Pollination is pollen reaching a receptive stigma of the same species. About 88% of flowering plant species depend on animals for it — but the cereals that feed the world are wind-pollinated, and an insect sitting on a flower is not necessarily pollinating it.
Pollination is where NatureHQ’s animal and plant intelligence meet, and it is full of distinctions that popular coverage collapses. Pollination is not fertilisation — it is pollen arriving, and fertilisation may follow hours or months later. Visiting is not pollinating: a honey bee can spend all day on a tomato flower and transfer nothing, because that pollen only comes out under vibration and a honey bee cannot vibrate at the right frequency. And "bees" is doing a great deal of unearned work, since the pollinators concerned are tens of thousands of species of bee, fly, beetle, moth, butterfly, bird and bat, of which the honey bee is one managed species that is sometimes the wrong tool. Getting these apart is not pedantry: it changes what pollinator decline means, which crops are actually at risk, and whether putting a hive in a field helps.
Developed record · 80% complete · reviewed 2026-08-10
What this page covers
A relationship rather than an organism. Covers flowering plants and the animals, wind and water that move their pollen.
About 88% of flowering plant species need animal pollinators — and wheat, rice and maize do not
Well supported
Good evidence backs this, though some details remain open.
Community-level surveys across latitudes indicate that approximately 87.5% of flowering plant species are animal-pollinated, rising to about 94% in tropical communities and falling to around 78% in temperate ones. The wind-pollinated remainder includes the grasses, and therefore the cereals that supply most human calories.
Who this applies to
flowering plant species globally
Studied in
Angiospermae
You may have heard
“Without bees we would starve”
Two errors stacked. The cereals supplying most human calories are wind-pollinated and would be unaffected; what animal pollination supplies is much of the variety, nutrition and interest in a diet, which is a serious loss and not starvation. And "bees" undersells it — the pollinators concerned are tens of thousands of species of bee, fly, beetle, moth, butterfly, bird and bat, and the honey bee is one managed species among them.
Why we rate it this way, and what the caveats are
Well supportedHigh confidence
A systematic estimate from community-level data with an explicit method and stated uncertainty, and the figure most cited in the pollination literature since.
How far it can be extended
Estimated from community surveys spanning tropical and temperate regions, with a consistent latitudinal gradient.
Caveats
A count of species, not of abundance, biomass or crop yield, which behave differently.
"Animal-pollinated" covers everything from obligate dependence to marginal benefit.
Community surveys are unevenly distributed geographically.
Still unanswered
How much does the species-level figure translate into yield dependence for real agricultural systems?
Crop-level evidence that wild pollinators improve fruit set independently of honey bee visits.
Pollen must reach a receptive stigma of the same species. Anything that moves it counts — an insect, a bird, a bat, the wind, occasionally water — and from the plant’s point of view the vector is interchangeable so long as the pollen arrives.
The distinction from fertilisation matters practically. Pollination is arrival; fertilisation is what happens after a pollen tube grows down the style to an ovule. A flower can be pollinated and still set no seed, which is why "pollinator visits" and "yield" are different measurements.
Wind: cheap per grain, enormously wasteful, and how most cereals and many trees reproduce.
Animals: expensive in nectar and pollen, far more precise, and the majority condition among species.
Water: rare, and confined to a small number of aquatic plants.
Words used here
Stigma
The receptive surface where pollen must land for pollination to occur.
Vector
Whatever moves the pollen — an animal, the wind, or water.
An insect on a flower may be pollinating it, robbing it, or doing nothing at all
Established
Specialists would state this without hedging. Multiple independent lines of evidence agree.
Pollination requires viable conspecific pollen to reach a receptive stigma. Flower visitors vary from effective pollinators to nectar thieves and robbers that bypass the reproductive structures entirely. In buzz-pollinated species — roughly 6% of flowering plants, including tomato and blueberry — pollen is released only by vibration, which honey bees essentially cannot produce, so a honey bee visit transfers nothing.
Who this applies to
flower visitors generally; buzz pollination demonstrated in Solanum and relatives
Studied in
Solanum, Bombus, Apis mellifera
You may have heard
“Bees pollinate everything”
A honey bee can sit on a tomato flower all day and pollinate nothing, because the pollen only comes out under vibration and it cannot vibrate. Commercial tomato growing depends on bumblebees for exactly this reason. Counting insects on flowers measures visits, and visits are not pollination.
Why we rate it this way, and what the caveats are
EstablishedHigh confidence
Buzz pollination provides a mechanical demonstration that cannot be argued around: a bee physically unable to vibrate at the required frequency cannot extract the pollen, however often it lands.
How far it can be extended
The visitor-versus-pollinator distinction is general across pollination ecology; the buzz-pollination case is the cleanest documented instance.
