Evidence and experiment registry

Ocean iron fertilisation: Results, settings and evidence limits.

Field experiments, observations, models and assessments answer different questions. Source, setting, supported conclusion and inference boundary remain distinct.

Evidence records

Fourteen evidence records.

Field evidence

IronEx II: traced equatorial patch and biological response

Primary source ↗

Equatorial PacificField experiment

Source: Coale et al. (1996), Nature
Setting / window: Equatorial Pacific; repeated iron additions; tracked for 19 days.
What it supports: Surface nutrient and CO₂ drawdown, a phytoplankton bloom and rapid movement of a traced patch in the study setting.
What it does not support: Does not establish durable export, additional net atmospheric removal, ecosystem outcomes beyond the window or regional scalability.
Scientific implication: A moving treatment, matched reference and downstream corridor resolve physical attribution.

Field evidence

IronEx II: community shifted toward large pennate diatoms

Primary source ↗

Equatorial PacificField experiment

Source: Cavender-Bares et al. (1999), Limnology and Oceanography
Setting / window: IronEx II community observations in the equatorial Pacific.
What it supports: A shift in phytoplankton size structure and increased large pennate diatoms during the experiment.
What it does not support: Community change does not determine particle export, food-web outcome, durability or atmospheric attribution.
Scientific implication: Taxonomy and size structure gain interpretation when paired with grazing, particles and carbon pools.

Model result

ENSO alters equatorial iron limitation and the counterfactual

Primary source ↗

Equatorial PacificObservation + model benchmarking

Source: Browning et al. (2023), Nature
Setting / window: Equatorial Pacific physical and biogeochemical variability.
What it supports: Natural physical forcing changes iron limitation and biological conditions relevant to a counterfactual.
What it does not support: Does not predict the outcome or legality of a proposed intervention.
Scientific implication: Baseline and reference observations resolve physical variability on the experiment’s time and space scales.

Field evidence

SEEDS-I: strong western subarctic Pacific diatom response

Primary source ↗

Subarctic North PacificField experiment

Source: Tsuda et al. (2003), Science
Setting / window: Western subarctic Pacific mesoscale iron-enrichment experiment.
What it supports: A strong phytoplankton response dominated by diatoms in this setting.
What it does not support: One experiment does not establish a regional average, durable removal or ecological acceptability.
Scientific implication: Bloom magnitude is interpreted with season, nutrients, community and export observations.

Field evidence

SERIES: export increased near the mixed-layer base

Primary source ↗

Subarctic North PacificField experiment

Source: Boyd et al. (2004), Nature
Setting / window: Northeast subarctic Pacific; seasonal iron-enrichment study.
What it supports: Enhanced particle export near the base of the mixed layer during the study.
What it does not support: Only a small share of newly fixed carbon passed below the permanent thermocline in the reported window; export is not net atmospheric removal.
Scientific implication: Passage below depth horizons and subsequent remineralisation distinguish deep transfer from shallow flux.

Field evidence

SEEDS-II: a smaller response shaped by grazing

Primary source ↗

Subarctic North PacificField experiment

Source: Tsuda et al. (2007), Journal of Oceanography
Setting / window: Western subarctic Pacific repeat experiment under different initial conditions.
What it supports: A weaker response in which small cells and grazing influenced the observed outcome.
What it does not support: Does not provide a universal grazer response or permanence result.
Scientific implication: Replicates and food-web measurements are needed to characterise variability.

Field evidence

Pseudo-nitzschia and domoic acid are credible monitoring targets

Primary source ↗

Subarctic North PacificIncubation + field observation

Source: Trick et al. (2010), PNAS
Setting / window: Iron-enrichment incubations and observations at Ocean Station PAPA.
What it supports: Iron addition favoured toxin-producing Pseudo-nitzschia and increased cellular domoic acid in this setting.
What it does not support: Does not demonstrate the same response in every region or ecosystem damage from a Southern Ocean programme.
Scientific implication: Species composition and toxin monitoring need proposal-specific toxin thresholds and replicated ecological observations.

