Every summer, the water at the bottom of Florida’s Banana River Lagoon — a 40-kilometer (24.9-mile) sub-lagoon running parallel to the Indian River Lagoon system — quietly suffocates. Dissolved oxygen crashes. Fish die or flee. The sediments release pulses of nitrogen and phosphorus that feed the next algal bloom, setting the stage for another crash. Researchers from Florida Tech have now mapped this vicious cycle in greater detail than ever before. What they found should concern anyone tracking the health of the lagoon: the single state monitoring sensor inside the study area, operating at mid-depth as required under standard practice, consistently read the water as healthier than it actually was at the bottom — by 0.7 milligrams per liter on average, and by as much as 2.1 mg/L at the peak of the worst month. That gap means the official picture of the lagoon’s oxygen conditions has been systematically too optimistic, in exactly the layer where fish live and the feedback loops that sustain eutrophication operate.

The study, published in Frontiers in Marine Science on August 31, 2026, deployed a network of 80 continuous dissolved oxygen sensors directly on the lagoon floor, recording measurements every hour for a full year. It is among the most spatially comprehensive dissolved oxygen monitoring efforts ever conducted in a shallow estuary — and its core methodological finding may be its most important: the way Florida, and most other states, monitors water quality in systems like this one systematically misses the worst conditions, at the worst moments, in the worst places.

An Unprecedented Bottom-Water Picture Across a Struggling Lagoon

The Banana River Lagoon (BRL) is shallow — averaging just 1.7 meters (5.6 feet) deep — and nearly cut off from tidal flushing, with freshwater residence times of 230 days to a full year. There are no inlets within the lagoon itself, and wind is its primary mixing force. These conditions make it a near-perfect natural laboratory for watching what nutrient pollution and warm subtropical temperatures do to a coastal ecosystem over time.

Led by Austin Fox of Florida Tech’s Department of Ocean Engineering and Marine Sciences, the research team placed 80 dissolved oxygen sensors at the sediment-water interface — the layer immediately above the lagoon floor — recording data from July 2024 through June 2025. Sensors were positioned using a 2-kilometer hexagonal tessellation grid, a spatial statistics method designed to ensure even distribution across the full lagoon, not just near existing monitoring infrastructure.

Each sensor was a HOBO U26-001 or MX801-DO optical datalogger housed in a copper-treated PVC enclosure that prevented biofouling without blocking water flow. Units were retrieved approximately monthly, swapped for freshly calibrated instruments, and validated against secondary probes during each field visit — with any reading that failed a post-deployment accuracy check of ±0.3 mg/L excluded from analysis. Of 800 post-calibration checks, 23 did not meet standards and were removed.

The network’s purpose was to test whether high-resolution bottom-water monitoring could identify hypoxic events — oxygen below the 2 mg/L threshold at which fish communities begin to experience serious harm — that lower-resolution or mid-depth monitoring would miss. The answer was an unambiguous yes.

What “Hypoxia” Means and Why the Threshold Matters

Hypoxia refers to oxygen-depleted water, defined in the scientific literature as dissolved oxygen below 2 milligrams per liter. Below that level, most fish either flee or die. Below 0.5 mg/L, mass mortality occurs. The condition is caused by the same eutrophication cycle playing out in lagoon systems worldwide: excess nutrients (nitrogen and phosphorus from fertilizer runoff, septic systems, and urban stormwater) fuel algal blooms. When those blooms die and decompose, bacteria consume the remaining oxygen. The deeper problem is what happens next: hypoxic and anoxic sediments chemically release the phosphorus they had previously bound — through iron reduction under anaerobic conditions — and impair the nitrification processes that would otherwise remove nitrogen from the system. The result is a self-reinforcing eutrophication and hypoxia loop: eutrophication causes hypoxia, and hypoxia releases the nutrients that fuel the next round of eutrophication.

