Updated July 24, 2015 with the hypothesis that snakes are monophyletic, arrived at earlier when taxa were included to link burrowers with non-burrowers.
A recent paper by Longrich et al. (2012) reports on a “missing link” snake from the Late Cretaceous. The paper argues that all snakes descended from a burrowing terrestrial ancestor rather than a swimming marine ancestor. Coniophis precedens (Marsh 1892) is a tiny snake, previously only known from short-spined vertebrae. Now that the jaws are known, more can be said of this species. “It moves like a snake,” said Longrich, “but it doesn’t feed like a snake.”
From the Longrich et al. 2012 abstract, “Coniophis occurs in a continental floodplain environment, consistent with a terrestrial rather than a marine origin; furthermore, its small size and reduced neural spines indicate fossorial habits, suggesting that snakes evolved from burrowing lizards. The skull is intermediate between that of lizards and snakes. the maxilla is firmly united with the skull, indicating an akinetic rostrum. Coniophis therefore represents a transitional snake, combining a snake-like body and a lizard-like head.”
Figure 1. Heloderma (above), Coniophis (middle) and Cylindrophis (below). This correct lineage of burrowing snakes is separate from that of the non-burrowing snakes with a separate marine origin. So both hypotheses are correct when snakes are diphyletic, a current heresy.
In contrast to modern snakes, the jaws remained fixed, limiting the size of its prey, which were swallowed whole. Coniophis’s status as the most primitive snake does not make it the oldest known snake. Others, like Pachyrhachis, extend back another 35 million years. Rather, according to Longrich, “It appears to have been ‘a living fossil’ in its own time, co-existing with more advanced snakes, as chimpanzees and humans still do.”
“It’s not the direct ancestor of modern snakes, but it tells us what the ancestor looks like,” Longrich said.
Adaptations permitting cranial kinesis followed the divergence of Coniophis. Coniophis exhibits an intramandibular joint (1). More derived Serpentes exhibit a maxilla–premaxilla joint (2), loss of maxilla–vomer contact (3), a nasofrontal joint (4), a maxilla–prefrontal joint (5), a mobile dentary symphysis (6) and an articular saddle joint (7). Alethinophidia is characterized by a reduced postorbital bar (8) and a palatine–pterygoid hinge (9). Macrostomata is characterized by a hinged supratemporal (10).
Figure 2. Click to enlarge. Snake tree according to Longrich et al. (2012). I added Ardeosaurus and Adriosaurus, two taxa in the lineage of non-burrowing snakes. Circled numbers refer to traits mentioned in the text (above). Trees like this are problematic at the get-go because they include no suitable outgroup taxa that would make the point that Coniophis would indeed nest where it does.
Unfortunately…
The Longrich et al. (2012) tree nests the new taxon at the very base (as the outgroup) of the rest of the snakes. This is a problem because nearly every other taxon ever known would also nest as the outgroup (elephants, flies, pterosaurs). To make a better case several other outgroups would have to be shown with Coniophis representing the transition between legged and legless taxa, or any other transitional grade that would be more pertinent.
There are too few traits shown in Coniophis to nest it in the large reptile tree, but we’ll take the Longrich examination as valid. The dorsally concave dentary appears to bridge the gap between Heloderma and Cylindrophis (Fig. 1). Pachyrhachis and the larger non-burrowing snakes do not share this trait.
Size matters
The burrowing snakes are all tiny and so is Coniophis.
Ghost Lineages
Many lizard genera are long-lived taxa. Ghost lineages abound in this clade. Bahndwivici, ancestral to both Varanus and Heloderma, is only known from the Eocene, but evidently extended back to the middle Jurassic. Varanus and Heloderma are known today, but likewise must extend back to the Jurassic.
References
Longrich NR, Bullar B-A S and Gauthier JA 2012. A transitional snake from the Late Cretaceous period of North America. Nature 488, 205-208.
Marsh OC 1892. Notice of new reptiles from the Laramie Formation. American Journal of Science 43:449-453.
The Pterosaur Heresies 
First, Longrich et al. did use a non-snake outgroup, they just didn’t show it in their figure.
Second, just as I said regarding mammals in your other post, molecular analyses show your diphyletic snake idea is wrong. Again we have retroposons, which have no known way of being lost and are nearly impossible to converge. For instance, Piskurek et al. (2006) found burrowing snakes (e.g. Rhamphotyphlops, Anilius, Leptotyphlops, Cylindrophis, Uropeltis) to share the AFE SINE with non-burrowing snakes (e.g. Candoia, Thamnophis, two elapids and four viperids), while both Heloderma and Varanus share the VIN SINE. Whereas your phylogeny would have Heloderma by the clade of burrowing snakes and Varanus by the clade of non-burrowing snakes. Conveniently, the SINE data agrees with large morphological analyses like Gauthier et al. (2012) as well as with nuclear genes (Vidal and Hedges, 2005) and mitochondrial genes (Vidal and Hedges, 2004).
I’d like to know how you think three different molecular sources can be wrong in the same way. What mechanism could possibly skew both nuclear genes and mitochondrial genes; and SINES which molecular phylogeneticists claim can’t exhibit homoplasy? Are all of the molecular geneticists wrong about that (and remember with no homoplasy, it doesn’t matterthat we don’t have extinct SINES to test)? We know ways morphological analyses can be wrong, after all. And indeed, that yours disagrees with Gauthier et al., Conrad (2008) and others shows at least someone’s is wrong. Until you provide some realistic answer as to how multiple kinds of molecular data including homoplasy-free data can converge on the same wrong answer (and in so many clades too, like our prior example with bats), how can we take your competing ideas seriously?