Field evidence

SOIREE: surface bloom and carbonate-system response

Primary source ↗

Southern OceanField experiment

Source: Boyd et al. (2000), Nature
Setting / window: Southern Ocean Iron Release Experiment; short field observation window.
What it supports: A marked surface biological response after iron enrichment in the study setting.
What it does not support: Surface response does not establish deep export or atmospheric durability.
Scientific implication: Particle fate, air–sea exchange and water-mass evolution extend beyond the bloom window.

Field evidence

SOIREE: negligible particle export in the short observation window

Primary source ↗

Southern OceanField experiment

Source: Charette & Buesseler (2000), Geochemistry Geophysics Geosystems
Setting / window: Particle export measured during and shortly after SOIREE.
What it supports: Negligible export was detected over the reported short window.
What it does not support: The limited observation window cannot establish later export, lack of export, durability or an atmospheric result.
Scientific implication: Observation duration determines whether delayed export and seasonal mixing are observed.

Field evidence

EIFEX: sinking diatom aggregates in one eddy

Primary source ↗

Southern OceanField experiment

Source: Smetacek et al. (2012), Nature
Setting / window: Five-week EIFEX study inside a coherent Antarctic Circumpolar Current eddy.
What it supports: Multiple measurements indicated substantial sinking biomass, with at least half inferred below 1,000 m in this diatom-dominated setting.
What it does not support: One eddy and one bloom do not establish general export efficiency, net atmospheric removal or later return pathways.
Scientific implication: Deep aggregate evidence advances a carbon-fate question; it does not close atmospheric accounting.

Field evidence

LOHAFEX: biological response without enhanced export

Primary source ↗

Southern OceanField experiment

Source: Martin et al. (2013), Global Biogeochemical Cycles
Setting / window: 39-day, approximately 300 km² Southern Ocean study with low silicate and strong grazing.
What it supports: Productivity increased while fertilisation-induced downward particle flux did not increase during the study.
What it does not support: Does not establish a universal no-export response; it shows dependence on community and setting.
Scientific implication: Silicate, cell size, grazing and particle flux jointly constrain interpretation.

Assessment / guidance

OIF carbon fate, MRV and ecological effects require an observatory

Primary source ↗

All regionsScientific guidance

Source: Buesseler et al. (2024), Frontiers in Climate
Setting / window: Peer-reviewed research guidance spanning carbon, ecology and observation design.
What it supports: A multi-platform observatory from baseline through winter mixing, including air–sea CO₂, particle fate, deep remineralisation and downstream effects.
What it does not support: A recommended measurement architecture is not a validated crediting method or project approval.
Scientific implication: Protocol, QA/QC, uncertainty and independent data access support testable field conclusions.

Assessment / guidance

NASEM assessment separates biological response from durable removal

Primary source ↗

All regionsExpert assessment

Source: National Academies (2022), Ocean-based CDR
Setting / window: Independent assessment of efficacy, durability, MRV, impacts and research needs.
What it supports: The biological mechanism has a substantial knowledge base while carbon fate, efficiency, durability, MRV and unintended consequences need further research.
What it does not support: The assessment does not evaluate a Liquid Trees proposal or grant permission.
Scientific implication: A removal claim needs full-system carbon accounting and uncertainty, not chlorophyll or export alone.

Model result

Scenario models test regional trade-offs and return pathways

Primary source ↗

All regionsModel study

Source: Yu et al. (2026), Nature
Setting / window: Process-rich 60-year fertilisation scenarios across ten ocean biomes.
What it supports: Regional differences, non-local outcomes and long return pathways emerge in the model scenarios.
What it does not support: Modelled scenarios are not field validation, site selection, ecological acceptance or observed removal.
Scientific implication: Model output informs measurement design and falsification tests; it does not label a region.