Cacao Pollination by Midges: Why Yields Are So Low

A cacao pod begins with a tiny flower and successful pollen transfer. Midges are a major part of that story, but the familiar claim that every chocolate bar depends on one kind of fly is too simple. Researchers are still establishing which insects contribute most in different growing regions, and how their visits translate into harvested beans.

That distinction matters for growers. A flower can receive a visitor without receiving pollen; receive pollen that is incompatible; or begin forming a fruit that later stops developing. Improving one stage does not guarantee a proportional increase in harvest. Nor does pollination biology, by itself, explain why so much cacao is grown by smallholder farmers.

How Cacao Flowers Make Pollination Difficult

Cacao flowers emerge from cushions on older branches and trunks, a pattern called cauliflory. They are small, roughly a centimeter across. Petal hoods shelter the anthers, while five staminodes surround the central female reproductive structures. An effective visitor must contact the right parts and transfer pollen, not merely land on a petal. The CACPOL research group’s illustrated explanation shows why access and behavior matter.

Midges Are Important; Visitors Are Not Automatically Pollinators

Forcipomyia belongs to Ceratopogonidae, the family commonly called biting midges. That family name does not establish the feeding behavior of every cacao-associated species. Identifying a fly to family, or observing it on a flower, is also insufficient to establish its contribution to yield.

An August 2026 study in Basic and Applied Ecology, using observations from Malaysia and French Guiana, found Forcipomyia midges to be the strongest pollinator candidates among the visitors examined. They were active on reproductive structures and more often carried pollen. The authors explicitly did not directly measure successful pollen deposition and subsequent fertilization. Their results support the importance of midges without proving that only midges can pollinate cacao everywhere.

Other work asks different questions. A 2025 Ecuador study combined visitor observations, video and pollen measurements. Ant-associated flowers had more pollen on their styles, and the authors identified several potentially useful visitor groups. But prior visits could contribute to the pollen counted, so this was not a clean measurement of each insect’s individual contribution to mature pods.

The useful distinction is between visiting, carrying pollen, depositing compatible pollen, and producing fruit. Different methods measure different links in that chain. One study need not invalidate another merely because its most abundant visitors differ.

Do Bees Visit Cacao?

Yes. The statement that bees cannot fit, cannot visit, or never touch cacao’s reproductive parts is too broad. A 2024 field study in Colombia observed the stingless bee Tetragonisca angustula carrying substantial cacao pollen loads and interacting with both male and female flower parts. Those observations make a blanket exclusion untenable, but do not establish that honeybee hives can replace natural cacao pollination.

Bee size, species, behavior and contact with receptive surfaces matter. Collecting pollen can remove it without delivering useful pollen elsewhere. Managed-bee pollination is therefore a question for evidence on the particular species and farming system, not a recommendation that follows from seeing bees on flowers.

There Is No Single Flower-to-Pod Percentage

A percentage needs a numerator, denominator and observation period. “Flowers visited,” “flowers with pollen,” “flowers forming young fruits” and “flowers producing harvested pods” are not interchangeable.

In Vansynghel and colleagues’ 2022 Peru experiment, natural pollination produced young fruits from 39 of 1,952 flowers, about 2.0%. Hand pollination produced 70 from 968, about 7.2%. These were young fruits counted six days after pollination, pooled across seven weekly rounds—not ripe pods at harvest. The low hand-pollinated result also contradicts a universal 50–70% success promise.

A 2025 analysis across Brazil, Ghana and Indonesia estimated effective natural pollination at 11–27%, depending on country, using a model calibrated against hand-pollination treatments. Those estimates are a different measure from the Peruvian early-fruit counts. The experiment treated flowers below two meters; its percentages should not be relabeled as an annual census of all flowers on entire trees.

Hand Pollination Can Help—and Has Been Used Commercially

Hand pollination transfers pollen directly to receptive flower parts using suitable donor flowers and tools. It is labor-intensive, but a farmer does not need to pollinate every flower a tree produces to investigate a useful gain.

