Passive acoustic monitoring of Asian hornbill: A case study of Oriental Pied Hornbill (Anthracoceros albirostris convexus) in the Lower Kinabatangan Wildlife Sanctuary, Sabah

Yusni, Ashraft Syazwan Ahmady, Kaur, Ravinder, Goossens, Benoit ORCID: https://orcid.org/0000-0003-2360-4643, Ancrenaz, Marc and Liew, Thor Seng 2026. Passive acoustic monitoring of Asian hornbill: A case study of Oriental Pied Hornbill (Anthracoceros albirostris convexus) in the Lower Kinabatangan Wildlife Sanctuary, Sabah. Journal of Tropical Biology & Conservation (JTBC) 23 , pp. 150-166. 10.51200/jtbc.v23i.6667

PDF - Published Version Available under License Creative Commons Attribution Non-commercial. Download (4MB)

Abstract

Asian hornbills are keystone species in tropical forests for their various ecological functions. However, they are faced with threats including habitat degradation, which calls for effective monitoring of these iconic birds. This case study of the Oriental Pied Hornbill (Anthracoceros albirostris) demonstrates the application of passive acoustic monitoring (PAM) with a deep learning model (BirdNET) in hornbill monitoring, comparing its efficacy to manual surveys. We also compare operational costs in data collection for both approaches. The study took place in Lot 6 of the Lower Kinabatangan Wildlife Sanctuary, Sabah, Borneo. We utilized PAM device – AudioMoth Dev 1.0.0 combined with manual surveys to evaluate hornbill occurrence across 23 monitoring stations within a 5 KM2 study area. A Bayesian multi-method occupancy model was utilized to estimate detection probability and site occupancy, integrating data from visual, aural, and PAM-based surveys. A. albirostris demonstrates high occupancy estimate across sites (ψ = 0.832) and notable detectability by aural (θ = 0.929) and PAM approaches (θ = 0.694). The assessment of the BirdNET deep learning model for the automated identification of A. albirostris, attaining a high precision of 0.96 with recall and F1 score of 0.46 and 0.62, respectively. Furthermore, PAM reduced monitoring expenses by almost 71% relative to manual surveys, mostly owing to decreased personnel and logistical demands. Our research indicates that PAM, enhanced by convolutional neural networks such as BirdNET, provides a scalable and economical approach for future monitoring of Asian hornbills.

Item Type:	Article
Date Type:	Publication
Status:	Published
Schools:	Schools > Biosciences
Publisher:	Universiti Malaysia Sabah (UMS)
ISSN:	1823-3902
Date of First Compliant Deposit:	3 August 2026
Date of Acceptance:	12 May 2026
Last Modified:	03 Aug 2026 10:45
URI:	https://orca.cardiff.ac.uk/id/eprint/188674

Actions (repository staff only)

Edit Item