This article explores the opportunities of utilizing Acoustic monitoring Technologies and Machine Learning to monitor the ecologically and economically important bumble bees. Their distinguishable acoustical characteristics during specific sound-producing behaviors enables effective identification, classification and monitoring.
Title : What’s the Buzz About? Progress and Potential of Acoustic Monitoring Technologies for Investigating Bumble Bees
Authors: Zachary J. Miller
Journal: IEEE Instrumentation & Measurement Magazine
Citation: Z. J. Miller, "What's the Buzz About? Progress and Potential of Acoustic Monitoring Technologies for Investigating Bumble Bees," in IEEE Instrumentation & Measurement Magazine, vol. 24, no. 7, pp. 21-29, October 2021, doi: 10.1109/MIM.2021.9549234
Open Access: No, restrictions apply
Pollination is a vital ecosystem service for both natural and agricultural ecosystems. Most flowering plants—including many staples in the human diet such as fruits, vegetables and nuts— require animals for pollination. The majority is done by over 20,000 species of bees, some of which are managed commercially to provide these services. The global industry of pollination is valued at US $153 billion annually and is increasing in demand to meet the needs of a growing human population. However, both managed and wild bees, especially the ecologically and economically important bumble bees (Bombus spp.), are suffering alarming declines worldwide, presenting serious implications for food security and biodiversity. Despite numerous pleas by farmers and scientists for improved management and monitoring methods, precision techniques for these essential pollinators are still lacking. Recent advances in acoustic monitoring technologies (AMT) show promise for bumble bee investigations. Bumble bees create a range of distinguishable sounds while flying, sonicating (buzzing on flowers to eject pollen) and interacting within the colony, making them amenable for acoustical surveys. While acoustic-based techniques have been used to study bumble bees, most of these efforts pre-date advancements in computer programming, machine learning and automation, and thus have not been widely adopted. Current standard practices in bumble bee monitoring include netting, trapping, and in-person observations, which are laborious, costly and often require lethal collection of bumble bees. AMT offer an alternative approach that is affordable, scalable and non-destructive, with potential to augment conservation and agricultural practices. The sounds produced by bumble bees may be useful to researchers and farmers regardless of their implications for survival and reproduction. The types of questions that can be answered will differ, as some bumble bee sounds produced have roles in communication with other bumble bees or with predators—e.g., ‘buzz runs’ and defensive ‘hisses’—making them amenable for eavesdropping on colony behavior; others are by-products of non-communicative activities— e.g., flight buzzes—providing opportunities to remotely track and monitor bees foraging on flowers. This review focuses on audio signals that are distinguishable rather than biological signals that are communicative; the former includes the latter, but not vice versa. Application of AMT to investigate bumble bees is still nascent in development, and improvements are needed across all stages of the AMT process, from sensor technologies and data transfer to audio classification and user interfaces. Here, I review the sound-producing activities of bumble bees, highlighting extant research and underscoring opportunities for further investigation. For each sound or soundscape, I emphasize the acoustic features that make it unique to particular behaviors and discuss how AMT could benefit bumble bee research and agriculture. In particular, I examine sounds produced from within bumble bee colonies and from bumble bees on or near flowers; I then discuss the potential application of AMT to study a major threat to bumble bees, and I conclude by reiterating the importance of cross-disciplinary collaboration between ecologists and computer scientists to monitor and manage species of conservation concern.
Key words: Sociology, Ecosystems, Collaboration, Machine learning, Acoustics, Biodiversity, Audio systems, Monitoring
For interesting and engaging discussions on Acoustic monitoring, join our Acoustic monitoring group.