Circular reasoning works because circular reasoning works (that special issue on paraphyly)

(The following is the second part of a series of posts on an Annals of the Missouri Botanical Garden special issue on “Evolutionary Systematics and Paraphyly”. All posts in this series are tagged with “that special issue”.)

The first paper is titled EvolutionarySystematics and Paraphyly: Introduction, and was written by Tod Stuessy and Elvira Hörandl, the organisers of the original IBC symposium. It can be read not only as an introduction to the topic but also a kind of summary of the whole special issue, and thus it might potentially make more sense to discuss it after the others. Also, it is densely packed with a great variety of claims, generally without developing or supporting them, because that job is quite reasonably left to the individual contributions to the special issue. Still, I will follow the sequence of papers as presumably intended by the editors.

Unfortunately, where one can discuss the central argument of more focused papers, in this case due to its nature there is hardly any alternative to going through the piece claim by claim and rebutting them individually, which makes for a less pleasant reading experience.

The first three paragraphs are meant to summarise the development of biological classification over time. The way it is presented here, the history of systematics went through the following phases: intuitive phenetics when people still believed in unchanging, divinely created kinds; intuitive combination of phenetics and phylogenetics after the discovery of common descent; quantitative phenetics with the rise of Numerical Taxonomy in the 1950ies and 1960ies; and now quantitative phylogenetics with the rise of Hennig's Phylogenetic Systematics (and of computer and DNA based phylogenetic analyses, which the abstract of the paper appears to conflate with the former).

What the authors would like to have as the next step in this meandering history is a regression to the half-phenetic, half-phylogenetic approach used before ca. 1960 but with the addition of quantitative methods. As quantitative methods for their preferred way of classifying (which strangely “have not received widespread use”) they mention something that I must admit to never having heard of, convex phenetics, but also Stuessy's own invention of patrocladistics, which is a rather idiosyncratic way of clustering by branch length distances on the phylogenetic tree in an attempt to justify the formal recognition of paraphyletic residues at the same taxonomic level as clades nested within them.

An obvious problem is that here as always the paraphylists cannot provide an objective and universal cut-off for how much branch length is acceptable, a problem that is exacerbated by the vast differences in tree lengths between different groups, different data matrices and different analyses. More importantly, of course, long branches are still and will always remain nothing more than an artefact of our ignorance of all the intermediates that did in fact exist along those branches and which may still turn up as currently undiscovered species or fossils. And already I start to sound like a broken record because I wrote the same in my previous post.

In the same section of the paper, the authors claim that “cladistics ignored divergence within lineages”, which is nonsense except in the question-begging sense that cladists won't use it to elevate a clade to the same rank as the grade around it. It can easily be used to recognise the clade at a lower rank. They further claim that “cladistic classification is too narrow because it does not reflect full dimensions of phylogeny” and “if representation of phylogeny is really the goal then we should aggressively seek quantitative means of obtaining more, rather than be satisfied with fewer, dimensions of phylogeny”, but they do not explain clearly what the dimensions of a phylogeny are supposed to be.

A phylogeny is simply a graph showing the evolutionary relationships of its terminals, and the system that best represents the phylogeny (and thus evolution) is by definition the phylogenetic system that they argue against. It is quite impossible to make an argument to the contrary without equivocating on the meaning of the word phylogeny; and that must be what is going on here, because by their own admission paraphylists do not want to represent the phylogeny faithfully but instead desire to re-introduce the criterion of similarity and dissimilarity into the practice of classification, making it less representative of phylogeny.

The next section discusses “the meaning of paraphyly”. It starts with the classic claim that “paraphyly is an evolutionary stage of monophyletic taxa”. I know that the authors have read a paper in which it has been pointed out to them that this is begging the question: The conclusion that they want to support – that paraphyletic “groups” are meaningful groups at all – is part of their assumptions, so the argument is fallacious, but they persist in using it.

Next, they define a paraphyletic group as “one that derives from a common ancestor but that does not contain all its descendents [sic] (Hennig, 1966)” and immediately start talking about paraphyletic species which, as I mentioned in my previous post, is a red flag indicating that somebody doesn't understand phylogenetic systematics. What they fail to mention is that what Hennig was concerned with are phylogenetic, that is tree-like, processes. The word paraphyletic as defined by Hennig is simply meaningless in a network structure such as existing within a sexually reproducing species, and thus a competent cladist can take the same stance towards the “paraphyly” of species as a mathematician does towards the corners of a circle. It is about what we do at higher levels, where relationships are mostly tree-like, where cladists and paraphylists differ.

