Snapdragon colour is not what it seems, and bees can tell

On 17 July 2026 a team at the University of Sussex reported a result that would sound modest if it weren't quietly unsettling: the colour of a wild snapdragon is not a single decision, painted on like one stroke, but a careful gradient, finer than a fingertip, that pushes bumblebees toward the tissue most likely to reward them.

The finding reframes a familiar piece of garden botany. Petals are not billboards. They are signposts, and the writing on them turns out to be more deliberate than the literature had assumed. The Sussex group, writing in the journal New Phytologist, shows that the surface of a single snapdragon petal can carry multiple, locally distinct colour cues, and that bumblebees read those cues as they land, walk, and probe. The plant, in other words, is not broadcasting a flat advertisement; it is curating the visitor's journey across its own face.

What the Sussex team actually measured

The researchers worked with wild-type Antirrhinum plants, the same genus sold as bedding in every garden centre but, in their uncultivated Mediterranean form, a long-standing model for the genetics of flower colour. Rather than scoring whole petals as one colour, they sampled reflectance at dozens of small patches and built high-resolution maps of hue, saturation and brightness across the flower's throat, its lower lip and the entry tube that a bee's head must enter to reach the nectar.

The maps showed consistent, repeatable patterns: a brighter, often more UV-rich zone around the lip that contrasts with a darker, redder throat. Critically, those patterns were not artefacts of pigment distribution alone. When the team blocked the UV component of the signal, either by genetic means or by experimental filters, the bee's ability to find the most rewarding tissue degraded. The conclusion, paraphrased carefully, is that the UV cue is doing navigational work, and that the plant is investing in producing it across the surface it most needs bees to visit.

A counter-narrative the field had settled on

The conventional textbook story treats petal colour as a coarse filter. A flower is "red" or "blue" or "yellow" because its pollinators have a preferred wavelength, and natural selection has dragged the flower toward that target. That framing has the virtue of simplicity and the defect of being mostly true at the wrong scale. A bee's eye, like ours, integrates over an aperture that is small relative to a petal. What matters, in the last metre of approach, is the local contrast at the millimetre scale: where the lip ends and the throat begins, where a vein of pigment draws the eye, where the entry to the nectar tube is announced by a step-change in reflectance.

The Sussex result pushes that insight into the literature proper. It says, in effect, that the textbook story missed half of the signal because nobody had mapped the petal finely enough. The fine structure was always there; the measurement wasn't. Once the measurement catches up, the picture changes: selection is acting not just on "what colour is this flower" but on "what pattern does this flower paint across itself".

Why this is bigger than snapdragons

Pollinator declines have made floral signalling politically as well as scientifically interesting. Policy debates over wildflower strips, hedgerow restoration and neonicotinoid restrictions all rest, somewhere underneath, on assumptions about how efficiently bees find nectar and pollen in mixed landscapes. If bees rely on fine-scale cues that current agri-environment schemes ignore, then the same field margin planted with the same seed mix may be delivering less forage than its cost implies.

There is a parallel argument on the crop side. Almonds, oilseed rape, blueberries and a long list of fruit and field crops depend on bee pollination, and growers routinely pay for hives by the box without much visibility into whether the bees in those boxes are actually working the target crop or wandering off into a more rewarding neighbour. A better understanding of how floral patterns steer that choice has obvious commercial value to seed companies breeding for attraction, and to ecologists trying to design landscapes that don't waste bee effort.

What remains uncertain

The Sussex study sits inside a long-running debate about how much of floral colour is signal and how much is constraint. Pigments are expensive to make, and a plant's palette is bounded by what its biochemistry can afford. The paper does not, and probably cannot, fully separate the hypothesis that the fine pattern is selected for its effect on bees from the alternative that it is an honest by-product of pigment distribution along veins and tissues. The authors acknowledge as much; they also note that the UV component, which is metabolically cheap relative to anthocyanin production, behaves more like a true signal than a by-product.

A second open question is generalisability. Snapdragons are a single genus with an unusually tractable genetic toolkit. Whether the same fine-tuning shows up in, say, oilseed rape, where petals are simpler and nectar access more uniform, is an empirical question the field will need to answer before any of this translates into breeding programmes or habitat prescriptions. For now the result stands as a clean demonstration that flowers, like cosmetics counters, can be subtler than the packaging suggests.

Desk note

This piece leads on the specific measurement and what it overturns, then widens to the policy stakes, in line with how the science desk covers primary research with downstream ecological and agricultural relevance.