Many species of marsh birds (e.g. rails and bitterns) are believed to be declining in North America, yet we lack an effective monitoring program to estimate their population trends. Broadcast of prerecorded calls to elicit vocalizations is a commonly used method in surveys of marsh birds, but whether gains in detection and index precision outweigh the drawbacks of call-broadcast is unclear. To evaluate the effectiveness of call-broadcast surveys, we pooled marsh-bird survey data from 8,047 point-count surveys contributed by 11 cooperators and compared numbers of birds detected and variation in numbers detected between call-broadcast and passive surveys. For most rails (particularly Virginia Rails [Rallus limicola]), call-broadcast surveys were effective at increasing the detection probability (e.g. average number of Virginia Rails detected per occupied point was 1.25 for call-broadcast surveys and 0.17 for passive surveys). The proportion of points at which no birds were detected was high for all species (range 74–99%) and was slightly lower on call-broadcast surveys as compared with passive surveys. Coefficient of variation (CV) among replicate surveys was higher for passive surveys, particularly for rails (average CV in number of birds detected per point was 209% for passive surveys and 189% for call-broadcast surveys). On the basis of those results, we recommend a marsh-bird monitoring protocol that includes an initial passive period followed by a period of call-broadcast to provide survey data that incorporate the benefits while avoiding the drawbacks of call-broadcast. We also recommend separating both the passive and the call-broadcast periods into 1-min subsegments that will allow estimates of components of detection probability within the monitoring effort.
Efectividad de Censos que Reproducen Vocalizaciones Pregrabadas para Monitorear Aves de Pantano