Try this tonight. Open your cupboard, take out your jar of honey and look at the label. If the honey comes from Moldova or Bulgaria, it contains up to 5,371 nanograms of neonicotinoids per kilogram. If it is French organic honey, it contains seven. These figures, published in July 2026 by Agir pour l’environnement and the French national beekeeping union (UNAF), are not an opinion. They measure, in a glass jar, the gap between a territory that has banned these molecules and a territory that still uses them.
What remains is to understand what these substances really are, and why they hold a place of their own in the history of pesticides.
The question has already received a collective answer. The most complete assessment available to date, the Worldwide Integrated Assessment, is based on the reading of more than eleven hundred peer-reviewed studies. Its overall conclusion fits in a few words: the scale of use, especially as systematic seed coating, combined with the persistence of the molecules in soils and water, continuously exposes environments and species that were never targeted. What follows sets out the essentials of this body of work, in the order in which things happen in the field.
A molecule copied from tobacco
Nineteenth-century gardeners sprayed tobacco juice on their roses. Nicotine is a natural insecticide, produced by the plant to defend itself against those that graze on it. For industry it had two drawbacks: it broke down quickly and it was dangerous to handle.
In the 1980s, chemistry rebuilt this molecule. It kept the active principle, the part that acts on the insect's nervous system, and gave it a stable architecture, soluble in water, able to circulate through plant tissue. Imidacloprid came on the market in 1991. It was followed by thiamethoxam, clothianidin, thiacloprid and acetamiprid. Flupyradifurone, more recent, belongs to another chemical family but hits exactly the same target. It is this functional kinship that matters, far more than regulatory labels.
How they act
In an insect's nervous system, information passes from one neuron to the next through a chemical messenger, acetylcholine. This messenger binds to a receptor, passes on its signal, then an enzyme destroys it within a few milliseconds. The circuit resets, ready for the next message. Everything depends on this reset.
Neonicotinoids bind to the same receptor. But the enzyme cannot destroy them. The signal stays locked open. The neuron fires endlessly, the insect trembles, becomes disoriented, is paralysed and dies. It is not a poison that burns or suffocates: it is a poison that prevents the silence between two messages.
The insect receptor binds these molecules far more strongly than that of vertebrates. This is the selectivity argument put forward by manufacturers, and it is partly true. It has a counterpart that is rarely stated: the lethal dose for a bee is counted in nanograms, in billionths of a gram. Weight for weight, these substances are several thousand times more toxic to a pollinator than DDT was. An invisible quantity, undetectable in the field, is enough.
The decisive difference: a molecule that enters the plant
This is where neonicotinoids leave the territory of conventional insecticides.
A contact insecticide is deposited on a surface. It acts for a few days, rain washes it off, sunlight breaks it down, and the insect can protect itself by hiding under a leaf. Neonicotinoids, on the other hand, are systemic. Coated around the seed and taken up by the root, they rise with the sap and spread through the whole plant. The stem, the leaf, the flower, the pollen, the nectar, the guttation droplets the plant exudes in the early morning.
The whole plant becomes toxic, for several weeks, without any visible treatment having been applied. The bee foraging on an oilseed rape flower has not met a single sprayer. Yet it receives its dose in the nectar. The hoverfly, the ladybird and the ground beetle crossing the field receive it too. There is no refuge inside a plant poisoned from within.
What never enters the plant
Between 80 and 95 per cent of the active ingredient coating a treated seed is never taken up by the plant. It stays in the soil.
These molecules were designed to be water-soluble, since they had to circulate in the sap. This agronomic quality becomes an environmental defect as soon as they leave the seed. Weakly held by clays, they migrate with drainage water towards ditches, rivers and aquifers, where they decimate the aquatic insects that form the base of the food chain in watercourses. Their breakdown, meanwhile, drags on. Depending on the molecule, the type of soil and the climate, the half-life ranges from a few weeks to more than a thousand days, which means that after nearly three years half the dose is still there. Flupyradifurone takes more than two hundred days to break down in water. From one season to the next, what has not disappeared adds to what has just been applied.
Nor does contamination stop at the edge of the field. Dust abraded from coated seeds spreads at sowing, drainage water carries the rest, and these molecules turn up in grass margins, hedges and the wild flowers at field edges, that is, precisely in the refuges where beneficial insects and pollinators are supposed to shelter. The treatment targets one crop; it soaks a whole landscape.
Lake Shinji, in Japan, provided the clearest demonstration. From 1993, the surrounding rice paddies were treated with neonicotinoids. Within a few years, the lake's zooplankton had fallen by 83 per cent, and fish catches followed, with collapses of up to 90 per cent. The study published by Yamamuro and colleagues in 2019 in the journal Science established the direct correlation. It was not a theoretical model. It was an entire ecosystem, observed in real time.
Doses that do not kill and yet are enough
The regulatory assessment of a pesticide is based on direct mortality. Individuals are exposed and the dead are counted. This method misses the essential point.
