Combining δ²H and δ³⁴S stable isotopes identifies diverse moult strategies in a short-distance migratory bird

Moult strategies in birds can vary considerably even within a single species, particularly in terms of where moulting takes place. Yet, the factors underlying this variation and its consistency across years and individuals remain poorly understood. One such strategy, moult migration, involves individuals moving to a specific moulting location amid migration to complete feather moult before continuing to their wintering grounds. Here, we used stable isotope analysis of hydrogen (δ²H) and sulphur (δ³⁴S) in feathers from 291 Common Starlings Sturnus vulgaris to estimate the prevalence of moult migration among birds breeding in the Baltics, where some individuals are known to perform moult migration to north-central Europe. To establish isotopic baselines, we included reference samples from Common Starlings with known moult locations either close to the breeding sites in Latvia (n = 24) or near expected moulting sites in Germany (n = 53). Additionally, we tracked migration of a subset of individuals by using multi-sensor geolocators to integrate isotopic and movement data for a comprehensive assessment of moult migration patterns and validation of isotope assignment accuracy. Feather stable isotope values differed significantly between Common Starlings breeding in the Baltic region and those breeding in north-central Europe. Assignments indicated that 86.2% of Latvian Common Starlings moulted near their breeding sites, while 4.5% undertook moult migration (9.3% remained unassigned). Both isotopes contributed substantially to the assignment accuracy, with their combined use improving the best discrimination between moulting regions. The isotope-based method correctly classified five of seven geolocator-tracked Common Starlings, with no incorrect assignments. Moult strategies did not differ between adults and juveniles, and repeated sampling of individuals showed consistent strategies across years. Our findings demonstrate that integrating δ²H and δ³⁴S, especially with reference samples, provides a reliable approach for identifying moult strategies even at small spatial scales, and highlight the promise of δ³⁴S as an emerging tracer in avian movement ecology.