Last week I tried to make my first post, and had a blowout. Here's take #2, with thanks to Erik and Maren for walking me through the details.
In short, the answer is "yes," because we have to study the evolution of something, and when one considers mechanism our understanding is greatly enriched. This perspective is explicitly developed in a paper by Autumn, Ryan, and Wake (2002) "Integrating historical and mechanistic biology enhances the study of adaptation". We read this paper a few months ago in lab meeting. To be fair, views on the paper were mixed. Yes, it was too much a response to a couple controversial papers by Sherman and Reeve, and it was overly pessimistic regarding the value of studies of selection in nature. But, it was also, in my view, a superb, empirical demonstration of the importance of mechanism and phylogeny. The authors show, very clearly, that when studying adaptation, one is one shaky ground when when one ignores such things. In the end, convincing studies of adaptation need to be integrative in scope (I would include studies of selection in nature here, too).
And that is why papers on how an isopod swims, or a limb develops, or a dragonfly flies are important.
The importance of mechanisms, I think, is that they might inform us about the costs of achieving some "optimal" solution. For instance,selection might not necessarily favour an adaptive behaviour, because the costs in terms of e. g. an extensive neurobiological machinery might outweigh the fitness benefits.
ReplyDeleteThese costs we might only be able to understand if we also investigate the mechanisms behind current adaptions. That is one, out of several good arguments, why evolutionary biologists should pay attention also to mechanisms and not treat them as a "black box".
Take, for instance, the famous "G-matrix". This is a statistical description of variances and covariances between traits AT THE POPULATION LEVEL. It does not, in itself, tells us anything about the mechanisms behind (say) genetic correlations (pleiotropy, linkage) or the developmental basis of trait correlations. The G-matrix is thus a pattern, that might reflect a multitude of different mechanistic underlying processes.