Aubert's analysis of phylogenetic terminology, part 2: definitions

Continuing the discussion of this paper from here. Originally I had the idea of going through the pages in sequence, but it may be more productive to tackle one by one what appear to be the main claims of the paper as I see them:

• The various definitions provided in the paper are in some way better than the ones that are currently accepted.

• There is no relevant difference between the systematics-relevant relationships and structures existing at any level of the diversity of life. (E.g. mother > daughter is completely equivalent to bony fish > land animals - they can all be drawn as diamonds and arrows, right?)

• A strictly phylogenetic classification is formally impossible.

• Cladism is part of structuralism and therefore characterised by "anti-realism and a metaphysical way of thinking".

• Cladism is built on biologically unrealistic assumptions that have been empirically falsified.

• There exists an objective approach to delimiting paraphyletic groups.

• It would be preferable to have two parallel classifications, one of clades and one that includes taxa that are allowed to be non-monophyletic.

So this might turn into seven separate posts, although the last few will perhaps be short. This is the first one, dealing with the definitions, and it got rather long.


The definitions themselves

About the first two thirds or so of the present paper consist of a mixture of "lemmas", "definitions" and "observations" interspersed with helpful figures. As mentioned before, the latter consist of diamonds connected by arrows and serve to illustrate the definitions. To get a flavour of the text, consider this example:

This is accompanied by a figure showing diamonds connected by arrows and a group of diamonds marked to illustrate the concept of "lineage". Here is an alternative way of providing the same information:

A lineage is a line of descendants without gaps.

This alternative is (a) shorter, (b) easier to understand, and (c) fully equivalent in the sense of not missing a iota of information or precision compared to the original. The question is then why one would choose to express what boils down to perhaps two to three pages of really simple, one-sentence-ideas with thirty pages of lemmas, indexed place holders, set theory and suchlike.

Perhaps this is a bona fide attempt to be 'precise', but a reader who suspected that it represents an attempt to intimidate and bamboozle the reader with math, to make them think that something much more profound is going on than there is, could at least not easily be called unreasonable. Curiously, Aubert himself asks in a later part of his paper "whether the criticisms against [sic] classical evolutionary systematics do not come from an illusion of precision due to the use of mathematical tools in cladistic analysis". (Pot, kettle, black?)

Anyway, some of the definitions are trivial or irrelevant and clearly only included for the sake of completeness, while others are directly relevant to the controversy around whether partial clades should be formally accepted as taxa. Concentrating on the latter:

Lineage = As mentioned above, a line of descent without gaps. Aubert illustrates it in a way that strongly implies a linear, unbranched shape, with only one item per generation. I find this odd because to me such a lineage would be incomplete, but I guess that just means that I am a cladist. What is even more unusual, however, is that Aubert also defines the term in a way that it is satisfied with a single individual. Surely that does not fit the way this word is generally understood.

Ancestor =  An item is an ancestor of another item if they are connected by a lineage. This is soon followed by an interesting "observation": An ancestor is apparently also its own ancestor, at least if I interpret the wording correctly. This is called at the same time "counter-intuitive" (true), "trivial" (not to me, perhaps because I do not believe that a single item is a lineage), and "greatly facilitat[ing] the formulation of many properties thereafter". It is nice that Aubert will find this to make his definitional jiu jitsu easier, but that does not necessarily mean that it makes biological sense. I mean, if I define rock as cheese I will also find it easy to prove that the moon is made of cheese, but most people would quite sensibly take issue with my definition.

Group = In Aubert's usage, any random assemblage of items in his graphs that he draws a line around. A cladist would, of course, consider anything non-monophyletic to not really be a group in any useful or meaningful sense. Often in this context people are asked to consider similarly nonsensical non-groups outside of the tree of life but likewise defined only by lacking a trait such as non-Hindus or shirts that are not white.

'Monophyletic' group = A group that contains its own common ancestor. Clearly here the currently accepted definition in the vast majority of scientific journals, conference talks, textbooks and university courses is a different one. I will subsequently scare quote this definition, so if the quote marks are not there I mean the widely accepted Hennigian definition.

