Plants cannot run, so they resist chemically and structurally instead. Some defences are permanent — spines, hairs, tough cuticles, stored toxins. Others are built only after damage, triggered by wounding and by chemicals in the attacker’s saliva, and among them are volatiles that predators of the attacker have learned to follow.
Being rooted in one place makes defence a chemistry problem. A plant under attack cannot escape and cannot fight, so what it can do is be expensive to eat and unpleasant to digest — and it does this in two modes. Constitutive defences are always present: thorns, stinging hairs, silica, thick cuticles, and standing stocks of tannins and alkaloids. Induced defences appear only after something starts eating, which saves the cost of maintaining them when nothing is. Induction is where the topic gets both interesting and badly reported. Damaged plants release volatile compounds, the blend differs depending on what is doing the damage, and predatory and parasitic insects orient towards it and find the herbivore. This is repeatedly described as the plant calling for help. Nothing is called and nothing is asked: injury has a chemical signature, and some insects have evolved to find their prey by smelling it. Both parties do better and neither intends anything, which is how most of what gets described as cooperation in nature actually works.
Developed record · 69% complete · reviewed 2026-08-10
What this page covers
How plants resist being eaten. Covers vascular plants and the insects and mammals that eat them, with most experimental evidence from crop and model species.
Quick facts
Two modes
Constitutive (always on) and induced (built after damage)
Induced signal
Coordinated largely by jasmonate
Why not always on
Defence costs carbon that could make seeds
Not intention
Volatiles are a consequence of injury, not a request
Plants keep some defences on permanently and build others only when bitten
Established
Specialists would state this without hedging. Multiple independent lines of evidence agree.
Plant defences comprise constitutive traits present regardless of attack — spines, trichomes, thick cuticles, stored secondary metabolites such as tannins and alkaloids — and induced responses produced after damage, including defensive proteins, altered metabolite profiles and volatile emission. Induction is understood as a response to the metabolic cost of maintaining defences that are not currently needed.
Who this applies to
vascular plants under insect herbivory
Studied in
Angiospermae
Why we rate it this way, and what the caveats are
EstablishedHigh confidence
Structural defences are directly observable and induced chemical responses are measured routinely; the cost argument is supported by trade-offs seen when defence is experimentally induced.
How far it can be extended
Both constitutive and induced defences are documented across the flowering plants; the specific compounds vary enormously by lineage.
Caveats
Costs of defence are easier to argue than to measure precisely in the field.
Many secondary metabolites have functions besides defence.
Herbivores frequently evolve tolerance, so a defence is effective against some attackers and not others.
Still unanswered
How much of a plant’s carbon budget goes to defence under natural herbivore pressure?
Constitutive and induced categories, defensive chemistry, and signalling.
Almost every plant on Earth is being eaten by something. The pressure is constant, and the responses fall into a small number of categories: make yourself hard to get at, make yourself unpleasant, make yourself hard to digest, or accept the damage and regrow. Most plants do several of these at once.
The reason not everything is defended all the time is cost. A compound synthesised and stored is carbon and nitrogen not spent on growing or seeding, and a plant that over-invests loses to a neighbour that does not. Induction — building the defence only when attacked — is the resolution, and it is why a leaf’s chemistry is different an hour after something starts eating it.
Words used here
Constitutive defence
A defence that is always present, whether or not anything is attacking.
Induced defence
A defence produced in response to damage, saving the cost of maintaining it.
The physical defences are the visible ones and they work on different scales. Thorns and spines deter large mammals. Trichomes — the fine hairs on a leaf — work against insects, either by making the surface hard to walk on, by being sticky, or by breaking to release irritants. Stinging nettle trichomes are miniature glass syringes: the tip snaps off and the rest injects.
Grasses take a different route entirely and load their tissues with silica. It is not toxic; it is abrasive, and it wears down the teeth of anything that grazes persistently. The high-crowned teeth of horses and cattle are the answering move in an argument that has been running for twenty million years.
Latex and resin count as structural defence as much as chemical. A cut that floods with sticky latex glues an insect’s mouthparts shut, which stops it eating regardless of what the latex contains.
Words used here
Trichome
A hair on a plant surface. Some are simply physical obstacles; others are glandular and release irritants.
Plants make an enormous number of compounds with no role in basic metabolism, and defence is where most of them earn their keep. The major classes behave differently and a plant usually employs several.
Terpenoids, including the aromatics of pines, mints and citrus, often deterrent or directly toxic.
Phenolics such as tannins, which bind proteins and make foliage hard to digest rather than poisonous.
Cyanogenic glycosides, stored inert and releasing cyanide only when the tissue is crushed.
Protease inhibitors, which interfere with an insect’s ability to digest protein.
