Turkey Snoods
Thanksgiving dinner. A time of gratitude and feasting. For some, a time of reflection ("Am I truly related to these people?"). But whatever else you experience this year, there won't be lulls in the conversation, because thanks to this newsletter, you're going to have so much to say about turkey snoods!
Actually, my purpose here is to present a case study on how weak statistics can yield findings that become enshrined as scientific fact. But this does require some discussion of snoods.
The snood
As you may know, a snood is the fleshy thing that hangs down over a turkey's beak. In young turkeys and hens, the snood may be no more than a bump. In adult males, the snood can be quite long, as shown in the photo below.
Observational data on this appendage is plentiful: Snoods get longer and redder when the male attempts to mate. Snoods retract when the turkey feels threatened. A threatened, angry turkey exhibits a redder snood than a threatened, frightened turkey. Et cetera.
Snoods serve no clear functional purpose. In the 1990's a biologist now at the University of Mississippi, Richard Bucholz, began to investigate whether they attract female turkeys, the working hypothesis being that females have a preference for larger snoods. (Here I will quietly tiptoe around the dirty jokes that could be made. Thanksgiving is a family holiday, after all...)
The possibility that snoods attract mates may sound pretty simple, but it represents a complicated, still-debated solution to a problem first identified by Darwin.
Sexual selection
The problem is this: Darwinian natural selection holds that only characteristics that promote survival get passed from generation to generation. If you accept this view of evolution, it's hard to explain features like turkey snoods, lion manes, and, of course, the feathers of male peacocks. These features don't help males survive. In fact, they may be counterproductive, because they require energy to produce, and they may even burden those who possess them. The flight of a male peacock is actually hampered by his tail feathers, for instance. If the bird in the photo below were fleeing from an aerial predator, that tail would actually decrease his chances of survival by slowing him down.
Darwin's solution to the problem was to propose that sexual selection accompanies natural selection. If female peacocks prefer mates with larger, gaudier tail feathers, males who fit that description just need to survive long enough to reach sexual maturity. Those tail feathers will then pay off, because the males will be selected as mates and pass their genes on to the next generation.
Sexual selection may sound plausible, but it begs the question: Why would females prefer long, gaudy tail feathers in the first place? Why would they prefer tail feathers with any particular appearance? Darwin struggled with this question (at one point, he wrote that "the sight of a feather in a peacock's tail makes me sick"), and modern scholars don't agree yet on an answer. The question isn't just about peacock tails, of course. It's a question about why, in some species, females tend to select mates with specific physical features.
One of the most commonly proposed answers is the "good gene hypothesis": Females prefer features that imply good genes. If larger, gaudier tails indicate better genes – e.g., those that promote better health – then female peacocks who select mates with such features are more likely to end up with healthy babies. This perspective informed Dr. Bucholz's research, which is responsible for, among other things, establishing the following scientific facts about wild turkeys:
1. Females prefer males with larger snoods.
2. Larger snoods indicate better health.
I call these "scientific facts", because you can find them in textbooks, in scientific articles, in science journalism (e.g., Smithsonian magazine), and on Wikipedia. As it turns out, these facts stem from a single study published by Dr. Bucholz in the journal Animal Behaviour in 1995. It's a clever study, in my opinion, and yet....
Does size really matter?
In this study, 23 female turkeys were introduced, one at a time, to a large "mate choice arena" (see diagram below) where they were allowed to move around freely in the presence of two cordoned-off live males who strutted however they pleased. The males were drawn at random from a small pool. Each female's preference was indicated by a species-specific behavior called "crouching", which indicates sexual receptivity.
As five of the females exhibited no preference, data analyses focused on the choices of the other 18 females. Bucholz found that these females tended to prefer males with longer snoods. No effects were found for other anatomical features of the males, or for the intensity of their strutting.
Now, you might expect me to complain about the tiny sample size, but I want to skip ahead to a more serious concern. The conclusion that females prefer larger snoods was based on two analyses. One (a t-test) showed that the 18 males who were selected had significantly larger snoods, on average, than the 18 males who weren't selected. Another (a principal component factor analysis) showed that a combination of snood size and width of skull was the only factor associated with female selection.
The problem with both of these analyses is that they collapse across females’ individual selections. Why this is a problem is easiest to explain regarding the first analysis. The fact that the selected males had larger snoods, on average, could be due to the influence of just one or two males with huge snoods that inflated the mean for this group. Mathematically, it's possible that most of the females actually chose the males with smaller snoods.
Unfortunately, the stat that would resolve this problem isn't presented anywhere in the article. This stat – how many of the 18 females chose the male with the larger snood – would've been highly informative. For instance, if 17 out of 18 females chose the snoodier male, we might agree with Bucholz that snood size matters. However, imagine that only 7 females chose the male with the larger snood, and that 11 chose the male with the smaller snood. The average snood length for these 18 chosen males might actually be longer than for the 18 who weren't chosen, if a few of the chosen ones were especially snoody. But, since only 7 males with larger snoods were chosen, the data would contradict Bucholz's conclusion.
The more sophisticated analysis that Bucholz ran addresses other potential problems. For instance, snood size is correlated with body mass, so if you want to claim that females prefer larger snoods, you have to control for mass, which Bucholz did. But we're still left with the problem of not knowing how many females chose the male with the larger snood.
Although Bucholz didn't report that number for this experiment, he did so in the second experiment of his study. This experiment had the same set-up as before, except that this time, Bucholz meticulously constructed two male turkey decoys, identical in appearance other than snood length. He also installed speakers behind each decoy that played recordings of male turkeys strutting.
