Underwater Microphones and Machine Learning Aid Right Whale Conservation

 

The North Atlantic right whale, one of the world's most endangered marine species, faces a dire population crisis. With fewer than 370 individuals remaining, these whales are threatened by ship strikes, entanglement in fishing gear, and climate-driven changes to their food sources. Monitoring their numbers is crucial for conservation efforts, yet traditional methods—such as aerial surveys—are costly, dangerous, and limited by weather conditions. However, a groundbreaking study by researchers at Cornell University has introduced a new approach that leverages underwater microphones and machine learning (ML) to provide a safer, more efficient, and cost-effective way to track these whales.

The Challenge of Monitoring Right Whales

For decades, researchers have relied on visual surveys conducted from airplanes and boats to estimate right whale populations. While these methods provide valuable data, they come with significant challenges:

• Expense: Aerial surveys require trained pilots and scientists, as well as specialized aircraft, making them costly to operate.
• Risk: Flying at low altitudes over the ocean poses risks to researchers and pilots.
• Limited Coverage: Aerial surveys can only be conducted in daylight and favorable weather, leaving gaps in data collection.
• Incomplete Population Estimates: Since right whales spend much of their time underwater, visual surveys can underestimate their numbers.

Given these limitations, scientists have turned to sound recordings as an alternative monitoring tool. Right whales produce distinctive vocalizations, including a signature “upcall,” which can be detected using passive acoustic monitoring.

How Underwater Microphones and Machine Learning Improve Whale Tracking

The new study, published in Endangered Species Research, demonstrates how integrating underwater microphones with ML algorithms can revolutionize right whale conservation. Researchers from the Cornell Lab of Ornithology’s K. Lisa Yang Center for Conservation Bioacoustics deployed an array of marine autonomous recording units (MARUs) across Cape Cod Bay—a crucial feeding habitat for right whales—to continuously capture whale sounds.

Step 1: Deploying Marine Autonomous Recording Units (MARUs)

The research team strategically placed MARUs throughout Cape Cod Bay to collect underwater sound data. These devices remained in the water for extended periods, continuously recording the ocean’s soundscape and capturing right whale vocalizations.

Step 2: Training the Machine Learning Model

Once the MARUs had gathered sufficient data, researchers trained a deep-learning model to identify right whale upcalls with 86% precision. The ML algorithm was validated against known whale recordings to ensure accuracy before being applied to the full dataset.

Step 3: Estimating Whale Presence and Numbers

By analyzing the frequency and distribution of upcalls, researchers could determine not only the presence of whales but also estimate their numbers in a given area. This represents a significant advancement over traditional acoustic monitoring, which primarily indicates only whether whales are present or absent.

Advantages of This New Approach

The combination of underwater microphones and ML offers several key advantages over traditional survey methods:

1. Continuous Monitoring: Unlike aerial surveys, which are limited to specific times and conditions, passive acoustic monitoring operates 24/7, providing uninterrupted data.
2. Safer and Cost-Effective: Eliminating the need for airplanes and human observers reduces costs and eliminates risks associated with low-altitude flights over open water.
3. Expanded Monitoring Range: ML-powered analysis allows scientists to monitor a much larger area than would be feasible with human observers alone.
4. Better Population Estimates: The ability to track whales continuously over time helps refine population estimates and understand movement patterns more accurately.

Implications for Right Whale Conservation

This breakthrough comes at a critical time for North Atlantic right whales, whose numbers continue to decline. Improved monitoring capabilities can directly contribute to conservation efforts by:

• Informing Ship Speed Regulations: Real-time acoustic data can help authorities implement dynamic speed limits for vessels in areas where right whales are detected, reducing the risk of ship strikes.
• Enhancing Fishing Gear Management: By understanding whale distribution patterns, regulators can adjust fishing zones to minimize the likelihood of entanglement in gear.
• Tracking Climate Impacts: As ocean temperatures shift, right whale feeding and migration patterns are changing. Continuous acoustic monitoring can help scientists study these changes and adapt conservation strategies accordingly.
• Expanding Conservation Efforts Beyond New England: While Cape Cod Bay is a known hotspot for right whales, these animals are found along the entire East Coast. The new method allows for broader monitoring beyond traditionally studied areas.

Future Directions and Challenges

While this research marks a significant step forward, challenges remain. Marissa Garcia, the study’s lead author, acknowledges that some uncertainties still exist in the whale population estimates derived from acoustic data. Future research aims to refine ML models further and address these uncertainties. Additionally, integrating acoustic monitoring with other data sources—such as satellite tracking and environmental modeling—could provide an even more comprehensive understanding of right whale populations.

Looking ahead, expanding the use of this technology to other critical habitats along the East Coast will be essential. Researchers hope that continued improvements in ML and increased deployment of underwater microphones will enhance conservation efforts for not only right whales but other endangered marine species as well.

The integration of underwater microphones and machine learning represents a major innovation in right whale conservation. By offering a safer, more cost-effective, and continuous monitoring solution, this technology has the potential to significantly improve our ability to track and protect these critically endangered animals. As researchers refine their models and expand monitoring efforts, the hope is that these advancements will contribute to reversing the decline of North Atlantic right whales and ensuring their survival for future generations.