(This article was originally published at Statistical Modeling, Causal Inference, and Social Science, and syndicated at StatsBlogs.)
Donald Kroodsma writes:
Birdsong biologists interested in sexual selection and honest signalling have repeatedly reported confirmation, over more than a decade, of the biological significance of a scatterplot between trill rate and frequency bandwidth. This ‘performance hypothesis’ proposes that the closer a song plots to an upper bound on the graph, the more difficult the song is to sing, and the more difficult the song the higher quality the singer, so that song quality honestly reveals male quality. In reviewing the confirming literature, however, I can find no support for this performance hypothesis.
OK, that sounds jargony, so let me make it clear: when Kroodsma says he “can find no support for this performance hypothesis,” what he’s really saying is that a sub-literature in the animal behavior literature is in error. Rip it up and start over.
How did everyone get it wrong? Kroodsma continues:
I will argue here that the scatter in the graph for songbirds is better explained by social factors and song learning. When songbirds learn their songs from each other, multiple males in a neighbourhood will sing the same song type. The need to conform to the local dialect of song types guides a male to learn a typical example of each song type for that population, not to take a memorized song and diminish or exaggerate it in trill rate or frequency bandwidth to honestly demonstrate his relative prowess. . . . There is no consistent, reliable information in the song performance measures that can be used to evaluate a singing male.
But the other side is not going down without a fight. Jeffrey Podos responds:
Kroodsma [in the above-linked article] has critiqued ‘the performance hypothesis’, which posits that two song attributes, trill rate and frequency bandwidth, provide reliable indicators of singer quality and are used as such in mate or rival assessment. . . .
I address these critiques in turn, offering the following counterpoints: (1) the reviewed literature actually reveals substantial plasticity in song learning, leaving room for birds to tailor songs to their own performance capacities; (2) reasonable scenarios, largely untested, remain to explain how songs of repertoire species could convey information about singer quality; and (3) the playback studies critiqued actually enable direct, reasonable inferences about the function of vocal performance variations, because they directly contrast birds’ responses to low- versus high-performance stimuli.
Where did the critics go wrong? Podos continues:
My analyses also reveal numerous shortcomings with Kroodsma’s arguments, including an inaccurate portrayal throughout of publications under review, logic that is thus rendered questionable and reliance on original data sets that are incomplete and thus inconclusive.
I have not read either paper because it just all seems so technical. I suppose with some effort I could untangle this one, but I don’t feel like putting in the effort right now.
Any ornithologists in the house?
Fun fact: Both authors in this discussion had the same academic affiliation of Department of Biology, University of Massachusetts, Amherst. Podos is a professor there, and Kroodsma is a retired professor. Either way, the story is compelling: youngster does shoddy research and the retired prof blows the whistle, or cranky old man can’t handle new methods. In some general sense, I’ve been on both sides of this debate: Sometimes I criticize what I see as flashy research with empty claims, otherwise I’m frustrated that traditionalists will seem to find any excuse not to take a new method seriously.
The post Bird fight! (Kroodsma vs. Podos) appeared first on Statistical Modeling, Causal Inference, and Social Science.
Please comment on the article here: Statistical Modeling, Causal Inference, and Social Science