Monday, February 3, 2014

Two possible meanings of "pseudoextinction"

Continuing to read Richard Zander's recently published Framework for Post-Phylogenetic Systematics.

I had originally intended to discuss general concepts instead of criticizing the book directly, but it probably cannot be avoided entirely; in the present case because I need to lead up to what I would like to discuss afterwards.

Repeatedly the book makes the argument that Hennig's internodal species concept is obviously wrong. To recapitulate, it is the idea that a species can be considered as existing through time between two lineage splits. On the tree of life, a species originates in a speciation event and ends at the next speciation event. Even if one of the products of that next split is morphologically and ecologically indistinguishable from it, it is still considered a new species. Because the ancestral species ends by definition at the split but is not really extinct (it survived, after all, in the form of both descendent species), it is sometimes called pseudoextinct.

Another concept of species on the tree of life is the composite species concept. It sees one species as occupying all internodes of the tree that have the same combination of (morphological, ecological, etc.) characters. Here, a species starts with the acquisition of a character and ends later when another change is fixed in its lineage. This process - of one lineage turning into what is afterwards seen as another species because of the fixation of a novel character - is often called anagenesis. But we have a similar situation here where a species is considered to end by definition but is not really extinct (it survived, after all, in the form of one descendent species with a different character combination), and so we could also consider it to be pseudoextinct.

On the other hand, under the composite species concept a species continues through a lineage split if only one of the two resulting lineages acquires a new character, an event that is sometimes called speciation through "budding". The one that has a new character is now a new species but the other lineage, the one that looks identical to the common ancestor, is considered to be the common ancestor. An example would be a widespread breeding group remaining unchanged while "budding off" a small isolated population that accumulates changes until it cannot interbreed with the others any more.

I do not want to turn this into a defence of the internodal species concept, but we should realize at this moment that there are at least two problems with the composite alternative. First, it simply seems absurd to consider only one of the two descendent lineages to be the ancestor. In reality, the ancestor has not survived as one of the descendants but as both of them. Second, the composite species concept comes with all the problems of any typological concept, in particular the question of what precisely counts as enough character difference to justify the conclusion that anagenesis has taken place.

Anyway, now that we have recapitulated the necessary background information, here is one of the various instances of Zander's criticism of Hennig's internodal species concept or pseudoextinction of the ancestral species (p. 41):
Cladists put their trust in anagenetic change that "disappears" the ancestor. One should, however, look for stasis first, because if it is there (and fits theory by biogeography, relative age of habitat, relatively generalized morphology, maybe even fossils), then postulating unknown, unnamed, and ad hoc shared taxa is not parsimonious. Deciding that pseudoextinction must have been the evolutionary  mechanism for cladogram splits because this allows sister-group generation by maximum parsimony put the cart before the horse.
More formally, we could clarify the argument as follows:

Premise 1: Because of Hennig's internodal species concept, cladists assume that all ancestral species are pseudoextinct.

Premise 2: Because they do not have the names of extant species assigned to internal nodes, cladograms show all ancestral species as pseudoextinct.

Premise 3: In reality, speciation very often happens through budding, with one of the descendent lineages indistinguishable from the ancestor (stasis). Speciation events at which both descendant lineages immediately undergo anagenesis after separation are probably very rare.

Conclusion (from 1-3): Cladism is built on false assumptions and a bad methodology.

There are several problems with this argumentation, specifically with the first two premises and with the implication that the third premise is somehow not shared by cladists.

For the first one, suffice to say that some cladists accept the composite species concept and ideas such as budding, and that many more don't care either way (and that is saying it gently). Hennig's pseudoextincion does not play any significant role in the daily practice of phylogeneticists who want to figure out whether this or that genus is a natural group or not.

The second premise appears confused about what cladograms and phylogenetic trees more generally are for. Their point is to show the branching order on the tree of life as well as we can infer it at the time. That's it. Once we have it, we can subsequently reconstruct ancestral characters, apply the composite species concept if we so desire, and then claim that this or that terminal is the same (composite) species as existed across this or that internal node of the tree. Alternatively we can apply the internodal species concept if we so desire and consider all internal nodes to have been the end points of pseudoextinct ancestral species. Both are potential interpretations of the tree, but the tree itself is just meant to show branching order, nothing else. It is silent on the question of pseudoextinction or asynchronous species concepts.

