Tropical forest covers 12% of the planet’s land surface, but is home to about two-thirds of all terrestrial species. The Amazon, spanning the vast Amazon River Basin and the Guiana Shield in South America, is the largest remaining stretch of tropical forest globally, home to more species of animals than any other terrestrial landscape on the planet.
Spotting wildlife in these dark, dense forests teeming with insects and thorny palms is always a challenge. This is due to the very nature of biodiversity in the Amazon, where there is a small number of abundant species and a larger number of rare species that are difficult to detect adequately.
Understanding which species are present and how they relate to their environment is of paramount importance for ecology and conservation, providing us with essential information on the impact of human-caused disturbances such as climate change, logging or wood burning. In turn, this can also allow us to take advantage of sustainable human activities such as selective logging, the practice of removing a tree or two and leaving the rest intact.
As part of BNP’s Bioclimate project, we are implementing a number of technological solutions such as photo traps and passive acoustic monitors to overcome these obstacles and refine our understanding of Amazonian wildlife. These devices surpass traditional surveys due to their ability to continuously collect data without the need for human interference, allowing animals to go about their business undisturbed.
Eyes in the trees
Camera traps are small devices that are activated by changes in activity in their vicinity, such as the movements of animals. They were essential to our fieldwork in the Tapajos National Forest in Para, northwestern Brazil, allowing us to investigate whether disturbances such as climate change have affected the presence and behavior of animals which are, in turn, necessary. to natural processes.
The dispersal of seeds by animals, which allows the regeneration of forests, is one of these processes. By eating fruit or carrying nuts, they typically expel or drop the seeds elsewhere. Our research has shown that at least 85% of all tree species on our plots have seeds scattered by animals.
We also know that many of these animals are heavily affected by the disorder. To better understand the impact of the loss of these seed-dispersing species, we need to know which ones spread which plants and to what extent.
We attempted to observe this by installing cameras at the foot of fruit trees at our study site, revealing which species ate which fruits and then carried seeds through the forest.
The survey produced over 30,000 hours of footage and we were able to ascertain that 5,459 videos featured animals. An impressive total of 152 bird and mammal species has been recorded, including rare records of endangered species such as the vulturine parrot (Pirilia vulturina).
The videos included incredible information about the behavior of the animals, such as an ocelot (Leopard) to hunt a common possum (Didelphis marsupialis), a giant anteater (Three-toed myrmecophages) carrying a baby on her back, and even a curious female capuchin monkey (Sapajus) who checked a camera and ended up throwing it to the ground.
Importantly, we also recorded 48 fruit-eating species, including species considered important seed dispersants, such as the South American tapir (Terrestrial tapir) which is capable of spreading large seeds over long distances due to its size.
Our research has shown that bird species such as the white-crested guan (Penelope pileata) and mammals such as the silver marmoset (Silver mico) and the Amazonian brown deer (Mazama and Morivaga) are regular fruit consumers. Many of these species are over-hunted in the study region, which can lead to cascading impacts on forest regeneration.
Sound recorders, on the other hand, are critical for compiling inventories of the species-rich bird community. Indeed, although the birds are rarely seen in dense forests, their vocalizations reveal their presence.
When ornithologists study tropical birds, they are limited by how often they can conduct counts as it is often logistically difficult to return to individual locations. As a result, traditional surveys are often quite long – between 5 and 15 minutes – with only a limited number of repeat counts at each site tested. This means that only a small portion of the time period when birds are most active can be detected, the two hours after sunrise, generally known as the dawn chorus.
Yet the birds don’t all sing at the same time: some species prefer to sing very early in the morning, most wait until it gets slightly warmer and the sun has fully risen, and some others get up late. Limiting ourselves to a few surveys, it is difficult to cover full time and detect all the species present. Furthermore, surveys conducted only in a handful of days mean that factors such as the weather or the presence of predators on certain days can completely change the species detected.
Our research found that by setting up standalone sound recorders to make 240 very short 15-second recordings for a total of one hour of detection, we could record 50% more species at each site we surveyed compared to four of 15 surveys. minutes that replicated the duration of human investigations. The extra polls allowed us to spread our survey period over several days, but more importantly, over the entire dawn chorus. We found that there was a small group of species that preferred to sing from 15 minutes before sunrise to 15 minutes after, and we were really likely to only detect them if we had multiple detections during that time, which is only possible with automatic recorders. .
These more comprehensive surveys allow us to provide better estimates of the species that live in these hyperdiverse regions, but also of those that vanish when forests are cut or burned. Using this method, we were able to detect 224 bird species in 29 locations with a total of just one hour of detection at each location.
Species found in intact and disturbed forests also confirmed our previous research which showed that undisturbed primary forests are home to unique bird communities that are lost when forests are damaged by selective logging or fires.
The sound recorders have also allowed us to collect data over long periods of time, with over 10,000 hours recorded so far.
However, the collection of data on this scale also means that it is not possible for a scientist to listen to all the recordings. Instead, the new field of echoacoustics has developed statistical techniques to characterize entire soundscapes. These acoustic indices measure the variation in amplitude and frequency to provide a metric of how busy or varied each soundscape is. By eliminating the need to identify individual sounds, they can efficiently process large volumes of acoustic data.
We used acoustic indices to show that undisturbed primary forests have unique soundscapes that can be identified with machine learning techniques. These data, in turn, allow us to contrast soundscapes that have been disturbed by phenomena such as fires or deforestation and to distinguish the groups of species that have been most affected.
Finally, camera traps and sound recorders allow us to have eyes and ears in the forest even when our researchers are not there. As technology develops, we will continue to use the latest techniques to better understand the behavior and ecology of animals and how to use them to better enhance and protect the habitats they live in.
In particular, we are trying to develop deep learning models to identify species and, in some cases, to differentiate between individuals of the same species. Images and sounds recorded by automatic recorders are opening up new ways of understanding animal abundance and behavior, providing new insights into the secret world of tropical forest fauna.
The “Bioclimate” research project of which this publication is a part was supported by the BNP Paribas Foundation as part of the Climate and Biodiversity Initiative program. It is coordinated by the Rede Amazonia Sustentavel (RAS).
Oliver Metcalf, Associate Postdoctoral Researcher, Manchester Metropolitan University and Liana Chesini Rossi, Invited User, State University of Sao Paulo
This article was republished by The Conversation under a Creative Commons license. Read the original article.