This dynamic has been well documented in large, deep, stratified systems like the Chesapeake Bay and the Baltic Sea. The Baltic is the canonical warning case: Conley et al. (2002) showed in Environmental Science & Technology that that the extent of hypoxia there correlated more closely with dissolved phosphate concentrations than with external nutrient loading — meaning the system had crossed a threshold where internal loading from its own sediments was sustaining eutrophication independently of what flowed in from shore. The Fox et al. study explicitly cites this Baltic precedent and warns that the Banana River Lagoon may be at or approaching a comparable threshold.

What has been far less studied is how hypoxia behaves in shallow, well-mixed systems — lagoons averaging under 2 meters (6.6 feet) deep, without the vertical stratification that traps low-oxygen water in deeper estuaries. The Banana River Lagoon study was designed specifically to fill that gap.

A Seasonal Swing of 4.8 mg/L — and a Monitoring Gap That Hid the Worst of It

Across the 12-month monitoring period, basin-wide monthly dissolved oxygen averages at the sediment surface ranged from 9.0 mg/L in January 2025 (when cool water holds more oxygen) down to just 4.2 mg/L in June 2025 — a 4.8 mg/L seasonal swing. Percent saturation followed the same arc, ranging from 102% in mid-winter to 65% at the summer trough.

The single SJRWMD (St. Johns River Water Management District) monitoring station within the study area — a standard mid-depth instrument sensor located near the center of the lagoon — read consistently higher than the bottom-layer network throughout the year. That gap averaged 0.7 mg/L across the full monitoring period and peaked at 2.1 mg/L in October 2024, at the tail end of the hot season when bacterial metabolism in the sediments was consuming oxygen fastest. In February 2025, the pattern briefly reversed: cold weather caused dense, oxygen-rich surface water to sink, and bottom readings actually exceeded mid-column levels by 0.8 mg/L — one of the few moments when standard mid-depth monitoring would have underestimated how good conditions were.

The researchers note that the mid-depth sensor was operating correctly — it measured what it was designed to measure. The problem is that it was measuring the wrong layer for this type of system. Traditional monitoring in Florida and most other states relies on mid-depth sensors and daytime discrete sampling. Both approaches, the study concludes, systematically overestimate average oxygen levels at the sediment surface in shallow systems like the BRL, because daytime photosynthesis pushes oxygen to its daily high just when most monitoring is conducted, and because mid-depth water is better mixed and less affected by the oxygen-consuming processes happening at the bottom.

Hypoxia on Nine Out of Every Ten Days

The bottom-layer data revealed conditions far more persistently dangerous than mid-depth monitoring would suggest. At least one sensor in the 80-station network recorded dissolved oxygen below the 2 mg/L threshold on 90% of all monitoring days — nearly every day of the year.

Individual sites spent anywhere from 0.2% to 40% of the entire monitoring period below 2 mg/L — between 18 and 3,500 cumulative hours. Eleven times over the year, the lagoon-wide average dropped below 2 mg/L simultaneously across the basin, with individual events lasting between one and eight hours. Three of those events followed algal blooms in July and August 2024 — identified by characteristic oxygen signatures, where DO first soared to extreme highs during the bloom, then crashed as the bloom decayed. The remaining eight events occurred between June 17 and 30, 2025, following the sudden, near-total collapse of the macroalga Caulerpa prolifera across the lagoon floor.

The diel (daily) oxygen cycle was itself a driver of near-daily hypoxia. At the sediment surface, dissolved oxygen swung an average of 5.1 mg/L between daily highs and lows over the study period — peaking at 7.0 mg/L/day in July 2024 and reaching a single-day maximum of 21.1 mg/L/day on July 29, 2024, during an intense algal bloom. The largest single diel range ever recorded at one site — from 290% saturation during the bloom to 0% saturation during its collapse — illustrates in a single number what happens when a dense algal bloom crashes overnight.

In the early morning hours, when respiration had depleted the overnight oxygen supply and photosynthesis had not yet resumed, bottom water across large portions of the lagoon regularly fell below the hypoxia threshold. This pattern of diel hypoxia — oxygen crashing every night and recovering during the day — is largely invisible to monitoring programs that sample during daylight hours.

