What morning surveys miss: passive acoustic monitoring reveals rare and ecologically significant bird species missed by standard survey periods

Passive acoustic monitoring is increasingly being used for bird surveys, but its contribution to detecting ecologically significant species remains poorly understood. We used autonomous recording units and automated acoustic analysis to assess the bird community of Lazarev Canyon, eastern Serbia, and to evaluate whether extended temporal sampling improves the detection of rare, habitat-specialist and other ecologically significant species. Fourteen autonomous recording units were deployed across the canyon, providing continuous acoustic recordings during three survey days in May 2024. Acoustic recordings were processed using BirdNET and all detections were manually validated. A total of 26,374 manually validated true-positive detections representing 71 bird species were retained. Species occupancy was highly uneven, with 66.2% of species recorded at five or fewer sampling locations. Twenty species were classified as species of special interest, including “forest specialists”, “rocky habitat specialists”, “rare species” and “likely transient migratory records”. Compared with expected species, rare species and transient records showed significantly lower spatial occupancy and detection frequency (P < 0.001). Comparisons of species detections between the full-day PAM dataset and the 05:00–10:00 h survey window showed that 18 species (25.4%) would have been missed if sampling had been restricted to this commonly used survey period. Species of special interest were disproportionately affected, with only 50% detected during the morning period compared with 84.3% of expected species (P = 0.005). Our results suggest that conventional survey schedules may underestimate the occurrence of rare and ecologically significant species, whereas extended passive acoustic monitoring can reduce this limitation and provide more complete and ecologically informative bird inventories.