Monday, January 11, 2016

Aubert's analysis of phylogenetic terminology, part 5: is cladism anti-realist?

Continuing the discussion of this paper from here, here, here, and here, and working through 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.

What assumptions is cladism based on?

The section of Aubert's paper most relevant to the claim of cladism being anti-realist and metaphysical is #10, Analysis of the Cladist Doctrine. But before we come to the supposed structuralist influence, it is interesting to examine how the cladist "doctrine" is described:
Cladism as it was founded by Hennig
Okay, let's stop right here for a moment. Aubert wants to recognise paraphyletic taxa, and mainstream systematics today recognises only monophyletic ones. Question is now, are we talking about cladism as founded by Hennig, or are we talking about Phylogenetic Systematics as practised today, by hundreds of people publishing papers to circumscribe groups as monophyletic? Because finding a mistake made by Hennig personally will not necessarily address the underlying disagreement with all the latter, who still want to make taxa monophyletic even as they may or may not agree with Hennig's species concept and phylogenetic methodology.

This is a bad start, comparable to a creationist rejecting modern evolutionary biology because of Darwin's now outdated ideas on inheritance. Let's see where this is going, but it is important to realise that today's phylogenetic systematists do not treat Willi Hennig like an infallible prophet. Practises have changed since 1966.

is based on three key assumptions. The first is that cladograms' nodes do represent real speciation events through splitting (Hennig 1966; see Figure 20C).
Cladogram nodes simply represent speciation events full stop. But whether that speciation was through splitting or budding is irrelevant for a cladogram representation of the tree of life (although it would not be for a phylogram representation). One suspects that some conflation between at least three different issues is going on here: (1) how speciation is assumed to work, (2) how to display the tree of life, as a cladogram, phylogram or whatever, and (3) how to classify organisms into species.
The second is that it is extremely unlikely, given the continual transformation of the living world and the vagaries of fossilization, that a species we know is the ancestor of another one. Even if by chance this ever happens, it would be anyway impossible to prove this parental link.
The point of Hennig's Internodal Species Concept is not that all speciation events are assumed to be splitting, but that the descendant species should both be treated as individually different from the ancestral species even if speciation was through budding, as should be unmistakably clear from reading his book. This is not about identity of ancestor and one of two descendants being "unlikely"; it is about such identity being nonsensical.

The chimpanzees are not our ancestors from 7 Myr ago, they are our sister lineage. The chimpanzees of today cannot be our ancestors, as they exist at the same time as we do and not 7 Myr ago. In reality, the ancestral species has turned into both humans and chimps and is thus identical to both together. Same for all other lineage splits. So arguing that many speciation events are really budding is missing the point.

What is more, there are even cladists who would agree with Aubert's species concept. I do not necessarily share their point of view, but they are still cladists and can still recognise monophyletic supraspecific taxa.
The third, more implicit, is that the evolution of species is occurring at a more or less constant rate (uniformitarianism, not to be confused with actualism). This would imply that the more recent is the divergence between two species, the more similar these two species are. Clades would therefore be the kind of groups with the maximum content in phenetic informations.
Cladism is not about maximising phenetic information content; phenetics is. That is why the latter is called phenetics and the former isn't, just in case that wasn't clear. Cladism is about reflecting relatedness, evolution, common descent. Thus all the rest is irrelevant. It is really interesting how often the text of the present paper not only assumes that phenetic similarity is the only criterion we should ever care about and argues from there, which is problematic enough, but also that it is already the only one that everybody does care about. That is just not the case.

Association fallacy

This misunderstanding of phylogenetic systematics does not bode well for the rest of the section. It could also be added that cladism = structuralism = anti-realism is not actually an argument for cladism = wrong or cladism = unacceptable. Maybe cladism is wrong, maybe anti-realism is bad. Personally, I would disagree with the former and agree with the latter. The point is that even if cladism = anti-realism, at least one of these two claims would have to be demonstrated independently from the present argument. In the absence of such independent demonstration it is at best something like an association fallacy.

Is cladism nominalist, anti-realist, metaphysical?

