Lynch, A., G. M. Plunkett, A. J. Baker, and P. F. Jenkins. 1989. A model of cultural evolution of chaffinch song derived with the meme concept. American Naturalist. 135:634-653.

(Chad Brassil)

This was one of the first empirical studies to utilize the meme concept. The defined memes loosely as any unit which can be transmitted. They studied Chaffinch populations which were introduced to New Zealand at the end of the 19th century. They recorded at least 20 songs from each of 30 birds at each of 19 sites across New Zealand. Characterizations of the songs were done visually using sonograms. They used two classification schemes, syllable variants for fine classification and syllable types for course classification.
To compare memes, they count all shared syllables and all shared syllable combinations, well redundant, this weights longer memes. They performed spatial autocorrelation analysis, fit the neutral alleles model, and used the rare alleles model to estimate gene flow.
The found a stunning 10,256 syllable variants and 4490 syllable types. Shorter memes were more likely to be shared than longer memes. Populations closer than 50 km were more likely to share memes than by chance; populations further than 120 km were less likely to share memes than by chance. For long memes, there is no correlation beyond 26 km as these memes are broken up by mutation and recombination. Only the syllable types display spatial clines; syllable variants are too highly mutated. They hypothesize that syllable type clines could be due to the historical colonization pattern.
They estimate that on average, 2 song types per generation leave the population, balanced by an equal number entering the population. This suggests that one migrating male enters and a population per generation. This also suggests that mutation rates are very high. Mutation rates are obviously higher for longer memes which results in greater population differentiation based on those memes.
They found that classic population genetic techniques applied fairly well to meme analysis. They found that high rates of mutation and drift oppose meme flow to maintain, and this case create, high diversity.