The Caulerpa Collapse and the Worst Month on Record

The most extreme oxygen conditions of the entire study period did not occur during the hottest weeks of summer 2024. They occurred in June 2025, following an ecological event that restructured the lagoon floor: the rapid, near-complete loss of Caulerpa prolifera, a macroalgal species that had dominated the BRL bottom since seagrass began collapsing after 2011 algal blooms.

Beginning in late April and early May 2025, the sea slug Elysia subornata — which grazes on Caulerpa — proliferated across the lagoon. Field observations and SJRWMD macrophyte monitoring confirmed a collapse of Caulerpa cover from roughly 45% of the lagoon in 2023 to just 3% by summer 2025. As decaying Caulerpa biomass decomposed, the heterotrophic oxygen demand of the decomposition process pushed bottom-water dissolved oxygen to its lowest levels of the entire monitoring year — lower even than the hottest August days, because the biological oxygen demand from decomposing organic matter compounded the already-strained summer conditions.

Eight of the 11 basin-wide hypoxia events occurred in those final two weeks of June 2025. The June 2025 monthly average DO across the entire lagoon floor — 4.2 mg/L at 65% saturation — was lower than any prior month in the study. On July 25, 2025 — after the study’s formal monitoring period had ended but while field work continued — a fish kill was recorded along the BRL’s eastern shoreline, where dissolved oxygen remained below 2 mg/L for at least 24 consecutive hours.

The Caulerpa collapse illustrates a broader vulnerability. The lagoon lost seagrass after 2011. Caulerpa partially filled that ecological niche. When Caulerpa collapsed, there was no resilient, oxygen-producing benthic community to take its place — only bare sediment with a massive pulse of decomposing biomass. The system’s response to that disturbance was the worst oxygen crisis on record.

Where Hypoxia Strikes Hardest: Muck, Macroalgae, and Bloom Corridors

The 80-station network made it possible to produce a detailed spatial map of where hypoxia is most severe and most persistent — information that could directly guide how restoration funds are spent.

Sites within 500 meters (1,640 feet) of “muck” — deposits of fine-grained, organic-rich anaerobic sediment that have accumulated in dredged channels and low-energy areas throughout the lagoon — had bottom-water oxygen 0.4 mg/L lower than areas more than 500 meters away. The effect extended beyond one kilometer (0.6 miles), with the deficit diminishing but remaining statistically significant with distance. Muck deposits act as persistent oxygen sinks, releasing chemically reduced compounds into surrounding water as anaerobic organic matter decomposes — an “oxygen debt” that spreads horizontally from the deposit footprint.

Spatial hotspot analysis (using the Getis-Ord GI* statistic in ArcGIS) identified two areas with significantly greater cumulative daytime hypoxia: southwest of the Cocoa Beach Golf Course and northwest of the Pineda Causeway. Both locations correspond to previously documented algal bloom hotspots in Brevard County. Both are near locations where fish kills have been historically reported.

Substrate type proved a more reliable predictor of hypoxia than water depth — a finding that inverts conventional monitoring assumptions built around deeper, stratified systems like the Chesapeake Bay. In the BRL, the shallowest sites (under 1 meter, or 3.3 feet deep) actually experienced more frequent hypoxia than deeper sites, because those shallow areas coincide with the dense macrophyte cover that drives the highest sediment oxygen demand. Sediment oxygen demand at vegetated sites averaged roughly three times higher than at adjacent bare sand — 10,300 versus 3,300 micromoles per square meter per hour, measured at eight paired sites.

What Does Oxygen Come from If Not Tides? Wind, at a Very Specific Speed

In a lagoon without tidal flushing, wind is the primary mechanism that mixes oxygenated surface water down toward the bottom. The study identified a critical threshold: when wind speeds exceeded 5 m/s (about 11.2 miles per hour), mixing pushed dissolved oxygen toward atmospheric equilibrium, sometimes producing oversaturation above 300% in fetch-exposed areas. Below 5 m/s (11.2 mph), bottom water oxygen tended to remain depressed because bacterial consumption in the sediments outpaced replenishment.

