In the face of environmental threats that are numerous and unpredictable, ecologists need new methods to monitor ecosystems at scale and in real-time. Soundscape ecology, through the use of passive acoustic monitoring, has efficiently characterized patterns of biodiversity in terrestrial and aquatic ecosystems. Yet, the question still remains: How can soundscapes be used to quantify the effects of ecosystem disturbance? This study assesses disturbance effects by analyzing long-term acoustic recordings. We introduce a theoretical model to describe how soundscapes respond to different disturbance types. In the context of this theoretical framework, we apply Soundscape Control Charts, a new method to measure the effects of natural and anthropogenic disturbances. Control charts are used to study how processes change over time; most commonly used in the industrial sector as alarm systems, control charts identify when a system deviates from its normal state. While control charts seem promising for detecting ecological change, they have proven difficult to implement because ecological data are typically collected at coarse temporal resolution, which greatly increases the variability of the data. Passive acoustic monitoring addresses this limitation by collecting data near-continuously, and minimizing variability through sub-sampling and time-series modeling techniques.
Results/Conclusions
In this study, Soundscape Control Charts detected the presence of sudden pulse disturbances, including the haying of a Nebraskan prairie and the impact of a storm near a coral reef in the Dry Tortugas, FL. We calculated twenty-four acoustic measurements that we assembled into distance-based multivariate control charts. In response to each disturbance event, the upper bound of the control chart, which was determined through bootstrapping methods, was exceeded. By obtaining acoustic snapshots at regular temporal intervals, we can document the effects of abrupt disturbance events as well as gradual deviations from ecosystem baselines, equipping land managers with a statistically rigorous “alarm bell” that alerts them when a system may require management intervention.