Duckweed Starch Content: Why It Ranges From 3 to 75 Percent
Duckweed starch content runs from 3 to 75 percent of dry weight. Starvation, not species, explains most of that range. What the papers actually measured.
Duckweed starch content is quoted in almost every paper on the plant as a single range: 3 to 75 percent of dry weight, depending on species and growing conditions. The range is correct. It is also close to useless on its own, because a 25 fold spread is not a property of a plant. It is a property of what was done to the plant before it was weighed.
This article walks through what the papers behind that range actually measured, which levers move the number, what the number costs, and how far any of it has been taken outdoors. The short version: starch in duckweed is a stress response, it is paid for in protein, it differs between clones of the same species, and nearly every figure in the literature comes from a flask run of one to two weeks.
The 3 to 75 percent range is a range of conditions, not of plants
Start with what a duckweed frond holds when nothing is done to it. In the broadest screening study published, 21 clones covering 11 species and all five genera were grown in complete nutrient medium, and the initial starch content was always between 4 and 10 percent of dry weight. That is the plant growing normally: fast, nitrogen rich, storing little.
A meta-view of published work by genus shows the same thing from the other side. Across the studies it evaluated, Spirodela ranged from 0.7 to 49 percent starch, Landoltia 1 to 17, Lemna 0.2 to 10, Wolffiella 6 to 11 and Wolffia 5 to 14 percent of biomass. Read those as what you get when you sample ponds and cultures as they are, with fresh biomass that is 92 to 94 percent water and fronds built for growth rather than storage.
The high end of the headline range comes from a different kind of experiment. It is induced. Every figure above 40 percent in the literature follows a deliberate intervention that stops the plant growing and leaves photosynthesis running. The low end is the plant. The high end is the protocol.
One mechanism behind every lever
The clearest model of what is happening comes from a 2015 review of duckweed for biofuel. It describes starch content as the balance of two processes: generation by photosynthesis and consumption by metabolism. Pushing the first is expensive with current technology, so almost all research effort has gone into suppressing the second. Every lever below works by cutting consumption, slowing growth, or both.
Starvation: the lever with the most data
Remove nitrogen or phosphate and the plant stops dividing but keeps fixing carbon. The carbon has nowhere to go but starch.
The best time series is for Landoltia punctata moved into nutrient-free water. Starch stood at 3.0 percent of dry weight at the start, tripled within 5 hours, reached 18.3 percent at 24 hours and 45.4 percent at 168 hours, with the activity of ADP-glucose pyrophosphorylase, the committed enzyme of starch synthesis, rising in step. That is one week, in a flask, under lights.
The 21 clone screen confirms the direction and sets the ceiling for the method. Two weeks on medium lacking phosphate or nitrogen produced the highest values, between 40 and 50 percent starch on a dry weight basis, while sulphate deficiency had only minor effects, and removing any of the other 14 ions in the medium did nothing. Starvation is a nitrogen and phosphorus story, which matters for anyone whose duckweed sits on a water that is rich in both.
CO2, photoperiod, salt and metals
A review in GCB Bioenergy collects the other levers. Growing duckweed in a closed container at 100,000 ppm CO2 for 7 days raised starch from 9.6 to 24.7 percent. Continuous light, a 24:0 photoperiod, combined with nitrogen starvation reached 60.03 percent starch and a starch yield of 76.45 grams per square metre. And the 2026 review in Sustainable Food Technology adds that salinity, phosphate or nitrate lack and the presence of heavy metals raise starch in some species, and that swine medium lowers protein while raising starch.
Those read like separate tricks. They are one trick. CO2 and continuous light push generation. Salt, metals and starvation stop growth and cut consumption. The plant is not making a choice. It is accumulating what it cannot spend.
Starch is paid for in protein
The Landoltia series measured protein alongside starch, and the two lines cross. Over the same 168 hours in which starch rose to 45.4 percent, protein fell from 29.6 to 11.3 percent of dry weight. The absolute protein in each flask stayed at 0.015 grams. The plant did not lose protein. It stopped making any, and diluted what it had with starch.
That is the fact to hold onto when reading supplier sheets or pitch decks. A duckweed with 45 percent starch is not a duckweed with 30 percent protein that also happens to store energy. It is a plant that has been starved for a week, and its protein number is now roughly a third of what the same clone shows under growth. The literature on what crude protein content actually measures and on which levers move duckweed protein content describes the same system from the other end: the conditions that give high protein are nutrient-rich water and fast growth, which are exactly the conditions that keep starch at 4 to 10 percent.
So a cultivation system is set up for one target. Protein or starch. A system engineered for a chosen goal, whether that is nutrient and contaminant removal, protein, or starch, before it is built is the design logic SERAPH describes on its own site. Starch is the case where the trade-off is sharpest, because the lever that produces it is the removal of the thing protein is made from.
Clones of one species differ as much as species do
The number most often attached to a species is a number for one clone. The screening study makes this uncomfortable in detail.
Two clones of Lemna minor were tested. Clone 9441 reached 46 plus or minus 2 percent starch under phosphate deficiency and 36 plus or minus 1 percent under nitrogen deficiency. Three clones of Spirodela intermedia were tested. Clone 7797 reached about 40 percent under either deficiency. Clone 9394 stayed under 25 percent even under nitrogen lack, and a third clone, 7125, reached 30 percent. Landoltia punctata clones landed between 30 and 40 percent, but which nutrient triggered the maximum varied from clone to clone.
