Wrinkled fingers, spandrels, and an adaptationist approach to applied evolutionary biology
I was asked to talk to a local TV reporter about the winkled-finger study. Didn't happen, but it got me to read the paper and think about writing about it here. But Ryan Gregory has already done a much better job than I would have. He used this paper as an example of the widespread tendency to assume that, if some trait serves a useful function today, that must be why it evolved.
To test such hypotheses, he suggests that we (among other things):
"3) Explicitly address the necessary assumptions about ancestral traits, habitat, selective coefficients, and population size.
4) Use the comparative method within a phylogenetic context. Identify other species that have the trait and those that lack it. Has the trait evolved independently under similar conditions? Is it found in species that would not be expected to exhibit it if it evolves adaptively for a particular function?
5) Consider and rule out non-adaptive explanations (developmental constraints, pleiotropy, etc.) as much as possible."
This seems reasonable, if our goal is to figure out why some trait evolved. But, as an applied evolutionary biologist, I am less interested in why solar tracking evolved or how legumes first came to impose sanctions on rhizobial "cheaters" than in figuring out whether it would be possible and useful to enhance (or perhaps eliminate!) these traits in crops.
But many of Gregory's points still apply. Consider Ryoko Oono's recent work in my lab. We knew that some legume species make their symbiotic rhizobia swell up, as the rhizobia differentiate into nitrogen-fixing bacteroids. She found that a given strain of rhizobia fixes nitrogen more efficiently (more N, relative to its respiration cost) in a host where its bacteroids are swollen. She repeated this with another rhizobial species in another pair of hosts.
But can we generalize, from only one strain each of two rhizobial species? Is this enough evidence to justify trying to breed this trait into soybean? Probably not, but there's only one other reported example of a rhizobial species that swells in some hosts and not in others. Despite considerable effort, she couldn't get that strain to nodulate one of its reported hosts.
So Dr. Oono took a phylogenetic approach, analogous to that advocated by Gregory. Based on ancestral state reconstruction, she concluded that legume traits that make make bacteroids swell have evolved at least five times. These independent evolutionary transitions suggest that imposing bacteroid swelling (or doing something else that consistently has swelling as a side effect) might be broadly useful to legumes, at least under past conditions.