What is now the argument for cladism being anti-realist? On pages 33 and 34, it proceeds as follows:
The presumed speciation mechanism requires taxonomists not to consider any species as ancestral to another one, and thus to classify them only as sister groups. This dogma according to which reality is not accessible to us is characteristic of a nominalist philosophy of knowledge: science aims to model and predict manifestations of reality but not to understand reality itself.
As I have explained above, this is simply not about speciation mechanisms. In addition, cladism does recognise species as ancestral; indeed it recognises only species as ancestral where 'evolutionary' systematics, in marked contrast to all of evolutionary biology, also considers genera, families, orders and classes as ancestral. Cladists just don't consider one of the daughter lineages to be identical to the ancestral lineages, that's all.

Admittedly, there are cladists who would consider ancestry to be un-knowable (see blog post here), and so at least that minority could be considered nominalist. I must admit that I do not understand their view, and it appears at first sight similar to certain postmodernist armchair philosophers claiming that science can't know anything ever about anything. In my eyes we can always draw tentative conclusions given the best evidence currently available and wait for somebody to come up with contrary evidence.

There are certainly some cases, especially in groups like diatoms, where the fossil record is so ridiculously complete that it is feasible to reconstruct direct ancestry. So I am a cladist who would not rule ancestral species out in principle, and thus the charge of nominalism does not apply to me. More importantly, the same is true for many other cladists, indeed the majority as far as I can tell. Consider E.O. Wiley for example, who describes how he would classify ancestral species in the Linnean system in one of his papers (Wiley 1979, Syst. Zool. 28: 308-337). And he just happens to have written what is perhaps the textbook of Phylogenetic Systematics.

That being said, this is mostly a theoretical consideration. Taking into account the whole diversity of life, cases like the aforementioned diatoms or even complete skeletons are rare. I would really like to know how the 'evolutionary' systematists who insist on classifying ancestral species would do that in practice given only what is usually available, for example a pollen fossil or half a jaw-bone. At least in my eyes, somebody who claims that these fragments provide sufficient proof for an ancestor-descendant relationship at the species level is over-reaching.

Finally, note that the idea that we cannot truly understand the nature of reality but only describe it as best as possible using our limited minds and language is not self-evidently wrong. In fact it might be self-evidently correct, regardless of how we treat ancestral species. I don't think most, if any, of us can really comprehend 13 billion years or gravity, we can just tell stories and draw analogies. Either way, in my eyes at this point the argumentation has already collapsed under the weight of its misunderstanding of cladism.
Hennig did recognize the material nature of reality and its independence of the subject. This can be seen through his attachment to the fact that successive branchings of cladograms do represent an evolutionary process. This is clearly a materialist position.
I see no problem here.
Finally, the Hennigian conceptual framework can be described as metaphysical since it is both rigidly static and reductionist. Indeed, its phyletic arrangement does not reflect any transformation (no species becomes another, no group emerges from another group, etc.) and is reduced to represent only the structure of the process supposed to be at work (the splitting of a mother species into two daughter species).
The difference between cladism and 'evolutionary' systematics lies precisely in the cladist understanding that the ancestral species is becoming the descendant clade. It is the 'evolutionary' systematist who has a rigid and static, pre-evolutionary understanding of systematics, as when they consider a small part of a diversifying clade to be the unchanged ancestor.

And that's it. That was the argument. In summary, I am not convinced that cladism as such is structuralist, anti-realist, metaphysical, or whatever other association fallacies one would like to throw at the approach.


  1. First point, should I only "talk[ing] about Phylogenetic Systematics as practised today"? No, most of them practice phylogenetics without knowing the philosophical justifications underlying it. My aim is not to characterize their almost robotic practice but to characterize the philosophical justifications. So, I am not missing my target.

    Secondly, I can't see your point of trying to rebut the fact that cladism is supported by a nominalist philosophy of science, this is not even a claim against cladism since I would have to rebut nominalism, which I haven't done (through I am clearly not nominalist).

    "whether that speciation was through splitting or budding is irrelevant for a cladogram representation of the tree of life" Yes, this is a nominalist point of view.

    "the descendant species should both be treated as individually different from the ancestral species even if speciation was through budding" Again, even a stronger nominalist claim.

    "Cladism is not about maximising phenetic information content; phenetics is." Historically wrong, it was one of the main arguments, and it is still one that I regularly read in textbooks. Clades were supposed to be more "natural" in Gilmour's meaning of this term. See Farris (1979) The Information Content of the Phylogenetic System.

