Habitat selection is influenced by landscape composition, resource distribution and social behaviour. Habitat selection is scale-dependent, and the spatial extent and the scale at which processes are measured are therefore of paramount importance. This study focused on the habitat selection of the little bustard (Tetrax tetrax). We developed habitat selection models with the aim of disentangling the relative effects of local habitat parameters (crop), landscape features and social factors (to take into account the exploded lek mating system of this species). We used the SILand package in R, which allows the estimation of simple and interactive effects, using spatially accurate counts of male little bustards over a 1074-km2 study suite (over 11&nbdp;consecutive years). Local and landscape-scale habitats were taken into account, including agri-environment schemes, allowing a quantitative analysis of these conservation measures. The predictive power of the models that only included local habitat was poor, but including landscape habitats –&nbp;interacting with local habitat or not – statistically increased the model's predictive power (by 16%). Adding the social factor further improved model quality (by 14%). From a conservation perspective, the presence probability of a male increases by about 25% when the field is managed with an agri-environment scheme. If the habitat around the field is of high quality (50% of grasslands within a 500-m radius), this probability rises to 0.8, and if a male is already present within 250 m, it reaches 1. This confirms that conservation measures should preferentially target hotspots of male presence (i.e., larger leks).