At this point Aubert worries about this definition allowing groups that contain items and their common ancestor but not the intermediate ancestors, so he consequently clarifies:

Continuous 'monophyletic' group = A group that contains its own common ancestor, and all of whose members are are connected to the common ancestor by descent through other members

Ancestral group = A 'monophyletic' group containing a common ancestor of another group. We are now really, really, really deep in begging the question territory as far as the controversy between cladists and their opponents is concerned. Not only is such a group not really a group in any meaningful sense, most of it isn't ancestral to the subclade Aubert would exclude from the 'group' in the first place.

Directly ancestral group = The same but the ancestor-descendant relationship between the ancestral item in the so-called 'ancestral group' and the common ancestor of the other group is direct, without intermediaries.

Exclusively ancestral group = The descendant group has only this one ancestral group, i.e. it is not derived from a cross between two different ancestral groups. Here Aubert seems particularly concerned with endosymbiosis. I will never understand why so many people believe that this is relevant. The chloroplasts are a monophyletic group that has colonised eukaryotes, and the plants are a monophyletic group of eukaryotes that has been colonised by a clade of cyanobacteria. This appears to be another case of conflation, in this case of systematics-relevant lines of descent with systematics-irrelevant horizontal gene flow. One could just as well claim that a human needs a new family name after getting a heart transplant. Curiously, once more there is a place in the paper where Aubert warns against precisely that kind of conflation himself.

A mere four pages after 'monophyletic' we come to paraphyletic, polyphyletic, and holophyletic, the former two defined the usual way, the last the suggested alternative to what nearly everybody today calls monophyletic. Aubert now provides two 'utilitarian' arguments for changing monophyletic to holophyletic.

First, if the paraphylists cannot use the word monophyletic to mean either monophyletic or paraphyletic, they "[have] no other term to account for their concept" and "cannot assert their point of view". From a cladist perspective, that is a feature, not a bug, because they consider the concept unhelpful and the point of view wrong. Again: like non-Catholics, not really a group, etc. One wonders also, given how frequently paraphylists make up new terms, why they could not just have made one up for this eventuality.

Second, the either monophyletic or paraphyletic meaning would be "useful in studies of unrooted phylogenetic trees where precisely [sic] it is de facto impossible to distinguish holophyly and paraphyly". Maybe there are dozens of people out there who have that issue, but I am not aware of a single situation like that. Towards the end of the relevant section Aubert mentions that somebody has already made up a new word for that situation (clan), thus undermining his own utilitarian argument.

Heterophyletic = What the rest of the world calls non-monophyletic.

Canonical holophyletic group = Associated with a paraphyletic group, what it would turn into if all its missing sublineages were added.

Complementary group = If I understand correctly, the subclades left out of a clade to circumscribe a paraphyletic group.

Degree of paraphyly = Number of subclades that had to be left out of a clade to circumscribe a paraphyletic group. A similar degree of polyphyly is introduced later.

On page 19 Aubert defines holoclady and heteroclady to solve the same imaginary problem in exactly the same way as Podani and Vanderlaan et al. before him. From a cladist paradigm, the question whether a valid group includes only extant terminals or goes right down to the ancestor is a complete non-issue; it is only if one desperately wants to accept paraphyletic taxa that it even arises. What is more, we now have three sets of terms for the same thing. Decrease the confusion? Improve communication? I wish.

It is interesting to note that Aubert comes perilously to a cladist view when discussing monophyletic, paraphyletic and polyphyletic groups of terminals:

This means that a holocladic pattern suggests the existence of a real situation of holophyly, whereas a heterocladic pattern only indicates a lack of holophyly: it is not then possible to formally decide between a real situation of paraphyly or polyphyly.

Translated: para- and polyphyletic groups have the same shape on a tree, while monophyletic groups are significantly different from both. Yes. That is one of the cladist arguments for accepting only the latter, and for considering para- and monophyletic so different that one shouldn't lump them into a term like 'monophyletic' sensu Aubert.

