Genotype predicts quantitative song variety in a chickadee hybrid zone despite limited sampling

Abstract In the avian suborder Passeri (the songbirds), song develops according to both a flexible neural template and auditory input from conspecifics, making innately constrained characters of song difficult to isolate. In a hybridizing population of Poecile atricapillus (Black-capped Chickadee) and P. carolinensis (Carolina Chickadee), we found that genetic ancestry was weakly predictive of a multidimensional measure of song variety (a continuously distributed quantitative alternative to categorical song repertoire size) but did not successfully predict one-dimensional song variety. We used species-diagnostic autosomal markers to genotype 55 individuals inside and outside of the atricapillus/carolinensis hybrid zone in Missouri and Kansas. Using active recording methods, we then obtained high-volume, high-quality song recordings of 10 genotyped chickadees from a single hybrid zone population on a small, lake-bounded peninsula in west-central Missouri. We extracted acoustic data from these recordings to generate measurements of song variety across 1, 2, and 3 dimensions of multivariate acoustic space for each individual. We tested how well, and in what direction, genetic ancestry predicted song variety for each of these dimensionalities, after predicting that song variety would increase with carolinensis ancestry. Linear models predicting song variety in 2 and 3 dimensions from genetic ancestry ranging from carolinensis-like backcrosses to pure carolinensis explained 41% and 43% of the variation, respectively, with slope values in the predicted direction, suggestive of genetic predispositions for multidimensional song variety although p-values fall marginally short of classical thresholds of statistical significance. A linear model predicting song variety in 1 dimension from genetic ancestry explained 12% of the variation and did not approach statistical significance. Our findings provide support for the continued use of multidimensional song variety measurements and offer future directions for tackling the question of the genotype-song relationship in hybrid zones between species with vocal learning.