Friday, February 20, 2015

Stemmy large-evolutionary changes (that special issue)

(The following is the ninth part of a series of posts on an Annals of the Missouri Botanical Garden special issue on “Evolutionary Systematics and Paraphyly”. All posts in this series are tagged with “that special issue”.)

The final contribution to the special issue advocating paraphyletic taxa is Richard Zander's Support measures for caulistic macroevolutionary transformations in evolutionary trees. There are two ways of addressing it, and with previous papers in this issue I have sometimes taken one and sometimes the other: Either one can go through the paper bit by, carefully analyse the argumentation, rebut one claim here but concede another there, and so on; or one can take a step back, point at the fundamental assumption underlying the whole line of argumentation, and explain in a few words why one considers it to be wrong.

Because I am tired and have much else to do, I will mostly use the second approach and then spend just a bit more time addressing other random aspects of the paper that stick out to me.

Really it is very simple: Richard Zander sees groups of organisms that exist today as the ancestors of other groups of organisms that exist today. I, and with me presumably most systematists and evolutionary biologists, believe that a group of organisms that exists today cannot possibly be the ancestor of another group of organisms that exists today.

Unless one were to push them forcibly into a working time machine, today's chimpanzees are not going to become our ancestors, today's fish are not going to become the ancestors of the land animals, and today's ferns are not going to become the ancestors of the flowering plants. Instead, these groups have common ancestors in the past, and thus, no matter how much Zander ridicules the concept, “unknown hypothetical ancestor → (one extant group, another extant group)” remains the most appropriate way of describing evolutionary history. As a group of individuals in a time slice, the ancestral taxon is separate from all of its descendants, and as an evolutionary lineage through time it is identical to all of them, but it does not make sense to equate it with only some.

So again, I do not accept the premise that would enable us to even start thinking in terms of what Zander calls “caulistic macroevolutionary* transformations”, and thus for me the entire argumentation of this paper never even gets onto its feet. Conversely, Richard Zander does not accept the premise that ancestors should actually be ancestral to their descendants, and so nothing I can write would ever convince him. Agree to disagree and all that, I guess.

Apart from that, as the first two thirds of the paper summarise Zander's recently published A Framework for Post-Phylogenetic Systematics, it seems appropriate to point out again its biggest problem. Despite the claim that “evolutionary systematics ... cherishes molecular cladograms as informative of aspects of evolution”, the Framework would, if adopted as common practice, in effect insulate subjective, authoritative statements of individual taxonomists completely against test or refutation through, among other things, molecular data.

Phylogenetic analysis of morphological data disproves your concept? The Framework tells you to reweigh the characters and try again until it doesn't. Molecular data disproves your concept? The Framework tells you to postulate “deep ancestry”. Read it for yourself. What it never seems to tell you is how you would figure out that your original taxonomic concept was wrong. And, to directly quote the present paper,
Molecular analyses, when they match morphological results, can only be said to be ‘‘not incompatible’’ and never cited as supportive because exemplars are usually insufficient to stand by themselves as well-documented representatives of coherent entities in nature; or, if they do not match, they can only be incongruent and never truly falsificatory.
Part of the reason we are supposed to distrust molecular data is, by the way, that they are unreliable anyway:
Sources of uncertainty, none of which are unfamiliar to phylogeneticists, include alignment, wrong gap costs, differential lineage sorting, hybridization, polyploidy, recombination, non-clocklike behavior, rates other than gamma distributed, differences between the results of ‘‘total evidence’’ and evaluations based on separate gene studies, possible strong selection pressure on noncoding promoter sequences, persistent pseudogenes, too few exemplars, endogenous retroviruses, gene conversion, self-correction of flawed DNA, paralogy, codon bias, chloroplast capture and other horizontal gene flow, novel clades, saturation, third codon bias, wrong identifications, long-branch attraction, model insufficiency, and other problems
I have taken the liberty of marking all those sources of uncertainty that affect morphological data to the same degree in red, although in that case some of them would come under different names, for example 'establishing homology' or 'homoplasy'. Those where I simply do not see why they are a problem for phylogenetics or where I do not understand what Zander means are in blue. But credit where credit is due: in contrast to several other contributors to the special issue, he gracefully acknowledges that phylogeneticists are fully aware of all these issues and take them into account. (The ones that aren't marked blue, at any rate.)

Among the issues mentioned above is poor sampling, and in fact the claim that molecular phylogenetics suffers from it is made in several places throughout the manuscript. The following, for example, is the very first sentence in the second section of the paper: “An exemplar of a taxon in a molecular cladogram is often the sole representative assigned to that taxon.” Unbelievably, the very next paragraph then accuses cladists of the opposite: “this is not true for molecular distributions, which focus on sampling sequences, not specimens. Such sampled sequences may even be from different specimens.” This seems to be one of those heads I win, tails you lose scenarios.

Finally, a perhaps minor stylistic issue:
Theoretical macroevolutionary transformations are then synthetic, emergent properties. Looking for deep ancestors linking molecular and morphological inferences is equivalent to the search for ‘‘hidden variables’’ in physics, such as the so-far unsuccessful search for a non-obvious classical explanation for the nonsensical rules of quantum mechanics. [...]
The molecular cladogram is commonly analyzed on a Markovian model in that present-day data are taken as sufficient to retrodict the relevant past (if a large enough data set), and analysis is by a reverse Laplace’s demon through coalescent theory. In integrable analyses, such as prediction of the future position of an ideal billiard ball on an ideal elliptical pool table, uncertainty is not increased beyond the initial uncertainty. [...]
The problem becomes nonintegrable and can lead to chaos, much as the position of even an ideal caroming billiard ball is not predictable after about seven bounces off a cushion on a rectangular pool table (Manzotti, 2011) as it transits into Brownian motion-like chaos. [...]
As Einstein said to Heisenberg (Gilder, 2008: 87): ‘‘It is theory which first determines what can be observed.’ [...]
In that Markov chain analysis is np-hard, Hastings–Metropolis sampling speeds the computation. [...]
Following Ekeland’s (2006) example of the rectangular pool table that yields nonintegrable predictions of a pool ball’s path, there are two paths that are in fact integrable, namely bouncing back and forth orthogonal to the sides. [...]
This is much like Haack’s (1993: 81) analogy of the crossword puzzle that is solved by mutual support from two rather different belief or experiential systems.
In case anybody forgot, this paper is basically making the argument that one extant group of organisms should be seen as the ancestor of another extant group of organisms.

Then again, maybe I should play billiard again one of these days. I rather enjoyed it when I was a student.

Footnote

*) Note that Richard Zander's definition of macroevolution appears to be rather idiosyncratic. When asked, most biologists are likely to explain that macroevolution is evolution above the species level. In the debates around creationism, evolutionary biologists often point out that macroevolution is nothing more than microevolution happening over a long time. Zander, on the other hand, defines macroevolution as what happened at that point in the past when a lineage changed enough for an 'evolutionary' systematist to consider one (now paraphyletic) plant family or genus to have produced another taxon of the same rank that is phylogenetically nested inside it. This redefinition of macroevolution into paraphyly then makes possible the claim that cladists do not accept macroevolution.

References

Zander RH, 2013. A Framework for Post-phylogenetic Systematics. Zetetic Publications, St. Louis; CreateSpace Independent Publishing Platform, Amazon.

ZanderRH, 2014. Support measures for caulistic macroevolutionary transformations in evolutionary trees. Annals of the Missouri Botanical Garden 100: 100-107.

No comments:

Post a Comment