Brett L. Walker (2024)
Applying citizen science data to quantify differences in song between controversial avian taxa, the Sagebrush and Timberline subspecies of the Brewer's Sparrow (Spizella breweri). Western North American Naturalist 83: 550-568.
Applying Citizen Science Data to Quantify Differences in Song between Controversial Avian Taxa, the Sagebrush and Timberline Subspecies of the Brewer's Sparrow (Spizella breweri)
Abstract
Divergence in the acoustic structure of mate-attraction signals between closely related taxa can contribute to reproductive isolation and speciation, and acoustic differences are often useful for identifying otherwise similar taxa from recordings. The Brewer's Sparrow (
Spizella breweri) is a migratory oscine songbird with 2 recently diverged subspecies. The nominate subspecies (
S. b. breweri) primarily breeds in sagebrush (
Artemisia spp.) shrublands in the western United States and southwestern Canada, whereas the Timberline Sparrow (
S. b. taverneri) breeds in alpine shrubs and conifer
krummholz near treeline in mountain ranges from east-central Alaska to northwestern Montana. Quantitative comparison of the acoustic structure of Brewer's Sparrow mate-attraction songs is needed to help resolve the outstanding debate regarding the taxonomic status of
S. b. taverneri and improve subspecific identification from song recordings. I compared the acoustic structure of 181 short song types from 180 male
S. b. breweri and 22 short song types from 20 male
S. b. taverneri recorded from across each subspecies' breeding range. Despite extreme variation in acoustic structure of short songs among individuals, 19 acoustic variables differed between subspecies. Short song types of
S. b. taverneri averaged 19.3%–22.1% lower maximum frequency, 4.9%–15.9% higher minimum frequency, 36.7%–42.6% narrower frequency bandwidth, and 12.9%–26.0% lower aggregate entropy (a measure of sound complexity) than those of
S. b. breweri across acoustic elements (song, section, syllable, and note). Random forest classification analysis identified 10 acoustic variables important for distinguishing between short songs of the 2 subspecies and correctly classified 89.2% (95% CI 84.1% to 93.1%) of songs overall. Principal component scores representing overall acoustic structure showed an abrupt change near the breeding boundary between
S. b. breweri and
S. b. taverneri but overlapped sufficiently that they did not meet an established “75%–99% rule” for reciprocal diagnosability of subspecies. An easy-to-measure logistic regression threshold based on song bandwidth and mean note bandwidth accurately classified >90% of short songs to subspecies. Additional recordings of
S. b. taverneri are needed to improve our understanding of geographic variation in the song of that subspecies. Data on responses of males and females to songs of the other subspecies would inform whether acoustic differences are sufficient to impede mate recognition and attraction in areas of potential breeding overlap.