From Brain Connectomes to Soundscape Connectomes: Representing Acoustic Landscape Heterogeneity Using Graph-Based Networks
A connectome is a map that illustrates the connections between different regions of the brain, enabling the explanation of some behaviors and diseases through the graphical representation of neural relationships. In neuroscience, connectomes are constructed using signal processing and graph-based algorithms, which have proven to be powerful tools for understanding the complexity of the human brain. Inspired by this idea, we propose transferring these concepts to ecological analysis, specifically through graph inference from acoustic signals, with the aim of constructing a map that represents the heterogeneity of soundscapes.
Specifically, we decompose each acoustic recording into sonotypes—unique acoustic entities characterized by their time-frequency space. By analyzing the occurrence of each sonotype at each sampling site, we generate an acoustic structure that corresponds to the distinctive acoustic footprint of each location. Comparing these footprints allows us to identify similarities between sites within the studied geographical area. To analyze the relationships between acoustic structures, we applied various graph inference algorithms, where nodes represent the geographical points where recorders were placed, and edges indicate acoustic relationships between these nodes. In addition to comparing graph techniques, we employed landscape metrics, commonly used in ecology, to identify landscape heterogeneity.
This approach was applied to two datasets from distinct regions in Colombia, each characterized by different types of land cover. The results reveal that similarities between sites can be identified based on their acoustic structure, highlighting a relationship between the acoustic footprint and the land cover. Furthermore, the analysis underscores the potential of using acoustic data as a proxy for understanding ecological patterns, offering new insights and perspectives for graphically representing acoustic heterogeneity and landscape dynamics. These findings pave the way for future research in constructing a soundscape connectomes, which could serve as an indicator of environmental conditions and biodiversity.