Systematics today: Why study comparative (insect) morphology
A New Year’s post whose motivation derives in part from an engaging discussion of a paper by Wright & Hillis 2014 we had during the past Fall semester. As I recall, we concurred that the paper was adequately executed in terms of running from premises to methods to conclusions. But we also thought (well, at least some of us) that it is yet another part of an unfortunate legacy in our field that tends to separate issues of ‘generating good evidence’ from issues of ‘identifying the right method of inference’. Lots of simulations were carried out when the matrix and characters in question were taken as is and essentially invisible.
A good (if seemingly vague) initial answer to the question “which method of phylogenetic inference should I use?” is: “Understand your evidence as much as possible, and get the right evidence for your targeted method”. Sound systematic practice is about optimizing the fit between (1) the evolutionary phenomena which the evidence embodies and (2) the inherent algorithmic assumptions of the method of inference (and robustness to violations thereof). Finding the right fit is often challenging, and only a posteriori verifiable. Still, there is only so much one can learn by evaluating (2) (largely) in isolation.
Almost inevitably the discussion turned to the issue of parsimony and morphology – clearly these two subjects have a storied past in systematics. And we saw the actual matrix which looked indeed rather challenging (and had been kindly shared by Michel Laurin – it was otherwise unavailable [not good]). As I recall, there were lots of multi-state characters, (subset) polymorphisms, inapplicables, missing data, etc. So then there was a legitimate (if somewhat baby/bathwater) argument made to just limit the use of morphology component in phylogenetic systematics to (and I paraphrase) ‘key characters in question’, letting the molecular evidence and non-parsimony inference methods drive the tree topology and character state optimizations. Something like that anyway.
I make no effort here to connect our discussion to the ‘morphology versus molecules’ literature, which is (1) vast and (2) the ‘versus’ is already not a good start to review it, and so on. Mainly I was a little underwhelmed with our graduate students’ ability to make a sound case for morphology, which after all is a substantial part of their thesis research projects. “Why are you (still) studying comparative (insect)* morphology to infer phylogeny and character evolution?” is a very valid question that one ought to have a suitable answer for. Here are some suggested answers.
Why are you studying comparative/phylogenetic insect morphology (in this day and age)?
- Studying comparative insect morphology satisfies my very personal need to express my appreciation (dare I say love) for the organismal group that I study better than studying non-morphological evidence.
- A significant part of the legacy of systematics is linguistic in nature. This is not accidental: human comprehension of natural phenomena in general has a strong language-centric tradition. Studying insect morphology perpetuates, reassesses, and refines this tradition. And almost invariably it turns out that immense refinements are needed.
- Morphology is only important if one is interested in addressing “why?” questions in systematics. Question: why is this specimen a representative of the monophyletic clade Perelleschus? Answer: the female spermatheca has a synapomorphic projection. Why would this other specimens also pertain to Perelleschus? It is legitimate to think that formulating and seeking answers for such (predictive) “why?” questions in systematics (or in science more generally) is worthwhile.
- Morphology is part of a sophisticated notion of ‘total evidence’ in systematics (which would mean: total relevant evidence; and what is ‘relevant’ is certainly subject to research and differential weighing). In enough cases it is the only, or the most powerful, evidence available.
- (Optional): Morphology, language, and parsimony often ‘work together well’. Those characters/states that permit cognitive and linguistic individuation and make reliable reference to deeper (phylogenetic) nodes also tend to perform well enough under parsimony, and in turn parsimony-based evaluation of wide ranges of characters allows successive and successful filtering of ‘the right evidence’ (which has the aforementioned properties). There seems to be an underlying (causal?) connection between (1) the amount of stability that evolutionary phenomena need in order to rise to the level of being individuated by us as evidence via language/predicates, and (2) their ability to act as reliable signals of monophyly under parsimony. Looking at the history of our field, it might be that much of what can be said sensibly about morphological traits with phylogenetic relevance can be said under parsimony-based inference (with Wittgensteinian pretentions, anyway).
And that may be good enough for now. Notice that none of these answers require one to weigh alternative kinds of evidence or methods of analysis less favorably. Morphology can make a positive case for itself, all by itself.
Of course this is not the complete picture. Surely one can make a wide-ranging, positive case for focusing on other data/inference methods. Comments on- or off-line are welcome.
* I cannot go into this, but the actual organismal group can and should have a large role in such discussions. I would like to think that bacterial systematists have a point in focusing on the evidence they tend to focus on, which looks a little different than many insect synapomorphies.
Related references (by NMF only):