    "In addition, cladism does recognise species as ancestral; indeed it recognises only species as ancestral" Cladism recognize them only nominally, there are not treated as ancestral but classified as a sister-group. You can have a realist point of view about evolution and being nominalist in your way of classifying species. This is quite schizophrenic to me, and that's partly why I find pattern cladists a lot more self-consistent.

    "Cladists just don't consider one of the daughter lineages to be identical to the ancestral lineages, that's all." It's metaphysics to me. Besides, not every cladists in fact. Contrary to what you insinuate, I am far from putting every cladists in the same bag. There are many ways to amend cladism to make it a little more or a little less nominalist.

    "It is the 'evolutionary' systematist who has a rigid and static, pre-evolutionary understanding of systematics, as when they consider a small part of a diversifying clade to be the unchanged ancestor." Characterizing evolutionists as "pre-evolutionary" systematists is quite provocative and really far from truth, since we assume for example that one family can give birth to another family, how would it be possible without evolution? On your side, you say that if your ancestor is a dinosaur, no matter how much your descendants will change, it will never be enough so that they could become a non-dinosaur. So, static, yes. And this also makes "taxa" nominal groups (many even says "individuals"), not classes.

    Nominalism emphasizes that the label is arbitrary, so that there is no "concept of a dinosaur". Realism emphasizes that the label matches a real cohesive class, so that one item of this class can leave it if it does not fit the definition anymore. Since you reject the second one and adhere to the first one, yes you have a nominalist point of view.

    1. No, most of them practice phylogenetics without knowing the philosophical justifications underlying it. My aim is not to characterize their almost robotic practice but to characterize the philosophical justifications. So, I am not missing my target.

      Has it occurred to you that there might be several independent justifications for phylogenetic systematics, not even necessarily philosophical? This is like arguing that nobody can properly accept punctuated equilibrium without accepting Gould's entire worldview, non-overlapping magisteria and all.

      Just don't expect people to stop circumscribing genera as monophyletic in Taxon just because you called them nominalists in Phytoneuron...

      As for the rest, maybe I am somehow nominalist by your logic, fine; seems like you always know better than cladists what cladists believe, even when cladists explicitly promote the opposite of what you claim. The above post still explains why I believe that the argument as originally presented in your Phytoneuron paper doesn't work.

      We will have to agree to disagree about what the true underlying concern for cladists was and is. Makes me curious though why Hennig, Farris et al. wouldn't just have been a pheneticist then, if you were right.

    2. "Just don't expect people to stop circumscribing genera as monophyletic in Taxon just because you called them nominalists in Phytoneuron..." I didn't expect that. Let me cite myself, page 42: "It would be vain to believe that a fight that lasted for 25 centuries can be suddenly completed."

      "seems like you always know better than cladists what cladists believe" Sarcasm or anti-intellectualism? I don't get it. Being or not being a cladist (besides, I was) does not make my analysis better or worse.

    3. Of course, sarcasm.

      For starters it would be helpful if you could at some point define 'cladist' and then stick to that definition. When you say that you were once a cladist yourself, or when you criticise current mainstream practice, you appear to use a very wide definition: cladist = everybody who wants stuff to be monophyletic. Then when I react to your claim of cladists not accepting ancestors by pointing out that E.O. Wiley for example has no problems with classifying species as ancestral, you appear to play a No True Cladist card, and when I argue that you can't defeat contemporary practice by criticising some individual cladists' personal opinions from 1970ies, you imply that cladist = only those who agree with Farris on everything.

      Trying to understand what you mean with cladist, and how you figure what 'the cladists' believe, is like nailing a pudding to the wall.

    4. I don't care about cladists, only about cladism. So, not sure who is really playing the No True Cladism card. In the limit case, you could even find a "cladist" who rejects every possible justifications of cladism and still classify according to the holophyletic principle. Could such a blind practice be called cladism? Well no.

      Cladism states that only holophyletic taxa can be natural for the reasons I have detailled. Section 10 of my paper is a characterization of what cladism is, so I won't copy/paste it here.

    5. "Finally, note that the idea that we cannot truly understand the nature of reality but only describe it as best as possible using our limited minds and language is not self-evidently wrong."

      Yes, as you point out Part 3, budding and splitting are both simplifications of reality. The realism of those philosophical abstractions matters for science only when it limits the reliability our inferences and predictions.