The stored-inert design deserves attention because it appears repeatedly. Cassava, cherry leaves and clover all keep a harmless precursor in one compartment and the enzyme that activates it in another. Chewing breaks both and mixes them. The plant is not poisonous until it is bitten, which means it never has to store the poison itself.
None of this is a barrier to specialists. Given long enough, herbivores evolve tolerance — and then dependence: monarch caterpillars sequester milkweed cardenolides and become toxic themselves, and koalas eat a diet that would kill almost anything else. A defence keeps out generalists and creates a niche for whatever can crack it.
Words used here
Alkaloid
A nitrogen-containing plant compound, often bitter and often active on animal nervous systems. Caffeine and nicotine are alkaloids.
Tannin
A phenolic compound that binds proteins. It makes leaves hard to digest rather than poisonous.
Sequester
To store a plant’s toxin in your own body and use it for your own defence, as monarch caterpillars do.
The part that is real, and the sentence about it that is not.
Chewed plants release chemicals that predators of the chewer can find
Well supported
Good evidence backs this, though some details remain open.
Herbivore damage induces plants to emit blends of volatile organic compounds distinct from those released by mechanical wounding alone, triggered in part by elicitors in herbivore oral secretions and coordinated largely through jasmonate signalling. Predatory and parasitoid insects orient towards these blends, and can discriminate between blends induced by different herbivore species.
Who this applies to
crop and model plant systems where volatile blends and predator responses were both measured
Studied in
Angiospermae, Insecta
You may have heard
“Plants call for help when attacked”
Nothing is being called and nothing is being asked. Damage produces a chemical signature; insects that eat the herbivore have evolved to detect it, because doing so finds them food. Both parties benefit and neither intends anything — which is how most of the interactions described as cooperation in nature actually work.
Why we rate it this way, and what the caveats are
Well supportedModerate confidence
The emission and the orientation behaviour are both directly measured. Confidence is moderate on the fitness benefit to the plant under field conditions, which is much harder to demonstrate than the response.
How far it can be extended
Induced volatile emission is documented across many plant families; predator recruitment has been demonstrated in a smaller set of well-studied systems.
Caveats
Dominated by crop species and laboratory bioassays.
Herbivores can also use these volatiles, so emission is not unambiguously beneficial.
Whether the response raises plant fitness in the field is demonstrated in relatively few systems.
Still unanswered
How much does volatile-mediated predator recruitment actually improve plant fitness in the field?
Induced defence, jasmonate signalling, and volatile-mediated attraction of predators and parasitoids.
A plant can tell the difference between being torn and being eaten. Mechanical damage produces one volatile profile; damage accompanied by chemicals from an insect’s oral secretions produces another, and the second one triggers a much stronger induced response, coordinated largely through the hormone jasmonic acid.
Parasitoid wasps — which lay eggs inside caterpillars — orient towards the blend produced by the caterpillar species they attack. In several well-studied systems they can discriminate between blends induced by different herbivores on the same plant. The interaction is real, repeatable, and used commercially in glasshouse pest control.
What it is not is a message. There is no sender selecting a recipient and no request. Injury produces a chemical signature because injured tissue is chemically different; a wasp that finds caterpillars by that signature eats better than one that does not, so wasps that detect it become common. The plant benefits and the benefit is a consequence rather than a purpose.
The same caution applies to the neighbouring plants that respond to these volatiles by priming their own defences. That looks like warning and it is better read as eavesdropping: a plant that reacts to the smell of its neighbour being eaten is reading a cue about its own immediate future.
How we know
Clipping sagebrush to see if the neighbours notice
Plant-to-plant signalling had been shown in sealed laboratory chambers and dismissed as an artefact. Does it happen outdoors, at natural spacing, in a real season?
Sagebrush growing wild in the Great Basin was experimentally clipped to simulate herbivore damage. Wild tobacco plants growing beside the clipped and unclipped sagebrush were then monitored for the rest of the season, and the herbivore damage they sustained was measured. Crucially, air contact and root contact were manipulated separately, so the route any effect travelled could be identified rather than assumed.
What happened
Tobacco growing next to clipped sagebrush sustained significantly less herbivore damage across the season, and the effect travelled through the air rather than through soil or roots.
What it shows
Volatile signalling between plants is a real field phenomenon and not a chamber artefact. An initially weak study proposing a real thing is a different animal from a wrong one, and this is what rescued the idea.
What it does not show
It does not show that sagebrush is warning tobacco, or benefiting from doing so — these are unrelated species, and the receiver is best described as eavesdropping. Later work suggesting volatiles evolved for signalling within a single plant makes the neighbour effect look like a leak rather than a message.
The controls — what makes this evidence rather than a story
Tobacco beside unclipped sagebrush gave the baseline — the plants differed only in whether their neighbour had been damaged.
