Monday, December 29, 2014

Island endemic genera and phenetics (that special issue)

(The following is the sixth 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”.)

Following an introduction and the contributions of Lockhart et al., Hörandl, and George, the fifth full paper in the special issue is Stuessy et al.'s “Paraphyly and endemic genera of oceanic islands: Implications for conservation”.

The main argument is quickly summarised, and it has actually already been made before by the same author, only then in a considerably more concise manner (Hörandl & Stuessy, 2010). When new species arrive on oceanic islands via long distance dispersal, in the most extreme cases as a single seed or a single pregnant female, they may find themselves presented with many new possibilities. Some selection pressures from their original habitat may not exist on the island, and there may be unused niches ready for the taking. The new arrivals also undergo a severe genetic bottleneck, carrying only a small fraction of the genetic diversity of the mainland population in themselves.

This means that island colonisers often have the chance of undergoing spectacular adaptive radiations in a short time. Echium and Sonchus in the Canary Islands, fruit flies or the Silver Swords in Hawaii are just some examples. In the words of Stuessy et al.,
Because of the speed of the divergence, it might be that the island genera are genetically not so divergent from the continental relatives, but they are usually very divergent morphologically, hence their recognition at the generic level.
So because they looked superficially very distinctive after their adaptive radiation into new niches, island lineages were traditionally often treated as genera distinct from the mainland genera they evolved out of. With the advent of phylogenetic systematics, however, they are sunk into these mainland genera, so that these island lineages are not island-endemic genera any more; they are just the island's representatives of the widespread mainland genus.

So what? So, according to Stuessy et al., this:
these actions could have a substantial effect on world island conservation.
This is, as far as I can see, as explicit as the paper makes the argument for paraphyletic taxa, but it is still clear what this is about. The idea is undeniably that one should keep island endemic genera because they make a better sell for conservation politics than mere endemic species.

First, I seriously doubt that people are more likely to conserve endemic genera than endemic species. Second, and more importantly, one will notice that this is not a scientific consideration. I can do no better than to reiterate what I wrote when I was first faced with this argument:
The damage it would do to taxonomy if its end-users were to gain the impression that political expediency instead of objective criteria underpins our work cannot be overstated. The logical extension of this approach would be to randomly elevate endemic species to generic rank wherever a taxonomist would like to see the creation of a protected area.
Really this whole line of argumentation makes me quite uncomfortable.

Beyond the political argument, the paper also mentions a somewhat scientific one, albeit without developing it in any detail:
Furthermore, as genetic [sic – probably “generic” was meant here] endemism is a criterion for delimitation of floristic zones, a reduction in endemic genera in oceanic islands might also have an impact on the classification of world vegetation.
This is certainly true, but as far as I am concerned, the impact would be to make them more scientifically accurate.

Finally, if one persists through more than twenty pages of discussions of individual islands and their endemic genera (often, strangely, focussing on whether they are monophyletic, which is besides the point anyway because generally the real question is whether they make something else non-monophyletic), one is rewarded with an attempt to provide an objective criterion for the recognition of paraphyletic taxa.

If anybody is interested enough to browse through my earlier ramblings on this blog, they will quickly find that I have repeatedly tried to make the point that no such criterion is possible. 'Evolutionary' systematists want to take a branch of the tree of life and find a place where they can say, this sub-branch is so very significantly divergent from the rest of the branch around and below it that they should both be treated as separate genera.

But the problem is, evolution is the gradual change of allele frequencies in populations; that is just how it works. Therefore you can go along the branches of the tree of life as far as you want, but unless the creationist straw man of the Theory of Evolution as a cat giving birth to a dog is true you will never find a situation in which such very significant divergence obtains. The illusion of such divergence is only possible thanks to the present day absence of the intermediates that we know must have existed in the past, but it is still that: an illusion.

With this in mind, under the understanding that whatever is suggested will only work on extant species and only by ignoring intermediate fossils, what does the present paper suggest?
Predictive quality (P) involving two genera, therefore, can be defined as:
P = (dAB/cA + cB) / 2
[sic - I assume that it should have been P = (dAB/(cA + cB)) / 2]
where d equals the distinctness between genera A and B, and c is the cohesiveness of genera A and B. Distinctness is measured by the character distance between the centers of the two groups (using a centroid method; Sneath & Sokal, 1973). Cohesiveness is the average character distance among group members (easily computed mathematically).
Okay, so this is phenetics, classification based on overall similarity, a school that was popular in the 1960ies and 1970ies. Not so 'evolutionary' after all, this approach. But the authors have anticipated this criticism.
Measuring predictive quality in this way is not phenetics. Phenetics is estimation of relationship by means of overall similarity, using as many characters as possible and without regard for evolutionary significance
Well, isn't that just what this paper is suggesting? “Regard for evolutionary significance” is what cladists have: They see the task of a systematist as that of finding out which characters are really indicative of relatedness and which aren't. In the approach described here, however, characters are simply used for phenetic clustering. And hello, Sneath & Sokal (1973)? That is the seminal publication of the phenetic school. Of course we are talking phenetics here!

I must also say that I kind of wonder about the behaviour of this formula. What, for example, if P is high because one genus is very cohesive indeed, but the other one isn't? Shouldn't both of them be cohesive to be recognised as distinct on the same level even under phenetic logic? And isn't a genus of one species always extremely cohesive? One could justify some pretty strange classifications that way.
This statistic provides a yardstick for evaluation for generic recognition. Clear genera should give a predictive quality in the neighborhood of 5.0 or higher. Weak genera are at 1.0–2.0, and if P drops below 1.0, there is no argument for recognizing more than one genus.
Why 5.0, and why 1.0-2.0? What about those around 3.0? Where do those numbers come from? What if I add a few more characters, or throw some out, and the results change? And should we now recognise weak genera or not? So really we still do not have an objective, reproducible approach comparable to phylogenetic systematics, we are still in the realm of subjective preference backed up by personal authority.

Anyway, the main point of this paper was the political one, and that is what occupies most of its pages.


Hörandl E, Stuessy TF, 2010. Paraphyletic groups as natural units of biological classification. Taxon 59: 1641-1653.
Stuessy TF, König C, Lopez SepulvedaP, 2014. Paraphyly and endemic genera of oceanic islands: Implications for conservation. Annals of the Missouri Botanical Garden 100: 50-78.

No comments:

Post a Comment