Thursday, August 3, 2017

Basal and transitional taxa

Shortly before I left for China I received an alert on an interesting paper:

Bronzati M, 2017. Should the terms 'basal taxon' and 'transitional taxon' be extinguished from cladistic studies with extinct organisms? Palaeontologia Electronica 20.2.3E: 1-12.

As can be expected from this title, Bronzati argues that the terms are misleading and confusing, and that they should not be used. I find myself tending to disagree, at least in part, and not only because of an allergic reaction to being told what words I am not supposed to use because it might confuse 'the public' (cf. free will debate). Before I go over the arguments, however, I would like to clarify where I agree:

First, there are clearly cases where it would be desirable not to use a concept or term because it is really wrong or incoherent, and in some cases even because it is misleading. At the recent conference I flinched at a speaker who said "this individual is paraphyletic". Although I understand what they meant (the utterly trivial and commonplace observation that an individual had two different alleles at a gene locus they had sequenced) such a sentence is Not Even Wrong and has to be based on confusion about, well, pretty much everything that matters in molecular and phylogenetic systematics beyond perhaps how to hold a pipette the right way and click "run analysis" in a few programs. But it is not necessarily the case that the terms basal and transitional suffer from the same problems.

Second, I obviously agree that supraspecific taxa should be monophyletic.

Third, I also agree that evolution is not teleological (with a caveat I will go into below) and that terms such as primitive or advanced are to be avoided, in particular when talking about organisms that live(d) in the same time-slice. And in fact there are very few people left who still think that e.g. mosses are primitive compared to seed plants. Both lineages as they exist today have evolved for precisely the same time. The mosses are certainly not more primitive as mosses than seed plants would be as mosses, they just went completely elsewhere in terms of morphospace and adaptive peaks.

Evolution is a story not of progress but of diversification, and it only looks to us as if there was progress from morphospace position A to position B because life necessarily had to start in some position, and even after a pure random walk some extant organism may still (or again) occupy that starting position or something close to it. A good analogy I once read is to imagine a bunch of people all starting in front of a wall and then milling about aimlessly. Although their movement is random the group will still expand in one direction, away from the wall, because they cannot go in the opposite direction; conversely then, the fact that they are now further away from the wall does not mean that they meant to move in that direction specifically.

It is consequently important to keep in mind the "studies with extinct organisms" part of the paper title, because on the question of sorting extant organisms into a ladder of progress all competent evolutionary biologists are agreed anyway. Okay, but what now of extinct taxa, which had in their time not yet undergone the same amount of evolution as the taxa we have today? Are they basal or transitional to the latter?

Bronzati starts by examining whether basal taxa are those that are older than the non-basal ones and observes that fossil ages do not necessarily reflect the ages of the lineages they belong to. He suggests instead to use "'early' and 'late' in an explicit comparative framework". That is very clear, but I do not think that this is how the word basal is meant by most people anyway, and it is certainly not how I would use it. As Bronzati soon observes himself, "'basal' is a relative term regarding the base of the tree" and thus refers to the relative age of lineages, not to the age of fossils.

I am not quite sure I understand the next part, where he writes that "different people certainly have different assumptions of what a 'basal' taxon is" and discusses whether something outside of clade A can be a basal A or not. I'd say not, but again, I think basal is a relative term along a tree topology and not something that I would use in this way.

Now Bronzati turns to the question that I consider the most relevant: "Basal taxa are closer to the root" - precisely that is how I understand the word - "but how to measure it?" But this is also where I think the argumentation becomes a bit odd, because he argues against the use of the term by comparing apples and oranges, and then throwing incomplete sampling into the mix. This will now need an illustration. Consider the following phylogeny:

Bronzati argues against the use of 'basal' by looking at species A, which people would supposedly (?) consider to be basal because they read the tree like a ladder from left to right, and then observing that this species is actually more distant from the root in terms of internal tree nodes than species B. I hope the problem is immediately obvious: species A is not the unit we would be talking about when saying "more basal than B". Would anybody ever actually say that A is basal in the tree? It is clearly fairly nested. Instead, the only use of such terminology that makes sense would be to say that the entire genus Ales (the red box) is basal in the entire family also consisting of the other genera Beles, Celes and Deles (the other three boxes), or more basal than Beles.

And this is where I am willing to be convinced otherwise but at the moment happy to continue using the term basal: if and only if we are talking about the branching order along a phylogeny backbone, along a grade. I will be the first to agree that all supraspecific ranks are arbitrary, but we also have to appreciate that we are using them, or alternatively unranked clade names, nonetheless. This is not so much about evolutionary theory as about having at our disposal non-atrocious language to describe a tree topology. When talking about these genera, what is so problematic about saying "Ales is basal in its family" compared with "Ales is sister to the rest of its family"? At least in my eyes the two statements are equivalent and neither is more misleading than the other. Making it about species A feels like a red herring.

