While browsing environmental news in The Conversation, I came across an article that truly stood out. It covers critical information that absolutely needs to be brought to light and shared with a larger audience. In March 2025, surfers at Waitpinga Beach on South Australia’s Fleurieu Peninsula reported coughing, stinging eyes, sore throats, and breathing discomfort after entering waters affected by a foamy surface slick. Within weeks the same thing was happening at every beach from Kangaroo Island to the Coorong, and dead fish were appearing on the sand by the truckload.

What began as a localised algal bloom became the largest marine wildlife mortality event in Australia’s recorded history. Over the next twelve months, citizen scientists logged more than 117,000 dead animals across 430 species (iNaturalist community, 2026). Beach clean-up crews removed 9,400 kilograms of carcasses in a single week at the December peak (Government of South Australia, 2026). South Australia’s tourism and fishing industries lost an estimated $200 million.

This was not a freak event. Three slow-moving climatic disasters had been quietly stacking the deck for years. By the time the bloom arrived, the surprising thing was not that it happened. It was that it had not happened sooner. And the same conditions are loading again right now.

A coastline goes red, then quiet again

The first sign of trouble was a microscopic dinoflagellate called Karenia mikimotoi, which damages fish gills and triggers human respiratory irritation. Marine biologists watch for it at 100,000 cells per litre, the threshold above which serious ecological harm becomes likely. SA Health’s water-testing program tracks 630 sites along the South Australian coast and tests them weekly (Government of South Australia, 2026).

The chart tells a brutal story. Adelaide Metro readings peaked at 23 million cells per litre in early September 2025, which is 230 times the danger threshold. By December the bloom had largely retreated from the metro beaches, only to flare up again in offshore waters around Kangaroo Island in January 2026. The pattern is consistent with what marine biologists call a “rebloom”: the algae never fully die off, they just wait for the next nutrient pulse.

What dies when a sea turns toxic

South Australia does not have a national agency dedicated to counting dead marine animals. The job fell to ordinary people. A volunteer-run iNaturalist project, SA Marine Mortality Events 2025–2026, became the de facto national record (iNaturalist community, 2026). Anyone with a phone could photograph a dead animal on a beach and upload it. The project verified 117,366 observations across 430 species in fifteen months.

Bony fish make up the bulk of the deaths, but the most striking pattern is breadth. Crustaceans, octopuses, sea urchins, snails and even seabirds appear in the data, animals that occupy entirely different parts of the food web. Karenia attacks gill tissue and is not, on its own, supposed to kill seabirds. That it shows up here at all is one of the loose threads scientists are still pulling at.

The triggers, in slow motion

Three things happened before the bloom that should have happened separately, but did not. The first was a flood. Through 2022 and into 2023 the River Murray ran higher than it had in a decade, dumping nutrients into the Coorong and the Southern Ocean (Bureau of Meteorology, 2026). The second was a marine heatwave: from 2024 onwards the surface of the Great Australian Bight sat persistently 1 to 2 degrees warmer than its long-term average (National Oceanic and Atmospheric Administration, 2026). The third was the bloom itself, but it could only happen because the first two had primed the system.

Read top to bottom. The first panel shows the 2022–23 flood pulse arriving downstream. The second shows sea surface temperatures climbing well above the 2020–23 baseline through 2024 and into 2025. The third shows the bloom arriving, on cue. None of these triggers is unusual on its own. The unusual thing was their order.

Geography of a kill

The mortality observations (iNaturalist community, 2026) are not evenly spread. Some beaches were buried under thousands of dead pipis. O thers fifty kilometres away barely registered a fish. The pattern matters because it tells scientists how the bloom moved.

Click any taxon group on or off to see how different parts of the food web were affected. The densest hotspot sits in Gulf St Vincent, the body of water that wraps around Adelaide. The current systems that should normally flush this gulf were weakened by the marine heatwave, turning Gulf St Vincent into a slow-motion trap. The Eyre Peninsula coast, which faces the open Southern Ocean, was hit later, but its species composition tells a different story: more crustaceans, more seabirds, fewer demersal fish.

Did intensity predict damage?

If the bloom killed by direct toxic exposure, you would expect a clean line: the higher the Karenia concentration, the more dead animals nearby. Most of the public conversation has assumed exactly this. The data is more complicated.

