In the last paragraph of the materials and methods section we find what I consider the most crucial element of this case study in "evolutionary" systematics, its criteria for classification. Phylogenetic systematics has one simple criterion for accepting a suggested supraspecific taxon as valid: monophyly. That means a taxon can be accepted as valid if and only if it includes all descendent species of one common ancestral species, and not only some of them. What I would need to see to take "evolutionary" systematics seriously is an alternative criterion that is just as testable, objective and universal. Why do I stress these three properties so much?
Because that is what makes biological systematics and classification a science as opposed to a pseudoscience. Testable is pretty obvious, and outside of theology and perhaps some of the more fruitless branches of philosophy no academic deals with ideas that cannot be tested anyway, although the ways they are tested vary - deductive logic, empirical evidence, mathematical proof, thought experiment, whatever. In the case of phylogenetic systematics, the status of any group of species as a valid taxon can be tested by reconstructing a phylogeny and verifying the monophyly of the group, as evidenced by at least one synapomorphy.
Objective means that given the same evidence and the same method, somebody else will be able to arrive at the same result as long as they are rational and operate in good faith; you could also call it reproducibility. To illustrate, the objectivity of the scientific method means that even if all our knowledge were obliterated and science had to be reinvented from scratch, our distant descendants would easily be able to rediscover the fact that water consists of two hydrogen atoms and one oxygen atom, how to build a functional airplane or that our planet rotates around the sun. It also means that a Chinese Taoist, a European Catholic and a Pakistani Sunni could all independently verify whether a given population of a species of plants suffers from inbreeding depression or not. In contrast, the lack of an objective method for the generation of religious belief means that it is, to say the least, highly improbable that our distant descendants would, after a similar obliteration of all religion, "rediscover" the idea that Jesus is the savior. (They might well not even come up the concepts of sin and salvation in the first place, as demonstrated by the many religions that happily do without it.)
In the case of phylogenetic systematics, monophyly is of course an objective criterion that can be used, by anybody with access to samples of the study group and a knowledge of phylogenetic methods, to independently and reproducibly arrive at the same conclusion.
Finally, with universal I mean that the same criterion can be used for all groups of organisms. For example, the idea that flowering plants should be classified primarily by the number and fusion of their floral organs, as in Linneaus' Systema Sexuale, is no help for the classification of ferns, much less snails. Monophyly, on the other hand, has the merit of universality: as long as there is phylogenetic structure, i.e. in all groups of organisms above the species level, it works. You can apply it to classifications of fish or insects as well as to those of grasses or liverworts. What is more, you can apply the same criterion at any taxonomic level above that of species, where phylogeny turns into tokogeny.
Let us see how the criteria suggested in the present paper measure up. That we are off to a bad start can already be seen from the plural: As discussed before in this series, having more than one criterion is a bad idea because the resulting classification does not tell a coherent story (Hennig, 1975).
Imagine sorting all your books by topic and family name of the author at the same time. Of course, your immediate solution will be to do one after the other, i.e. first by topic and then by author within the topics, but that is cheating because it is not really at the same time at all. To approximate the effect of an evolutionary classification, you would have to randomly switch the criterion from time to time, so that the biology textbooks form one section, except for those written by authors with names starting with A, B, F, I, T and W, which are intermingled with novels and cooking books written by authors with names starting with those letters, and then within the biology section you have three subsections on botany, zoology and genetics, but in each case authors with different letters are excluded and form a separate fourth subsection that is sorted alphabetically... and so on. And now imagine somebody else trying to find a book in your library.
More importantly, due to the nested, one might even say fractal, nature of biological diversity that necessarily results from common descent, even the more sensible approach of sorting books first by topic, then by author would not work in systematics. We cannot say: we classify by similarity up to the family level, and beyond that by descent, not least because the family level (like all others except species) is completely arbitrary. In that sense, even phenetics makes more sense than evolutionary systematics because at least it is consistent.
But okay, despite these problems the authors of the present paper insist on using several different criteria in the same classification. Which ones, and where?
For bifurcating parts of the tree topology, we follow the principle of monophyly s.l. (including holophyly and paraphyly; Ashlock, 1971, 1984; Hörandl and Stuessy, 2010; Blöch et al., 2009; Stuessy et al., 2011) as the main criterion for classification.Ah yes, evolutionary systematists have their idiosyncratic terms: they call paraphyletic "monophyletic in the wider sense", and then rename monophyletic into holophyletic. (And that is causing hardly any confusion at all!) No matter, let's focus on the criteria. What the authors say here is that in a group showing phylogenetic structure, they will circumscribe taxa to be monophyletic or paraphyletic. Or in other words, after spending much of the paper trying to justify paraphyletic taxa as necessary because of reticulate evolution, they now say they will accept paraphyletic taxa even if there is no reticulate evolution. Why? They don't say, or at least not here.