A worker bee exposed to a sublethal dose of acetamiprid does not die in the hive. She loses her bearings, cannot find her way home, and dies alone in a field. No corpses at the hive entrance, no alarm, but a colony that wears itself out replacing foragers that never return. These doses also impair olfactory memory, learning ability and immune defences, which opens the door to the parasites and viruses the colony had kept in check until then.
The same silent mechanism affects beneficial insects. An adult lacewing no longer lays its eggs on infested leaves. A ground beetle crossing a treated field absorbs the molecules through its cuticle and sees its reproductive capacity reduced before its larvae have even hatched. It is not only ecological niches that are erased, it is breeding individuals, carrying the next generation, eliminated before they have passed anything on.
Below the surface, the soil
The earthworm ingests treated soil, mixes it, digests it. Populations decline, even in soils that have been restored for years. Deeper still, the mycorrhizal network, that age-old chemical dialogue between the root and the symbiotic fungi that give it access to nutrients and water, is disrupted even at the lowest concentrations tested.
There is a contradiction here that needs to be named. This living network is the real engine of carbon sequestration in farm soils, the very sequestration that it is now proposed to pay for. Destroying the soil microbiome while promising to pay farmers to store carbon amounts to sawing off the branch one claims to be sitting on.
At the scale of the landscape
In 2017, Hallmann and colleagues established, in protected areas in Germany, the disappearance of 75 to 80 per cent of flying insect biomass in twenty-seven years. The figure is all the more striking because it comes from nature reserves that were simply next to treated fields. In forty years, 60 per cent of farmland birds have disappeared. Not because they are hunted, but because they have nothing left to eat.
For birds, starvation is not the only route. A coated seed is still a seed, and the seed-eaters gleaning behind the drill swallow them. A few treated seeds are enough to make a lethal dose for a small songbird. Two mechanisms are therefore stacked: the direct poisoning of those that eat the seed, and the slow starvation of those that eat the insect.
This disappearance is not merely an aesthetic impoverishment of the landscape. The skylark, the corn bunting and the grey partridge regulated populations of plant-eating insects. Their collapse dismantles the main natural mechanism of pest control. And among the insects that come back after years of reduced chemical pressure, those that survived carry an acquired resistance. Resistant pests in an environment stripped of predators: that is the mechanical outcome of the sequence. More powerful molecules will then be demanded.
The cocktail effect, the blind spot of assessment
One last difficulty must be raised, because it weakens the whole regulatory edifice.
A molecule is approved on its own. It is given to individuals under controlled conditions, a no-observed-effect dose is established, and a threshold is derived from it. But no living being in a farmed landscape ever encounters a single molecule. In one season, the ground beetle, the earthworm and the bee pass through dozens of substances, insecticides, fungicides, herbicides, adjuvants, each compliant with its own authorisation.
Yet these substances interact. Some fungicides block the liver enzymes that allow the bee to detoxify insecticides, which multiplies the real toxicity of the ingested neonicotinoid by several dozen times. The mixture then proves far more toxic than the sum of its components. This phenomenon is documented, but it is not built into approval tests, which continue to reason molecule by molecule.
In other words, the thresholds we consider safe were set in a world that does not exist.
Why these molecules are dangerous
None of their properties is new taken on its own. It is their combination that makes this chemical family unique.
They are toxic at infinitesimal doses, invisible in the field. They are systemic, and therefore present in the pollen and nectar the plant offers its pollinators. They are soluble, and therefore mobile towards surface water and groundwater. They are persistent, and therefore cumulative from one season to the next. And they are applied as a preventive seed coating, that is, regardless of any actual presence of a pest, over millions of hectares at a time.
An insecticide that is toxic but short-lived remains a local accident. An insecticide that is persistent but not very mobile remains a field-level problem. A molecule that combines all five characteristics becomes a constant background chemical pressure across an entire territory. And the cocktail effect is a reminder that this pressure never acts alone.
Moving beyond neonicotinoids
One objection comes up in every debate, and it is legitimate: what do we replace the molecule with? The honest answer is that we replace it with nothing, if by that we mean a one-for-one substitution. No available product reproduces the effect of a systemic coating applied preventively over a whole crop area. What does exist is combined control, in which several levers that are modest on their own together produce stable regulation. It is no longer a matter of treating a field, but of designing it as an ecosystem able to regulate itself.
The first lever plays on the pest's senses. An aphid does not find its host plant by chance; it locates it through visual signals and volatile molecules. Sowing a companion plant between the rows, oats or barley, scrambles these cues and sharply reduces the landing rate on the crop. The so-called push-pull strategy takes the logic further by combining, within the field, repellent species that drive the pest away and, at the edge, attractive species that draw it in and hold it outside the production area. Ground mulch works on the same principle, breaking the visual contrast that some flying insects use to land. Ibn al-Awwam planted garlic and rosemary between his fruit trees in the twelfth century without knowing the word kairomone. He had observed that insects avoided these areas, and that was enough for him.
