How are we going to get out of this mess ?Deep phylogeny of reptiles in crisis
This is the first in a series of posts. I will probably go back and revise these posts over time or re-write them as a single post. Right now, I am using this to draft my thoughts on the problem of reptile deep phylogeny.
I should define what I mean by “reptile deep phylogeny”. I am using “reptile” in its loosest Romerian sense, i.e. including even reptilomorph-like taxa and non-therapsid synapsids. I try to use “modern reptile” to refer to what we think of as living reptiles and everything closer to them than mammals. In the past, “modern reptile” has sometimes been used to distinguish traditional diapsids — lepidosaurs and crocodilians from turtles, but since all molecular evidence is firmly stacked in support that crocodilians and turtles form an exclusive group without lepidosaurs, we need to get past thinking of them as “primitive”. In fact, almost every group of “primitive” reptiles associated with turtles appear to have nothing to do with them. This is a topic for another post. Again, this is a topic for another post. Right now, I want to focus on a group of Permian secondarily aquatic reptiles: the mesosaurs.
You’re probably asking “why mesosaurs?” It’s simple really. Including mesosaurs into our current data sets for resolving reptile relationships reveals some very big issues.
I’m not going to get into the pre-cladistic suppositions on mesosaur relationships in this post. Modesto (1999: 16) did an excellent job summarizing the history of the inclusion of mesosaurs in phylogenetic analyses.
- Gauthier et al. (1988) found mesosaurs as the sisters to taxa we now regard as parareptiles (millerettids, pareiasaurs, and procolophonids).
- Laurin and Reisz (1995) found mesosaurs basal to all other Sauropsida but Modesto (1999) criticized their character scorings and found support for parareptilian mesosaurs.
Since the publication of Modesto (1999), other authors have included mesosaurs in various analyses of reptile relationships.
- Hill (2005) found mesosaurs outside of Amniota.
- Muller and Reisz (2006) reported multiple positions depending on the method they used to analyse their data set — basal parareptiles, basal eureptiles, or basal sauropsids. It’s worth noting that Bayesian analysis supported a basal eureptile position in both Muller and Reisz (2006) and Modesto et al. (2015).
Now for the monster in the room, a much misunderstood or VERY ignored paper by Michael Maisch. Maisch (2010) included both ichthyosaurs and mesosaurs in the matrix of Laurin and Reisz (1995) and the matrix of Rieppel and Reisz (1999), the latter of which no authors have ever bothered to do, just as no one has considered mesosaurs in the diapsid analysis published by Müller (2004) or its derivatives (Bickelmann et al., 2009 and so on).
Maisch (2010) found that when both ichthyosaurs and mesosaurs are included in the analysis, they both nest together, either as parareptiles OR surprisingly as basal diapsids (much more derived than the position found by any of the previous authors). However, Maisch (2010: 171) had a different perspective on these results rather than taking them at face value:
In my opinion the resulting cladogram can only interpreted in one way: there is a high number of characters in the data matrix that are related to or strongly affected by secondarily aquatic adaptations, which are very suspect of being due to convergences rather than to common ancestry. Most of these characters are reductional, too. Due to this, totally unrelated secondarily aquatic groups, as exemplified by the Mesosauria, and possibly also the Ichthyosauria, are forced into a position within basal diapsids, close to the aquatic eosuchians (taken here as a grade or possibly clade, of stemgroup diapsids including Youngina, Acerosodontosaurus and the tangasaurids) and Claudiosaurus and not too far from the sauropterygians and placodonts. The phylogenetic relevance of the resulting cladogram, at least concerning aquatic forms, is therefore in my opinion highly doubtful.
Earlier in the text before discussing the results of his analysis, Maisch (2010:170) had hypothesized that existing data matrices focused too heavily on secondarily aquatic adaptations:
[T]he matrix overemphasizes convergences to aquatic adaptations. Should the mesosaurs — which have not been considered as diapsids before — occupy a similar place in the resulting cladogram as the ichthyosaurs, something must be doubtful about the entire data matrix, at least with respect to secondarily aquatic taxa.
Interestingly, the observations of Maisch (2010) have elicited almost no published comments. But Chen et al. (2014) did attempt to test the influence of secondarily aquatic adaptations on diapsid phylogeny by running their analysis with and without these characters. Their analysis did not include mesosaurs.
Despite no evidence that the authors are aware of Maisch (2010), Sobral et al. (2015) echoed his criticism of the existing diapsid data matrices:
It could be argued that the matrix of Chen et al. (2014) is biased towards taxa with features viewed as adaptations for an aquatic mode of life, and thus we decided to include Elachistosuchus in the only other recently published character-taxon matrix for basal diapsids and early saurians (Ezcurra et al., 2014), which focuses on terrestrial taxa.
I would suggest perhaps the best way to determine where mesosaurs go in the tree might be to pursue multiple tests:
- Include mesosaurs in a matrix focused on terrestrial taxa like Ezcurra et al. (2014) just as Sobral et al. (2015) did.
- For the inclusion of mesosaurs in a matrix like Chen et al. (2014) or Schoch and Sues (2016), multiple representative basal eureptiles should be included as well as characters from Muller and Reisz (2006) that are relevant for their phylogenetic resolution.
Interestingly enough, in a result that has not really been discussed anywhere, Modesto (1996) found mesosaurids (actually represented by multiple OTUs, not just lumped together as Mesosauridae) were the sister to basal diapsids (represented by Claudiosaurus and the “younginiforms”). In this tree, captorhinids, araeoscelids and paleothyridids were the most basal taxa. This is intriguingly similar to the trees that result using the data matrix of Maisch (2010) if the matrix is run first without the inclusion of mesosaurs or ichthyosaurs, reweighted, and then mesosaurs and ichthyosaurs re-added.
More work clearly needs done here, even if it is from behind a computer screen to better understand what is going on with our current data matrices for reptile relationships.