Autonomous sensors, drones, eDNA, and machine learning models transform biodiversity monitoring in remote Peru.

A scientific expedition to Yaguas National Park in the Peruvian Amazon tells of a transformation that goes beyond environmental conservation. Perù northern, near the indigenous community of Puerto Franco, Conservation International coordinated a field test to understand if Artificial intelligence, autonomous sensors, camera traps, bioacoustics, environmental DNA and drones can build a continuous representation of life in a remote forest. The news of April 6, 2026 is significant because it shifts the topic from occasional discovery to repeatable measurement of ecosystems.

The area is one of the largest and least accessible protected areas in the Peruvian Amazon. Yaguas National Park was classified as a national park in 2018 and, according to the National Service of Protected Natural Areas of the State of Peru, it extends for 868.927,84 hectares in the Loreto region. Its ecological value is linked to the lowland forest, river systems, and rich aquatic and terrestrial fauna. This very vastness makes traditional surveillance fragile: few rangers, considerable distance, illegal activities, and high costs to reach the camp.

The original plan called for entering the heart of the park. However, the arrival of illegal miners, followed by the closure of the area by Peruvian military authorities, forced the group to change their base of operations. The expedition therefore worked on the edge of the park, in the Puerto Franco area, where, according to Conservation International, the forest remains intact for approximately 90 percentThe retreat brought the test closer to the reality of tropical protected areas, where technology, security, logistics, and local collaboration must proceed together.

“We don’t really know what’s there.”

This quote from Ali Swanson, nature technology lead at Conservation International, sums up the problem. Many tropical ecosystems are not only difficult to protect: they are difficult to understand consistently. A traditional survey offers a snapshot of a moment, but it doesn't always produce a time series useful for measuring changes, impacts, and outcomes of conservation policies.

Sensors and biological samples to read a distant forest

The innovative value of the Yaguas mission lies in the integration of multiple instruments. camera traps record the passage of mammals and other animals; the recorders bioacoustics intercept songs, calls and sound signals; the samples of Environmental DNA, or eDNA, search for biological traces left by species in the water; drones They map the forest canopy; insect monitoring systems use light, cameras, and algorithms to count what the human eye can't process in a timely manner.

It's an approach from Research and development applied to conservation: the field becomes a laboratory, but remains within the social complexity of the territory. The expedition was led by Conservation International-Peru, with monitoring design entrusted to Okala and technological contributions from Conservation International, Okala, Limelight Rainforest, and Selvatek. The Frankfurt Zoological Society is listed as a long-term partner of the protected area.

The autonomous component is crucial. In a forest without infrastructure, the main cost isn't buying sensors, but carrying them, powering them, recovering them, transmitting the data, and making it interpretable. Ecologists, technicians, botanists, and members of the local community were present at the Puerto Franco camp, equipped with drones, water sampling kits, batteries, camera traps, and a Starlink terminal to send data outward. Connectivity thus becomes part of theindustrial architecture of monitoring.

“If we succeed here, the implications go far beyond Yaguas.”

The key is scalability. If a system works in a remote, humid, and logistically complex location, it can become a model for other tropical forests, increasing the amount of information produced without displacing researchers and communities.

160.000 insect observations in just five days

The most obvious example concerns nocturnal insects. For decades, the traditional method has been almost artisanal: a light shines on a sheet, specialists observe, collect, and identify what they know. It's an effective but partial practice, because the diversity of tropical arthropods exceeds the capacity for manual classification in the field. Tom Walla, co-founder of Limelight Rainforest, described the limitation bluntly.

“You collect maybe one percent.”

In Yaguas's experiment, the cloth and the light remain, but become part of a computational system. High-frequency cameras photograph the organisms for several consecutive nights; in the laboratory, models of machine learning algorithm catalog and count the observations. In five days, according to Conservation International, the system generated 160.000 observations of insects and identified 854 taxa, mostly at the family or genus level. It's not yet a complete taxonomy at the species level, but it's already a leap in scale in interpreting ecological indicators.

The innovation here isn't in declaring that the algorithm "discovers" the forest. It's in building a data stream large enough to make biological groups usable that were previously too complex, variable, or expensive to monitor. Insects can become living sensors of the habitat's state, as long as the data is collected in a standardized way.AI It serves as an analytical infrastructure, not a substitute for the scientific method.

The same logic applies to other tools. Camera traps, bioacoustics, eDNA, and drones produce Ecological Big Data: heterogeneous information, collected in parallel, useful only if there are clear models, protocols and responsibilities.

Verifiable data to fund the protection of protected areas

The most interesting point concerns the relationship between data and conservation funding. According to Conservation International, Yaguas' annual budget in 2023 was approximately dollars 250.000, while Yellowstone, a comparable-sized U.S. park, operates on an annual budget estimated at approximately $77 million. The comparison highlights the structural gap between ecological value and administrative capacity in many tropical protected areas.

Swanson links the issue to nature credits, still emerging tools for financing verifiable conservation outcomes. To work, however, these models require evidence: forest integrity, species presence, evolution over time, and the effectiveness of protection measures.

"You can't monitor the entire park. With the right tools in the right places, it's not necessary."

The statement of Jeremy Cusack, chief scientist of ok, brings the topic to the industrial field: monitoring must not cover every square meter, but must sample representative places, repeat the measurements and make the extrapolation credible. This is the logic of scalable systems: distributed sensors, replicable protocols, trained algorithms, local communities involved in maintenance and operational management.

In Puerto Franco, six community members accompanied the scientists, walking over ten kilometers a day, installing sensors, activating traps, and collecting data. When the researchers left, the garrison did not disappear: some of the expertise and responsibility remained local.

A reminder against an extractive vision of technology

“It’s very important for us to learn more about what we’re already protecting.”

The words of louis perdomo, a member of the Puerto Franco community and the expedition, are a reminder against an extractive vision of technology. Biodiversity monitoring cannot be a simple remote data acquisition: it requires sustainability institutional, operational continuity and a proper relationship with those who live in the monitored territories.

For tech companies, conservation becomes a testing ground for edge sensors, computer vision, acoustic analysis, satellite connectivity, and data pipelines. These are areas of the digital industry, but here they are measured on robustness, low impact, repeatability, and scientific verifiability.

The Yaguas case shows a broader trajectory: from protection based on maps and episodic inspections to conservation based on continuous data. In this sense, the technological monitoring It is not an end in itself; it is a knowledge infrastructure at the service of public decisions, research, and protection.

For Perù, Yaguas represents a natural and political laboratory. For the innovation sector, it is a concrete case of Artificial intelligence applied to a physical, remote, and non-standardized problem. The most important result, for now, isn't an absolute promise: it demonstrates that biodiversity, local communities, and data can be incorporated into the same operating model without reducing the forest's complexity to a single metric.

Yaguas, the camera traps that reveal the Amazon's hidden biodiversity.

Here are three insights that might interest you:

Starlink among the Marubo: the network arrives in the Amazon rainforest.
Ocean Innovation in Brazil: The Blue Amazon Frontier
Manaus: The Innovation Lab at the Heart of the Amazon