Tracking echolocation, which is the sound that certain animals such as cetaceans make to find their way around, can be used to track the movements of two endangered species of freshwater dolphins in the Amazon , and carry out conservation strategies for these natural habitats. This is demonstrated by a study published in 'Scientific Reports' in which researchers from the UPC Bioacoustic Applications Laboratory participated.
The boto (Inia geoffrensis) or pink river dolphin and the tucuxi (Sotalia fluviatilis) are two species of freshwater dolphins in danger of extinction that live in inaccessible places in the Amazon. These species and their interactions with people are threatened by human activity from fishing, agriculture, mining and dam building. During the wettest period of the year, between the months of April and August, both species of dolphins move to the rivers of the rain forest areas of the Amazon (várzeas) in search of freshwater fish. However, the floodplain and vegetation make it extremely difficult to observe dolphins with boats or drones.
The researcher Florence Erbs and other scientists from the Laboratory of Bioacoustic Applications of the Universitat Politècnica de Catalunya - BarcelonaTech (UPC) , linked to the Higher Technical School of Vilanova i la Geltrú (EPSEVG) , have studied about 800 square kilometers of the Reserve of Sustainable Development of Mamirauá, in Brazil. For this they have used the data obtained through five hydrophones submerged at depths of between three and five meters in the Solimões and Japurá rivers. The results of the study, which are published in the landmark journal Scientific Reports (Nature), reveal new knowledge about the movements of river dolphins that could contribute to improving the conservation of these species. The work has been developed together with scientists from the Instituto de Desenvolvimento Sustentável Mamirauá (IDSM) and the National Marine Mammal Foundation, from San Diego, in the United States.
In the study, bioacoustic data was analyzed from the river channels, bays, lakes and flooded forests of this reserve. It has been carried out in different periods during the wet and dry seasons, between June 2019 and September 2020. The researchers have used deep learning algorithms - what is known as a Convolutional Neural Network or convolutional neural network - and manually collected bioacoustic data, to automatically classify dolphin echolocation sounds, ship noises and also rain, with accuracy levels of 95%, 92% and 98%, respectively.
The researchers found that dolphin presence increased from 10% to 70% in the bay and river as water levels rose between November and January. The authors of the study explain that the dolphins were using these waterways to enter the floodplain of this Amazon basin. Young bottlenose dolphins and females with pups of these dolphin species spent more time in the floodplains than males, either because of the rich abundance of prey or as protection against the aggressive behavior of the males.
The researchers are urging the methodology they used to better understand and protect the habitat preferences and needs of Amazon river dolphins as part of the Providence project, which is conducting extensive bioacoustic monitoring of aquatic and terrestrial biodiversity in one of the most threatened tropical forests on the planet.