OECD Lemna Test (TG 221): What Regulators Use It For
The OECD Lemna test (TG 221) from the regulator's side: where it is required, how an ErC50 becomes a legal limit, and its limits on real water.
The OECD Lemna test is the one aquatic plant study almost every chemical regulator in the world recognises. It appears in pesticide dossiers in Brussels and Washington, in biocide files, in some REACH registrations and in Canadian effluent rules, and the number it produces can end up as a legal concentration limit in surface water. Yet the pages that rank for the term are the OECD abstract and a handful of contract laboratories selling the study. None of them explains why the test exists, who requires it, or what happens to the result once the laboratory signs the report.
This article does that. We have already written a plain-language guide to the mechanics of the Lemna growth inhibition test: species, media, endpoints, the validity rule. Here we take the regulator’s side. SERAPH builds duckweed cultivation systems for real outdoor water, so we spend a lot of time reading what an ecotoxicology number legally means before we compare it with anything that happens in a pond.
What the OECD Lemna test is, in one paragraph
OECD Test Guideline 221 exposes exponentially growing monocultures of Lemna, usually Lemna gibba or Lemna minor, to at least five concentrations of a substance for seven days and compares vegetative growth with untreated controls. Frond number is counted, at least one other variable (frond area, dry weight or fresh weight) is measured, and the laboratory reports the concentration that inhibits growth rate (ErCx) or yield (EyCx) by a chosen percentage, with a no observed effect concentration if the design allows it. A limit test at a single dose of 100 mg/L is permitted. The guideline was adopted in 2006. That is the whole test. Everything below is about what surrounds it.
Why the OECD wanted a higher plant test at all
Before 2006, aquatic plant hazard was assessed almost entirely with unicellular algae. The report of the international ring test that produced the guideline explains why that worried regulators. Algae lack certain sites of action present in higher plants, so they can be insensitive to compounds that are specifically phytotoxic to vascular plants, which is exactly what a herbicide is designed to be. Algal tests also struggle with coloured and turbid samples and with low dissolved oxygen, which describes a great many real effluents. A floating macrophyte that is easy to handle and sensitive to a wide range of contaminants solved both problems. The initiative came from the Swedish chemicals agency KEMI and was taken forward by WRc in the United Kingdom, and the 1998 OECD review prioritised Lemna because draft methods from ASTM, the US EPA and others already existed.
The idea was older than the OECD process. Environment Canada’s method notes that duckweeds have been used to detect phytotoxicity since the 1930s, and that the US EPA proposed Lemna minor as a representative aquatic macrophyte in 1979. What the OECD added was standardisation at scale: the ring test involved 37 laboratories across Europe, North America and the Far East, and most data sets met the draft criteria. That reproducibility is the asset. A test that 37 laboratories can run the same way is a test that 40 governments can agree to accept.
What the OECD label buys: tested once, accepted everywhere
The reason “OECD” matters in the name is a legal instrument most people outside regulatory affairs have never heard of. On 12 May 1981 the OECD Council adopted its Decision on the Mutual Acceptance of Data, which requires that data generated under OECD Test Guidelines and OECD Principles of Good Laboratory Practice must be accepted by every adhering country for assessment purposes. The OECD summarises it as tested once, accepted for assessment everywhere, with one important caveat: the receiving country must accept the study, but how it interprets the result remains its own prerogative. The same page draws a distinction worth remembering. Test Guidelines, of which there are now more than 160, describe how to run a study. They are not data requirements. Deciding which studies a dossier must contain is a national or EU decision.
The practical effect is large. A Lemna study run to TG 221 under GLP in one country is accepted in over 40 others, and the OECD estimates the system saves governments and industry more than EUR 309 million a year in avoided duplicate testing. Three conditions apply: the study follows an OECD guideline and GLP, the facility has been inspected by a national GLP monitoring programme, and that programme has passed an OECD evaluation. Non-OECD countries can join; Argentina, Brazil, India, Malaysia, Singapore, South Africa and Thailand are full adherents.
So when a contract laboratory sells “OECD 221”, what the customer is buying is portability. One study, one invoice, every market.
Where the test is actually required
EU pesticides: Regulation 283/2013 and the Commission test-method list
For plant protection products the data requirements for active substances sit in Regulation (EU) 283/2013, and the Commission publishes a companion list of the test methods that satisfy each point. The 2013 list names OECD Test Guideline 221 under effects on aquatic macrophytes, next to an ASTM Myriophyllum method and a proposed rooted-plant protocol. The updated 2023 list for the same regulation carries OECD 221 alongside OECD 239 and OECD 238, the two 2014 Myriophyllum spicatum guidelines, which is how a floating plant and a rooted one came to sit side by side in a dossier. Contract laboratories also cite the EU’s own copy of the method, C.26 in the Annex to Regulation (EC) 761/2009, and state the trigger plainly: the test is required for herbicides and plant growth regulators.
