Schille, L.; Poirier, V.; Raspail, F.; Chaumeil, P.; Bordenave, P.; Herrault, P.-A.; Paquette, A.

Urbanization is a major driver of avian biodiversity loss, primarily through habitat fragmentation and the degradation of food resources, leading to the homogenization of bird communities that are often assumed to share increasingly generalist dietary traits. However, the interaction of urbanization gradients with local habitat features in shaping dietary adjustments remains poorly understood, both at the species and community levels, and it is unclear whether these adjustments reflect prey availability or active food preferences to meet energetic needs. We conducted a study across 25 plots distributed along a controlled urbanization gradient in Montreal, Canada. We quantified habitat variables at both landscape and local scales, sampled arthropod prey guilds, identified insectivorous bird communities using acoustic monitoring, and estimated their potential insectivory from trait-based approaches. In parallel, we assessed realized insectivory using cafeteria experiments offering three types of artificial prey (lepidopteran larvae, spiders, and ants), monitored with custom-built cameras developed specifically for this study to record bird-prey interactions. Along the urbanization gradient, we predicted that (i) functional diversity of bird communities declines and foraging-related traits converge toward more generalist strategies; (ii) profitable arthropod prey availability such as lepidopteran larvae decrease, while other guilds (e.g., Hymenoptera, Araneae) increase; (iii) realized insectivory increasingly diverges from potential insectivory; and (iv) food preferences vary due to both prey availability and active prey selection. We found a strong decline in avian biodiversity and in the availability of high-quality prey along the urbanization gradient, with a convergence toward generalist dietary traits. Yet, the avian biodiversity loss was buffered by canopy cover and tree diversity. Impervious surfaces, canopy cover, local vegetation cover, and lepidopteran abundance were key drivers of the composition of foraging communities observed at cafeterias. Interestingly, realized insectivory exceeded potential insectivory under high local vegetation cover, but the opposite pattern emerged in sparsely vegetated sites. Attack probability on larvae models increased with impervious surfaces, whereas attacks on ants decreased with tree diversity, suggesting active selection of nutritionally profitable prey independent of actual prey availability. Overall, our results highlight the critical role of small, unmanaged vegetation patches, alongside larger and structurally diverse canopy-covered areas, in sustaining avian biodiversity and insectivory functions in cities.