Acoustic monitoring for evidence-based conservation in tropical forests

Abstract

Tropical forest biodiversity is declining at unprecedented rates due to accelerating human impacts including overexploitation and habitat loss and fragmentation. Protected areas are central to conserving biodiversity within increasingly human impacted landscapes, however their effectiveness is undermined by illegal hunting in reserves and habitat loss within key biological corridors. The ability to monitor threats and biodiversity is central to improving effectiveness of conservation management. However, efficient and cost-effective monitoring methods for tropical forests are lacking, particularly for cryptic threats such as hunting. As a result, site managers are confronted with knowledge gaps that hinder their ability to address key challenges effectively.
The emergence of affordable conservation technology methods, such as acoustic monitoring, has the potential to provide accessible, systematic, wide-scale, and information-rich insights into ecosystem health. While these low-cost methods aim to democratize monitoring technology for conservation, the necessary methodology to efficiently utilize the collected data is largely inaccessible to the conservation community, thus limiting their application. This thesis addresses the challenge of monitoring for conservation management in tropical forests, with the main aim of supporting evidence-based conservation in Belize through the advancement of acoustic monitoring tools.
Chapter 1 provides the background and context to the issue. Chapter 2 develops and tests an open-access, reproducible workflow for efficiently processing acoustic data and detecting target sounds, focusing on gunshot detection. Chapter 3 assesses the impacts of hunting on wildlife populations in Belize using hunting metrics derived from acoustic data and a distance-based proxy, integrated with camera trap data of game species. It reveals that hunting activity is primarily driven by the spatial distribution of large-bodied game species, with hunters penetrating deeper into protected areas where game availability is highest. Chapter 4 evaluates alternative methods for integrating acoustic monitoring data and machine learning outputs with occupancy models, using a case study of howler monkeys (Alouatta pigra). It finds that the accuracy of occupancy estimates varies widely depending on the model type, with false positive occupancy models commonly producing unstable occupancy estimates that are sensitive to multiple parameters. Chapter 5 evaluates the effects of forest cover and fragmentation on howler monkey occurrence in Belize using multiscale techniques, demonstrating a practical approach to monitoring a vocal primate species and revealing a consistent negative effect of fragmentation. Chapter 6 provides a forward-looking synthesis of the main findings from Chapters 2-5.
Overall, this thesis demonstrates the value of acoustic monitoring for conservation in tropical forests. It advances monitoring methods by addressing methodological challenges and demonstrating practical uses. Nevertheless, persistent challenges hinder the widespread adoption of this technology, including the absence of user-friendly software, the demand for substantial computational resources, and the absence of clear guidance for end-users. Addressing these issues will be crucial for fully realizing the benefits of acoustic monitoring for tropical forest conservation.

More information

Identifiers

ORCID for Patrick Doncaster: orcid.org/0000-0001-9406-0693

Catalogue record

Export record