The authors then make the argument that polyphyletic groups are very different from paraphyletic ones. The argument is more fully developed in a later contribution, so I will postpone its discussion except to note that the complement of terminals on a phylogenetic tree is exactly the same for paraphyletic and polyphyletic groups.

The following section is called “paraphyly as part of the evolutionary process”, following up on the question-begging claim mentioned before. “Paraphyly and holophyly [=monophyly] are alternating conditions that cycle forward and change over many generations” - I am unable to unlock the meaning of this sentence, but I suspect it is just another variant of the same:

1. We really, really want to recognise taxa based on overall similarity.
2. We map these similarity-based taxa onto a phylogenetic tree, and it turns out they are paraphyletic.
3. Look, there are paraphyletic taxa on the phylogenetic tree, right in front of you.
4. This shows that paraphyletic taxa exist in nature.
5. Therefore, we should recognise paraphyletic groups. Why are you cladists so unreasonable?

For illustration, “paraphyletic” species are mentioned another time. Argh.

This is followed by yet another way of expressing the same circular argument: “the progenitor is rendered paraphyletic”. Once more, there are no paraphyletic groups out there in nature, progenitor or otherwise. What is out there in nature is a phylogenetic tree of life on which we can circumscribe taxa to be either monophyletic or non-monophyletic. Cladists want to do the former. Paraphylists want to do the latter, yes. That and five dollars will get them a bar of chocolate. But they don't get to claim that paraphyletic groups just are. They do not exist a priori, we either decide to recognise them or not.

(Monophyletic groups, clades, on the other hand, have a much better claim to being out there in nature. They are, by definition, what a species turns into when it diversifies. In other words, a species from 50 million years ago is now a clade, and a clade of today is one species from 50 million years ago. One can coherently doubt the naturalness of clades if one is willing to doubt the objective existence of species in the sense of distinct evolutionary lineages, but although there may be some who do that I am not sure if most paraphylists are willing to go that far.)

The section ends with a mention of allopolyploidy as a process that “results in paraphyletic parental taxa”. I will freely admit that excessive hybridogenic speciation would indeed make phylogenetic systematics impossible. But it is clear that such reticulating events only happen across small phylogenetic distances. You can easily make allopolyploids inside Mentha section Mentha, but you will try in vain to produce them even between different sections of that genus let alone with other genera. Zoom out a bit and you have a tree again, and thus the problem is smaller than paraphylists would like it to be. Apart from that, if there were such excessive reticulation the result would not be paraphyly but the absence of a phylogenetic structure in which the term paraphyly makes sense. So the claim that hybridogenic speciation makes the recognition of paraphyletic taxa necessary is false either way.

The final section of the paper summarises the objectives of the symposium. There are two thoughts here that are of interest. The first is that the paraphylists see their approach as “progress”. It has never been clear to me in what sense that would be the case even if they were to prevail, because in this very paper they freely admit they would just return to pre-1960 classifications: “in a sense this represented returning to evolutionary classification prior to phenetics and cladistics, but now also in a completely quantitative framework”. (And the quantitative framework is only smoke and mirrors without an objective and universal criterion for the acceptance of paraphyletic taxa, but such a criterion is still nowhere to be found.)

The second is the complaint that some plant families are morphologically heterogeneous after things have been made monophyletic: “Plantaginaceae, for example, now contain genera with virtually no morphological resemblance to the genus Plantago.”

So what if this heterogeneity is what evolution produced?, one might be tempted to say, but the funny thing is that the example is singularly badly chosen. Plantaginaceae are superficially heterogeneous now because they were assigned genera that had previously been accommodated by the Scrophulariaceae. And that family was, in its pre-phylogenetic circumscription, one of the worst, most illogical waste basket taxa I have ever seen. Even as a first year student, before I had ever heard of phylogenetic systematics or monophyly, I leafed through a flora of Germany and wondered how addled the taxonomist must have been who thought that Scrophularia, Verbascum, the parasitic Rhinanthoideae, Calceolaria, Veronica and Antirrhinum should be in the same family – but the parasitic Orobanchaceae separate despite obviously being the closest relatives of and most similar to the Rhinanthoideae.

Sorry, but the new, phylogenetic classification makes way more sense even from the perspective of morphological homogeneity. The heterogeneity of the current Plantaginaceae has nothing on the heterogeneity of the old Scrophulariaceae.

As some of the next few posts will show this is not the only time that paraphylist arguments fail even by their own standards.

Reference

Stuessy T, Hörandl E, 2014. Evolutionary systematics and paraphyly: Introduction. Annals of the Missouri Botanical Garden 100: 2-5.