Skip to content

Real-time island biosecurity surveillance: Evaluating a wireless camera network for AI-assisted early detection of invasive mammals on Santa Cruz Island, CA

Published Article

California

Publication date: May 27, 2025

File format: URL

View resource

Rapid detection of nonnative mammals is essential for preventing biological invasions, yet traditional camera traps slow response times due to manual image retrieval. On Santa Cruz Island, a wireless camera network transmitted images instantly from remote sites to the cloud, where machine‑learning models flagged potential invasive species. High recall for rodent detections and faster, more reliable data review greatly improved early‑warning capacity compared to SD‑card systems. Although initial equipment costs were higher, networked cameras proved more cost‑effective over long deployments. Future versions could integrate cellular or satellite links, expanding biosecurity and conservation applications across vulnerable island ecosystems.

Subject Tags

  • Conservation Technology
  • Invasive Species

Abstract

Early detection of nonnative mammal incursions enables rapid management actions that are needed to prevent full-scale invasions. As biosecurity monitoring tools, camera traps can aid in the detection of nonnative species; however, the burden of image management and resources required to access cameras regularly for image collection both inflate costs and extend the latency period between invasive animal detection and manager response. Here, we describe a wireless camera network on Santa Cruz Island (SCI) that enabled instantaneous transfer of camera images from remote field sites to the cloud. Initial classification of images by machine learning allowed human reviewers to prioritize examining photos of possible concern. Comparison of AI predictions and human-validated image labels confirmed that machine-learning models had high recall (or a low false negative rate) for image sequences containing rodents. Comparisons with a prior SD-card-based camera system on SCI revealed significant improvements in data review frequency and reliability, improving the likelihood of prompt nonnative species detection. Despite higher initial equipment costs, networked cameras were cost-effective over time, outperforming traditional methods in long-term deployments. Future iterations of the network could leverage cellular or satellite networks for broader scalability, enhancing biosecurity and general conservation efforts on islands and other vulnerable protected sites.

Citation

Brenner, L. J., Rindlaub, N., Matos, J., Meyler, S., Pollock, S., Schuetzenmeister, F., & Holmes, N. D. (2025). Real-time island biosecurity surveillance: evaluating a wireless camera network for AI-assisted early detection of invasive mammals on Santa Cruz Island, CA. Western North American Naturalist, 85(2), 365-375. https://doi.org/10.3398/064.085.0220

TNC Authors

TNC Authors

  • Nick D. Holmes
    Associate Director, Oceans Program, California
    The Nature Conservancy
    Email: nick.holmes@tnc.org