Birdsong is an important secondary sexual trait which may vary between but also within species. Intraspecific variation is generally studied either on the geographical or on the temporal scale; most of the studies exploring the variation of song over time, however, focused on species with rather simple songs. In this study, we explored the temporal changes in song of a complex songster, the Thrush Nightingale (Luscinia luscinia), recorded after 33 years (in 1986 and 2019) at the same locality in south-eastern Finland. Our analysis revealed a complete turnover of song types over the study period, with no song type shared between the two recording years. In contrast, 40% of the originally recorded syllable types were still found in the repertoires of recently recorded males. Their song type repertoires were significantly smaller but the songs themselves were on average longer compared to the 1986 recordings. Repertoires of both syllables and song types were more shared between males recorded in 1986 than between those from 2019. We discuss the processes that may have contributed to these temporal changes in song and call for more detailed studies of song evolution in wild populations.

Dialects are a specific form of geographic variation of birdsong with relatively sharp boundaries between distinct song characteristics, which provide opportunities for focused studies of processes underlying the emergence of spatial patterns in vocalization. Several songbird species that exhibit dialects became models for such research, and for some of them large-scale datasets were assembled that included recordings provided by the public. Among them, the Yellowhammer (Emberiza citrinella, Emberizidae) is particularly prominent, as it has been recently a subject of dedicated citizen science projects focusing on its dialect distribution. The most successful, in terms of public participation as well as the number and density of obtained recordings, was the Dialects of Czech Yellowhammers (DCY) project, which assembled detailed data at the whole-country level. A mosaic of almost all common song variants known across Europe was observed in Czechia, but the results indicated that some of the traditionally recognized Yellowhammer dialects may not represent geographically clustered song variants, at least not in Central Europe. We quantitatively analysed variation (frequency and temporal characteristics and modulation) of the terminal song element in three dialects defined by arbitrary frequency thresholds in DCY. Multivariate analyses indicated that pooling these to two distinct groups reflects the variation in the songs, as well as their spatial distribution, better than retaining the current classification to three dialects or their finer splitting to even more categories. We provide simple measures that may be used for classification of these Yellowhammer song variants in Central Europe. However, we warn from indiscriminate transposing of results from one region to another, as that may lead to substantial biases. Future studies of birdsong variation will benefit from big data assembled by citizen scientists, but to maximise their usefulness for further dialect research, careful delineation of dialect boundaries is essential.