I must admit that I got somewhat lost around the definitions of stem group, crown group and basal group. Stem and crown appear to be used in the same way as by everybody else, although there is a strange excurs arguing that because people find it useful to have a word for the branch between the lineage split that created a clade and the first surviving lineage split in the clade we should formally accept paraphyletic taxa. This does not follow, to say the least. But 'basal group' seems to be defined as all species of the clade that have already produced other descendant species. As Aubert believes that species can split off other species and still stay the same species (see Composite Species Concept) and thus be terminals themselves, the utility of this term is unclear to me.

Pages 26-27 reintroduce another previously seen paraphylist attempt at defining oneself to victory: Cladists do not classify, they only arrange, because a classification is about affinity, and of course the only meaning of affinity that Aubert accepts is phenetic similarity. Being related is not an affinity, apparently. Also, classifications should contain "similar objects". Like species, for example, which just happen to be the basal units of phylogenetic systematics?

The definitions peter out with 'cladogram', and are followed by a more structured discussion of history and philosophy of science.

So, are Aubert's definitions preferable to the currently accepted ones?

The real question is then, if the definitions provided by Aubert are supposed to be better, better in what sense? Explicitly the abstract claims:

First, that the current terminology "prevents proper communication between the proponents of either side". This is a red herring, because any set of definitions facilitates communications as soon as it is universally accepted. As the people using Haeckel's meaning of monophyly, for example, are clearly in the minority, the best course of action for somebody genuinely concerned about communication would be to tell them to use Hennig's meaning instead.

Second, that "consequently, the research in phylogenetics is globally erratic and the taxonomic classification is highly unstable." I have already given the only possible answer: Instability arises necessarily from progress in our scientific understanding of the diversity of life, regardless of what terminology we use. Classifications changed before cladism, only there was no clear criterion to settle the issue once enough data are in.

The rest of the paper supplies two additional claims, sometimes implicitly.

Third, that Aubert's definitions are to be preferred for 'morphosemantic' reasons. It is true that monophyletic is perhaps not ideal from that perspective compared to holophyletic, because the latter does stress the inclusion of all descendants of the common ancestor.

However, at least in my eyes such arguments are handily trumped by the practical consideration of minimising the confusion and disruption that would be caused by changing a widely accepted meaning. It could also be pointed out that Aubert himself is not consistent in that regard: hetero- means different, so to use heterophyletic as a synonym for what is now called non-monophyletic (~non-holophyletic sensu Aubert) seems odd from a 'morphosemantic' perspective. By his own logic, it should probably be a synonym for polyphyletic.

Fourth, that the new definitions and terms capture something or clarify or add precision to something in a way that had previously been overlooked. This and the first one (to facilitate communication) are also what I personally would consider to be the two legitimate reasons for any attempt at altering a terminology.

As should be clear from the above, I don't think they do. Several of them are clearly instances of circular argumentation, like lineage or ancestral group; more on that perhaps in subsequent posts dealing with the supposed impossibility of phylogenetic classifications etc. Others lump very different biological realities into one term, such as 'monophyletic' sensu Aubert, so they do the opposite of capturing something about nature as it is. Yet others would only lead to more confusion, such as again redefining 'monophyletic' or the new words for describing group shapes in synchronous classifications where others already exist. Many simply appear to be irrelevant.

Most importantly, however, it is all besides the point. No matter how much definitions are shuffled around, the real questions are on the lines of what classification would be most predictive, most useful, most natural, most objective, and so on. The real question is not whether I can write down "rock is a kind of cheese" and get it published somewhere, but whether I can demonstrate that the moon is indeed made from milk products.

As far as I can tell, there is essentially one actual argument in favour of paraphyletic taxa in the 54 page paper (spoiler: it is the same as Brummitt's). Of course, one argument would be enough if it was sound. But enough for now, this has got long enough.