The practical implication is that residential canals, areas shielded by spoil islands, and other low-fetch areas within the lagoon received consistently less wind-driven oxygen mixing — keeping them chronically more hypoxic than open-water sites. Even during the two hurricanes that struck the area during the study — Hurricane Helene (September 24–27, 2024) and Hurricane Milton (October 5–10, 2024), which peaked at winds of 15.7 m/s (35.1 mph) and 18.1 m/s (40.5 mph) respectively — the mixing effect was temporary. Hurricane Milton, which passed directly over the study area, elevated turbidity so severely that dissolved oxygen remained depressed for several days after wind speeds fell, as the clouded water reduced photosynthesis.

Why Restoration Progress May Be Harder to Measure Than Managers Thought

This is where the study’s methodological finding carries its most serious policy implication.

Florida’s ongoing investment in Indian River Lagoon restoration is substantial. Brevard County’s Save Our Indian River Lagoon (SOIRL) program — funded by a voter-approved half-cent sales tax since 2016 — had collected nearly $497 million of a projected $586 million by early 2026, with 119 projects completed and 100 more underway. In May 2026, the St. Johns River Water Management District released encouraging seagrass monitoring data showing IRL seagrass coverage nearly doubled between 2023 and 2025 — from 9,924 to 17,042 hectares, an increase roughly equivalent to 13,000 football fields. One of the primary muck dredging projects — targeting roughly 248,000 cubic yards of accumulated organic sediment northeast of the Eau Gallie Causeway — launched in April 2026.

These are genuine, measurable improvements. But the Fox et al. study raises a question that restoration programs need to answer: if the monitoring system used to track progress systematically reads oxygen 0.7 to 2.1 mg/L higher than the bottom-water reality, what does measured improvement actually mean? The Central Indian River Lagoon Basin Management Action Plan draft released in April 2025 described a monitoring network of 45 stations for the entire IRL system. The Fox et al. study deployed 80 sensors in the BRL alone — one small sub-lagoon. The discrepancy in monitoring resolution is not a bureaucratic detail. It determines whether small but ecologically critical improvements in bottom-water oxygen are actually detectable.

The Baltic Sea parallel is instructive here too. Research has shown that in systems where internal loading from sediments has become the dominant nutrient source, reducing external inputs — stopping runoff, upgrading septic systems — may not produce measurable oxygen improvement in the layer that matters until sediment conditions themselves change. That takes time. If the monitoring framework does not measure bottom-water oxygen where and when it is worst, apparent improvement in the metrics may precede actual improvement in the ecosystem.

This does not mean restoration is failing. It means the ecosystem’s response to restoration may be harder to see than the monitoring framework currently allows — and that expanding the spatial and vertical resolution of oxygen monitoring should be considered alongside the restoration work itself.

Does Cutting Nutrients Even Help if the Sediments Have Taken Over?

The study’s synthesis and perspectives section explicitly cites the Baltic Sea as a cautionary precedent where internal loading from hypoxic sediments can sustain eutrophication even when external nutrient inputs are reduced. The key mechanism is the chemical release of phosphorus from iron compounds under anaerobic conditions — once sediments become persistently anoxic, they stop functioning as nutrient sinks and start functioning as nutrient sources. The study’s data show that the BRL’s worst hypoxia zones are already producing this effect, particularly near muck deposits. Muck removal — dredging — is therefore not a supplementary restoration tool but a necessary complement to nutrient source reduction. Neither alone is likely to be sufficient.

An Ecological Blueprint for Targeted Remediation

The study’s authors argue that the 80-station dataset functions as a spatial targeting tool for restoration investment — something that coarser monitoring could never provide.