The third premise, however, is the one where the greatest confusion reigns. Again, Zander says that cladists assume speciation events to usually result in two lineages that are morphologically and ecologically different from the common ancestor because only then does it make sense to consider them to be sister lineages instead of ancestor-descendant pairs. He then goes on to (correctly, I would wager) point out that in reality most speciation results in one lineage that is different from and one that is similar to the common ancestor. Ergo, cladists are wrong.

The problem with this can be phrased in more than one way. One could point out that Zander completely misrepresents the cladist position. No, they do not assume that speciation events usually result in two lineages both morphologically different from the ancestor. The entire point of Hennig's internodal species concept is, after all, to consider both descendants new species EVEN IF one of them is indistinguishable from the ancestor, so cladists are well aware that that will often be the case. No, it does not make sense to treat the two lineages as sister groups only if they are both morphologically different from the ancestor. It makes sense in any case because they quite simply are each others sister lineages, and it also makes sense because the ancestor hasn't turned into one of them while the other poofed into existence out of nowhere. It has turned into both of them.

Another way of looking at the problem is that Zander is equivocating between two different meanings of pseudoextinction. There is pseudoextinction (Hennig) which means treating species as ending in a lineage split regardless of morphology, and pseudoextinction (anagenesis) which means treating species as ending only when they change morphologically. Those are two completely different concepts forced into the same name. It should be clear that one cannot refute the first by pointing out that the second rarely happens.

Worse, one cannot refute pseudoextinction (Hennig) anyway because it is simply a definitional convenience. It is not an empirically testable claim. It does not depend on any observations beyond those that Zander would share (i.e. common descent). It cannot be "wrong" because it is just a definition. The best one could do would be to show that the concept is less useful a definitional convenience than the composite species concept, and I have already mentioned above that the latter is not that unproblematic itself.


On a tangentially related note, among the many things that I do not understand about the Framework is why Zander wants the interior nodes of the phylogeny to be assigned to existing species, as opposed to the internodes.

Remember, a tree consists of two types of elements: internodes (branches) and nodes (the points where the branches connect). Logically, a branch on a species tree would be interpreted as one ancestral species evolving through time, and so it makes sense under the composite species concept to assign some of the internal internodes to known species.

But the nodes, as desired by Richard Zander? In a way they are quite ill-defined: they are the places where two or three species meet (composite and internodal concepts, respectively). They are not really one species, they are two or three at the same time.


  1. Alex:
    I appreciate your serious consideration of my proposals. Note that I did not simply complain about the faults of cladism. There are lots of publications by us paraphyly-huggers pointing out such problems . The challenge for anyone is to propose a workable substitute for what is not working. I presented a substitute method of doing evolutionary systematics using the (few) positive elements of morphological and molecular phylogenetics but also incorporating evolution-informative but not sister-group-informative information.

  2. Thanks for your kind comment.

    Unfortunately, we will have to agree to disagree about the main issue because I simply fail to see what is not working about phylogenetics. The biggest problem you perceive appears to be that we cannot describe serial evolutionary transformations or ancestor-descendant relationships where the ancestor is a contemporary group of organisms.

    To me, that is not a problem but working as intended. First, because I do not think that evolutionary biology is necessarily about what you consider it to be about, although that admittedly is more of a definitional squabble. Second, because I hold the idea of an ancestor being contemporaneous with its descendants to be self-evidently absurd; the real ancestor existed before our time and then turned into all members of the clade, not only some of them. It is hard for me to imagine how this observation could be addressed except by begging the question.

    More generally, I have read dozens of papers arguing for paraphyletic taxa in botany plus a few by zoologists and the only half way convincing argument I came across was (sorry to say) Brummitt's. And that one works only if one accepts several assumptions that are at a minimum dubious (we can be sure that a fossil is an ancestor instead of a side branch; keeping Linnaean ranks is more important than an accurate description of nature, etc.).

    So it is not as if I did not do my homework, but I still fail to see what is wrong about cladism.