This is a cautionary tale
The following blog reminds all workers to score the entire specimen if possible, and to score as many more-or-less-complete specimens as possible. Why?
It is of vital importance to use as much data as possible
when scoring each taxon in a phylogenetic analysis to remove any trace of attraction by convergence that happens when just using bits and pieces of cherry-picked taxa.
From Pittman et al. 2020,
“Generally during early avian evolution, the furcula, coracoid, and sternum become more craniocaudally elongate, while the manual digits become reduced and fusion between the metacarpals increases.”
Not true. In a valid phylogenetic context (Figs. 1–3), like the wide gamut large reptile tree (LRT, 1729+ taxa; subsets Figs. 2, 3), some taxa developed birdy traits quickly while others dawdled or reversed. In this way some bones demonstrated convergence with other less related clades. With this in mind, start with a valid unbiased topology, then let the taxa tell their own story. Avoid the temptation of an easy diagram. Do the necessary work.
Due to taxon exclusion
Pittman et al. mixed up the order of the pectoral girdles + hands of basal birds (Fig. 1), hoping to tell the story they wanted to tell: gradual evolution. Not only did they skip about a dozen pertinent taxa, they got the order wrong by eyeballing a few traits on cherry-picked taxa.
With more taxa, as in the LRT,
(Figs. 2, 3) the girdles and limbs are phylogenetically re-ordered here (Fig. 1, layer 2 with colors). If Pittman et al. wanted to show gradual evolution, they needed to first establish a valid tree topology by adding more taxa. Instead, by cherry-picking certain traits to show gradual evolution, Pittman et al. were “Pulling a Larry Martin“, putting individual traits on cherry-picked taxa ahead of an entire suite of traits and a wide gamut of taxa.
When the phylogenetic order is corrected
based on unbiased results recovered by the LRT (subsets Figs. 2, 3), what seemed to Pittman et al. a gradual transitional series is here revealed to be an example or two of convergence. Note the similarly elongate coracoids on the enantiornithine Parabohaiornis and the unrelated ornithurine, Yanornis (Fig. 1`), derived from an Early Cretaceous sister to a living taxon, Megapodius.
Time after time paleontologists cherry-pick taxa.
That has to stop. Add more taxa and let the software decide the tree topology. Similarly, don’t rely on parts alone (Fig. 1) to illustrate hypotheses, unless they represent taxa already nesting together based on all of their parts and a wide gamut of taxa. Body parts, like hands and girdles, can converge, as they do here.
On a similar note, basal mammal workers
have put too much reliance on tooth traits. Unfortunately, sometimes that’s all they have. If so, what should they do? They should build a tree topology based on complete or more nearly complete specimens. THEN fit it in those tooth and mandible taxa once the tree topology is established in a broader sense, as in the LRT. Earlier (Fig. 4) you saw how odontocete and archaeocete traits brilliantly document a step-by-step reversal to a simple cone shape, like those of basal pelycosaurs. The addition, subtraction and modification of tooth cusps in mammals occurred much more widely than shown by this one example.
Pittman M, O’Connor J, Field DJ, Turner AH, Ma W, Makovicky P and Xu X 2020. Pennaraptoran Systematics. Chapter 1 from Pittman M and Xu X eds. 2020. Pennaraptoran theropod dinosaurs. Past progress and new Frontiers. Bulletin of the American Museum of Natural History 440; 353pp. 58 figures, 46 tables.