Towards the automated detection and occupancy estimation of primates using passive acoustic monitoring

摘要: Abstract Recent advancements in technology have made possible the use of novel, cost-efficient biomonitoring techniques which facilitate monitoring animal populations at larger spatial and temporal scales. Here, we investigated using passive acoustic (PAM) for wild primate living forest Tai National Park, Cote d’Ivoire. We assessed potential a customized algorithm automated detection multiple species to obtain reliable estimates occurrence from data. First, applied on continuous rainforest recordings collected autonomous recording units (ARUs) detect classify three sound signals: chimpanzee buttress drumming, loud calls diana king colobus monkey. Using an occupancy modelling approach then what extent automated, probabilistic output needs be listened to, thus manually cleaned, by human expert, probabilities derived ARU data fully verified human. To do this explored robustness probability simulating datasets with various degrees cleaning false positives negative detections. further validated comparing it observers point transects located within same study area. Our demonstrates that data, combined processing methods such as our algorithm, can provide results comparable humans require less effort. show are quite robust effort, particularly when is high, suggest some even naive occupancy, without any cleaning, could quick indicator guide efforts. found most influenced time day drums while temperature and, likely, poaching pressure, affected monkey calls. None covariates appeared strongly call detection. Finally, conclude semi-automated presented here used early-warning system activity additional improving its performance.