pseudoseparans exists, for instance), but this is really unambiguous. Naming them may be difficult, however.
  3. Holy smokes look at MO279831!!!! Major evolutionary novelty there. What a discovery! This taxon has MASSIVE implications for biogeography…

One additional observation I want to make is that the typical 3% species-level divergence metric commonly used to delimit “Operational Taxonomic Units” (OTUs) would vastly underestimate the number of species of porcini. The graph below shows the range of pairwise divergences for a selection of species in the tree above. The bottom “whisker” indicates the minimum pairwise distance between species. The red line indicates the typical 3% divergence cutoff used to delimit OTUs. Most species have minimum pairwise divergences far below the 3% threshold (and many also below a “stringent” 1% threshold). pairwise_dist

I want to acknowledge the important expertise and efforts of the citizen scientists who contributed material for this study (and I hope will continue to contribute material!), my students in BIOL 5425 who generated the sequence data, and especially my PhD student Alex Bradshaw for managing the student activity and doing the bulk of the work. I also want to acknowledge Jean-Marc Moncalvo, who is my collaborator on the Sarawak and Vietnam material, Sven Buerki who contributed the enormously important specimen of Boletus phaeocephalus, and QMM in Fort Dauphin, Madagascar, for facilitating fieldwork that resulted in the undescribed Boletus that falls near B. albobrunnescens (admittedly, this may be an artifact of sampling bias, which is being revisited with whole genome sequencing…).