Monday, September 17, 2012

Is this the first network from conflicting datasets?

There are no new problems in science, only old ones that obtrude themselves on you in new ways. One biological example of this truism is the discovery by molecular biologists that different genetic datasets produce different phylogenetic trees. This has been known at the phenotypic level for centuries, so that systematists have used reticulating structures such as maps, networks, webs etc to try to display the many different ways in which biological organisms appear to be related to each other, with different relationships shown by different parts of the phenotype.

This multi-faceted nature of relationships became problematic after 1859, when biologists more-or-less settled on trees as a means of displaying genealogy, because trees show only one set of relationships. Indeed, Darwin himself (1851, 1859) noted that different organ systems suggest different relationships. However, he believed that a taxonomic classification should be based on the structure of the entire body, not just one organ system, so that all relationships are considered. He thus concluded that it is necessary to give each kind of feature a different "weight" in contributing to the overall scheme, both for working out phylogenies and for erecting classifications based on them.

Darwin did no explicit phylogenetic work of his own, and so it was St George Mivart who first had to deal with this situation empirically (as discussed this previous blog post). His early work was principally on the comparative anatomy of primates, for which he provided very detailed comparisons of the skeletons of a large number of species, notably in Mivart (1865), based on the axial skeleton (or spinal column), and Mivart (1867), based on the appendicular skeleton (or limbs).

In the 1865 paper Mivart noted that the data for the spinal column "lead to an arrangement of groups and an interpretation of affinities somewhat differing from, yet in part agreeing with, the classification founded on cranial and dental characters". Moreover, the 1865 and 1867 studies did not produce the same phylogenetic tree. Mivart explicitly noted in a letter to Darwin (1870): "The diagram in the Pro. Z. Soc. [1865] expresses what I believe to be the degree of resemblance as regards the spinal column only. The diagram in the Phil. Trans. [1867] expresses what I believe to be the degree of resemblance as regards the appendicular skeleton only" (Darwin Correspondence Project, letter 7170).

In the modern world, one way to deal with this sort of data conflict is to use a phylogenetic network rather than a phylogenetic tree. That is, we do not need to produce a consensus classification based on weighting different datasets (as suggested by Darwin), and we do not need to produce a series of conflicting trees (as done by Mivart). We do not even need to produce a consensus tree, which would be a combination of these two approaches. We can, instead, display the conflict among the different data sources in a single diagram, as a network.

Click to enlarge.

Here, I have produced a network based on Mivart's 1865 and 1867 trees. There are 24 taxa of Primates, although the Gorilla was not in the 1865 dataset, and Inuus, Cynocephalus, Chrysothrix were not in the 1867 dataset. From these two trees I have produced a SuperNetwork, using the SplitsTree program. This is thus a splits graph, interpreted in the usual manner, so that the reticulated parts represent conflict between the two trees and the non-reticulated parts represent agreement.

The main conflict between the two trees is in the relationship between the Nycticebinae and Cebidae, shown as the centre reticlulated area in the network. Basically, the root of Mivart's trees is between these two groups, and they swap sides of the root between 1865 and 1867! Another cause of this netted area is whether Homo is within the Simiinae (as in the 1865 tree) or sister to the Apes (as in the 1867 tree). These two sources of conflict are quite major, from the biological point of view.

The bottom netted area of the network is caused by conflicts about relationships within the Lemuroidea, which are of less consequence. The top netted area refers to whether Simia is sister to either Hylobates (in 1867) or Troglodytes (in 1865), which is a relatively minor point.

It is tempting to see Mivart's change in the position of Homo (from 1865 to 1867) as psychological rather than empirical. Mivart came to reject the idea that humans should be placed in a phylogenetic tree, and expressed this strongly in Mivart (1871). He then became a strong opponent of Darwinian evolution for a time. Nevertheless, he returned to the fold at least once, in Mivart (1881), where he presented a phylogenetic tree of much of the Mammalia, based on dentition. However, he conspicuously excluded the Primates from this tree, thus dodging the theological problem entirely.

Note: For another early example see Fritz Müller and the first phylogenetic tree.


Darwin, C. (1851) A Monograph of the Sub-Class Cirripedia, with figures of all the species. The Ray Society, London.

Darwin, C. (1859) On the Origin of Species by Means of Natural Selection. John Murray, London.

Mivart, St.G. (1865) Contributions towards a more complete knowledge of the axial skeleton in the primates. Proceedings of the Zoological Society of London 33: 545-592.

Mivart, St.G. (1867) On the appendicular skeleton of the primates. Philosophical Transactions of the Royal Society of London 157: 299-429.

Mivart, St.G. (1871) On the Genesis of Species. Macmillan, London.

Mivart, St.G. (1881) The Cat: An introduction to the study of backboned animals, especially mammals. John Murray, London.

No comments:

Post a Comment