      Aside from reconstructing branching patterns, cladistics is useful because we can often assume that more closely related species will have more similar anatomy, physiology, ecological adaptions, or pharmaceutical properties than more distantly related species.

      However, because of different selection pressures acting on sister lineages, this correlation does not always hold, causing problems for conservation and evolutionary studies that use relatedness as a proxy for other variables. In these cases, concepts like composite morphospecies and evolutionary grades could be a useful corrective, provided one is careful not to confuse a grade with a lineage, a modern species with an extinct ancestor, or fuzzy sets with crisp sets.

      "Phylogenetic signal" (or perhaps more accurately, "phylogenetic autocorrelation") is a testable hypothesis. Here's a study that looked at diversification within squamates (lizards and snakes). One could use this data to argue for raising the unusually diverse snake clade Alethinophidia from an infraorder to a suborder and lumping lizards and thread snakes into a paraphyletic group. However, there would be no justification here for maintaining paraphyletic squamate families or for cleaving snakes from lizards.

    6. Damien,

      Having different reasons than the ones you find convenient to discuss is not the same as doing something blindly.


      I would still say that to be meaningful an evolutionary study must deal with clades, no matter how morphologically heterogeneous they are. Assume you are studying whether heterostyly is correlated with higher diversification rates in the genus Primula - if you don't include morphologically specialised but nested Dodecatheon, your analysis is invalid. And helping people to avoid the mistake of making conclusions about the evolution of incomplete groups is one of the arguments for phylogenetic classification.

  2. I used to think Elvira Hörandl had identified some overlooked flaws of cladism, but have a learned a lot from this blog and see the validity of most of your arguments. In particular I now understand that:

    -A classification system that recognizes only monophyletic taxa (sensu Hennig) facilitates research on evolutionary patterns and processes

    -‘Evolutionary’ classifications convey no information other than “this is a monophyletic or paraphyletic group that the taxonomist thought was distinctive enough in some unspecified way to be ranked at the family level.”

    -In mainstream phylogenetic systematics, Linnean rank conveys no information other than “this family is a more inclusive clade than the genera included within it.” Most of the problems that evolutionary systematists attribute to cladism are actually limitations of Linnean ranks and binomials, and could be avoided by switching to something like PhyloCode with hyphenated uninomials.

    However, one can be a cladist in most respects while still speaking of informal paraphyletic groups like “bryophyte” or “fish” that many people seem to think are unsuitable for scientific discourse. Steve Mount discussed the taboo against “prokaryote” in a post titled "Can We Not Speak of Fish?"(google it, my link isn't working) and advocated what is essentially a flipped version of Damien Aubert’s proposal. Please consider this when you discuss parallel classifications in part 8 of your series.

    In summary, I fully support the definition of taxa as monophyletic groups, and I would like to see them used to more rigorously define paraphyletic groups. Scientists will continue to refer to paraphyletic groups, and for good reasons. When they do, it would be useful if those groups were understood to be the relative complements of monophyletic taxa rather than informal categories, form classes or sloppy and unscientific categories.

    1. I believe we do not disagree about much. I also constantly refer to poly- and paraphyletic groups, but that is not the same question as how the official classification should be organised.

      The interesting issue here would be why we need words to refer to groups like paraphyletic dinosaurs or fish. In some cases it is clearly because of some ecological similarity or because they are a resource category, like fish in the sense of what I could catch with a fishing rod or net. But it is just a historical accident that paraphyletic fish are a useful group in that sense, and even then only if we ignore all that are too big to catch. If there were swarms of 20 cm dolphins swimming through the ocean, fish in a resource sense would have to be polyphyletic, like game, weed, crop plant or seafood.

      Or put another way, we don't need names for paraphyletic groups for these practical purposes, we need names for phenetic or ecological clusters. To me that does not supply a justification for a classification that recognises paraphyletic but not polyphyletic groups.

      So in most cases I fear that the only reason we do in fact often need names for paraphyletic groups is tradition.

    2. The main advantage of a paraphyletic grouping like "fish" over a resource category like "seafood" or a phenetic descriptor like "fusiform" is that helps translate between vernacular and taxonomic language. Anyone can recognize a fish, and that can be refined to "bony fish" and then narrowed to a clade, the ray-finned fish.

