Taxonomic revision of Thraupidae.

For each species an x,y pair is created with the x - the sequence number of the taxon in the IOC world bird checklist version 10.1 (labelled “pos01”); the y value is the sequence number in version 10.2 (“pos02”). Thus, in version 10.1 the first taxon is Orchesticus abeillei while in 10.2 this species is third so its corresponding point is plotted at (1,3). In contrast, in 10.1 Parkerethraustis humeralis is the 394th species in the list; in 10.2 it is the fourth species in the sequence giving (394,4).
Obviously if no changes in sequence occurred the points would lie along the diagonal identity line. Each genus has a single color – the colors are assigned by an algorithm in the plotting routine and there are so many genera that the color list is cycled through multiple times. Each genus has a single color, but the colors are not unique to a particular genus.

I labelled various genera/point-groupings.

Points, or clusters of points that moved earlier in the sequence are below the identity line, those that moved later in sequence are above.

If a genus cluster maintains largely the same within-genus sequence but is moved within the Thraupidae, it will be a diagonal group of points parallel to the identity line but moved above or below it. An example is Poospiza.

In contrast the genus Tangara has moved within Thraupidae, and the sequence of species within the genus is largely reversed.

What does it all mean?

1. Between IOC 10.1 and 10.2 there has been an extensive revision within the family. This fact is evident from casual inspection, but the plot shows how widespread the changes are.
2. The revision includes both reordering of genera and reordering some, but not all within-genus species sequences.
3. As is obvious from the plot, the position in sequence in 10.2 has minimal relationship to position-in-sequence in 10.1, with a corrected R-squared correlation coefficient of 0.1429, implying the position in 10.1 overall contributes only 14% to the position in 10.2.
4. I think the above approach is probably nonsense, as it represents correlation of essentially ordinal data. "2" is not twice "1", and "300" is not in some sense 3 times "100". It seems nonsensical to argue that the 10.2 sequnce-position is a response variable and the 10.1 sequence-position is the control variable. Probably more reasonable is the result of Pearson ordinal correlation, the result of which is -0.38. The Spearman correltation is -0.39. These both suggest there is little or no or correlation -- exactly what the eye tells us.

In contrast, this similar plot shows the change across 9 years (2009-2018) for the generally accepted sequence of Tyrannidae.

In this case the only substantial change was moving Calyptura cristata from the Cotingidae to the Tyrranidae.
We can, I believe, expect extensive revisions soon in this group.

Comments & Suggestions to Data Steward :