And this is then also all that needs to be said about sampling, because it is based on a similar argument. Bronzati describes a hypothetical tree of all dinosaurs with all of the huge bird clade represented only by the chicken and then jokes he "would hope that no one would suggest that the bird is a basal dinosaur ... based on the number of intervening nodes to the root". No, I don't think anybody would. But maybe we would say the birds (!) are. If, hypothetically, part of the topology were (birds, (dinos2, (dinos3, dinos4) ) ) then yes, I would not have any problem saying that the birds as a whole are more basal in the tree than that other named clade dinos2, for example, because the birds as a whole are quite simply branching off one more inclusive ancestor closer to the root than dinos2.

What is really puzzling to me is that Bronzati himself makes the same point two paragraphs later: "it is not terminal taxa (...) that can be 'more basal' in relation to other terminal taxa, but the nodes (i.e. hypothetical ancestors) of the tree in relation to other nodes".

Concluding his discussion of basal 'basal', Bronzati examines the question whether basal taxa have more plesiomorphic traits and concludes no, but again this is based on considering in isolation a very derived descendant of the entire clade I would call 'basal'.

He then turns to the term 'transitional'. Here he appears to make two main arguments against its use. First, that evolution has no goal, and second, that phylogenetic trees are branching diagrams instead of ladders.

I have already mentioned above that I agree completely that evolution is non-teleological, but with one caveat, which is this: lineages may discover, for the first time, a new peak in the adaptive landscape, and when that happens we can expect them to evolve up that peak, so that earlier forms would be more poorly adapted to the new situation than their later descendants. Bronzati himself mentions the colonisation of dry land, focusing of course on vertebrates, his specialty. Using the group that I am more familiar with as an example, it seems clear that the early vascular plants started out without roots, and that the lineages that descended from them evolved roots because having those was a pretty good idea on dry land. In fact there are none left that are primarily without roots, presumably because they were out-competed (although there are a few secondary losses under unusual circumstances, e.g. Cuscuta).

I would argue that this, and only this, and only along a time axis, is where we can perhaps meaningfully speak of primitive and advanced, but that is not even the point here, because the term we are dealing with is transitional. More important seems the second argument. Yes, phylogenetic trees are branching diagrams, but they do not merely consist of terminals, they also consist of hypothetical ancestors. It is a bit unclear to me where Bronzati stands on the question of those; on page 6 of his paper, as mentioned, he talks about hypothetical ancestors himself, but here he spends considerable time arguing in a way that suggests that he does not want to identify actual species or fossils as ancestral:
It is important to stress that the absence of autapomorphies in taxa [sic] B does not indicate that it is transitional between A and C-F. Firstly, this might be just a reflex [sic] of the lack of ability to translate different morphologies into phylogenetic characters. Furthermore, the study of living species shows us that even if there is no recognisable morphological difference between [sic], they can differ at the genetic level.
Of course they can, but remains unclear to me what should keep us from tentatively concluding that some fossil may represent an ancestor until we get additional evidence that shows otherwise, just like pretty much every other conclusion in science is also tentative. And if we have a presumed ancestor we can say that it is transitional between an even earlier presumed ancestor and descendants further down the line. There is no teleology involved here, but the internal nodes of a phylogeny can indeed be read as a ladder of ancestor-descendant relationships.

I am sorry to say I just don't see the problem here either.

Bronzati ends with making four recommendations:

Tree toplogies should be described with sister-group statements, avoiding terms like basal or early diverging. My concern is that this will lead to very ugly and repetitive language when describing anything but a very small phylogeny: "our results indicate that A is sister to the rest of the study group. B is then sister to the rest of that rest, and then C is sister to the rest of that rest we just mentioned; now D is sister to the rest of that last rest ...", and so on for another four clades. That is just not very aesthetic. So why not a much more concise "the earliest diverging lineage is A, followed by B, C, and D"?

Instead of calling a terminal taxon a basal member of clade A, we should say it is a non-A member of the next larger named clade around it, as in non-avian dinosaurs. That makes sense, but again, I would never have used basal for a deeply nested terminal anyway but only to discuss the relative position of several clades along a grade.

We should say "this taxon fills a gap in the fossil record" instead of "this taxon is transitional". As mentioned above, I don't see it, perhaps because I have a different approach to internal nodes and species without autapomorphies.

Finally, we should avoid teleological language. No disagreement from me on this one!

No comments:

Post a Comment