Why can't we all just get along?: Morphological vs. Molecular Data

Hello everyone, staying on top of things in a timely manner is far harder then I care to admit. Still, here is a new post. These may still be somewhat infrequent until I can finish up my thesis, or a least the first draft of it. But once February hits, perhaps these will be more on track. Here's hoping.....

http://txtwriter.com/Backgrounders/Dinosaurs/dinoBG2.html

So, for those of you who have looked into phylogenies, especially those concerning modern taxonomic groups and modern genera in particular, you may have run into some problems (or you may not have, I don't actually know for sure). I will focus on a specific case, but this will not be the end.
 




Many people are aware of phylogenies, cladograms, and the concept of systematics in general. These are ways to convey the relationships between organisms and help (or possibly help) show evolution through time. These are commonly used in biology and paleontology today. These two fields don't always use the same data sets though. This is not really surprising, since biologists working with modern taxa have the ability to get far more data, while paleontologists commonly only have morphological characters from the fossils they can find. While biologists commonly have several types of data they can use to acquire phylogenies, a relatively common one is molecular data. And this makes sense, DNA should be able to be used for discussing and discovering the relationships of living things to each other.

http://www.onlpharmacy.com/blog/tag/genomic-information

Sure enough, phylogenies based on molecular data are quite common. A simple search for "molecular phylogeny" in Google Scholar gives >326,000 results. While these can deal with anything from bacteria, to plants to animals, to distinct parts of animals. These have allowed people to discover far more information about relationships of living things. The concept is simple enough, use the same basic thing to compare, say a certain nucleotide or protein, pick your set of organisms with one or more outgroups, align them, then compare them. As a note, you want to have roughly the same numbers for each individual you use. So, say you investigate the cytochrome b protein in felids, you would want the total number of proteins in each taxon to be roughly, say all around 200 total aa proteins (or whichever number you decide upon).

After aligning all of them and running them through a program like PAUP, you will get at least one cladogram. It may look something like this:

Stephens and Wiens 2003, fig. 3

The above figure is figure 3 from Stephens and Wiens (2003). It shows two molecular phylogenies, one as maximum parsimony (a) and the bottom as maximum likelihood (b). I won't be getting into the difference between the two at the moment, just realize they are two slightly different ways of analyzing the same data. The two phylogenies actually do combine three separate data sets though and combine them. They both use cytochrome b, 16S and ND4 data sets.

At first glance, it seems that these do a very good job of agreeing with the known taxonomy presented by the different genera. The only problem is that Clemmys is strongly split at the bottom of the phylogenies. This was cleaned up after this paper was published, however, with the Western pond turtle (Actinemys) and the wood turtle and bog turtle (Glyptemys) being placed in different genera.

Actinemys marmota (left) Glyptemys muhlenbergii (top), Glyptemys insculpta (bottom)

 So, in just looking at the molecular data, the family Stephens and Wiens (2003) were investigating seems to be pretty well off. But they didn't leave it to just that data. They also looked into the morphological data as well:

Figures 4 and 5 from Stephens and Wiens (2003)

Graptemys flavimaculata

These two figures above [figs 4 and 5 from Stephens and Wiens (2003)] show the authors' results on there morphological data alone. They did use soft tissue, osteological, morphometric, and statistical data as part of this set. The use of any of those sets can be debated, and has by some people. They also use two ways of attempting to scale the data so that weighting would be, presumably, the same. I won't get into those here, but be aware of them and that they are not agreed upon fully. What can be easily seen is that the two phylogenies above do not agree with each other. The relationships shown are not the same. In the phylogeny on the left, Pseudemys comes out as the most derived member, while the one on the right has Graptemys in the same position.

Trachemys scripta elegans

I would like to focus on Trachemys though. The slider turtles are the subject of my Master's thesis at the moment, and have been the topic of much debate for quite a while now. Without getting too detailed at the moment, it is apparent that many researchers have been well aware of the problems, but no one has taken it upon themselves to completely clear it up. Dale Jackson did some work on the group in the 1970's and 1980's, but their is still far too little known about them. What can be seen is that the members of Trachemys in the phylogenies above does not come out as a cohesive or monophyletic group. Not only the genus, but, many times, the supposed species (with their respective subspecies) don't group together well either. In these instances, Trachemys comes out as either paraphyletic or polyphyletic. It can even be seen in the phylogeny on the right that the Pseudemys group is well nested within Trachemys.