In a 2020 Indonesian field experiment, hand-pollinating easily reached flowers—about 13% of all flowers per tree—increased yield per tree by about 51%. A separate, more extensive treatment increased yield by about 161%. The study’s economic calculation estimated a 69% increase in annual net income after including operational, pollination-labor and opportunity costs. These were results under the study’s conditions and prices, not a current profit forecast for every origin.

Use also extends beyond research stations. Ghana Cocoa Board’s project record describes a nationwide hand-pollination initiative launched in 2017. COCOBOD subsequently reported farmers and cooperatives employing trained pollinators in its 2024 account of productivity programs. These first-party reports document implementation; they do not independently establish every claimed yield or income benefit.

Labor availability, training, donor compatibility, weather and the extra crop’s value still determine whether it pays. A sensible comparison is the cost and return of a specific treatment against an untreated baseline—not the cost of pollinating an imagined 100,000 flowers on every tree.

Compatibility Matters as Much as Moving Pollen

Some cacao genotypes are self-incompatible; others can set fruit using their own pollen. Different trees can also be cross-incompatible. Planting several unspecified genetic lines does not automatically solve the problem.

A 2017 genetic study identified different mechanisms involved in cacao self-incompatibility, including gametic non-fusion and fruit drop. It identified self-compatibility-associated alleles in Amelonado and Criollo material and used CCN-51 as a self-compatible control. It therefore does not support calling most Criollo inherently self-incompatible or CCN-51 merely partially self-compatible.

Use tested compatibility information for the actual planting material. A 2023 Peruvian pollen-supplementation experiment found that donor genotype mattered, while genetic distance alone did not predict fruit-set success. The broad Criollo, Forastero and Trinitario categories are not a substitute for that information. High-yield material such as CCN-51 should also be assessed for its complete agronomic characteristics, rather than attributing its production to one compatibility trait.

Habitat Management Needs Local Evidence

Moist decomposing plant material, including leaf litter and banana pseudostems, can provide larval habitat for cacao-associated midges. The CACPOL group describes adding suitable breeding habitat as one possible intervention. It is not a prescription to scatter all pod waste indiscriminately or create standing water everywhere; local crop-management advice and substrate condition matter.

Shade has several effects at once. It changes the flower visitors’ environment, tree temperature, available light and competition among plants. The 2022 Peru study found different relationships between canopy closure and visitation in its dry northern and wetter southern sites. That result does not support a universal rule that more shade always brings more pollination, or that full-sun plots necessarily lose it altogether.

The 2025 multi-country study linked yield to pollination, temperature and tree/soil-resource factors. Its leaf-litter association was not evidence that midges caused the yield benefit; the authors considered soil-resource effects an explanation. Habitat protection and tree nutrition can both matter without being the same mechanism.

What This Means for Yields and Chocolate

Cacao’s flowers are numerous, but harvested output must be measured directly. Multiplying 50,000 annual flowers by an unrelated 2% figure gives 1,000 mathematically; it does not establish that an average cacao tree produces 1,000 mature pods. UF/IFAS’s home-growing guide, for example, describes up to about 50 pods per plant per year under its stated conditions, with strong dependence on vigor, climate, management and variety. That is a contextual horticultural estimate, not a global limit.

Large farms also exist: a Brazilian hand-pollination experiment worked within a 150-hectare cacao agroforest. Pollinator dependence does not biologically mandate tiny farms or explain farmers’ average age. Likewise, cocoa-price volatility cannot be reduced to one insect’s life cycle.

For buyers, useful questions concern documented farm practices and outcomes: what habitat is maintained, how planting material is selected, and whether productivity claims come from measured harvests. Forest-grown cacao, including Bolivian wild-harvest cacao, should not be presumed free of pollination constraints merely because it grows in forest.