Phylogeny of the Carnivora – its topsy-turvy!

The large reptile tree
(LRT) presents a novel topology for many clades within the Reptilia. Among them is the Carnivora (Fig.1). The LRT uses fossil taxa and, you’ll note by comparison, is virtually upside-down (topsy-turvy, backwards) when it comes to trees recovered in molecular studies. That major difference MIGHT be traced to the choice of outgroup, as you will see…

Using molecular phylogenetics
(no fossils) Eizirik et al 2010 recovered a cladogram of the Carnivora that used Manis, the pangolin (Fig. 2), as the outgroup. Does this surprise you? …especially considering the fact that Manis has bounced around various nodes on the mammal family tree for decades. …and since it is toothless! And since it has scales instead of hair! etc. etc.

That, in itself, is very strange
to have a highly derived taxon used as a plesiomorphic outgroup. By contrast, in the LRT the outgroup is Monodelphis (Fig. 3), a tiny very plesiomorphic, opossum-like basal placental with origins in the Jurassic. And it has teeth!  And hair!

Using a pangolin as the outgroup
the Eizirik team recovered a basal split between feliforms and caniforms.

Feiliforms include Nandinia, then a split between cats and civets + hyenas + mongooses + fossas.

Caniforms include a basal split between wolves and bears + seals + raccoons + minks. Essentially these topologies are quite similar to the LRT, only in the opposite order with cats and dogs nesting in basal nodes, while minks and mongooses nest in derived nodes.

Notice the relatively flipped topologies
Can we blame this on the choice of an outgroup? On the lack of fossil taxa? On the inadequacies of DNA analyses across large clades? Or a little of all three?

Note that
Talpa, the extant Eastern mole, and Mondelphis, the extant gray short-tailed opossum were excluded a priori from the Eizirik study, but revealed by the large gamut analysis of the LRT, which minimizes a priori assumptions such as these.

Also using molecules
Wesley-Hunt and Flynn 2005 found a similar topology to the Eizirik study, turning the order recovered by the LRT on its head, with opossum-like carnivores (civets, minks) in derived nodes. This study used a variety of outgroups (Manis, Elephas, Loxodonta, Equus, Bos, Sus, Homo) rather than Monodelphis. Results did not change the topology within the Carnivora.

Now is a good time to ask yourself,
Why did they use such silly, useless and obviously wrong outgroups rather than seek the one true plesiomorphic outgroup?

This is exactly why
this blog and ReptileEvolution.com were created — to throw back the curtain on such odd practices, methods and choices — AND produce viable alternative answers. These are experiments you can repeat yourself, BTW.

Let’s not forget
moles (Fig. 3) are carnivores, too!

References
Eizirik E, Murphy WJ, Koepfli KP, Johnson WE, Dragoo JW and O’Brien SJ 2010. Pattern and timing of the diversification of the mammalian order Carnivora inferred from multiple nuclear gene sequences. Molecular Phylogenetics and Evolution 56:49–63.
Wesley-Hunt GD and Flynn JJ 2005. Phylogeny of the Carnivores. Journal of Systematic Palaeontology. 3:1–28.