Since the two decoys sounded and looked like each other (except for snood length), this seems like a rigorous way to test the hypothesis that snood length matters. Unfortunately, out of 23 females tested, only 8 showed a clear preference. According to Bucholz, 7 of the 8 preferred the male decoy with the longer snood.
Here I think the problem of small sample size is inescapable. Scientific "facts" should not be based on sample sizes of 8. More importantly, what Bucholz found was not a 7-out-of-8 preference, but rather a 7-out-of-23 preference. In other words, 7 out of the 23 females (i.e., 30%) showed a preference for the decoy with the longer snood, while one showed a preference for a shorter snood. We don't know anything one way or the other about what kind of mate the other 15 females preferred. For all we know, all of them actually preferred smaller snoods!
In sum, the results are inconclusive, and we don't really know whether females turkeys prefer larger snoods.
(In a later study, Bucholz showed that among male turkeys, larger snoods are recognized as a sign of dominance, but in this study the stats are once again weak and fail to properly disambiguate larger snoods from larger body mass. This study too is the only one of its kind. In the end, I believe that the purpose of the snood remains unknown.)
Conclusion
At this point you may be thinking: So what?
In other words, if you accept my argument in this newsletter, all we need to do is to insert a bunch of "maybes" in front of scientific facts about the purpose of turkey snoods. Why should we care whether these odd little facts get demoted to the status of possibilities?
Here are three reasons why I consider the issue important:
1. It matters for zoologists.
Researchers continue to debate the basis of female mate selection in particular animal species. The good gene hypothesis has competitors, and in some species, there's not even consensus as to the nature of the preferences these hypotheses are meant to explain. In short, for zoologists (particularly those who study ethology, or animal behavior), the purpose of snoods matters.
2. It matters for the integrity of science.
It's not uncommon for scientific facts to be grounded in a single study that later turns out to be flawed. This suggests a need for replication and vigilance. Perhaps it's no big deal (outside of zoology) if we're wrong about the purpose of turkey snoods, but if we keep accepting "facts" based on single studies, we're going to be wrong a fair amount of the time.
To illustrate what "a fair amount of time" might mean, consider the Reproducibility Project. In 2015, Brian Nosek of the University of Virginia led a collaborative effort to replicate the significant findings of 97 studies published in three of the most prominent psychology journals in 2008. Nosek and colleagues took great pains to duplicate the methodologies used in the original studies, often consulting with the authors of those studies to get the details exactly right. In the end, they only obtained significant findings in 35 of the 97 cases. In other words, roughly two-thirds of the time, Nosek and colleagues were unable to replicate the findings of the original studies. This is a disturbing finding, given the prominence of the three journals and their influence on what constitutes scientific "fact" in psychology.
3. It matters for all of us.
Evolutionary accounts of animal behavior have a way of escaping their cages. What I mean is that there's a long and not altogether lovely history of sociobiologists and other scholars applying what they've learned from animals to our particular species.
The issue here concerns all people, not just students of animal behavior. It's one thing to debate the source of "crouching" behavior in female turkeys and how that indicates sexual receptivity to the brain of a male turkey. As long as you're talking about turkeys and no other species (other than birds), that's innocuous. It's another thing altogether to say anything about human sexuality that's informed by observations of other species. And yet much has been said, and some of it has been deeply offensive to women.
One of the most famous examples is controversy over the book Sociobiology: The New Synthesis, first published by Harvard entomologist E. O. Wilson in 1975 and still in print. The title of this book spawned a new discipline focusing on the role of evolution in animal behavior. Arguably, the first 600+ pages of this book are fine, in the sense that they only rile up experts who disagree with Wilson about technical issues. It's the final chapter, entitled "Man" (and only 33 pages long) that creates trouble. According to some, this chapter presents a kind of biological determinism that includes, among other things, the idea that behavior is inescapably gendered. Thanks to our genes, the idea goes, men and women each exhibit distinctively gender-specific behaviors, regardless of how we're socialized.
The issues are complicated, in part because scholars disagree with each other, and with Wilson, about what Wilson was actually claiming. The key point is that because humans are animals too, any broad claims about animal behavior matter. The more wrong we are about the role of evolution in shaping animal behavior, the more wrong we're likely to be about human behavior. The fact that males compete for females in some species doesn't mean they do so in the same way in all species, or that females aren't also competing for males, or that competitiveness isn't a learned, malleable behavior. Reductionism is a deceptively easy path. It's easy to think of human males, for example, as hard-wired to strut and spar and otherwise "peacock" as they compete for female attention. (You know the cliches…. the middle-aged guy who wears an expensive snood and boasts about the expensive snood he drives, or the famous snoods he hangs out with, or the time he spends in the gym toning his awesome snood...). But maybe human behavior is more complicated than that - and more readily changed as younger generations of kids are raised differently.
In short, in a small way, it matters for us whether larger snoods automatically attract female turkeys (one more datapoint in favor of biological determinism) or whether they don't (one more datapoint in favor of a complex, malleable interplay between genes, environments, and behavior).
Finally, although I believe that the purpose of snoods isn’t known yet, I’ve noticed that in recent years the word “snood” has been appropriated and misused, anatomically speaking, to describe a kind of head covering that’s used to protect (or humiliate) some of our favorite species (see below).
Poor guy. I’m not sure dogs consider this the most attractive look…
Thanks for reading!