1. Passive acoustic monitoring (PAM) is a powerful tool for wildlife monitoring, but the time and expertise required to process large volumes of data pose significant challenges. Automated acoustic detectors improve efficiency by speeding up data processing. Class imbalance, resulting from fewer target signals relative to noise, complicates development and can negatively impact performance. However, training datasets should also reflect the conditions of real-world PAM datasets. 2. We developed an automated acoustic detector for chimpanzee loud calls while addressing class imbalance. We predicted that greater data diversity and high-quality data (clear signals, minimal noise interference) would enhance network performance and that class imbalance, by supporting diversity, is essential for functionality. We built training datasets with data recorded in wild settings and applied a temporal convolutional neural network approach using Deep Audio Segmenter (DAS). We trained networks using datasets containing varying levels of noise (50%, 75%, 90%, 99%) and also tested the effectiveness of frequency removal in improving performance. 3. The network performances varied significantly, with F1 scores of 0.44 to 0.86, exceeding a previous study (5% F1). The most imbalanced dataset produced the best performing network, capturing 90% of pant-hoot events and annotating them with 90% (SD = 20.9) accuracy. The results showed that increased class size was associated with greater intraclass diversity and improved precision (0.41–0.83). The networks showed consistently high recall rates, especially when frequency removal was not applied (0.89–0.92). 4. This study stresses the importance of class size and diversity in developing automated acoustic detectors. It also highlights the value of high-quality data for accurate pattern recognition of the target signal and the importance of the noise class for effective class decoupling and detector functionality. This research supports the advancement of PAM in chimpanzee studies, opening new opportunities to integrate remote sensing for efficient wildlife monitoring.

Forty years of deforestation and logging have degraded and fragmented much of Borneo’s lowland forest. This poses a threat to the island’s unique biodiversity, which can be exacerbated by hunting and killing. Although orangutans sometimes persist in small forest patches, it is unclear if such highly fragmented habitats can sustain viable populations, and whether they facilitate movements across modified landscapes over the long-term. Since longitudinal population data are unavailable, inferences must be made from modelling. We applied a spatially-explicit individual-based model to predict the potential long-term viability of orangutan populations across Borneo. Specifically, we examined how population dynamics and dispersal could be affected by the loss of habitat fragments and removal of individuals through hunting, retaliatory killings and capture and translocation. Small forest fragments facilitated orangutan movement, increasing the number of individuals settling in non-natal patches. However, large rivers remained a substantial barrier, and limited the capacity of orangutan populations to recover from decline. Orangutan populations were also highly vulnerable to even small amounts of offtake, with annual removal of >2% diminishing the positive role that small fragments played in sustaining population connectivity and long-term viability. Our results imply that orangutan populations could grow and recover from recent declines across Borneo if further habitat loss within human-modified landscapes is minimized. However, this will only be achievable if efforts are made to reduce the removal of orangutans by promoting coexistence with people, limiting killings, and only engaging in translocations in rare cases where no suitable alternative exists.