Caveats
Effectiveness varies by flower, visitor, weather and time of day, so single-visit measures generalise poorly.
Nectar robbing is not always costly to the plant, and can be neutral.
Honey bees are effective pollinators of many crops; the point is that visits are not interchangeable.
Still unanswered
How much of measured "pollinator visitation" in survey data represents actual pollen transfer?
King et al., 2013 · Methods in Ecology and Evolution
Measures what a single identified visit actually deposits on a virgin stigma, which is the measurement the whole distinction rests on.
How we know
Why a honey bee cannot pollinate a tomato
Some flowers refuse to release their pollen to most visitors. What does it take to get it out?
Buzz-pollinated flowers — tomato, aubergine, blueberry and their relatives — hold pollen inside tubular anthers that open only at a tiny terminal pore. Across the studies this review assembles, bees were recorded gripping anthers and vibrating their flight muscles while the acoustics were measured, and pollen release was quantified against vibration frequency and amplitude. Artificial vibration of flowers, at controlled frequencies, separated what the flower requires from what any particular bee happens to do.
What happened
Pollen comes out only under vibration in a particular range. Bumblebees and many solitary bees produce it by decoupling their flight muscles and buzzing against the anther; honey bees essentially cannot. Around 6% of flowering plant species hold their pollen this way.
What it shows
Visiting a flower and pollinating it are different events, and here the difference is mechanical rather than behavioural. A honey bee on a tomato flower is not an inefficient pollinator — it is not a pollinator of that flower at all, which is why commercial tomato growing depends on bumblebees.
What it does not show
It does not mean honey bees are poor pollinators generally; they are effective on a great many crops. Nor is the buzz-pollinated fraction of flowering plants precisely known — 6% is an estimate — and which bee species can sonicate effectively is still incompletely catalogued.
The controls — what makes this evidence rather than a story
Mechanical vibration at chosen frequencies tested the flower directly, with no bee involved.
Visits by bees that cannot sonicate provided the negative case — visitation without release.
Pollen release was measured rather than inferred from visit counts.
Buzz pollination is the cleanest demonstration because it is mechanical rather than behavioural. Some plants — tomatoes, aubergines, blueberries, cranberries, around 6% of flowering species — hold pollen in tubular anthers that release it only under vibration at particular frequencies. Bumblebees grip the anther and vibrate their flight muscles; honey bees essentially cannot. Commercial tomato production depends on bumblebees for exactly this reason.
There are other ways to visit without pollinating. Nectar robbers bite through the base of a flower and take the reward without touching the reproductive parts. Thieves reach the nectar legitimately but are the wrong shape to pick up pollen. And a visitor carrying the wrong species’ pollen delivers nothing useful — which is why floral constancy, a forager sticking to one species per trip, matters as much as visit counts.
Conservation
Why this changes what helps
Adding honey bee hives is often described as helping pollinators. Honey bees are one managed species, they compete with wild bees for forage, and they cannot pollinate buzz-pollinated crops at all. Where wild pollinator populations are the limiting factor, habitat — flowering margins, nesting sites, reduced pesticide exposure — does what extra hives cannot. Consult a local conservation body rather than generalising from this page.
Where this applies: General ecological principle; local conservation advice should take precedence.
Words used here
Buzz pollination
Shaking pollen from tubular anthers by vibrating flight muscles. Also called sonication.
Floral constancy
A forager sticking to one flower species per trip, which is what makes the pollen it carries useful.
Nectar robbing
Taking nectar without contacting the reproductive parts, often by biting through the flower base.
Six steps that popular coverage compresses into one.
Pollination is one link in a chain, and almost every confusion in this area comes from collapsing two links into one. Laid out in order, each step can fail independently.
Pollination: a pollen grain reaches a receptive stigma of the same species. Nothing has been fertilised yet.
Pollen germination: the grain grows a tube down through the style — hours in some species, months in others.
Fertilisation: sperm cells delivered by that tube fuse with the egg inside an ovule.
Seed development: the fertilised ovule becomes a seed; the ovary around it becomes a fruit.
Dispersal: the seed is moved away from the parent, by wind, water or an animal that is not trying to help.
Germination: given water, oxygen, a suitable temperature and whatever cue its dormancy demands, the seed produces a root.
A flower can be visited and not pollinated, pollinated and not fertilised, fertilised and produce seed that never disperses, and disperse seed that never germinates. Counting insects on flowers measures the first step only, which is why visitation and yield are different measurements.
The single most consequential distinction in pollination ecology, and the hardest to photograph.
Almost every statement in circulation about which animals pollinate what rests on visitation counts: someone stood by a plant and recorded who landed on it. The measurement that matters is different — how much of the right pollen a visitor leaves on a stigma — and when both are measured on the same plant they rank the visitors differently.
How we know
Letting one insect touch one flower, once
Pollinator surveys count visits. Does a visit from one insect deliver the same pollination as a visit from another?