The second lever works below the surface, and it is both the slowest to establish and the most lasting. A plant fed by a living soil is not the same plant. Restoring symbioses, by adding ramial chipped wood and rebuilding the mycorrhizal network, improves the root's access to nutrients and water. Adding activated biochar holds water and shelters a competitive rhizosphere microbiome, where pathogenic micro-organisms struggle to find a free niche. The result can be read in the very structure of the plant: thicker cell walls, biochemical defences that can be mobilised faster, tissues less attractive to sap-sucking insects. A living soil does not kill the pest; it produces a plant the pest exploits less well.
The third lever brings back the predators. Creating permanent habitats for ladybirds, lacewings, ground beetles and parasitoid wasps, in the form of flower strips, hedges or untreated margins, amounts to reinstating the regulation that chemistry had removed. Entomopathogenic fungi complete the arsenal against certain pests. To this is added a form of prevention that is often neglected because it cannot be sold: strict destruction of volunteer plants and management of field edges, which deprive aphids of the reservoirs where they keep viruses from one season to the next. Much of the damage blamed on the pest is in fact down to what it was left to overwinter in.
The fourth lever is genetic, and it is the one that was missing when the ban came in. Varietal breeding is speeding up, particularly on sugar beet, the most exposed crop. The identification of hardiness genes from exotic varieties makes it possible to develop hybrids that maintain their yield even when infected by a virus. The plant does not become resistant to the aphid; it becomes tolerant to the virus the aphid carries, which moves the problem to where it can be solved.
Ten years without pesticides: what RésOpest shows
That leaves the question all this raises. Do these combined levers last the distance on real areas, under ordinary conditions, over years? A French trial finally makes it possible to answer with something other than conviction.
The RésOpest network, set up in 2012 by Inrae with the Purpan engineering school and Cirad, ran nine cropping systems for ten years without any pesticide at all, on nine sites across mainland France, from the oceanic climate of Brittany to the Mediterranean climate of Mauguio. Five arable systems and four mixed crop-livestock systems. The results were published in February 2026 in the journal Plant Disease.
It is worth making clear what the trial was not, because that is what gives it its value. It was not organic farming. Mineral fertilisers were still allowed, and so was tillage, including ploughing. The only constraint imposed was the complete abandonment of pesticides. A third way, in short, between conventional and organic, which isolates the chemical question from all the others.
Management rested on three pillars, in which the levers described above can be recognised. Prevention first, with management of infected residues, use of healthy seed and cleaning of equipment. Making the most of plant diversity next, with long rotations of five to nine years, varied crop families and sowing dates, intercrops and variety mixtures. Soil health finally, with cover crops and reduced tillage.
The result that directly concerns our subject fits in one sentence from Inrae's press release: damage to crops from diseases and pests did not increase significantly over time. Ten years without a single insecticide, and no pest explosion. That is the exact opposite of the disaster scenario brandished in every debate on derogations.
Honesty requires giving the full picture. Yields were most often below those of chemically protected conventional systems, even though some situations reached equivalent or higher levels. And the real difficulty came not from insects but from weeds, especially docks in temporary grassland, to the point of sometimes requiring a return to ploughing. The hard point of pesticide-free farming is therefore not the insecticide. It is the herbicide.
Economically, the four arable systems whose performance could be costed generated a net margin allowing an income of more than twice the French monthly minimum wage in four cases out of five, and more than three times in a third of cases. The authors' conclusion is clear: these systems are technically and economically feasible, provided there are diversified crop sequences, suitable marketing channels and proper value for the products that come out of them. In other words, the lock is no longer agronomic. It is commercial and political.
That is what Jean-Noël Aubertot, research director at Inrae and initiator of the study, says when he justifies the approach by the need for a public research body to take risks to explore what is possible. What is possible has been explored. The results exist. They now need funding.
It remains to name what these levers really change. They do not aim to eradicate the pest, and that is precisely what makes them hard to adopt. A well-managed field keeps its aphids; it holds them below the damage threshold. The clean slate promised by seed coating was never an agronomic objective, only a commercial promise, and the price of that promise can be read in the preceding pages. These techniques do, however, demand what the molecule made unnecessary: observation, training, technical support in the field, and funding that matches the service rendered. They do not demand a miracle.
What France had shown
Go back to the jar of honey. Seven nanograms against five thousand three hundred and seventy-one.
These two numbers say it all. When chemical pressure eases, the load in the environment collapses, and the dialogue between plant, soil and insect is restored. Slowly, partially, but really. The proof existed, and it was French.
It still exists, and so do the levers for doing without these molecules.
Ibn al-Awwam observed in the twelfth century that the earth always answers those who know how to listen to it. He had understood what we are laboriously rediscovering: soil is not an inert medium but a living partner, and breaking it costs infinitely more than respecting it.
Source of the trial cited: Ortiz-Vallejo D., Cellier V., Deytieux V. et al. (2026), “Pesticide-free agriculture: is a third way possible besides organic and conventional agriculture?”, Plant Disease. Inrae press release of 19 February 2026.