Managed hives and wild pollinators

Why can adding more beehives to a landscape leave its crops less well pollinated over time?

A grower worried about fruit set does the obvious thing and brings in more hives. Bee numbers on the blossom go up; anyone can watch them. Yet across years, in some landscapes, fruit set per flower does not follow, and can drift down. That is a strange shape of failure: "diminishing returns" will not produce it, because diminishing returns flattens a curve rather than bending it downward. So something is being subtracted at the same time as bees are being added.

Begin by refusing to treat "a bee visit" as a unit. What the flower needs is a specific physical event: compatible pollen delivered onto a receptive stigma. A visit that takes nectar without touching the anthers accomplishes nothing, and visits differ enormously in how reliably they achieve it.

Consider what a honeybee is. A generalist with a short tongue, a preference for nectar, a habit of working a flower the easy way, and no capacity for sonication — the buzzing that shakes pollen from the tube-shaped anthers of tomato, blueberry and cranberry, which several wild bees manage and honeybees cannot. Alfalfa is the sharpest case: the flower is spring-loaded and strikes the visitor's head when tripped, and honeybees learn quickly to rob nectar from the side without tripping it. Seed growers use leafcutter and alkali bees instead, because the honeybees keep visiting while ceasing to pollinate. So abundance and effectiveness are separate quantities, and a landscape can gain the first while losing the second.

What would losing the second require? Wild bees declining as hives increase — so, competition for the same flowers, between two sides that are not symmetrical. That asymmetry is the whole of it. A honeybee colony is a large, long-lived, mobile foraging unit that flies kilometres, recruits nestmates to a rich patch, and can be fed sugar through a dearth. A solitary bee is a single female with a flight season of weeks, a range often of hundreds of metres, and in many species a narrow diet — some take pollen from one plant genus and nothing else. She provisions a fixed number of brood cells from whatever is within reach while she is on the wing. Draw down the pollen in that radius and she does not switch crops or wait for next month; she provisions fewer cells, and there are fewer bees next year.

Then the timing, where the trap closes. The crop blooms for two or three weeks; the hives are in the landscape most of the year. So the competition that matters happens largely outside the bloom, on the margin flowers that carry wild bees through the rest of their life cycle while the colonies are still foraging hard. "There are plenty of flowers during bloom" is true and beside the point.

Two pressures compound it. Managed colonies are dense pathogen reservoirs — deformed wing virus among them — and flowers are shared surfaces; spillover into wild bumblebees is demonstrated, with prevalence higher near apiaries. And a landscape reduced to one pollinator species loses its insurance: honeybees fly poorly in cold or wet weather when bumblebees still work, so a bad spell during a short bloom removes the whole service rather than part of it.

Diversity also did more than add its own visits. Where the two forage together, honeybees are displaced more often and move between rows more, carrying pollen further — exactly what an outcrossing crop like almond needs. Remove the wild bees and the remaining honeybees become individually less useful.

Take the folk accounts one at a time, because each fails for a reason that teaches something.

"The bees are dying — that's the story." This gets the sign wrong. The global stock of managed honeybee colonies has risen substantially over the past six decades even as pollination shortfalls grew; colony losses are a beekeeping cost, not a shortage of managed bees (direction recalled from long-run agricultural statistics; I would not lean on a percentage). And the question here is what happens when hives are added.

"It is just saturation." Saturation predicts a plateau. It cannot produce a fall in fruit set per flower or in effectiveness per visit, and both are what the puzzle describes.

"Pesticides and habitat loss are killing wild bees." Both are real and probably larger drivers overall. But neither predicts a decline grading with apiary density while pesticide use and habitat are held constant, which is the pattern this account exists to explain.

That gives the test. If competition-plus-substitution is right, the fall in service should be strongly crop-specific — largest where buzz pollination or precise anther contact is needed, near zero in crops honeybees pollinate well — and should appear as visits rising while fruit set per visit falls, with wild bee abundance tracking distance from apiaries in an otherwise uniform landscape. Service declining equally across all crop types, or wild bee decline flat with apiary density, would sink it.

The synthesis I lean on compared wild-insect and honeybee visitation across dozens of crop systems worldwide and found wild insects improved fruit set in essentially all of them, honeybees in a minority, and wild insects markedly more effective per visit. That is recalled; treat the asymmetry as robust and the multiplier as approximate. I decline to put a number on pathogen spillover's contribution — the prevalence gradients are documented, the causal share is not.

One honest limit: this is landscape- and crop-dependent, not a law. Where non-crop forage is abundant all season, added hives can raise total service at no cost to wild bees. The paradox needs a resource bottleneck outside the bloom to bite.

• What margin planting must look like — in bloom timing, not just area — to relieve the bottleneck rather than move it.

Every term the collection defines is gathered in the glossary.