Vocal communication plays a key role in the behaviour of many animals, mediating social interactions from mate choice to species recognition. In songbirds, vocal signals like mating songs are socially learned. This process may drive rapid song divergence across populations and species, potentially helping to establish and maintain species boundaries. However, social learning may also lead to song convergence if individuals from different populations or species come into contact and learn from one another. This raises a fundamental question: how are differences in acoustic signals maintained if individuals are flexible in what they learn?

In this thesis, I address this question by investigating patterns of song learning in the pied flycatcher (Ficedula hypoleuca), and consider how social and intrinsic factors may contribute to the persistence or loss of song variation at multiple levels. Chapter I focuses on a recent  contact zone between the pied and the closely related collared flycatcher. Here, I first show that pied flycatchers learn song elements from collared flycatchers to produce mixed songs that resemble both species. Further, individual differences in mixed singing did not vary with the relative abundance of collared flycatchers in their breeding environment. This suggests that convergence may not be driven by exposure to heterospecifics during the breeding season, but rather by innate differences in learning biases or early-life song exposure. In Chapter II, I test for innate learning biases at the population level using an egg translocation experiment. Adult males hatched and raised in a foreign population were successfully able to learn the songs of this host population. However, they preferentially learned those host song elements that were most similar to their source population. This suggests that population-specific song learning is shaped not only by social exposure but also by evolved learning biases. In Chapter III, I investigate what juvenile pied flycatchers learn when they are simultaneously exposed to songs of their source and foreign populations using a controlled tutoring setup. While tutored individuals learned songs from all populations, they reproduced song elements from their source population with greater accuracy. Together, Chapters II and III demonstrate that learning biases influence both what individuals learn, and how precisely they reproduce them. These biases may play a key role in maintaining population-level song variation, even when birds are exposed to a wide range of foreign songs. In Chapter IV, I address the often-overlooked role of call variation within species. I show that even structurally simple alarm calls vary geographically across pied flycatcher populations. Moreover, nestling flycatchers  discriminate between alarm calls of their own and foreign populations, depending on the context in which they are produced. This shows that nestlings are attuned to call variation from an early age, and highlights the potential role of call divergence, alongside song divergence, in social recognition and population-level interactions.

Together, these findings illustrate the processes by which socially learned signals evolve and persist, offering new insights into the interplay between learning, communication, and reproductive isolation in the wild.