Wednesday, June 27, 2012

Paradigms are like glue

Bateman; John Innes website
Uh oh.  An iconic study of sexual selection bites the dust. Well, with a bit of help. Researchers at UCLA attempted to replicate a 1948 study by English geneticist Angus John Bateman, but without success.  The original study, said to be second only to Darwin's Origin of Species in importance to the study of sexual selection, suggested an evolutionary advantage for males who are promiscuous and females who are choosy in selecting mates.  Indeed, one of the foundational paradigms of the field is named after Bateman.  "Bateman's principle" is (quoting Wikipedia) "the theory that females almost always invest more energy in producing offspring than males invest, and therefore in most species females are a limiting resource over which the other sex will compete." As the work was the basis of much subsequent research in the field, the failure to replicate is big.

A June 26 story in ScienceDaily cites the lead author of the new study, published June 11 in the online edition of PNAS:
"Bateman's 1948 study is the most-cited experimental paper in sexual selection today because of its conclusions about how the number of mates influences fitness in males and females," said Patricia Adair Gowaty, a distinguished professor of ecology and evolutionary biology at UCLA. "Yet despite its important status, the experiment has never been repeated with the methods that Bateman himself originally used, until now.
"Our team repeated Bateman's experiment and found that what some accepted as bedrock may actually be quicksand. It is possible that Bateman's paper should never have been published."
In the original experiment, Bateman isolated selected fruit flies in jars, 5 males, 5 females, or 3 of each, and allowed them to mate freely.  He then determined the number of offspring of each mating, which he was able to do because he chose adults with distinct mutations, such as narrow eyes or curly wings or thick bristles, that appeared in the offspring and were markers for parentage.  From Mendelian expectations, and assuming that the mutations did not affect survival, one-quarter of the offspring should be double-mutants, carrying a mutation from each parent, half would be single mutants, and one-quarter would be mutation-free. But the proportions of offspring in each class found by the replication done by Gowaty et al. "departed strongly from Mendelian expectations." 

Fly with curly wing mutation
Gowaty says that while this identification technique was clever, it turns out it also was responsible for the study's fatal flaw.  For his identification purposes, the offspring that Bateman counted had to be double-mutants but these flies turn out to be less viable.  Thus, Bateman "overestimated subjects with zero mates, underestimated subjects with one or more mates, and produced systematically biased estimates of offspring number by sex."  Indeed, he assigned more offspring to fathers than mothers, which is why he concluded that male fruit flies produce many more viable offspring when they have multiple mates but that females produce the same number of children that survive to adulthood whether with one mate or many.
In their repetition -- and possibly in Bateman's original study -- the data failed to match a fundamental assumption of genetic parentage assignments. Specifically, the markers used to identify individual subjects were influencing the parameters being measured (the number of mates and the number of offspring). When offspring die from inherited marker mutations, the results become biased, indicating that the method is unable to reliably address the relationship between the number of mates and the number of offspring, said Gowaty. Nonetheless, Bateman's figures are featured in numerous biology textbooks, and the paper has been cited in nearly 2,000 other scientific studies.
This is not the first time Bateman's study has been questioned -- in fact so many species violate the principle that many evolutionary biologists have suggested that it shouldn't in fact be considered valid.  But, as Gowaty points out, "Bateman's results were believed so wholeheartedly that the paper characterized what is and isn't worth investigating in the biology of female behavior."  And, she says, "Paradigms are like glue, they constrain what you can see.  It's like being stuck in sludge -- it's hard to lift your foot out and take a step in a new direction."

The term 'paradigm shift' was first proposed by Thomas Kuhn in his influential The Structure of Scientific Revolutions, first published in 1962.  He argued that most of the time we are doing 'normal science', designing studies and interpreting results in terms of an accepted theoretical framework.  We cling to that theory the way a religious zealot clings to a sacred text, resisting any breech of the accepted truth.  Difficult results are explained or rationalized away, until a truly better idea comes along.

