A UPC team listens to endangered river dolphins in the Amazon

Amazon river dolphin

Amazon river dolphin. Image: Marina Gaona

Dolphin swimming underwater

Dolphin swimming underwater. Image: Wezddy Del Toro

Echolocating animals, such as cetaceans, use sound as a form of navigation. Echolocation clicks can be used to track the movements of two endangered freshwater dolphin species in the Amazon, contributing to conservation strategies for their natural habitat. This is demonstrated by a study published in Scientific Reports featuring researchers from the UPC’s Laboratory of Applied Bioacoustics (LAB).

Sep 21, 2023

The boto (Inia geoffrensis) or pink river dolphin and the tucuxi (Sotalia fluviatilis) are two species of endangered freshwater dolphins that inhabit inaccessible parts of the Amazon. These species are threatened by human activity related to fishing, agriculture, mining and dam construction. During the wet season—April to August—both dolphin species move into the Amazon seasonal floodplain forests or várzeas in search of freshwater fish. However, flooded fields and vegetation make it extremely challenging to survey dolphins from a boat or using drones.

Researcher Florence Erbs and other scientists at the Laboratory of Applied Bioacoustics (LAB) of the Universitat Politècnica de Catalunya - BarcelonaTech (UPC), which is linked to the Vilanova i la Geltrú School of Engineering (EPSEVG), surveyed about 800 square kilometres of the Mamirauá Sustainable Development Reserve in Brazil. They used data from five hydrophones operating at depths of three to five metres in the Solimões and Japurá rivers. Published in the prestigious journal Scientific Reports (Nature), the study provides new insights into the movements of river dolphins that could contribute to conservation strategies for these species. It was conducted in collaboration with scientists from the Mamirauá Institute for Sustainable Development (IDSM) and the National Marine Mammal Foundation in San Diego, USA.

The study is based on bioacoustic data taken from river channels, bays, lakes and flooded forests in the reserve. It was carried out over several periods during the wet and dry seasons between June 2019 and September 2020. The researchers used deep-learning algorithms—specifically a convolutional neural network—and manually collected bioacoustic data to classify automatically dolphin echolocation clicks, shipping noise and rain, with an accuracy level of 95%, 92% and 98% respectively.

Dolphin presence was found to increase from 10% to 70% in the bay and river as water levels rose between November and January. The authors explain that dolphins used these waterways to enter the floodplains in the Amazon basin. Immature botos and females with calves spend more time in the floodplain than males, because either it provides a prey-rich habitat or it offers shelter against the males’ aggressive behaviour.

The researchers are calling for their methodology to be used to better understand and protect habitat preferences and requirements of river dolphins in the Amazon as part of the Providence project, which conducts extensive bioacoustic monitoring of aquatic and terrestrial biodiversity in one of the most threatened tropical forests on the planet.