This study aims to identify the potential benefits of a landscape description that encompasses both preferred and non-preferred habitats, to study biodiversity response to landscape characteristics. The 'landscape-centred approach' (LCA) is meant to overcome important issues in landscape ecology: (i) describing the complexity and heterogeneity inherent in anthropogenic landscapes, (ii) disentangling composition and configuration effects without multicollinearity, and (iii) allowing the simultaneous study of species with different ecological requirements. In this paper, we evaluate the capacity of the LCA, which includes all land covers, (i) to provide better fitted models explaining the spatial distribution of ecologically distinct species and (ii) to allow a better understanding of the relationships between species distribution and landscape characteristics, compared to a classical 'habitat-centred approach' (HCA) focused on the preferred habitat. We used a dataset of 47 ecologically distinct bird species (generalist, specialist of farmland, forest or urban areas) distributed on 841 observation plots to compare the performances and results in species distribution modelling between LCA-models and HCA-models. LCA-models had a lower AIC for 72% of the species. Based on AIC, the LCA performed better for a majority of species in each of the four groups (69% of farmland species, 56% of forest species, 72% of urban species and 86% of generalist species). For about 25% of the species, the LCA revealed species sensitivity to landscape composition while the HCA did not. With the LCA, responses to the amount and configuration of land covers other than the preferred habitat were found for 70% and 55% of all specialist species, respectively. Our results show that providing a unique landscape description relevant for ecologically distinct species is possible. Beyond potential implications for designing multi-species conservation projects, we suggest that the LCA could thus allow analysing species interactions and other mechanisms underlying community structure. Furthermore, the LCA allows understanding with more accuracy the respective influence of different landscape facets, which is crucial in widespread anthropogenic landscapes characterised by high heterogeneity.