In the case of holophyly (=monophyly s.s.), we classify clades with posterior probability values >75 in the molecular tree or at least 50% BS support in the MP analysis of the total evidence approach; a minimum of five shared characters is set to accept a clade for formal classification. Paraphyletic groups were only considered if shared characters appear as symplesiomorphic;So this is their criterion for when to accept paraphyletic groups: whenever monophyletic groups that they could alternatively recognize have less than five morphological* synapomorphies and/or receive no bootstrap support. Why 50% and not 70%? Why bootstrap and not Jackknife, or why not Bayesian posterior probability of 0.95 and above? Why five synapomorphies? Why not four or six? Or why not any number above zero, as in phylogenetic systematics? And for the purposes of deciding whether there are four or five, what is one character? Morphological characters are notoriously controversial in their delimitation. If they had built their character table slightly differently, two of their current characters might be one or vice versa, or they might have examined twenty additional characters and thus found additional synapomorphies for many clades that they did not recognize in the present case. And why not molecular synapomorphies?
This is all totally arbitrary. What it is not is a universal and objective approach that could be transferred to entomology or phycology, to classify beetles or algae; the methodology could not even be transferred to somebody else building their own, independent morphological character matrix for Ranunculus.
polyphyletic groups based on homoplasious characters were not accepted.That appears like the most reasonable sentence in this part of the methods section so far. But to put it into the relevant context, note that the difference between polyphyletic and paraphyletic groups is much smaller than that between monophyletic and non-monophyletic ones. In fact, one can easily draw a hypothetical non-monophyletic "group" onto a phylogeny that is both polyphyletic and paraphyletic at the same time, depending merely on the inferred character state of the common ancestor for the various characters that might be used for its definition. That is because the difference between monophyla and non-monophyla is actually structural, whether they cover complete or incomplete branches of the tree of life, but the difference between polyphyla and paraphyla is only the choice of the morphological spotting character. Considerations like this are, of course, what lead to the insistence on monophyly for the circumscription of taxa in the first place.
In the case of reticulate evolution, ancestry remains uncertain. We hypothesize that shared characters in genetic clusters do reflect synapomorphies of either the original parental taxa, or synapomorphies that the first hybrid has acquired and passed on to all its descendants. We classified paraphyletic groups only if they meet the above-mentioned criterion of five shared characters, and if alternative cladistic concepts would be not applicable according to the above-mentioned criteria.So if I understand them correctly, H&E define paraphyletic groups if they cannot be entirely subdivided into monophyletic groups with bootstrap support and five morphological synapomorphies, and they define them based on a minimum of five morphological symplesiomorphies. Quite apart form the fact that paraphyletic groups are still non-groups by definition and symplesiomorphies are still non-characteristics (compare the heterogeneity of a group of non-Mercedes-drivers defined only by the symplesiomorphy of not driving a Mercedes), settling on five instead of two or eight symplesiomorphies is just as arbitrary as with five synapomorphies.
On top of that, remember that the presence of five synapomorphies or five symplesiomorphies is an illusion brought about by extinction. This is the long branch fallacy again or, in other words, another variation of the emotional "but they look so different" argument that all evolutionary systematics ultimately reduces to. In reality, evolution gradually produced one new character after the other, and there never was a long branch but a long sequence through time of slowly evolving populations. The intermediates could still pop up as newly discovered species, and they could be found as fossils. In many other cases we will never actually see them, but the fact remains that a classificatory approach that only works in groups with a fortuitously high level of extinction and a fortuitously poor fossil record is of very limited use indeed.
I am not entirely sure how the part about reticulate evolution fits here. As discussed previously, I believe that the authors may be confused about the difference between tokogeny and phylogeny and may not realize that a severely reticulating system does not contain paraphyly but instead no type of -phyly whatsoever. Mostly, however, I cannot shake the impression that the relevant sentence was perhaps inserted as an afterthought when they realized that they had originally justified the need for paraphyletic taxa with hybridization, polyploidy and anagenesis (not that any of those are good justifications in the first place) but forgotten to mention these processes somewhere in the vicinity of the sentences that go "we accept paraphyletic taxa when...".
In summary, at least in my eyes this case study on Ranunculus does not demonstrate the application of the testable, universal and objective criterion for the circumscription of taxa that "evolutionary" systematics would have to provide.
Next up, the results and discussion sections of the paper. Continue reading here.
*) Again, H&E use "character" as synonymous with "morphological character", implying that molecular and biochemical characters are for some reason not characters.