Right now, salmon farmers in southern Norway are doing something that no treatment protocol, no veterinary intervention, and no amount of operational experience can fully solve. They are watching the water and waiting.
A Pseudochattonella bloom (a species of toxic microalgae) has been tracking north along the Norwegian coast since February. By mid-March 2026, it had already killed more than 105,000 salmon across five sites in Rogaland. By the final week of March, Norway's Institute of Marine Research (Havforskningsinstituttet, HI) had upgraded the risk to HIGH across three full production areas covering the Skagerrak coast, Ryfylke, and Bømlo. At one site in Flekkefjord, 41 percent of the entire stock died within a single event window.
The industry is currently in response mode. Mowi, Grieg, and Bremnes Seashore have all activated contingency measures — lowering nets, halting feeding, coordinating with slaughter facilities. And yet there is a fundamental tension at the heart of every decision being made right now: the bloom does not negotiate.
This is what makes algae events so distinct from almost every other challenge in salmon farming, and why they deserve more sustained attention than they typically receive outside of the weeks when a bloom is actively killing fish.
Most mortality events in salmon aquaculture follow a logic that the industry has spent decades learning to interrupt. Pathogens can be detected, treated, or vaccinated against. Sea lice can be counted and treated. Oxygen levels can be measured and corrected. Even chronic stressors (water temperature, gill disease, feed issues) have response levers that farms can pull.
Algae blooms don't work that way. Pseudochattonella, Chrysochromulina, Karenia — these are free-floating microorganisms driven by ocean physics. They grow in the warm, stratified surface layer when spring conditions align: snowmelt lowering salinity, lengthening daylight accelerating growth, calm winds preventing mixing. They can travel dozens of kilometres on coastal currents. And they arrive at cages with very little warning.
The toxins some species produce (haemolytic compounds that destroy gill cell membranes) act in hours. A fish exposed to lethal concentrations of Chrysochromulina leadbeateri begins dying long before any response can be mobilised. By the time mortality is visible, the window for intervention is often already closing.
There is no antidote. There is no treatment protocol. The standard response measures (lowering cage depth, stopping feeding, emergency harvest) are not cures. They are attempts to reduce exposure.
The current Rogaland outbreak is serious. But to understand why algae blooms deserve to be treated as a structural risk and not just a bad spring, it helps to look at the historical record.
In 2019, a Chrysochromulina leadbeateri bloom spread across approximately 450 kilometres of northern Norway's coastline. It killed an estimated 7–8 million farmed Atlantic salmon — roughly 14,500 tonnes — in the span of a few weeks. Financial losses ran to approximately NOK 800 million. One company, Ballangen Sjøfarm, lost around 2.7 million fish. Northern Lights Salmon lost close to 2.4 million. Insurance covered only about 20 percent of production costs.
HI researchers described the bloom as unprecedented in scale. When asked why 2019 produced such an extreme event, the institute's scientists acknowledged they had no single unambiguous explanation.
Further back: 1991, 1998, 2008. Different years, same fjord systems, same species, similar patterns.
And in Chile, in 2016, a Pseudochattonella verruculosa bloom, the same genus now spreading through Rogaland, killed approximately 27 million salmon and trout across the fjords of Los Lagos Region. Roughly 39,000 tonnes. Losses estimated at US$800 million. It remains the largest farmed-fish algae mortality event ever recorded globally. The bloom coincided with a strong El Niño that produced warm, stratified, essentially stagnant water in fjord systems with flushing times of up to 200 days. The conditions were, in a sense, a waiting catastrophe.
Chile has since seen major Pseudochattonella events in 2021, 2023, and 2024. The scale varies. The pattern does not.
Norway has the most developed algae monitoring infrastructure of any salmon-farming nation. HI's Algestatus service publishes regular risk assessments throughout the spring and summer season. The institute runs satellite remote sensing, operates FerryBox sensors on commercial shipping routes, tracks current models, and coordinates with Denmark and Sweden to catch blooms originating in the Skagerrak before they track north toward Norwegian farms. Private monitoring companies like SeaEco AS have supplemented public systems with dedicated surveillance networks in northern production areas.
