Firmiana daweishanensis can exceed 30 meters in height and bears unisexual flowers in clustered panicles on exposed flowering branches. Individual flowers open yellow, shift to orange, and finish red before senescence, so yellow and red florets coexist within the same crown and often on the same branch. The research team, led by Bo-Qiang Wang and Jing Yang, spent March and April 2024 observing 16 flowering branches across the accessible crown of one large adult tree near Nanxi Town. Each branch was watched from 9:00 to 17:00 for eight hours, producing 128 branch-hours of observation. The observers recorded the identity of every visitor, the number of visits, the cumulative time spent on flowers, and crucially, whether the floret visited was yellow or red, female or male.

The pollinator assemblage turned out to be dominated by leafcutter bees. Six hymenopteran species showed stable, effective flower-visiting behaviour, repeatedly contacting reproductive structures during their visits. A single taxon, Megachile sp., accounted for 46.2 percent of all visits and 39.9 percent of cumulative visitation time, and the three Megachile categories together represented most pollinator activity. Nomia incerta, Apis mellifera, and Xylocopa appendiculata contributed smaller shares. For a tree whose adult individuals are scarce and spatially isolated, the identity of these visitors matters enormously: when pollinators struggle to connect isolated flowering adults, successful visitation may not translate into effective fertilisation, a problem that is especially acute for conservation-priority species with tiny populations.

The behavioural data revealed a striking association with floral colour stage. Yellow flowers received an average of 0.170 visits per flower per hour, compared with just 0.022 for red flowers, and cumulative visitation time followed the same pattern, at 0.631 seconds per flower-hour for yellow versus 0.116 for red. To test whether this association held after accounting for flower availability and the grouping of observations within branches, the team fitted generalised linear mixed models with floral colour, floral sex, and their interaction as fixed effects, flowering branch as a random intercept, and the logarithm of flower-hours as an offset. Visit counts were modelled with a negative-binomial distribution and visitation time with a Tweedie model, both with log links. The omnibus colour tests were highly significant for both responses, with Holm-adjusted p values on the order of 10 to the negative 6, while the omnibus sex tests and colour-by-sex interactions were not significant.

Secondary contrasts from the full models quantified the preference precisely. Yellow flowers attracted 6.40 times more visits than red flowers among female florets and 4.80 times more among male florets, with corresponding ratios of 5.56 and 4.06 for cumulative visitation time per flower-hour. All four colour contrasts remained significant after Holm correction across the eight secondary comparisons. Sensitivity analyses strengthened confidence in the qualitative pattern: when the assumption that visitation scales proportionally with flower number was relaxed, the estimated flower-number coefficients fell below one, yet the colour association remained significant, and leave-one-branch-out refitting preserved the result in every case. Interestingly, at the branch level, neither the abundance nor the relative proportion of yellow and red flowers correlated significantly with pollinator activity, suggesting that colour-dependent selection operates at a fine spatial scale, among individual florets, rather than across whole branches, although the modest sample of 16 branches may have limited power to detect broader correlations.

Why would bees prefer yellow? The study measured a suite of floral traits that change in concert with colour, and the picture that emerges is of a coordinated multimodal state signal rather than a purely visual transition. Floral scent sampling using solid-phase microextraction and gas chromatography-mass spectrometry showed that linalool, the dominant volatile compound, was emitted more strongly by yellow flowers than by red ones in both sexes, with yellow male flowers producing the highest per-flower emission and red male flowers the lowest. Linalool is widely involved in plant-insect communication, so stronger emission from young flowers may reinforce colour-based discrimination, particularly for bees that integrate visual and olfactory information while foraging. Nectar measurements told a similar story: female flowers accumulated more nectar during the yellow-stage 10-hour interval than during the red-stage interval, and male flowers yielded no recoverable nectar by early on the second day, even though the flowers remained attached.

Reproductive function itself shifted through the floral lifespan in ways that aligned with the visitation pattern. Pollen viability, assessed by in vitro germination in a medium of 30 percent sucrose and 0.003 percent boric acid, was highest early in flowering and declined progressively thereafter. Stigma receptivity, tested with the benzidine-hydrogen peroxide method, rose after flower opening, peaked around mid-anthesis, and then decreased. In other words, the yellow and early-transition stages coincided with both greater reward availability and stronger reproductive function, consistent with the hypothesis that the yellow-to-red change functions as an honest signal of declining floral value. Bee visual modelling added a mechanistic visual dimension: reflectance spectra measured at the yellow, orange, and red stages showed that yellow flowers generated significantly greater green-receptor achromatic contrast against a standard green background than red flowers, at 0.295 versus 0.083, a difference confirmed by Welch’s t-test with a p value of 4.05 times 10 to the negative 6. In the honeybee hexagon colour space, the yellow-stage spectrum also sat farther from the adapted-background origin, indicating stronger modelled visual conspicuousness under standard illumination.

