Friday, 9 August 2013

Alternative views on the origin of placentals

The first placental mammal
From O'Leary et al. Science 2013; 339: 662-7
Reprinted with permission from AAAS

Hot on the heels of the controversy about crown mammals comes a dust up over crown placentals. This time in Science and in response to an important paper by O'Leary et al. that combined morphological and molecular data (see previous post). One conclusion in that paper (here) was that ordinal diversification of placentals occurred after the Cretaceous-Paleogene boundary (Explosive Model, cf. Wednesday's post).

This has elicited a response from an eminent group of molecular phylogeneticists (Springer et al. here). Their best estimate of the origin of placentals (here) posits a common ancestor 100 Mya with most orders arising in the Cretaceous (Long Fuse Model). They contend that that molecular phylogenies are more reliable than phenome-based (morphological) phylogenies even when they imply ghost lineages (i.e. with no fossil record).

In their rebuttal, O'Leary et al. (here) note that molecular phylogenetics has failed to resolve the basal split in Placentalia (see here) or to establish a sister taxon for primates. Thousands of nonplacental Cretaceous fossils are known so why are placentals absent? 

A seemingly weightier point made by Springer et al. is that ordinal diversification over a short time span in the early Paleocene implies high substitution rates. Maybe so respond O'Leary et al. Let's redo our tree but impose average substitution rates as a limitation. The result is to extend the origin of orders past the K-Pg boundary but only by some 6 My. That puts the common ancestor at 70 Mya which is still 30 My later than the molecular estimate.

1 comment:

  1. It's worth remembering that the divergence dates estimated by molecular phylogenetics are the dates at which reproductive isolation took place, not (necessarily) the dates at which morphological diversification occurred. Apparent incongruity between molecular and morphological studies might arise from this difference between what the two approaches are actually measuring.