This is a bit disturbing to someone researching the group. What exactly is going on? One key aspect to keep in mind is that convergence due to similar lifestyles can be a big problem. If different taxa have evolved the same or similar features to do the same things, they may be seen as more closely related then they truly are. This depends on the characters and features that people use to make their phylogenies though. The hope is that, with enough characters and features being used, the true relationships will come to light. If that is the case or not is not actually known at the moment.

Paleontologists only have the luxury of morphological data sets. Perhaps one day we will get some DNA from a dinosaur or other long extinct species, but that is not so today. They use what they can, and come up with phylogenies and relationships based on that data alone. It has been said by some researchers, including John Wiens on numerous occasions, that combining the molecular and morphological data sets is a good idea. The idea being that they will strengthen each other, assuming that they are both working towards the same end (correct?) phylogeny. It was suggested to me that, even working with fossils and modern taxa, and even only having molecular data for the modern taxa, that combining the data sets would be recommended and should strengthen all the relationships in a phylogeny of fossil and modern taxa. Stephens and Wiens (2003) did this as well with their modern taxa, combining all their data sets with the following phylogeny:

Figure 7 from Stephens and Wiens (2003)

Malaclemys terrapin

In the figure above, which shows all taxa in their study, Trachemys first appears to fall out all together. All Trachemys species appear within a "group", all their names above or below each other, with no other nested genera present. If you look more closely though, you will see that Trachemys is still not monophyletic. In this instance, they should either be split apart into 5 (yes 5!) different genera, or have other genera combined as Trachemys. This seems somewhat crazy, since most researchers have felt that the group was either fairly well set, or that they should be combined with Pseudemys and/or Chrysemys. In fact, with this grouping, it would be Graptemys and Malaclemys that should be grouped with Trachemys to make it a monophyletic group. As a note, I DON'T believe that the idea of subgenera is a good idea and/or the answer.

They also had a second phylogeny which only took into account the taxa with complete data sets (Stephens and Wiens, 2003, fig. 8), which is to the right here-------------------------------->

These groupings seem to become far nicer and follow known taxonomic groups. It is of note, though, that only a single Trachemys species is present in this table. So, is it Trachemys alone that is causing all these problems?







I am not so sure of that. It makes more sense for it to be a problem with a number of taxa within the Deirochelyinae, although Trachemys may be the biggest trouble-maker. I believe that convergence is one of the biggest problems here. Many of these turtles live in similar environments and do similar things. Because of this, they have developed many of the same features. A key area to look at may simply be what features are being used for these phylogenies. Another key area could be that more molecular data needs to come to light on this group in general.

Regardless, I have talked with Dr. Jacques Gauthier, who has informed me that this kind of concept, where the molecular and morphological data do not agree, is a relatively common occurrence. Not just with this group, or even just turtles, but many groups. This is especially true when dealing with fossils and fossil groups.

The molecular data seen above does not agree with the morphological data. In essence, because they don't agree, they are working against each other. When combining the two different sets, the one with the most data will simply win out. This is not what we are scientists are striving for. We want the two different kinds of data to agree and give us a stronger idea of the true relationships present. Sometimes this does happen and they data sets essentially agree, but all too often that is not the case.

Ray-finned fish evolution from Hurley et al. (2007), image from Dracovenator blog

So, why can't the molecular data and morphological data agree? While I don't know the exact answer, I believe that eventually they will. We just haven't gotten enough of either. We must believe that the true relationships will eventually come to light with enough good data. So keep plugging away, and we will figure these things out. Even if the organisms themselves do not care how they fit in together; Trachemys doesn't care how it relates to the Alligator that may be trying to eat him, many people, myself included, do care.

Trachemys and alligatorid

Look to the Dracovenator blog post titled: Problems in Ray-Finned Evolution to see another case of disagreement with a different group, the ray-finned fishes.


References:

Stephens, P. R. amd J. J., Wiens. 2003. Ecological diversification and phylogeny of emydid turtles. Biological Journal of the Linnean Society 79:577-610.

Hurley, I.A., Lockridge Mueller, R, Dunn, K.A.,Schmidt, Friedman, M., Ho1, R.K., Prince, V.E., Yang, Z., Thomas, M.G. and Coates, M.I. (2007)A new timescale for ray finned fish evolution. Proc. R. Soc. B 274, 489–498