The difficulty with answering this in the field is that a flower in the open receives many visits from many species, so nothing can be attributed. The design removes that entirely. Flowers were bagged before opening so they had received no pollen at all, then unbagged and watched until exactly one identified insect visited, then rebagged. The stigma was removed immediately, mounted, stained, and the pollen grains of the correct species counted under a microscope — which also means counting past the pollen of other species the insect was carrying.
What happened
Deposition varied by more than an order of magnitude between visitor species on the same plant. Ranking visitors by how often they came produced a different order from ranking them by how much pollen they left, and several of the most frequent visitors were among the least effective.
What it shows
That visitation and pollination are different measurements, and that surveys reporting the first are not reporting the second. It is the methodological basis for refusing to call an animal a pollinator of a plant because it has been photographed on the flower.
What it does not show
Deposition is not seed set. A flower can receive ample pollen and still set nothing, for reasons of resource limitation or incompatibility. A single visit is also not a season: a mediocre depositor that visits constantly may deliver more in total than an excellent one that comes rarely, and this design cannot see that. Labour-intensive enough that sample sizes per plant–visitor pair are small.
The controls — what makes this evidence rather than a story
Flowers bagged before anthesis, so the count starts from zero rather than from an unknown baseline.
Unvisited bagged flowers processed identically, to establish that the bagging itself deposited nothing.
Conspecific pollen counted specifically, so a visitor arriving covered in the wrong pollen is not credited.
Visitors identified to species at the moment of the visit rather than inferred from what was common.
The gap has several causes and they compound. A visitor can drink nectar without touching an anther. A bee grooming pollen into its baskets is removing exactly the material the plant wanted carried. An animal that visits many species at once arrives covered in pollen of the wrong plant, which occupies the stigma without fertilising anything. And a flower that requires vibration hands nothing to a visitor that cannot vibrate.
This is why NatureHQ will not describe an animal as a pollinator of a plant on the strength of it having been seen on the flower, and why "pollinator decline" and "fewer insects on flowers" are not the same statement.
Words used here
Single-visit deposition
The pollen a visitor leaves on a virgin stigma in one visit. The standard measure of how effective a pollinator actually is.
Pollen limitation
When a plant sets less seed than it could because not enough compatible pollen arrived.
Flower shape and colour predict the pollinator loosely, not reliably
Well supported
Good evidence backs this, though some details remain open.
Pollination syndromes — recurring suites of floral colour, shape, scent, timing and reward associated with pollinator groups — have real predictive value when interpreted as adaptation to the most effective pollinators. They frequently mispredict the actual visitor assemblage, because most plants are visited by many species and specialisation varies with community context.
Who this applies to
animal-pollinated flowering plants, mostly studied in temperate systems
Studied in
Angiospermae
You may have heard
“Red tubular flowers are for birds, white scented ones for moths”
The pattern is real and it is a tendency. It is taught as a lookup table because it is memorable, and the exceptions are numerous enough that a syndrome is a hypothesis about a flower rather than an answer.
Why we rate it this way, and what the caveats are
Well supportedModerate confidence
The convergence itself is well documented. Confidence is moderate on the predictive strength, because effectiveness data — as opposed to visitation data — exist for relatively few systems.
How far it can be extended
The convergent floral traits are found worldwide; the strength of the association with any one pollinator group varies by community.
Caveats
The syndrome literature has studied specialised systems disproportionately.
A syndrome describes the selective history, not necessarily who visits today.
Introduced pollinators can visit flowers whose syndrome does not match them at all.
Still unanswered
How often does a syndrome correctly predict the most effective pollinator rather than the commonest visitor?
Fenster et al., 2004 · Annual Review of Ecology, Evolution, and Systematics
Reconciles specialised floral form with generalised visitor assemblages by weighting the most effective pollinators.
Flowers pollinated by similar animals tend to converge on similar features, and the resulting suites — the pollination syndromes — are genuinely useful. They are also taught as a lookup table, which they are not.
Syndromes as tendencies, with the caveat attached
Pollinator
Typical flower
Where it breaks
Bees
Blue, purple or yellow; landing platform; ultraviolet nectar guides; scented by day
Bees readily visit flowers of every other syndrome, including red ones
Butterflies
Upright, narrow tube, a platform to stand on, bright colour
Long tubes exclude short-tongued visitors, not butterflies specifically
Moths
Pale or white, strongly scented after dark, opening at night
Many "moth flowers" are worked by bees the following morning
Birds
Red or orange, tubular, robust, abundant dilute nectar, little scent
Red is only a bird signal where bees are the alternative; not universal
Bats
Large, pale, night-opening, strong musty scent, held clear of foliage
Reliable where nectar bats exist; the syndrome predicts nothing where they do not
Flies
Open and shallow, dull, sometimes smelling of dung or carrion
A very large and heterogeneous group; the syndrome is the weakest of these
Wind
No petals, no scent, no nectar; dangling anthers and feathery stigmas
The most reliable of the set, because there is nothing to attract
The reason syndromes work at all is that a flower is adapted to the visitors that matter for its seed set, not to everything that lands on it. So a syndrome is a reasonable hypothesis about the most effective pollinator and a poor prediction of who will be seen on the flower — which is exactly backwards from how it is usually used.
