Résumé
Abstract Understanding the selection of breeding sites in cities is key for the effective management of urban bird populations. We analysed 16 years of data collected within a relocation program for female mallards in Berlin (Germany), with the goal of (i) characterizing the breeding output and breeding sites in urban environments, (ii) identifying factors correlating with probabilities of individuals to return to previously used breeding sites (i.e. breeding site fidelity) and (iii) assessing the consequences of ongoing management strategies on the selection of breeding sites. Mallards using anthropogenic structures for breeding primarily chose balconies, roof terraces and courtyards. In 22.8% of cases, mallards selected the same sites in subsequent breeding events. Breeding site fidelity decreased with higher human population density at that location, and decreased as mallard families were relocated further away from the breeding site to a suitable water body. In contrast, return probabilities were higher for subsequent broods within the same breeding season than across two successive seasons. Above all, the identity of the breeding location and the identity of the female best predicted whether an individual would return to a given breeding site. We conclude that breeding site fidelity in urban mallards is strongly shaped by micro-scale location-specific properties (e.g. balcony characteristics), as well as individual variation due to an intrinsic propensity for site fidelity. We argue that adequate management for waterfowl in urban environments depends on why birds choose to breed on anthropogenic structures, on their conservation status and the availability of breeding sites around urban waters.