Integrating passive acoustic monitoring and point counts to assess occupancy of secretive marsh birds in Tennessee

Monitoring of wildlife populations is fundamental for evidence‐based conservation yet remains particularly challenging for cryptic species. Secretive marsh birds (hereafter marsh birds) are notoriously difficult to monitor because of their shy, unobtrusive nature and dense vegetation preferences. Existing approaches, such as call‐broadcast point count surveys, can result in low detection probabilities and increased uncertainty in population assessments. We integrated passive acoustic monitoring with traditional call‐broadcast point count surveys to evaluate how differences in survey design influence estimates of occupancy and detection for marsh birds. We further used dynamic occupancy models to examine how local and landscape characteristics influence patterns of site use while accounting for potential temporal changes in occupancy. During April–May 2022–2023, we conducted point count surveys at 78 wetland points in Tennessee, USA, for 6 marsh bird species including the king rail ( Rallus elegans ), least bittern ( Ixobrychus exilis ), common gallinule ( Gallinula galeata ), American coot ( Fulica americana ), pied‐billed grebe ( Podilymbus podiceps ), and sora ( Porzana carolina ). Prior to point counts, we deployed autonomous recording units (ARUs) at each point and recorded for 5‐day periods during times when marsh birds are most active (i.e., 3 hr after sunrise, 3 hr before sunset). We rarely detected our focal species using point count surveys; however, passive acoustic monitoring with ARUs increased detection probabilities, consistent with greater temporal sampling effort. Furthermore, we found that landscape characteristics were stronger predictors of marsh bird site use than local vegetation characteristics in Tennessee. Generalist species such as American coots, pied‐billed grebes, and soras occurred statewide, seemingly more tolerant of agricultural cover and forested wetlands. In contrast, king rails were rarely detected (1.3% of our surveys), suggesting current wetland conditions in Tennessee may not support breeding populations of this marsh specialist. Collectively, our study demonstrates that monitoring methodology can strongly influence estimates of occupancy and detection in low‐detection systems, and that differences among survey approaches are closely linked to temporal sampling effort and species‐specific detection processes. However, challenges in estimating occupancy of rare species at the edges of their ranges remain, and careful cost‐benefit considerations are warranted when incorporating passive acoustic monitoring into long‐term programs.