China's new national park system prioritizes the conservation of large-scale ecosystems to maintain ecological integrity, exemplified by Qilian Mountain National Park (QMNP) in northwestern China. QMNP straddles both pristine and pastoral landscapes and provides critical habitat for diverse migratory and resident birds as key stopover and breeding grounds. In this region, centuries of pastoralism have shaped grassland structure and biodiversity. However, spatial distribution patterns of avian species across the heterogeneous, coupled natural and pastoral landscapes of the Tibetan Plateau remain poorly understood due to persistently low detection rates resulting from high species richness, vast distribution ranges, and extensive landscape heterogeneity. To address this, we combined field surveys with functional guild-based Bayesian hierarchical occupancy modeling to elucidate guild-specific spatial distributions and their environmental drivers. Over the three consecutive summers of 2022–2024, we surveyed 72 line transects and 27 fixed observation points, recording 40 species classified into three functional guilds (small-sized, medium-sized and raptors). Vegetation cover and slope gradient were the primary determinants of guild distributions: small-sized birds strongly preferred patches of dense vegetation and avoided hotspots of medium-sized birds, whereas raptors favored gentle slopes. Furthermore, effects of growing season precipitation, distance to water, and human activity varied substantially among guilds in both magnitude and direction. These findings reveal distinct guild-specific habitat selection patterns driven by environmental filtering. Our findings demonstrate that extensive natural-pastoral landscapes remain crucial for avifauna despite traditional pastoral practices, and that functional trait-based modeling can effectively address the significant low-detection challenge in remote, biodiversity-rich grassland ecosystems of the Tibetan Plateau.