It is, by global standards, a serious and well-funded system.
And yet the current outbreak has still produced extensive, rapid mortality. The alarm was raised. Farmers were warned. Risk was upgraded incrementally from moderate to high as the bloom intensified. At Buksevika (Mowi's worst-affected site) fish still died at a rate of more than 40 percent before the bloom passed.
The honest assessment from HI researchers is that the system tracks blooms as they develop but cannot reliably predict with site-level precision which farms will be hit first, or at what concentration, or on what timeline. Ocean physics doesn't cooperate with the resolution that operational decisions require.
Scotland's HABreports system, which is ran by the Scottish Association for Marine Science and built around the EU ASIMUTH project, is Europe's most advanced shellfish biotoxin early warning platform, but finfish-specific algae forecasting remains underdeveloped. Chile runs hundreds of monthly monitoring stations across Patagonian waters, but monthly sampling across geography of that scale is structurally too infrequent to function as early warning.
All three of the world's major salmon-farming nations share the same fundamental gap: the monitoring is better than it was, and it's still not predictive enough to eliminate the response lag that costs fish.
Part of the challenge is that algae blooms sit at the intersection of oceanography, climate science, and farm operations, three domains that rarely speak the same language.
The conditions that produce dangerous blooms are getting harder to forecast, not easier. Warming ocean temperatures are extending the seasonal window for certain species. In Chile, HAB events that once concentrated in summer are appearing in other seasons. In Norway, Pseudochattonella appears to be tracking further into Rogaland's production areas than it once did.
The other challenge is economic. A comprehensive real-time early warning system would require dense sensor networks, continuous water sampling, high-frequency satellite data, and operational protocols that can be activated within hours. That infrastructure is expensive. And because algae blooms are, from an individual farm's perspective, infrequent, the ROI case for sustained investment can be difficult to make in a good year.
The consequence is that preparation tends to happen reactively, in the weeks after a major event, when the losses are fresh and the political pressure is high. Then the cycle repeats.
The hardest number to look at isn't the mortality count. It's the time between when a bloom is first detected at moderate concentrations and when it hits a specific farm site at lethal density.
Right now, that gap is often measured in hours, not days. And most of the operational responses available to farmers (lowering nets, early harvest, wellboat transfer) require hours of logistics to mobilize. By the time a farm can act at scale, it is often already inside the mortality window.
Closing that gap (even partially) is where better data and better models make a material difference. Not by predicting the unpredictable, but by shifting the response window even slightly earlier. If an operator knows with higher confidence that their specific site faces HIGH risk in the next 6–12 hours rather than the next 24–72, the decision to pre-position a wellboat, halt feeding, or bring forward an emergency harvest changes. That's not a technology pitch. It's an arithmetic one. In a mortality event measured in hundreds of thousands of fish, hours matter.
Algae events tend to arrive when farms are most exposed — spring, when cages are stocked, when fish have just gone through the stress of seawater transfer, and when slaughter capacity is stretched across a regional industry all facing the same risk simultaneously.
The 2019 crisis overwhelmed regional slaughter capacity at multiple points during the peak mortality period. Multiple companies were trying to emergency-harvest simultaneously, in the same fjord systems, with the same limited number of wellboats. The coordination failures of that event were as expensive as the bloom itself. The 2026 Rogaland outbreak is still developing. The final mortality count isn't known. But the pattern is recognisable to anyone who has been watching this industry for more than a decade.
A bloom arrives and it intensifies faster than expected. The response infrastructure proves just sufficient or just insufficient, depending on the site. Researchers note that conditions were unusual. And then, eventually, the water turns clear again, and the industry begins calculating the cost.
Harmful algal blooms are not a new problem. They are a recurring one. The question isn't whether the next event will come. It's whether the data infrastructure to respond earlier will be in place when it does.
Manolin's Fish Disease Library includes detailed entries for anyone looking for deeper reference material on the biology, monitoring, and farm response protocols behind these events.