Scale dependence of avian functional rarity reveals mismatches between global and local conservation priorities

Aim. Spatial scale shapes both how rarity is defined and how conservation priorities are derived from it, yet most functional rarity assessments rely on a single (often global) reference pool. We test how the scale dependence of functional rarity affects hotspot identification and the alignment of global versus local conservation priorities in birds. Location. Global. Time period. Contemporary. Major taxa studied. Birds (10,906 species). Methods. Using AVONET morphological traits and global distributions, we quantified functional rarity as the average of two components: trait distinctiveness (mean distance to other species in trait space) and geographic restrictedness (1 − occupancy). We computed both components at four nested spatial scales (global, biogeographic realm, biome-within-realm, and local assemblage context). We then mapped assemblage-level functional rarity–weighted richness (Frwr) on a ~100 × 100 km grid, defined hotspots under two policy-relevant thresholds (top 7.1% and top 30% of cells), and quantified hotspot congruence among scales. Results. Trait distinctiveness was nearly invariant from global to biome scales, whereas restrictedness strongly decreased at finer scales, driving marked scale dependence in functional rarity. Hotspots identified using Frwr showed very low spatial overlap across scales (<4% shared), indicating that globally derived priorities often miss assemblages that are functionally rare in regional or local contexts. Multiscale hotspot convergence was concentrated in a limited set of regions, including major tropical mountain systems and island arcs. Main conclusions. Functional rarity offers a trait-based perspective on conservation value, but its interpretation depends on the reference pool and thus on scale. Integrating multiple scales is necessary to reconcile global biodiversity targets with the protection of regionally and locally distinctive ecological strategies.