Hotspot maps of cumulative hypoxia duration, proximity analysis of muck deposit impacts, substrate-type associations with oxygen demand, and the identification of specific algal bloom corridors all translate directly into prioritization decisions: where to dredge first, where to deploy aeration systems, where to focus habitat restoration to break the diel hypoxia cycle, and how to track whether any of those interventions are actually improving conditions where conditions are actually worst.

The researchers also call for further investigation of what they term a poorly characterized ecological risk: the chronic effects of repeated short-duration hypoxia. The BRL documented near-annual basin-wide hypoxia events, and at least one station experiencing low oxygen on 90% of monitoring days. Whether fish communities, benthic organisms, and the nutrient cycling pathways that support lagoon health can withstand that level of chronic stress — as distinct from the episodic hypoxia already well-studied — remains an open question.

Frequently Asked Questions

Why does the Indian River Lagoon keep having fish kills, and is the oxygen problem getting better or worse?

The fish kills are caused by dissolved oxygen crashes — typically overnight, when algal blooms or decaying organic matter consume the bottom-water oxygen that photosynthesis built up during the day. The Fox et al. study documented a direct fish kill on July 25, 2025, along the BRL’s eastern shoreline that resulted from oxygen remaining below 2 mg/L for more than 24 hours — a consequence of the Caulerpa prolifera collapse that month. Whether conditions are improving overall is genuinely difficult to say, because the answer depends on which layer you measure and when. Seagrass coverage increased dramatically between 2023 and 2025, which is an encouraging sign. But the worst bottom-water oxygen conditions recorded in the study period occurred in June 2025, not during the summer heat of 2024 — driven by the macroalga collapse that had nothing to do with external nutrient inputs. The feedback loop connecting hypoxia to internal nutrient release from sediments means that even genuine improvements in water quality metrics may not translate immediately to reduced hypoxia at the bottom.

What is the monitoring gap the study identified, and does it mean official data is wrong?

The monitoring gap is real but reflects design limitations, not error. The single SJRWMD mid-depth sensor inside the study area was operating correctly — it measured what it was placed there to measure. The problem is that mid-depth monitoring in a shallow, well-mixed system like the BRL does not capture what is happening at the sediment-water interface, where oxygen-consuming biological and chemical processes are most intense, and where fish communities are most exposed. On average, the bottom water was 0.7 mg/L worse than mid-depth readings — with October 2024 showing a gap of 2.1 mg/L. Daytime discrete sampling adds to the bias, because photosynthesis is actively producing oxygen during sampling hours. The net effect is that official water quality data for shallow lagoon systems may systematically overstate oxygen conditions at the bottom.

What can be done differently as a result of this study?

The study provides both a methodological template and a spatial targeting tool. On methodology, it demonstrates that continuous bottom-water sensors on a hexagonal grid produce a fundamentally different and more accurate picture of hypoxia in shallow systems than standard practice — a finding relevant to any comparable lagoon monitoring program. On restoration, the hotspot maps of cumulative hypoxia, muck proximity effects, and bloom corridor identification give managers specific geographic targets for dredging, aeration, and habitat restoration investment. The broader implication for Florida is that restoration success metrics — including the milestones embedded in the Central IRL Basin Management Action Plan — should be evaluated against bottom-water dissolved oxygen data, not only mid-depth averages.

What happens to the nutrient cycle when the sediments go hypoxic?

When bottom water drops below the hypoxia threshold, the chemistry at the sediment-water interface changes in two ways that worsen the eutrophication problem. First, anaerobic conditions impair nitrification — the process that is the essential first step in converting nitrogen into harmless nitrogen gas and removing it from the system. Second, iron compounds in the sediment chemically reduce under anaerobic conditions and release the phosphorus they had previously bound. This means the sediment floor shifts from acting as a nutrient sink — absorbing and removing nutrients — to acting as a nutrient source, releasing stored phosphorus and nitrogen back into the water column where they can fuel the next algal bloom. This feedback loop is why the study’s authors warn that cutting external nutrient inputs alone may not be sufficient: the internal nutrient loading from already-hypoxic sediments may sustain eutrophication independently, as has been documented in the Baltic Sea.