Revisiting tree space searching with an example (Spindler et al., 2018)

From Bones to Books
Nov 3 · 2 min read

I wanted to revisit a discussion of tree space searching with a few examples, and I am starting off using Spindler et al. (2018). These authors described a new varanopid (possibly arboreal) from Germany. Their tree included some additional sauropsids (including parareptiles) to deepen their sampling.

The authors initially ran it with their entire taxon sampling, but their findings were as follows:

“Noticably stronger than eureptilian OTU, parareptiles appeared polyphyletic, with Australothyris (same as for Captorhinus) showing affinities to Caseasauria, whereas Milleropsis and Erpetonyx plotted among basal Varanopidae.”

If you’re wondering what that would roughly look like in their tree, it would be like:

Now, I was curious about this, I wondered what characters might support such groupings and are they ones that have been previously neglected in sauropsid matrices? So I downloaded their matrix and then re-ran it in PAUP* (v. 4.0a166). First, a heuristic search with the default settings was performed finding 60,057 shortest trees of 598 steps. In this, both Captorhinus and Australothyris were in a polytomy with Tseajaia and almost all other taxa included (the only resolved nodes were Milleropsis+Erpetonyx+Varanopidae, Caseasauria, and Echinerpeton+Ophiacodon). In a 50% majority-rule consensus, resolution was a little better (but remember that majority-rule consensus trees are not really evidence of better supported trees), while almost all the synapsid clades were better resolved, Captorhinus remained in the polytomy though Australothyris moved to the base of Caseasauria.

Then, I ran a tree space search, using Wilson’s recommended settings. This resulted in 17,010 shortest trees of 587 steps (so, 40,000+ fewer trees; 11 steps shorter). Aside from that computational time was much shorter, the strict consensus resolved Australothyris at the base of Caseasauria, Milleropsis and Erpetonyx stayed as sisters to Varanopids (but closer to varanopids than Apsisaurus or Ascendonanus), all the known ophiacodontids grouped together, and the haptodontiforms grouped together, rather than lying in a giant polytomy. The majority-rule tree (again, not evidence of better supported, just more frequent trees) found an interesting topology-

(Tseajaia(Australothyris+Caseasauria(Captorhinus(Protorothyris(Paleothyris(Eupelycosauria [as in Spindler et al., 2018](Apsisaurus,Ascendonanus(Erpetonyx(Milleropsis,Varanopidae))))))))))

So, my point mostly being with this, even with a lot of conflicting data, tree space searching outperforms the default heuristic search setting, finds shorter and fewer trees.

From Bones to Books
Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade