Friday, July 20, 2012

Standards for evolutionary inference: some happy place between Baba Brinkman and Dick Lewontin.


Last week I was at the Evolution 2012 meeting in Ottawa. A joint meeting of five evolution and ecology societies, it was the largest collection of evolutionary biologists (plus some ecologists) ever assembled in a single place. It was an incredibly dynamic and exciting meeting with tons of opportunity for discussion and argument lubricated with appropriate – if expensive – beverages. Two presentations in particular, keynote addresses of a sort, stimulated a series of arguments that led to the reflections contained in this post.

UBC biologist Rosie Redfield gave a public outreach talk that detailed the debacle surrounding a paper published in Science where a bunch of NASA scientists claimed to have found bacteria that could construct their DNA backbone with arsenate rather than phosphorous. If true, this would have incredible implications for our understanding of life on earth and beyond. Rosie had read this paper and written in her research blog (http://rrresearch.fieldofscience.com/) about how horrible it was – stated simply, the work was far from providing the standard of evidence necessary for the inferences it was attempting to make. Rosie quickly became a scientific icon of the blogosphere for her withering criticisms of the paper. Rosie then redid the studies with the proper standards and showed that the original result didn’t hold – all organisms do in the end probably use phosphorus. Rosie’s paper was also published in Science – and it appeared online precisely in the middle of her talk, which was not a coincidence.  It has been said that “negative reviews often give a frisson of pleasure the reader” (Houle 1998 - Evolution) and so do negative talks to the listener – at least to many of them.

The other “keynote” I want to mention was the performance by Baba Brinkman of his “Rap Guide to Evolution.” I had known of, and greatly appreciated, this work of science/art/poetry/music for a few years but this was the first time I had seen it in person. In addition to giving great performances of some of the work I already knew of, he had some new – and equally hilarious and insightful – stuff (Don’t Sleep with Mean People was a highlight). Everyone should hear and see Baba’s work: http://www.bababrinkman.com/. Coincidentally, I happened to run into Baba during one of the mixers and also at bars on two other nights, which formed the second inspiration for this blog about standards of evidence.

Baba dropping an evolutionary rap anthem.
Baba’s work summarizes in entertaining and generally correct form many of the basic underpinnings – and also unique nuances – of evolutionary biology. He also strays into evolutionary psychology. Although prefacing his dialog with a disclaimer that much of the field might well be bunk, he is clearly a fan – even if only because of how it makes one think about human behavior.  His favorite example is the correlation among human communities between mortality rates and teen pregnancy rates (and other things). He explains in a rapid rap flow the hypothesis of Daly and Wilson (1988) about how it makes evolutionary sense to reproduce early if your mate is likely to die soon. In short, teen pregnancy might be an adaptive reproductive strategy (or tactic) in response to high-risk situations. The graph relating mortality rates (x-axis) to teen pregnancy (y-axis) is extremely strong; indeed, it would be the envy of most empirical evolutionary biologists working on any organism. But (to continue the point of my previous post on Faith’s Conjecture: http://ecoevoevoeco.blogspot.ca/2012/07/faiths-conjecture.html) how can we be sure this is a causal relationship. Perhaps high teen pregnancy rates (x-axis) lead to fathers adopting riskier and more aggressive strategies that increase mortality rates (y-axis). Or maybe both phenomena are independently driven by some external factor. What standard of evidence would be necessary to make this conclusion with surety?

In my conversation with Baba, he argued that work on humans should be held to the same standard of evidence as work on other organisms. With other organisms, however, we would perform experiments. We would alter mortality rates and see if it changed the age of reproduction. Indeed just such experiments have been done: when guppies are introduced from high-mortality environments to low-mortality environments they evolve delayed reproduction (and vice versa). But we can’t do this sort of controlled experiment with humans, of course. So we clearly can’t hold work on humans to the same standard of evidence – but we should push as far as possible. Or should we?

