RESUMEN
Hypotheses on regional song variation ("dialects") assume that dialects remain stable within regions, are distinct between regions, and persist within populations over extensive periods [1-3]. Theories to explain dialects focus on mechanisms that promote persistence of regional song variants despite gene flow between regions [4-6], such as juveniles settling in non-natal populations retaining only those songs from their repertoires that match neighbors [7, 8]. It would be considered atypical for a novel song variant to invade and replace the established regional variant. Yet some studies have reported song variants shifting rapidly over time within populations [9-11]. White-throated sparrows, Zonotrichia albicolis, for example, traditionally sing a whistled song terminating in a repeated triplet of notes [12], which was the ubiquitous variant in surveys across Canada in the 1960s [13]. However, doublet-ending songs emerged and replaced triplet-ending songs west of the Rocky Mountains sometime between 1960 and 2000 [11] and appeared just east of the Rockies in the 2000s [14]. From recordings collected over two decades across North America, we show that doublet-ending song has now spread at a continental scale. Using geolocator tracking, we confirm that birds from western Canada, where doublet-ending songs originated, overwinter with birds from central Canada, where the song initially spread. This suggests a potential mechanism for spread through song tutoring on wintering grounds. Where the new song variant has spread, it rose from a rare variant to the sole, regional song type, as predicted by the indirect biased transmission hypothesis [10]. VIDEO ABSTRACT.