Separating airborne from below-ground contact identified the route, which the earlier chamber studies could not do.
Damage was measured over a full season in the field, rather than chemistry measured in a box over days.
Adding the smell of damage to an undamaged plant gets its caterpillars eaten
Well supported
Good evidence backs this, though some details remain open.
Supplying individual herbivore-induced volatile compounds to undamaged wild tobacco plants in a natural population increased predation on herbivore eggs and larvae by a generalist predatory bug and reduced herbivore numbers. One compound additionally reduced oviposition by the herbivorous moth. The effect was produced by the volatiles alone, in the absence of damage.
Who this applies to
one wild tobacco population with one predator and one herbivoreDo not extend this beyond the taxa listed — the popular version over-reaches.
A field manipulation that supplies the proposed cause and measures the proposed effect, rather than measuring emission and inferring a benefit. Confidence is moderate because it is one system.
How far it can be extended
Induced volatiles occur very widely; the demonstration that they reduce herbivory in nature exists for a small number of systems.
Caveats
Synthetic supply is constant where a real damaged plant releases in pulses.
Herbivore numbers were measured rather than seed set, so the fitness benefit is one step further on.
Herbivores can use the same volatiles to locate plants, so emission is not unambiguously good for the plant.
Still unanswered
Is emission selected for its effect on predators, or is it an unavoidable consequence of damage that predators exploit?
Frames the interaction as eavesdropping on a cue rather than communication, and reviews what survives field conditions.
The step that turns this from a plausible story into a measurement is running the experiment backwards: instead of recording what a damaged plant emits and assuming a benefit, supply the compounds to undamaged plants in a wild population and count what happens to the herbivores.
How we know
Making an undamaged plant smell attacked
Damaged plants release volatiles and predators arrive. Does the plant actually benefit, or is that an assumption?
Most work in this area measures what a damaged plant emits and what a predator does in a laboratory arena, then infers a benefit. This runs the experiment the other way. Individual volatile compounds identified from herbivore-damaged wild tobacco were released synthetically beside undamaged plants in a natural population, and predation on herbivore eggs and larvae was counted on treated and untreated plants — the proposed cause supplied deliberately, the proposed effect counted in the field.
What happened
Plants releasing the induced volatiles suffered greater predation of herbivore eggs and larvae by a generalist predatory bug, and carried fewer herbivores overall. One compound also reduced the number of eggs the moth laid on the plant.
What it shows
That herbivore-induced volatiles reduce herbivory in nature, by two routes: bringing in predators and deterring egg-laying. It converts a plausible story into a measurement, and does so without requiring the plant to be signalling to anyone.
What it does not show
Synthetic release is constant where a real plant emits in pulses, so the dose is unnatural. Fewer herbivores is not more seed, and the fitness consequence is a further step. One plant species, one predator and one desert population. And the same volatiles can be used by herbivores to find plants, which this design does not rule out.
The controls — what makes this evidence rather than a story
Untreated plants in the same population, exposed to the same weather, predators and herbivores.
Individual compounds released separately, so an effect can be attributed rather than credited to a blend.
Undamaged plants used throughout, so the volatiles are the only thing distinguishing them — the wounding itself is not a confound.
Predation and oviposition counted directly on the plants rather than predator arrival being scored as success.
It works, in two ways at once: more of the herbivore’s eggs and larvae are eaten, and the moth lays fewer eggs on a plant that smells as though something is already eating it. Neither route requires the plant to be addressing anybody. A caterpillar-hunting bug that can smell caterpillar damage finds more caterpillars, and a moth that avoids plants smelling of predation avoids losing its eggs.
Some plants do not make their own defences at all. They pay for them. Extrafloral nectaries — nectar glands on leaves and stems, nowhere near a flower — supply sugar to ants, and the ants patrol the plant and attack insects on it. It is defence bought with the same currency used to pay pollinators.
A few take the arrangement further and provide housing: swollen hollow thorns in some acacias, with ants living inside them, protein-rich food bodies produced on the leaflets, and ants that attack anything landing on the plant and prune encroaching vegetation. The relationship is close enough that the plants do badly without their ants.
Does volatile-mediated predator recruitment actually improve plant fitness in the field?
Why it matters: The emission and the wasp response are both well measured. Whether the plant ends up with more seed as a result has been shown in far fewer systems than the story is told about.
How large is the cost of defence in a real plant’s carbon budget?
Why it matters: The cost argument underlies the whole constitutive-versus-induced framework, and precise measurements are scarce.
How much do herbivores exploit induced volatiles to find food?
Why it matters: If herbivores use the same cues, induction is not a straightforward benefit, and the balance is not well quantified.