Research scientists are collaborating on a large-scale project to provide an updated estimate for Hudson River’s shortnose sturgeon population, a federally endangered species, using acoustic telemetry and side-scan sonar. Acoustic telemetry uses stationary receivers to detect signals emitted from tagged marine species such as sturgeon. The receivers store the unique tag number and the date and time that a fish swims past a receiver (like e-z pass for sturgeon). The side-scan sonar uses sound waves to create an image of the river bottom and objects in the water column, such as fish.
During winter side-scanning surveys, we also use gill nets to capture fish allowing us to validate the species that we are “seeing” on the side-scan imagery. Shortnose sturgeon congregate in large numbers in over-wintering locations. The relatively small size of adult shortnose sturgeon coupled with dense clusters of the fish make manually counting the side-scan sonar images very time consuming. As a result, we are now using automated image-processing analysis, followed by the application of machine learning to help speed up the process. Both programs are open-source and easily modifiable to fit the specific size of the fish and image quality. Pairing two open-source image and data processing programs gave us the ability to count large numbers of fish much more quickly than the previous human counts.
The side-scan sonar derived counts will be merged with the river-wide acoustic telemetry data to estimate the proportion of individual fish in the overwintering areas, and by extension, in the overall Hudson River. This provides a robust and relatively low-cost means to track recovery of America’s first endangered fish. By early 2024, we hope to have population estimates for two years.
Funding for this project comes from the Hudson River Foundation and the Hudson River Estuary Program. The shortnose sturgeon were collected and tagged under a National Marine Fisheries Service Endangered Species Act Research permit.