A pilot assessment of avian communities and soundscapes along an Amazonian fluvial corridor

Abstract

Quantifying biodiversity patterns in remote Amazonian ecosystems remains constrained by the limitations of traditional field surveys. We combined passive acoustic monitoring (PAM), machine learning, and ecoacoustic metrics to assess the taxonomic and functional structure of bird communities along a riparian gradient in the Kapawi River, Ecuadorian Amazon. A total of 2,030 recording hours were acquired using 16 Autonomous Recording Units (ARUs) deployed along a river-to-interior forest gradient (0-800 m from the riverbank). Automated detection with BirdNET yielded 92,137 records corresponding to 379 bird species. Species richness was highest at the river edge (325 species), which also harboured the greatest number of unique taxa (71 species), while interior sites showed lower but more consistent local richness. Multivariate analyses showed clear spatial segregation between riparian and interior communities. Despite this turnover, the trophic structure remained highly homogeneous (>90% similarity), dominated by insectivorous and frugivorous guilds. Generalized linear models (GLMs) indicated strong positive associations between avian species richness and key ecoacoustic metrics, with particularly pronounced effects for the Acoustic Diversity Index (ADI) and the Bioacoustic Index (BI). Spatially explicit analyses further demonstrated marked heterogeneity in acoustic structure along the fluvial gradient, reflecting fine-scale variation in soundscape composition. Together, these findings show that riparian habitats structure avian communities primarily at the taxonomic level, while functional organization remains largely conserved across the gradient. This mismatch indicates that biodiversity components respond unevenly to environmental variation, with taxonomic richness being more sensitive than functional composition. Our results underscore the potential of ecoacoustic approaches as scalable, non-invasive tools for detecting spatial patterns in biodiversity and habitat-driven community assembly in tropical systems.