I’ve claimed on several occasions
that my cladogram was fully resolved and all of its subsets were also fully resolved. While that is typically true, everyone prefers a cladogram with more strength, where the taxa are lumped and separated by at least three points in Bootstrap Analysis. When that happens the Bootstrap tree will also be completely resolved (= all scores 50-100).
I found a quirk
And as I write this I am going to figure out why this happened and wonder if it can be repaired. Typically a bad score or several bad scores are responsible for any lack of resolution. Not sure if that’s the case this time.
- A heuristic search algorithm using PAUP of the entire cladogram: fully resolved.
- The same deleting all anamniotes and lepidosaurormorpha — the new Archosauromorpha (Eldeceeon and all derived taxa) remains: fully resolved.
- The same deleting all basal archosauromorphs and synapsids: the protodiapsida (Myceteosaurus and all derived taxa) remains: fully resolved.
- The same deleting all basal protodiapsids and basal diapsids: the new younginiforms (Spinoaequalis and all derived taxa) remains: fully resolved.
- The same deleting all aquatic younginiforms: the terrestrial younginiforms: Spinoaequalis + the SAM K7710 specimen(s) of Youngina and all derived taxa) remains: fully resolved.
- Now, working backwards: The same deleting all dinosaurs: 6 trees retained. Loss of resolution in the Protodinosauria. Several incomplete taxa based on drawings there. We’ll reexamine that clade in a later post.
- The same adding Herrerasaurus: fully resolved. A more complete data specimen solves the problem in #6.
- The same deleting all archosaurs: fully resolved.
- The same deleting the protoarchosaurs (the Gracilisuchus clade): fully resolved.
- The same deleting all the poposaurs: fully resolved.
- The same deleting all the Arizonasaurus/Ticinosuchus/Aetosaurus clade: fully resolved.
- The same deleting all remaining Rauisuchia (Vjushkovia through Postosuchus): fully resolved.
- The same deleting all erythrosuchidae and ornithosuchidae: 2 trees retained. Loss of resolution at the Euparkeria node.
- The same adding Garjainia: fully resolved.
- The same deleting all Choristodera: fully resolved.
- The same deleting all Chanaresuchidae and Parasuchia: fully resolved.
- At this point with a single tree scoring 447, I attempted a Bootstrap analysis. Basically all that is left here are the basal terrestrial younginiforms including several Youngina and Youngoides specimens, protorosaurs and basal archosauriforms up to and including all tested proterosuchids + Euparkeria, Osmolskina and Garjainia. Here the small Proterosuchus skull (BPI/1/4016) and Elaphrosaurus rubidgei (RC59) are apparently mucking up the works, even though they do not nest together otherwise.
- The same adding Diandongosuchus, a basal parasuchian: fully resolved.
- At this point with a single tree scoring 488, warrants another Bootstrap analysis. Again Elaphrosuchus scores insufficiently distinct from several other included taxa to lack a score of 50+ across four other clades. And a very low score of 51 separates Elaphrosuchus from the BPI/1/4016 specimen of Proterosuchus. Both are skull only taxa and both have a certain amount of damage.
- The same deleting the new taxon, Teyujagua: Bootstrap scores all above 50. So that addition caused problems.
Are there scoring errors here?
Or do these taxa converge? Or do two sisters lack any data points in common? Let’s find out by taking a closer look at the offending parties.
[About a day or two elapses at this point in the narrative]
There were many errors around these nodes, hopefully all are now repaired. I reexamined several drawings, photos and tracings. Unfortunately fossil bones don’t come with overlying colors, so they have to be interpreted.
I also added a taxon
(Figs, 1, 2). It turned out to be a key transitional taxon. Score corrections and the new taxon boosted 5/6 of the Bootstrap scores.
Adding the ‘other Euparkeria‘: SAM PK K6047A
While reexamining the images of the Euparkeria holotype in Ewer 1967, I compared the rostrum that has a naris (SAM PK K6047A) with the classic holotype (SAM PK K 5867) that lacks a naris (Fig. 1). The fossils did not match. The dimensions were off (orbit vs antorbital fenestra, etc.) and the teeth were different in length. Ewer provides two images of the 6047 specimen, lacking data for the middle rostrum between the pix. The tracing (Fig. 1) recovers a basal euarchosauriform with a longer rostrum and narrower orbit, more like that of its phylogenetic predecessor, the BPI/1/4016 specimen of Proterosuchus (Fig. 3). This taxon ties Euparkeria more closely (more gradual transition in traits) to Proterosuchus (Figs. 2, 3).
There is also a SAM PK K6047B specimen
and it has been named Browniella africana by Broom (1913) I have not seen it. I do not know if it resembles or was found with the 6047A specimen. Most workers consider this taxon a junior synonym of Euparkeria.
Sookias and Butler 2013
reviewed the Euparkeriidae, but did not mention the 6047A or B specimens, except, perhaps as two of the eleven specimens that comprise their hypodigm. They defined the clade in this fashion: “Euparkeriidae Huene 1920. Stem-based definition –the most inclusive clade containing Euparkeria capensis Broom 1913a but not Crocodylus niloticus Laurenti 1768 or Passer domesticus Linnaeus 1758. (new).” I have not tested all the taxa listed by Sookias and Butler, but their definition seems to be overly broad.
What do we learn here?
- Incomplete taxa can cause loss of resolution, as everyone knows. The addition of a more complete cousin can provide the remedy.
- Scoring errors also lead to loss of resolution.
- There is only one tree, the tree of Nature, that we are trying to model here. So there IS a correct solution to this problem.
- Adding taxa almost always provides traits that make phylogenetic transitions more gradual. The only exceptions are terminal taxa, those that lack descendants.
- Known proterosuchids still do not represent ontogenetic (maturation) stages. They are phylogenetically distinct taxa that lead to more derived clades (Figs. 2,3).
- There is still no evidence for the sisterhood of Euparkeria with the verified sisters Turfanosuchus and Gracilisuchus (Sookias and Butler 2013, Butler et al. 2014).
- Phylogenetic miniaturization preceded and was part of the basal archosauriform radiation.
- It is important for professionals not to assume that different specimens represent a single species. Minor differences might turn out to be key traits as demonstrated here.
Finally: the value of a large gamut cladogram
becomes more valuable with every added taxon. More gradual transitions become apparent and bias is further minimized.
Broom R 1913. On the South-African Pseudosuchian Euparkeria and Allied Genera. Proceedings of the Zoological Society of London 83: 619–633.
Butler RJ, Sullivan C, Ezcurra MD, Liu J, Lecuona A and Sookias RB (2014). New clade of enigmatic early archosaurs yields insights into early pseudosuchian phylogeny and the biogeography of the archosaur radiation. BMC Evolutionary Biology 14:1-16.
Ewer RF 1965. The Anatomy of the Thecodont Reptile Euparkeria capensis Broom Philosophical Transactions of the Royal Society London B 248 379-435.
Sookias RB and Butler RJ 2013. Euparkeriidae. Geological Society, London, Special Publications published online January 24, 2013 as doi: 10.1144/SP379.6