The species league table in the recent review reads: Landoltia punctata 72.2 percent, Spirodela polyrhiza and Lemna minor 45.68 to 57.23 percent, Lemna aequinoctialis 26 percent and Wolffia 24 percent, with an amylose to amylopectin ratio around 18 to 82 in Landoltia. Those are real measurements. They are also the best clone each study happened to hold, under the protocol that study happened to run. A different clone of the same species, under the same protocol, could land 15 points lower.
The practical consequence is that choosing a species for starch is a clone choice, and it has to be made by measurement on the clone you will actually grow. Species identification is the first step, and telling duckweed species apart is harder than it looks, but it is only the first step.
Where the starch is meant to go: bioethanol, and the firewall
Nearly all of this research exists because duckweed starch ferments easily. Low lignin means the biomass can be enzymatically converted to ethanol without thermophysical pretreatment, and the same review quotes an ethanol yield of 6.42 thousand litres per hectare for Spirodela polyrhiza against 4.31 thousand for corn.
The study behind that comparison is worth describing precisely, because it is also where the two halves of the duckweed story meet. Spirodela polyrhiza was first grown on diluted pig effluent, then transferred to clean water for 10 days. Starch content rose by 64.9 percent, biomass production was reported at 9,420 kilograms per hectare per year, and after enzymatic hydrolysis and yeast fermentation the theoretical ethanol yield came to 6,420 litres per hectare per year, about 50 percent above corn ethanol. An earlier study by the same group produced a hydrolysate containing 50.9 percent reducing sugar on a dry weight basis and an ethanol yield of 25.8 percent of biomass.
Two things follow. First, the effluent phase is doing the nutrient removal and the clean-water phase is doing the starch induction. This is the starvation lever again, applied as a process step after a treatment step. The nitrogen the plant stripped from the effluent is what it stops receiving when it is moved. For an operator the attraction is obvious: the polishing stage and the feedstock stage are the same pond, run in sequence, and the biomass that comes out of a system built for duckweed wastewater treatment has a fermentation route rather than a disposal cost, whether that is ethanol or duckweed biogas.
Second, that biomass grew on pig effluent. Under SERAPH’s published policy, biomass grown on wastewater, manure or contaminated water is remediation-grade and is never sold or blended as food or feed, and the two streams are separated from the first sensor reading. A high starch number does not change that. Neither does the observation, repeated in reviews, that the protein-rich residue after fermentation could be fed to livestock. If the input was effluent, the output is a fuel feedstock and a residue, and nothing on the production-grade side of the firewall. The route to bioethanol is covered on its own in duckweed starch and the route to bioethanol; the firewall is the part of it that does not move.
What was measured and what was extrapolated
Every number in this article shares one feature. It was measured in a flask, a tray or a small tank, over a period of 7 to 16 days, with a known clone, under a lamp or a greenhouse roof. The per-hectare and per-year figures are those flask results multiplied up. They are theoretical yields, and the papers say so.
Nobody has published a season of outdoor starch data on a real water with real weather, where nitrogen arrives when it rains and leaves when it does not, where temperature decides the consumption half of the balance, and where the starvation step cannot be scheduled by moving a flask. Turions, the dormant buds some species form before winter, are the one outdoor starch-rich state the plant produces on its own, and turions are a separate subject because they are a survival organ, not a harvest.
SERAPH states its own maturity as TRL 3, proof of concept, with outdoor field validation still ahead. The knowledge base the system reasons over, 256 indexed duckweed papers across 31 research areas, includes the starch literature summarised here, and it is why the brand’s line is that AI proposes and the bench validates. The starch range is a good example of why both halves are needed. The literature can propose that a two-week starvation step on a given clone will land between 40 and 50 percent. Only the pond can say what it does in October on the Canary Islands.
FAQ
Which duckweed species has the highest starch content?
Landoltia punctata holds the highest reported figure, 72.2 percent of dry weight. But the 21 clone screen shows clones of one species ranging from under 25 to about 40 percent under the same protocol, so the clone you hold matters as much as the species name, and the only reliable number is one measured on it.
How long does duckweed take to accumulate starch?
Hours to weeks. Landoltia punctata in nutrient-free water tripled its starch from 3.0 percent within 5 hours, reached 18.3 percent at 24 hours and 45.4 percent at 168 hours. Two-week protocols on phosphate-free or nitrogen-free medium are the standard for reaching 40 to 50 percent.
Can duckweed starch be converted to ethanol without pretreatment?
Reviews report enzymatic hydrolysis followed by yeast fermentation with no thermophysical pretreatment, because duckweed has little lignin. Reported outcomes include a hydrolysate with 50.9 percent reducing sugar and an ethanol yield of 25.8 percent of dry biomass.
Is starch-rich duckweed grown on wastewater usable as animal feed?
Not under SERAPH’s policy. Biomass grown on wastewater, manure or contaminated water is remediation-grade and is never food or feed, whatever its starch or protein content. It can be a fermentation feedstock. The production-grade and remediation-grade streams are kept separate from the first sensor reading.