EFSA’s 2013 aquatic guidance sets the logic. Lemna is the default macrophyte test species. If Lemna and algae are apparently insensitive to a herbicidal product, which EFSA pegs at an EC50 above 1 mg/L, or if the herbicide mimics a plant growth hormone, a rooted macrophyte is required, preferably Myriophyllum, and Glyceria is advised when terrestrial trials show the product mainly hits monocots. The first tier for a herbicidal mode of action therefore needs at least two algae and one macrophyte, and the macrophyte is nearly always Lemna.
EU biocides and REACH
The biocides framework copies the pesticide trigger. ECHA’s guidance for the Biocidal Products Regulation says a Lemna test to OECD 221 should be performed for herbicides, plant growth regulators and fungicides where there is evidence of herbicidal activity, reporting growth and yield on frond number plus a second variable.
REACH is different, and this trips people up. For industrial chemicals the Annex VII requirement is a growth inhibition study on aquatic plants, preferably algae. Lemna is not a standard REACH data point. ECHA’s guidance does name duckweeds as the most commonly used vascular plants when such a test exists, notes that Lemna designs run up to 14 days, and insists the ECx or NOEC should relate to growth rate. The data then feed three decisions: classification and labelling, the Predicted No Effect Concentration for water, and the toxicity criterion in a PBT assessment. In practice a Lemna study enters a REACH file when a registrant wants a second primary producer to refine the PNEC, or when a substance has herbicidal properties that algae would miss.
United States: FIFRA and TSCA
The US counterpart is OCSPP 850.4400, which the EPA uses to develop data under the Toxic Substances Control Act, the Federal Insecticide, Fungicide and Rodenticide Act and the Federal Food, Drug and Cosmetic Act. Two choices in that guideline reveal what the regulator is actually worried about. For pesticides the EPA prefers testing the typical end-use product rather than the technical-grade active ingredient, because the formulation is what reaches a ditch, and it requires the static-renewal design for pesticides. The results are used to set toxicity levels and, explicitly, to decide whether higher-tier testing is needed. Environment Canada’s survey of the legal landscape adds that FIFRA requires a duckweed growth test, TSCA can require one, and the test is optional for NPDES discharge permits.
Effluents and receiving waters
The ring-test report is candid that the guideline’s main purpose is hazard assessment of chemicals, but several national authorities said from the start that the method would be valuable for effluents, and the report appends guidance for testing effluents, leachates and receiving waters. Canada went furthest. Its Lemna minor method was used in federal and private laboratories to help meet the country’s metal mining effluent regulations, and it is written for single chemicals, industrial and urban effluents, leachates and surface waters alike. This is the version of the test a water operator is most likely to meet.
How a Lemna number becomes a legal concentration
The assessment factor of 10
In the EU pesticide scheme the laboratory’s EC50 does not become the limit. It is divided by a default assessment factor of 10, a value fixed by the Uniform Principles for evaluating active substances, and the result is the Regulatory Acceptable Concentration, defined as the maximum surface water concentration resulting from pesticide application that can be tolerated. The factor is there to cover two extrapolations the test cannot make on its own: from laboratory to field, and from individual organisms to populations. EFSA’s guidance sets the tier 1 factor at 10 for plants against 100 for the acute invertebrate assessment, because the plant tests already cover most of a life cycle.
The predicted exposure concentration is then compared with the RAC. If exposure is below it, the assessment stops. If not, the applicant climbs tiers: more species, a species sensitivity distribution, or a micro- or mesocosm study.
Growth rate or yield: a change with consequences
Until January 2015 the EU took the lowest of the three EC50 flavours a Lemna study produces, growth rate, yield or biomass, and divided that by 10. The 2013 EFSA guidance switched tier 1 to the growth-rate endpoint, the ErC50, because it does not depend on test duration or design. The scientific argument since then is whether that switch quietly reduced protection. A 2023 analysis of EFSA’s own lists of endpoints found the intended protection level reached for 69.2 percent of substances with ErC50 against 87.5 percent before, and recommended an extra factor of about 2.4 on top of the default 10. A Wageningen comparison of 17 herbicidal substances found tier 1 protective in 65 percent of cases with ErC50 versus 76 percent with the yield or biomass endpoint, and concluded the current practice did not need changing. Both camps use the same Lemna data. The disagreement is about the factor, not the plant.
