Tuesday, January 15, 2013

Ranunculus, part 3: Morphological versus molecular data

This is part of a series on a paper by Hörandl & Emadzade (subsequently H&E) suggesting an "evolutionary", i.e. pre-cladistic, classification of Ranunculus. See the previous installments here: part 1, part 2.

Continuing with the introduction, the second paragraph starts as follows:
Evolutionary classification further attempts to integrate information on structure and function into the taxonomic concept (Stuessy, 1987; Mayr and Bock, 2002; Hörandl, 2007, 2010; Stuessy and König, 2008).
As a counterpoint, I strongly suggest to read Hennig (1975). Dealing with the same claim then advanced by Ernst Mayr, he demonstrated handily that trying to accommodate two very different criteria of classification in the same system will only result in confusion. In the end, the system is completely useless because you cannot even consistently read one of the two stories - common descent or phenological similarity - out of it.

The remainder of the second paragraph introduces the issue of morphological characters. After stating that morphological data contains phylogenetically useful information, the authors raise the problem of homoplasy arising from parallel evolution or reversal. One could mention here that DNA characters, with only four different character states at each position, and considering that at least coding regions will also be under selective pressure, are not entirely free from the same problems either, but okay.

Staying with the topic of morphological data, H&E examine the different ways in which it is used.

Even at the risk of appearing like a bean counter, I have to say that I found it fairly unfortunate that throughout most of the relevant paragraph the authors use "character" as a shorthand for "morphological character", as if molecular characters aren't. And note that this happens not after a sentence that makes clear that we are reading a discussion of morphological data, but after a sentence that explicitly mentions both of them in the same breath. At a minimum, the section invites misunderstanding.

H&E list three ways in which morphological data are used in "taxonomic literature" (I assume here they mean phylogenetic studies instead of actual taxonomic literature like floras, monographs and descriptions of new species):
  1. Phylogenetic analyses of morphological data alone.
  2. Phylogenetic analyses of molecular data alone, with morphological data mapped onto the molecular phylogeny.
  3. Phylogenetic analyses of combined morphological and molecular data.
Which of these approaches is to be preferred, if any, is a fairly interesting point even quite apart from any consideration of classification. As a colleague recently argued in our journal club, the problem with #1 is potentially that any subsequent interpretation is somewhat circular. If, on the other hand, you use molecular data and then map morphology onto it (#2), the first provides an independent backbone on which you can interpret the changes in the latter, but some people don't like the idea of excluding morphology from the inference of the phylogeny. And as H&E themselves mention in the present paper, #3 has recently proved very successful: adding morphological data to a molecular dataset can apparently sometimes achieve much better resolution especially at very deep nodes, as I have learned last year when I heard a talk on lady beetle phylogeny. Food for thought, and those colleagues who never even consider using morphological data surely don't know what they are missing.

However, H&E mostly use this section to sow, as they see it, doubts about the current practice of phylogenetics. First, they mention the potential of reticulate evolution to increase homoplasy in morphological characters, then they make a rather hard to follow argument about morphological characters "differentiating" earlier or later than the molecular marker (?), and finally they point out that datasets can only be combined if they are congruent. It is not clear to me in what sense the first and the third are anything but universal problems that apply with the same force to any set of molecular markers, and the second, to the degree that I understand it, should be trivially resolved by selecting different or more molecular markers in the future.

Having now discussed the major addition the paper makes when compared to the authors' previous publications, i.e. morphological data (the molecular datasets had already been published before), H&E take a first stab at describing the aim of the study. As mentioned in my first post, it is to present a case study of "evolutionary" classification, which as they argue has rarely been done. To an extent, they are correct: The controversy of recent years in systematic botany is big on misguided criticism of phylogenetic systematics but small on developing a viable alternative.

But there are good reasons why that is the case. Unless evolutionary systematists can provide an objective, universal and testable (i.e. scientific) criterion for the circumscription of paraphyletic taxa, their school will remain nothing more than an attempt at turning the wheel back to the time of Haeckel, only that they would now have better phylogenies on which to circumscribe paraphyletic non-groups as artificial taxa. Consequently, every not explicitly phenetic classification from the decades before phylogenetics came to dominate systematics was a case study in "evolutionary" classification. The colleagues at that time already tried to take common ancestry into account as far as they could infer it based on fossils, morphology and a dose of intuition. Nonetheless the authors write that "the great majority of taxonomists still follow cladistic principles" instead of "the great majority of taxonomists have moved beyond pre-cladistic principles".

Most of the remainder of the introduction introduces the study group, the buttercup genus Ranunculus, and there is not much to say here with regard to the questions that interest me, i.e. whether the paper provides any reasonable justification for paraphyletic taxa or the aforementioned scientific criterion for their circumscription. However, before moving on to the materials and methods I would like to draw attention to the last full paragraph on the second page, where H&E list a number of "evolutionary processes" that are believed to be at work in Ranunculus. One has to know Hörandl's previous publications on the matter to realize that this is not merely some additional information on the study group but indeed part of her case for paraphyly: these processes supposedly cannot be captured sufficiently well in a phylogenetic classification, therefore paraphyletic taxa.

Still, the processes are simply the same already mentioned in an earlier part of the introduction: polyploidy, which is not a problem for phylogenetic systematics except in the case of allopolyploidy, and even then only if it is rampant and involves very distant parental species; hybridization, which the authors leave ill defined and which is not a problem except in the case of allopolyploidy, see previous item; and anagenesis, which is simply not a problem full stop. Next, the materials and methods section.

Continue reading.

No comments:

Post a Comment