In the paper, Gowaty et al. conclude
The paradigmatic power of the world-view captured in Bateman’s conclusions and the phrase “Bateman’ Principles” may dazzle readers, obscuring from view methodological weaknesses and reasonable alternative hypotheses...
Here we don't see a new theory but just a refutation of an iconic proof of the existing theory, but the clinging drive is the same. This is of course true in any field, but it may be even more true of beliefs we (including scientists, who we idealize as being objective) hold about ourselves -- our inherent nature and our behavior, and their evolutionary origins and value.  What is, or what we choose to see, must be good or 'natural' because it evolved.  It is why so many stories, including fossil accounts (see Holly's award-winnning post), lead to resistance or even resentment when they are questioned by those the majority typically denigrate as nay-sayers.

We as scientists like to believe we interpret our data free from ideology.  That is rarely if ever true.

3 comments:

JKW said...

Disclaimer: I have not yet read the Gowaty et al. paper. My immediate question, though, is whether the study refutes the Bateman Principle or just Bateman's original methods used to propose the principle. How many studies post-Bateman's original 1948 study have replicated Bateman's principle using different methods? The point I'm getting at is that the original methods may have been abandoned without the principle itself being undermined in any meaningful way. It was my understanding that Bateman's Principle (later extended by Trivers in the 1970s) simply explained the opportunity for intrasexual selection to be at play: the idea that the faster reproducing sex (which in humans is males) tends to have greater variance in reproduction than the slower reproducing sex (which in humans is females). It was distinct, I thought, from the operational sex ratio and intersexual selective forces that may also be at play simultaneously (and influencing phenotypes in similar or contradictory ways). Recognizing I'm not an expert in this area but believing I'm at least competent, it was my understanding that the Bateman Principle itself was not about "promiscuity" or "choosiness" at all, but, rather, just about the idea that there are differences in reproductive variance between the sexes upon which a host of sexual selective pressures (intra- and inter-) could act. I'd love to see a follow-up post on MT about how the "paradigmatic power" of any paper is different now than it was in the 1940s when Bateman did this (or even 1970s when Trivers renewed interest in Bateman's work), thanks to the ability to distribute one's work without a publisher and without the "normal scientists" (to use Kuhnian terms) giving the stamp of approval before ideas challenging the paradigm can see the light of day.

Anne Buchanan said...

Yes, thanks, very good point(s). This is not my area, but my understanding is that this failure to replicate the original experiment is not the first time that the Bateman Principle has been questioned, just perhaps the loudest. Enough species do not follow the principle that many evolutionary biologists have already questioned its status _as_ a principle (that is, it's not always true that females invest more in offspring than do males). And some have questioned Bateman's methodology in the past as well, including some of the authors of the new paper.

But, how is a principle refuted? Is one experiment that fails to replicate enough? Is refutation the result of a gradually accumulating body of data? Or a sudden change in what people are willing to accept or believe about the field? I think this is what's interesting about the culture of science (and other belief systems), and what the Gowaty paper means for that, as we tried to address in the post.

Indeed, despite some grumblings about Bateman's methodology over the years, and many known counter examples, his principle still stands. Again, this isn't my area and I don't know the literature in detail, but this is what interested me about the Gowaty paper. And of course the fact that data don't speak for themselves but must be interpreted and the interpretation generally fits, and emerges from a paradigm.

Do you think that changing ways to distribute one's work will change the way that paradigms shift? Or are created? Will they be easier to challenge? I don't know. My tendency is to say it won't make much difference, or even that it might be even harder to challenge existing paradigms because so much money is invested in current ones. We here on MT try our best to challenge paradigms, e.g., but can't say we've been very successful at it!

Ken Weiss said...

The point here would be that this kind of selection doesn't always happen, is difficult to demonstrate, and is therefore not a central law of life.

If it does happen in some, or even many, instances, then fine. That is just an interesting aspect of nature. But it does not qualify as a dogma or the kind of near ideological generalization that oversimplifies how evolution actually works.

Backing away from such oversimplifications is, in my view, the most important thing we can do to shore up our science.

Now, by the way, because it's been held as such a rule, or one must say as such an ideology, in various debates, it provides fodder for the nutcase Creationist community. And we deserve their scorn to that extent.

But the nutcase aspects are:

(1) To think that scientists ever get anything totally right and that this new result (even if wholly true) is a devastating blow to evolutionary theory.

(2) And the totally false syllogism by which it is inferred that because we get some aspect of evolution wrong, that some particular alternative explanation (Creationism) is therefore true.

We'll never convince the nutcase community. But we should, for the sake of science, be more circumspect when we make our striking adaptationist or causal claims about life