Each dot is one of the 630 monitoring sites. The horizontal axis is how intense the bloom got there. The vertical axis is how many dead animals citizen scientists found within 5 kilometres. Two things stand out. First, Adelaide Metro sites (in red) sit far above every other region at every level of bloom intensity, suggesting either better detection by urban citizen scientists, or a genuinely higher mortality rate driven by the gulf’s poor flushing. Second, several sites recorded high mortality at relatively low bloom intensity. That means Karenia concentration alone does not explain the kill. Other factors such as deoxygenation, secondary toxins, and food-web cascades are doing real work. I have used absolute counts rather than per-capita rates because the question here is geographic, not normalised: where did the most dying happen? A per-capita view would tell a different story about detection effort, not impact

The next bloom is already loading

Three triggers converged once. Two of them, Murray flow and sea surface temperature, are still elevated. Coastal monitoring shows Karenia cell counts in offshore waters rising again in early 2026. South Australia’s environment department is now treating bloom events as a recurring risk rather than a one-off disaster.

The harder question is whether Australia has the monitoring infrastructure to see the next one coming. South Australia’s 630-site network is comprehensive within state waters, but the federal marine-heatwave warning system and the freshwater-discharge gauges sit in separate agencies that do not routinely combine their data. The bloom of 2025 was visible in the data three years before it happened. Nobody was looking at all three signals together.

Citizen scientists with phones did the counting. They should not have had to.

Data and methods

This article combines four open datasets. Karenia cell counts and monitoring site locations are from the South Australian Government Algal Bloom dashboard (Government of South Australia, 2026), accessed via its public ArcGIS feature service. Mortality observations are from the SA Marine Mortality Events 2025–2026 citizen-science project on iNaturalist (iNaturalist community, 2026), exported as a CSV containing 117,366 verified records. River Murray flow at Lock 10 Wentworth is from the Bureau of Meteorology Water Data Online service (Bureau of Meteorology, 2026). Monthly sea-surface-temperature anomalies were computed from NOAA Optimum Interpolation SST v2.1 satellite gridded data (National Oceanic and Atmospheric Administration, 2026), with anomalies calculated against a 2020–2023 monthly climatology over the latitude range 38°S to 32°S and longitude 133°E to 141°E. All data preparation and visualisation was performed in R using the arcgislayers, rerddap, plotly, leaflet, and tidyverse packages.

Acknowledgements

Generative AI (Anthropic’s Claude, accessed May and June 2026) was used during this assignment in line with RMIT University Library (2026) guidance on the acknowledgement and referencing of AI tools. I used it as a sounding board when brainstorming the story angle, when troubleshooting plotly date-axis behaviour during chart construction, and when reviewing the colour palette for colour-vision safety. All data sourcing, narrative writing, chart-design choices, and R coding were my own.

I also acknowledge the volunteer iNaturalist community whose 117,366 observations form the basis of half this analysis, the ColorBrewer 2.0 project (Brewer & Harrower, 2013) for the colour-vision-safe palettes used throughout, and the Bureau of Meteorology and the United States National Oceanic and Atmospheric Administration for providing the freshwater-flow and sea-surface-temperature data freely under open licence.

References

Anthropic. (2026). Claude (Opus 4.7) [Large language model]. https://claude.ai

Brewer, C. A., & Harrower, M. (2013). ColorBrewer 2.0 [Web tool]. The Pennsylvania State University. https://colorbrewer2.org/

Bureau of Meteorology. (2026). Water Data Online: Murray River at Lock 10 Wentworth (A4260512) [Data set]. Commonwealth of Australia. http://www.bom.gov.au/waterdata/

Government of South Australia. (2026). Harmful algal bloom monitoring sites and water testing data [ArcGIS Feature Service]. Department for Environment and Water. https://www.algalbloom.sa.gov.au/

iNaturalist community. (2026). SA Marine Mortality Events 2025–2026 [Citizen science observations data set]. iNaturalist Network. https://www.inaturalist.org/projects/sa-marine-mortality-events-2025-2026

National Oceanic and Atmospheric Administration. (2026). NOAA Optimum Interpolation Sea Surface Temperature (OISST) v2.1 [Satellite gridded data set]. NOAA NCEI. https://www.ncei.noaa.gov/products/optimum-interpolation-sst

RMIT University Library. (2026). Artificial intelligence (AI): Acknowledgement and referencing guidelines. RMIT University. https://rmit.libguides.com/referencing_AI_tools