A worked example: metazachlor
Numbers make the tiers concrete. For the herbicide metazachlor, a field-data paper from the same Wageningen group lists the regulatory endpoints side by side: a tier 1 RAC of 0.872 micrograms per litre derived from the Lemna gibba data requirement, a tier 2 RAC of 5.54 from a species sensitivity distribution of 18 species divided by a factor of 3, and a tier 3 RAC of 5 from a mesocosm study. The Lemna number was the strictest by a factor of roughly six. That is the test doing its job as a conservative first screen, and it is why applicants invest in higher tiers.
When Lemna is not enough
Two follow-ups are common. If the standard study shows high risk, a recovery study transfers a dozen fronds into clean water after the seven-day exposure and compares their growth with controls, extending the recovery period until growth resumes. And where the mode of action suggests Lemna is the wrong plant, the rooted Myriophyllum spicatum tests take over. The control growth rates the guidelines demand show how different those organisms are: Lemna must manage at least 0.275 per day over seven days, Myriophyllum only 0.05 per day over 14.
What the regulator’s number does not say about a pond
Every element above is designed for comparability. One clone where possible, one medium, one temperature, seven days, exponential growth, 37 laboratories agreeing on the answer, 40 countries accepting it. The assessment factor exists because the people who wrote the scheme knew the vessel is not the field. They did not pretend otherwise; they put a number on the uncertainty.
That is the right way to read an OECD 221 result when you are not a regulator. It is a hazard figure, engineered to be the same in Las Palmas and in Ohio. It is not a prediction of what a mixed duckweed cover will do on a lagoon in November, how a population recovers after a pulse of effluent, or what removal an operator will measure. Nearly all of the published duckweed evidence, the regulatory test included, is indoor, and that gap is what SERAPH was founded to work on. The company describes itself as TRL 3, proof of concept, with outdoor field validation still ahead, and it designs for real weather and water rather than for the vessel. One more line from the same source applies to anyone reading an effluent toxicity report. Biomass grown on wastewater, manure or contaminated water is remediation biomass. It is never food or feed. Plants that have been through a toxicity test are not either. Our overview of duckweed for wastewater treatment describes how that separation is engineered, and our piece on whether duckweed absorbs heavy metals shows why it has to be.
Where SERAPH sits relative to the test
Plainly: SERAPH is not a GLP laboratory and does not run OECD 221. We read the test the way a systems designer reads a specification. What we bring is a mapped view of the evidence, a knowledge base of 256 indexed duckweed papers across 31 research areas, and a measurement stack built around the endpoint the guideline itself prefers, non-destructive canopy observation. SERAPH Vision reconstructs a cultivation tray in 3D from a single capture, 170,000 surface points drawn from 5.9 million, and distinguishes duckweed species in frame, which matters because duckweed species identification is the first thing a guideline asks a laboratory to report. Growth and stress reading from the image is still in development, and the system makes no claim to regulatory use. AI proposes, the bench validates.
Two sibling pieces complete the set: the mechanics of the Lemna growth inhibition test for readers who want the protocol logic, and a closer look at Lemna toxicity test endpoints and what each one is sensitive to.
FAQ
Is the OECD Lemna test required under REACH?
Not as a standard information requirement. REACH Annex VII asks for a growth inhibition study on aquatic plants and states a preference for algae. Lemna is the vascular plant most often used when a registrant adds a second primary producer, for example to refine a PNEC or because the substance has herbicidal activity that algae would not detect.
What is the difference between OECD 221 and OECD 239?
OECD 221 is the seven-day test with floating Lemna. OECD 239, adopted in 2014, is a 14-day water-sediment test with the rooted plant Myriophyllum spicatum, and OECD 238 is its sediment-free version. EFSA asks for a rooted macrophyte when Lemna and algae are apparently insensitive to a herbicidal product or when the herbicide acts like a plant growth hormone.
What assessment factor is applied to a Lemna ErC50 in EU pesticide risk assessment?
A default factor of 10 at tier 1, set by the Uniform Principles. The ErC50 divided by 10 is the Regulatory Acceptable Concentration against which predicted surface water exposure is compared. Several research groups have argued the factor should be larger now that growth rate rather than the lowest endpoint is used.
Does a Lemna test run in one country count in another?
Yes, provided it meets the Mutual Acceptance of Data conditions: conducted to OECD 221 under OECD GLP, in a facility inspected by a national GLP compliance programme that has itself passed OECD evaluation. The 1981 Council Decision makes acceptance for assessment binding across all adherents, although each authority interprets the result for itself.
Why is the Lemna test sometimes used instead of the algae test?
Two reasons given in the OECD ring-test report. Higher plants carry herbicide target sites that unicellular algae lack, so an algal test can miss a specifically phytotoxic compound. And strongly coloured or turbid samples interfere with algal growth measurement, while a floating plant counted by eye is far less affected, which is why contract laboratories offer Lemna as the alternative for coloured substances.