By Elizabeth McHuron, UW CICOES, with contributions from Sharon Melin, NOAA Alaska Fisheries Science Center

San Miguel Island

Perched atop a cliff, on an island less than 30 miles from the southern California coast, sits a small unassuming research station that—on a clear day—boasts million-dollar views of white sandy beaches, sparkling blue waters, and magnificent sunsets. While San Miguel Island is not physically far from one of the world’s most iconic metropolises, it is a world away from the traffic, lights, and sunny weather of the City of Angels. There is no running water, and instead of car horns, the sounds of barking sea lions can be heard from the research station overlooking the western tip of the island that is home to one of the world’s largest California sea lion colonies.

Human visitors to the island are few and far between. The most consistent presence comes from visitors to the Channel Islands National Park and a small group of researchers from the Marine Mammal Laboratory at NOAA’s Alaska Fisheries Science Center in Seattle. The researchers spend much of their time counting, weighing, and marking California sea lions and some of the other four seal and sea lion species that breed on the island. Their efforts have generated nearly a half-century of California sea lion data spanning numerous El Niño events and the infamous “Blob” that brought extremely warm sea surface temperatures to much of the North Pacific Ocean.

Now, under shifting federal research priorities, the fate of these long-term sea lion datasets hangs in the balance. These shifts have been the impetus for a new collaborative effort among researchers at CICOES and the Marine Mammal Lab’s California Current Ecosystem Program. Together, we are developing predictive tools to help mitigate the potential loss of further data collection while also informing cost-effective sampling strategies to maintain this valuable timeseries.

Learning from the past to predict the future

Long-term ecological datasets, such as the ones collected on San Miguel Island, are invaluable for understanding how ongoing and future environmental changes will impact individual species, communities, and entire ecosystems. In the California Current Ecosystem, which includes nearly 3,000 km of waters from Baja California to southern British Columbia,

California sea lions are considered ecosystem indicators because they respond rapidly to environmental changes, such as increases in ocean temperatures that can affect food availability or the production of algal toxins that can result in mortality. California sea lions are also one of the most visible and charismatic reminders of ocean conditions along the US west coast, as starving or sick sea lions often become stranded on very public shores. Stranded animals are a potential public health risk because of zoonotic diseases, and they also can strain limited resources of rehabilitation facilities that care for them.

Our goal is to build upon existing knowledge that, simply put, warm years produce small pups, presumably because their moms have a harder time finding food. This has cascading effects because starving pups tend to strand more during these years and survival rates tend to be lower. For example, during one of the very warm years of the “Blob,” only 11% of pups born were estimated to survive to their first birthday when pups grew very little and were 3-5 kg lighter than the long-term average. We plan to explore how well we can predict the weight of sea lion pups in the fall and their growth rates in late fall/early winter using just measurements of sea surface temperatures within foraging areas, and whether considering other factors, such as upwelling indices or sea lion diets, markedly improves our predictive ability.

Preliminary results indicate that sea surface temperature alone does a reasonably good job in predicting both pup mass and growth rates in many years. For example, models successfully predicted 40% of the years when pup masses fell below thresholds associated with above-average strandings of sea lion pups. Most additional metrics added little in terms of predictive ability, with the exception of sea lion diets that help explain why pups in some years are larger or smaller than expected based on sea surface temperatures alone. This comes at a trade-off though. While considering diets increases our ability to predict years with anomalously low pup masses (from 40% to nearly 70% for some prey species), it also results in more ”false alarms” or years in which models predict pup masses should be anomalously low but are not. This raises the question of whether, from a management or economic perspective, it is worse to be unprepared for something that eventually happens or to prepare for an event that never materializes.

Other values of long-term datasets

Research programs that generate long-term ecological datasets help train the next generation of scientists, extending their value well beyond their current ability to understand ecological processes and inform management decisions. Training can come through a variety of mechanisms, including use in undergraduate classrooms, as sources for theses and dissertations, and through hands-on training in the field. For example, several projects within the CICOES undergraduate internship program have involved long-term datasets on salmon and groundfish in Alaska.

I have first-hand experience with these benefits, having traveled to San Miguel Island in 2013 as a (somewhat) young PhD student. I spent more than a week studying moms and their pups at the tail-end of an unexplainably (still!) poor year. This was my first foray into the sea lion world, but certainly not my last. It provided the experience I needed to complete my PhD on California sea lions and created collaborative opportunities that have helped shape the trajectory of my career. I am just one of the numerous interns, volunteers, and students who have experienced the magic that is San Miguel Island; we each carry a small piece of the institutional knowledge from NOAA researchers, which is so vital for marine mammal research.

Bridging the gap

The ability to predict the mass and growth rates of California sea lion pups, based on sea surface temperatures—albeit imperfectly—helps provide an early warning of conditions in the California Current Ecosystem to state and federal managers and other interested stakeholders, as well as valuable information for stranding centers to anticipate and coordinate responses to large stranding events of emaciated sea lion pups. We can forecast these predictions months in advance because of the existence of regional ocean models that simulate ocean temperatures and other physical and biogeochemical conditions. For example, a suite of new regional ocean models developed under NOAA’s Changing Ecosystems and Fisheries Initiative aims to provide high-resolution past and future conditions across large swaths of the Atlantic, Pacific, and Arctic Oceans. While our focus is on a breeding rookery in southern California, results have ramifications for the entire California Current Ecosystem because California sea lions are wide-ranging within this ecosystem throughout the year, and changes in pup growth and survival that help drive population dynamics are connected to local ecosystem processes.

Long-term ecological datasets are, by definition, collected regularly and continuously. This regularity can be difficult to maintain due to the need for consistent support, in addition to unexpected events that may derail data collection. For example, California sea lion pups had been weighed annually from 1975 until 2020, when the COVID-19 pandemic resulted in the cancellation of fieldwork. Similarly, weather and funding delays in 2024 resulted in cancellation of the fall field season. Predictions from our models indicate that neither 2020 nor 2024 had a high probability of being poor years for sea lions, which is consistent with anecdotal observations that eight-month-old pups in the 2024 cohort were in good body condition.

Predictions cannot replace the need for ongoing data collection, crucial for validating and improving the accuracy of forecasts. They can help bridge the gap when annual data collection is not possible, as was the case in 2020 and 2024. Forecasts can identify years when it may be important to prioritize data collection, such as average sea surface temperature years when models struggle to accurately predict pup masses. Combining regular annual predictions with less frequent field sampling provides a path forward to help maintain this valuable long-term dataset under a shifting landscape of research and funding priorities.