Words used here
Pollination syndrome
A recurring combination of flower colour, shape, scent and reward associated with a group of pollinators.
Nectar guide
A marking directing a visitor to the nectar. Many are visible only in ultraviolet, which bees see and we do not.
Around 12% of flowering plant species are not animal-pollinated, and they include almost everything humanity eats in bulk. Wheat, rice, maize, rye and barley are wind-pollinated; so are most of the trees of the northern forests, and all the grasses. A conversation about pollinator loss and food security has to begin by separating these, because the staple cereals are not at risk from it at all — the crops that are, are the fruits, nuts, oilseeds and vegetables that supply most of the variety and much of the nutrition.
Wind: vast pollen production, no reward, drab flowers, feathery stigmas. Cheap per grain, hopeless per grain delivered.
Water: rare, and mostly in seagrasses, where pollen is released in strands that drift onto stigmas.
Self-pollination: some flowers pollinate themselves before opening at all, guaranteeing seed at the cost of variation.
Mixed strategies: many plants outcross when visitors are available and self-pollinate as a fallback late in the season.
Selfing is worth taking seriously rather than treating as a failure mode. It guarantees reproduction where visitors are scarce, and for a colonising plant arriving alone in a new place it is the only option available. The cost is genetic: repeated selfing exposes harmful recessive alleles, which is what inbreeding depression means and why so many plants have machinery to prevent it.
This history has one recurring feature: each generation established a real pattern and the generation after found that the pattern was looser than it looked. Sprengel discovered that flowers are built around their visitors and was ignored for eighty years; the syndrome framework he inspired then hardened into a rule that the last thirty years of measurement has had to soften again.
1793
First observation
Sprengel finds that flowers are built for visitors
Christian Konrad Sprengel describes nectar guides, the timing of anther and stigma maturity, and the relationship between floral form and insect visits. His book is largely ignored for eighty years.
1862
Reinterpretation
Darwin makes it about coevolution
Darwin’s work on orchid fertilisation rehabilitates Sprengel and reframes floral form as the outcome of selection by pollinators — including the prediction of a Madagascan moth with a foot-long tongue, found decades later.
Changes how the 1793 result reads
Sprengel had the observations and no mechanism. Selection supplied the mechanism, which is what made the observations worth returning to.
2004
Challenge
The syndrome framework is examined rather than assumed
A review sets the neat colour-and-shape categories against the observation that most flowers are visited by many species, and finds syndromes predict the most effective pollinator far better than they predict who turns up.
Changes how the 1862 result reads
Not a refutation of coevolution. A correction to how tightly it was assumed to specify one flower to one animal.
A synthesis of community-level surveys puts animal pollination at about 87.5% of flowering plant species, and finds that the widely quoted 80% figure had no traceable empirical source.
Across some 600 fields and 41 crop systems, wild insect visitation raised fruit set everywhere while honey bee visitation raised it in 14% of systems — and more honey bees did not substitute for wild insects.
Changes how the 2004 result reads
Shifts the practical question from which flower suits which animal to which animals are actually delivering, and the answer is not the one being managed.
Visitation is shown to be a poor proxy for pollination
Measuring what a single identified visit deposits on a virgin stigma finds deposition varying by more than tenfold between visitors, and ranking visitors differently from visit counts.
Changes how the 2011 result reads
The methodological correction underneath the ecological one: a great deal of what is reported as pollination data is attendance data.
Bees are the most important group by a distance, and there are around 20,000 species of them, of which the western honey bee is one. But flies pollinate more than most people realise — including in cold and high-altitude places where bees struggle — beetles were probably the original pollinators of early flowering plants, and moths handle a great deal of night-time work.
Specialisation runs from near-total generalism to relationships involving exactly one plant and one animal. Figs and fig wasps are the classic case: most fig species have their own wasp species and neither reproduces without the other. Those relationships are spectacular and rare, and most pollination is done by generalists.
How much recorded "pollinator visitation" represents actual pollen transfer?
Why it matters: Most monitoring counts visits because visits are countable. If the relationship between visits and transfer varies by species, a great deal of conservation data means less than it appears to.
Do managed honey bees measurably harm wild pollinator populations?
Why it matters: Competition for forage is documented; whether it drives declines is contested, and it bears directly on whether beekeeping is conservation.