Towards scalable modern marine ecosystem monitoring

The ocean covers 71% of Earth's surface yet remains profoundly undersampled. Climate change is driving warming, acidification and deoxygenation at unprecedented rates, while ecosystems, often represented by iconic umbrella species such as North Atlantic right whales, emperor penguins, and coral reef ecosystems, face mounting anthropogenic pressures. Addressing these data gaps requires fundamentally better ocean observation capabilities, yet progress is constrained by the high cost of field campaigns, and institutional risk-aversion toward novel technology. This dissertation addresses the “observation gap” in ocean science by exploring low-cost, IoTderived technologies and empirically driven modeling to create robust, scalable marine ecosystem observing systems. Chapter 2 presents an agent-based model that integrates empirical whale diving behavior and realistic vessel kinematics, to evaluate the detection performance of surface-based thermal infrared sensors under realistic operational conditions, providing an evidence base for speed-restriction and real-time avoidance policies. Chapter 3 introduces ALERT, an automated IoT system integrating custom VHF signal processors with LoRaWAN to deliver near-real-time notifications for efficient archival biologger retrieval. Deployed in Antarctica, it successfully detected returning emperor penguins, demonstrating that robust, polar-grade automation improves detection reliability and response time while reducing manual effort and freeing field teams for other scientific work. Chapter 4 extends this approach by introducing a cooperative, encounterdriven mesh network in which LoRa-enabled tags opportunistically exchange data at sea before wirelessly offloading to a fixed gateway. Empirically driven agent-based modeling and LoRa linkbudget analyses indicate that this architecture could recover nearly 100% of data from colonyreturning penguins and ~40% from non-returning individuals. Two independent prototypes and field tests further demonstrate the potential of low-cost LoRa (peer-to-peer and LoRaWAN) systems for ecosystem research, even under Antarctic conditions. Finally, Chapter 5 shifts scale entirely, demonstrating through systematic in situ measurements that individual coral colonies act as persistent point sources of dissolved dimethylsulfide (DMS), with potential ecological and climatological implications, a finding only possible by sampling at colony scale rather than bulk water. This thesis coheres around three themes, model-informed design, scale-aware sensing, and rigorous adaptation and planning of cost-effective IoT for high-stakes field science, offering a pathway to democratize and expand ocean observation while maintaining scientific rigor.

Grantor dc:publisher

Massachusetts Institute of Technology and Woods Hole Oceanographic Institution

Year dc:date.issued

2026

Author dc:creator

• Baille, Loicka Myriam R.

Advisor dc:contributor.advisor

• Zitterbart, Daniel P.

• Ocean
• Ecosystem
• IoT

DOI

10.1575/1912/72998

OAI identifier oai:identifier

oai:darchive.mblwhoilibrary.org:1912/72998