      Yes, a zoo or pet shop might want to talk about "reptiles" excluding birds because they are generally poikilothermic and terrestrial to semi-aquatic and so require different enclosures and care. A field biologist might want to talk about "reptiles" excluding birds because they generally can't be identified by their calls, can't be caught in mist nests, occupy distinct ecological niches owing to their poikilothermy, and as such require different strategies for sampling and identification. However, the point you are missing is that the many similarities between crocodilians and squamates relative to birds are not superficial, they are symplesiomorphies. That gives "reptile" some evolutionary content and predictive value that is lacking in an ad hoc polyphyletic functional group like "terrestrial and semi-aquatic poikilotherms."

      "Class Reptilia" glosses over the similarities between crocodilians and birds. "Class Crocodilia" would overstate the differences between crocodilians and squamates. It is not difficult to capture something about both common descent and branch length by qualifying a reference to a paraphyletic group with its bounding clades, saying something like "sauropsids share V, reptiles share W, only birds have X, only crocodilians have Y, and archosaurs share Z." "Reptiles" here is just a concise way of saying "crocodilians, squamates, Sphenodon, and testudines" or "non-avian sauropsids."

  3. You’ve said before that paraphyletic groups are “not real groups.” This is entirely a philosophical argument (although I'm not sure that it is nominalist). Your view as I understand it is that species are real synchronous phenomena theoretically approximated with the Biological Species Concept and operationalized by the Genotypic Species Cluster Concept. Clades are natural because they are species viewed across time. That might be true, but so what? The crown birds are a natural group, but they’re not what all systematists means by “bird”: Aves has also been applied to at least three apomorphy-based clades and a branch-based clade. Clade Dinosauria is a natural group but it’s not what everyone means by “dinosaur”: many paleontologists have used the word in its traditional sense with the full understanding that the immediate ancestors of Archeopteryx were a lineage of Coelurosaurs (rather than any or all “dinosaurs”) and that at least some members of the group were endothermic and had feathers.

    Clades can defined by shared traits, but so what? Some of those defining traits are uninteresting and others have been reversed or highly modified in some descendants. Some groups currently thought to be clades will turn out to be paraphyletic and their defining traits will turn out to be plesiomorphies. What’s so special about synapomorphies?

    This would all be academic if one could communicate effectively without ever speaking of fish, but I’m not sure that’s possible.

    1. Paraphyletic “stem groups” are useful for describing the upper and lower limits of soft-tissue character inferences in fossils or hypothetical ancestors.
    2. Paraphyletic groups allow one to refer to fossils without being overly general or overly specific: The most recent common ancestor of Velociraptor and Troodon was also an ancestor of birds. Velociraptor, Troodon, and ducks all have pennaceous feathers, so they are eumaniraptorans. Troodon has some distinctive tooth features not found in Velociraptor so it is a troodontid. Both Velociraptor and Troodon have pennaceous feathers, sharp teeth, stabilizing tails, and sickle claws, so they are “deinonychosaurians” (non-avialan eumaniraptorans).
    3. Paraphyletic groups simplify discussion of hypothetical ancestors. If someone thinks the ancestral Rubiae was similar to the extant Kelloggia except in a few minor respects, “Kellogginae” could be a convenient shorthand to refer to both.
    4. Paracladistic groups sensu Podani are useful for contrasting plesiomorphic and apomorphic features between living species on either side of a “long branch.”
    5. Paracladistic groups are useful for chopping up a speciose branch of the tree of life into manageable chunks for textbooks, field manuals, and such. The informal group is a placeholder for a more detailed description of various ‘basal’ clades.

    All those uses have serious limitations, but those limitations are best understood by contrasting (A-B) with (A) rather than defining (A-B) out of existence.

    1. Yes, I agree that the 'not real groups' argument is philosophical, but really every time we think about how to think about things we are doing something that can be called philosophy.

      I am afraid I do not understand all five of your points, perhaps because I am not a palaeontologist. It is not clear to me why one would want to refer to Kelloggia and the common ancestor of Rubieae as a group, or why one would need have a name for (A-B) to discuss apomorphies. At least 1 and 5 seem to bracket out the problem of gradualism of evolution, although it is not so much of a problem for 5 as you imply it is only about extant species, i.e. one time-slice.