Indeed, Longrich used Anguimorpha as the outgroup and Varnoidea as another outgroup in another study. The dangers of using suprageneric taxa are well known. In the Gauthier study (and others) the most derived burrowing snakes nest as the most basal snakes, which is great cause for concern. Don’t you prefer to have basal taxa with plain old features and derived taxa to have giant antlers, bizarre wings and jaws that move left and right rather than up and and down? How can we take such studies that nest Scolecophidia (no teeth in the upper jaw, blind, jaws move left and right, etc. etc.) as basal snakes seriously? It would have been great if Longrich et al. used Adriosaurus, Heloderma, Bahndwivici, Ardeosaurus, Eichstattisaurus, Lanthanotes, Cryptolacerta in their study. They made all the difference in mine.
You didn’t state your explanation of how three separate molecular systems can all get the same wrong result, nor how retroposons are supposed to exhibit homoplasy and why all molecular phylogenetecists are wrong about them.
I think your thoughts about basal taxa stem from your confusion about what cladograms show. Taxa are only “basal” because their branch has less species (or at least less examined species) than the “derived” branch. Scolecophidians are thus basal because there are less of them than there are alethinophidians. But the Scolecophidia branch has plenty of diagnostic characters just as the alethinophidian branch does. There’s no reason at all to think e.g. Leptotyphlops should be more normal looking than e.g. Vipera. They’ve both had equal time to evolve since their lineages split apart, after all.
Working on a lizard DNA blog, so will address that in just a few days.
Taxa are basal because their branch has fewer species?? Just the opposite. Taxa are basal because they have more descendants, including all derived taxa within that clade. If Leptotyphlops is basal it should have more traits in common with outgroups. It doesn’t. I will seek to understand how the blind pipe snakes became basal in other studies. I haven’t done so yet. It’s one of the great mysteries.
But Leptotyphlops isn’t an ancestor to anything. Nothing on a cladogram is an ancestor, and certainly no living genus is (yet). Leptotyphlops looks basal on a cladogram because most cladograms include more alethinophidians than scolecophidians-
(Agama(Leptotyphlops(Python(Acrochordus(Coluber(Vipera,Crotalus))))))
But you could also make the cladogram like-
(Agama(Vipera(Typhlops(Rhinoleptus,Leptotyphlops))))
And now Leptotyphlops looks derived and Vipera looks basal. None of the snake genera in either of these cladograms can be expected to have more traits in common with Agama (or with the first snake population) than with any other snake genus in either cladogram. After all, Leptotyphlops hasn’t been sitting around unchanged since the ancestral snake population split between what would become scolecophidians and what would become alethinophidians. It’s been evolving its own set of characters.
I’m with Mickey on this – how can the molecular data all be wrong in the same way? Parsimony dictates that we take the simplest answer that explains all of the data/observations at hand. Three different lines of evidence all pointing to the same answer with some pretty bulletproof data to back them up, or one alternate view with data that can easily be wrong (character homoplasy, etc. in morphological data) and no explanation of how the other data is wrong?
Quantitative data > qualitative data. “Feeling” that basal taxa should all look similar, even with millions of years of divergence from a common ancestor does not outweigh measurable, repeatable, and testable data derived from multiple independent sources.
First of all, there is no “feeling”. There’s maximum parsimony including fossil taxa, which DNA can’t touch. Maximum parsimony includes shared traits, of course, and these are expressions of proteins created by genes. So, morphology is genetics. The molecular DNA story, unfortunately, includes HUGE ghost lineages (back to any splits, like 320 mya for mammals/other reptiles). A recent paper (Hay et al. 2008), which I will blog on, reports that while Sphenodon has changed very little since the Triassic, and it matures very slowly, and breeds not too often, but the rate of change in its DNA is among the fastest among all vertebrates. This could be the reason for the discrepancy in DNA and morphology.
Hay JM, Subramanian S, Millar CD and Mohandesan E 2008. Rapid molecular evolution in a living fossil. Trends in Genetics, 24(3):106-109.
For your post on reptile molecular analyses, remember that missing fossils don’t matter for retroposons, since they cannot reverse and are basically impossible to converge. So if e.g. bats and horses share a retroposon, their common ancestor must have had it. And if lemurs don’t have that retroposon, they can’t be a descendant of the bat-horse common ancestor. We don’t need protobat DNA, since protobats must have had that retroposon. And indeed, if we had protobat DNA, it wouldn’t help, since we already know it had the retroposon. Of course retroposons can’t tell us if e.g. Phenacodus was a member of the bat-horse group, but they do give us a backbone cladogram to constrain our analyses with.
Also, while morphology is (mostly) caused by genetics, many genes analyzed don’t have any bearing on skeletal anatomy. Squamates have been analyzed with mitochondrial genes for instance, which mostly affect mitochondria of course. And even in the nucleus Vidal and Hedges (2005) looked at e.g. RAG1 and RAG2 which “encode enzymes that play an important role in the rearrangement and recombination of the genes of immunoglobulin and T cell receptor molecules during the process of VDJ recombination.” So they just affect the immune system, not morphology.