Forums | HES 健康与环境科学

25 November 2022, 11:24 AM

Speaker: Eben Goodale

Date: 2022-11-29

Time: 11-12am

Room: ES354

Exploring research direction in eco-acoustics

Eben Goodale, a talk for HES Departmental Seminar, 11/29/22

Ecoacoustics is an emerging field with the goal of detecting ecological patterns and processes through the analysis of audio-recordings. The development of autonomous recording units (ARUs), which can be placed in the field for long durations, may allow ecoacoustics to evolve into a kind of remote sensing, capable of monitoring a wide variety of vocal animals and the health of environments. However, while we now have the tools to collect large amounts of recordings, the tools for processing these sound files are still evolving. In this talk, I will describe my experiences with ecoacoustics, which began when a colleague asked me how to monitor the animal communities in forests fragments of Xishuangbanna, in southern Yunnan. A first project was aimed at testing the efficacy of “acoustic indices”, which is a way of measuring the complexity of sound files as a proxy for biodiversity. We found some acoustic indices to be better than others, and none to have very strong correlations with actual counts of animals. Subsequent projects have focused on inter-relating ecoacoustic data with information from vegetation plots, which are one of the largest investments in ecological research. In the latest project, we used machine-learning algorithms to detect individual bird species on a grid of ARUs placed across a 20 ha forest plot, and then used the plot data to identify potential frugivorous bird / fruiting tree relationships. Finally, I will describe some beginning thoughts about building a three-dimensional array of ARUs next to a canopy crane, to look at how a community of chorusing tree frogs partitions acoustic space. In these projects we see some of the clear benefits of the technology, in capturing voluminous data about the behavioral synchrony and relative proximity of animals, while at the same time some limitations in either low resolution or time-consuming methods for processing the data.

For online participants, please join us via the Tencent Meeting below:

Link: https://meeting.tencent.com/dm/Dj8q64n0iGx5

Meeting ID: 903-4467-2926