In 1999, I was in the audience for a symposium called “Darwinian Evolution Across Disciplines” held at Dartmouth College. A collection of speakers from a variety of disciplines (religious studies, anthropology, medicine, etc.) spoke about how evolutionary thinking had influenced their disciplines. (The representative real evolutionary biologist, Dick Lewontin, spoke about cosmic evolution – go figure.) At the end of the symposium, all of the speakers lined up their chairs at the front of the room and the audience was allowed to ask questions. My question started with the preface that evolutionary biologists were heavily criticized (by Dick) for telling “just so stories” to explain the adaptive significance of trait variation. (That is, inferences weren’t based on an appropriate standard of evidence.) I then noted out loud that it seemed like most of these other disciplines were still in the “just so” phase, and I asked what the prospects were for transitioning into the rigorous hypothesis testing now employed in evolutionary biology. A few speakers gave some comments and then one of them turned to Dick, who had not yet spoken, and asked something like “What do you think Dick?” After pausing for effect, he turned to face all of the other speakers and said something like “I don’t think that any of you know how evolutionary biology works.” And then he went on to disparage their fields for a few minutes before turning to me and saying that he still felt evolutionary biology was in the “just so” phase. Then he walked out. Frisson! (A video exists of Dick chastise us [although I can’t seem to access it right now]: http://www.dartmouth.edu/~dead99/Disc/Disc.htm. )

The lesson I think we should take from all this arguing and frissing is that we should push evolutionary psychologists to base their inference on the highest possible standard of evidence THAT IS APPROPRIATE FOR WHAT IS POSSIBLE. That is, we simply can’t perform evolutionary experiments on humans. We need to find some happy point between Baba Brinkman and Dick Lewontin – probably a lot closer to the former than the latter.

And, of course, we should strive for exactly the same thing in evolutionary biology in general: the highest possible standard of evidence. But striving is different from achieving, and so it is reasonable to ask should we also HOLD evolutionary biologists to those standards – by, for example, blocking the publication of studies that don’t perform manipulative experiments in nature or that don’t use the latest genetic methodologies? I don’t think so. First, incredible inequities exist in the manipulability and genetic resources for different taxa. We simply can’t expect people who work on mosquitofish and walkingsticks to deploy the same genetic tools as people who work on flies or mice or stickleback or humans. And we can’t expect people who work on elephants or hippos or albatrosses or humans to perform the manipulative evolutionary experiments typical of flies and mice and bacteria. This doesn’t mean that we should all work on those “ideal” model systems – we really do need to study the full pageantry of life. The solution then must be to hold work on a given taxon to the highest possible standards for that taxon – standards must be scaled to taxa.

Or is that really the case? Can we really expect everyone who works stickleback to fully sequence all the individuals in their study populations? Can we really expect everyone who works on guppies to perform manipulative evolutionary experiments? Clearly not, as the resources and opportunities to do so are out of reach of most people working even on these groups. Thus, we can’t apply a particular standard of evidence to all papers from a given taxonomic group – even in Science and Nature! Otherwise, we are always chasing the latest technology or the most money or the biggest collaborations – and we are saying, in essence, that work published just a few years earlier would not be worth publishing in the same form today. This is nonsense. The major advances in evolutionary biology were ideas – and ideas are what stand the test of time (V!). So I would encourage all of us, whether acting as authors or reviewers or editors or bloggers or rappers to strive for the highest possible standards of evidence in our own work while also judging the work of others according to what is reasonable and possible for them – and to what extent their ideas are interesting and stimulating. (Just think how much less exciting the last year would have been without those NASA scientists.) We need to find some happy place between Baba Brinkman and Dick Lewontin.

Baba, Mr. Simmonds, and Hendry lab graduates Xavier (far left) and Erika (far right)

2 comments:

  1. I think all we really need is conclusions that don't stray beyond our evidence.

    ReplyDelete
  2. That is indeed an alternative way to say it!

    ReplyDelete

Predicting Speciation?

(posted by Andrew on behalf of Marius Roesti) Another year is in full swing. What will 2024 hold for us? Nostradamus, the infamous French a...