Low-cost open-source recorders and ready-to-use machine learning approaches provide effective monitoring of threatened species

Passive acoustic monitoring is a powerful tool for vocally active taxa. Automated signal recognition software reduces the expert time needed recording analyses and allows researchers managers to manage large datasets. The application of state-of-the-art techniques automated identification, such as Convolutional Neural Networks, may be challenging ecologists without informatics or engineering expertise. Here, we evaluated use AudioMoth — low-cost open-source sound recorder monitor threatened patchily distributed species, Eurasian bittern (Botaurus stellaris). was carried out across 17 potential wetlands in north Spain. We also assessed performance BirdNET an freely available classifier able identify over 3000 bird species Kaleidoscope Pro user-friendly detect vocalizations presence target species. percentage presences automatically detected by compared manual annotations 205 recordings. effectively recorded up distances 801–900 m, with at least 50% uttered within that distance being manually detected; this reduced 601–700 m when considering using Pro. 59 63 (93.7%) recordings known while 62 (98.4%). At vocalization level, BirdNet were between 76 78%, respectively, human observer. Our study highlights ability detecting distances, which increases technique spatial scales. According our results, single could useful species' 150 ha. proves utility passive monitoring, coupled Pro, accurate, repeatable, cost-efficient method temporal Nonetheless, further research should evaluate on larger number under different conditions (e.g., more closed habitats), improve knowledge about BirdNET's perform monitoring. Future studies aim develop adequate protocol effective bittern.