The pollination treatments delivered results with real conservation weight. Of 100 female flowers bagged before anthesis to exclude pollinators, none produced fruit. Of 100 flowers hand-pollinated with self-pollen from male flowers on the same individual, an artificial geitonogamous self-pollination treatment, none produced fruit either. Yet naturally exposed flowers achieved a fruit set of 6.19 percent, yielding 54 fruits from 872 flowers. This contrast provides experimental support that floral visitors contributed to fruit production under the study conditions, and it shows that self-pollen did not result in successful fruit set. Because no hand-cross-pollinated positive control was included, the team could not conclusively distinguish self-incompatibility from other causes of self-pollination failure, but for a species with extremely small populations, an inability to set fruit through self-pollen could heighten reproductive vulnerability if compatible pollen donors are scarce. Qualitative observations added a developmental note: bagged, unmanipulated female flowers underwent the same yellow-to-orange-to-red sequence as exposed flowers, suggesting that visitation is not required for the colour transition, in contrast to species such as Saxifraga fortunei, where reddening is associated with pollen-tube progression.

One puzzle remains the retention of red-stage flowers, which accounted for an average of 44.54 percent of observed floral units. In many colour-changing species, older, less rewarding flowers are kept on the plant and may contribute to long-distance display while younger flowers guide close-range choices. In Firmiana daweishanensis, the colour-changing region is largely restricted to the inner part of the calyx lobes, while much of the outer lobe area stays pink throughout anthesis, so the flowering crown appears predominantly pink from a distance even when many individual flowers have red centres. The authors hypothesise that retaining red flowers helps preserve crown-level display, while the yellow-to-red transition in the central region provides a finer-scale cue within the crown. Immediate abscission of nearly half the floral units would substantially reduce the visible display, but because red-flower retention was not manipulated and long-distance attraction was not measured, this proposed functional separation between crown-level display and within-crown signalling awaits experimental verification.

The study’s limitations are acknowledged frankly: all behavioural observations came from one accessible adult tree on one day, reflecting the logistical constraints of studying a PSESP but restricting population-level inference, and scent, nectar, and reflectance were measured on flowers separate from those observed for visitation to avoid disturbing natural behaviour. Future work should combine manipulative experiments, such as colour alteration, scent addition, and nectar supplementation, with pollinator choice assays to separate the contributions of visual signals, olfactory cues, and reward status. Even so, the findings point to practical conservation priorities: protecting flowering individuals and nearby potential pollen donors, maintaining canopy and forest-edge habitats used by Megachile bees, reducing pesticide risks during flowering, and monitoring dominant pollinators. Where natural fruit set remains low, assisted outcross pollination and ex situ propagation could complement these measures, provided compatibility tests and genetic information guide the actions. More broadly, by using a scaffold to observe flowers within the crown of a tall rainforest tree, the study adds a rare canopy perspective to floral colour-change research, a field historically biased toward lower, more accessible vegetation, and offers a biological foundation for conserving one of China’s most narrowly distributed and threatened trees.

Subject of Research: Pollinator Visitation Within a Single Crown of the Rare Rainforest Tree Firmiana daweishanensis Is Associated With Floral Colour Change

Article Title: Pollinator Visitation Within a Single Crown of the Rare Rainforest Tree Firmiana daweishanensis Is Associated With Floral Colour Change

Article References: Wang, B.-Q., Xiao, S.-Y., Li, C.-J., Ping, Y.-M., Zhang, J.-G., Sun, W.-B., & Yang, J. (2026). Pollinator Visitation Within a Single Crown of the Rare Rainforest Tree Firmiana daweishanensis Is Associated With Floral Colour Change. Ecology and Evolution, 16(10), Article e74490. https://doi.org/10.1002/ece3.74490

Image Credits: AI Generated

DOI: 10.1002/ece3.74490

Keywords: pollination ecology, floral colour change, Firmiana daweishanensis, PSESP conservation, bee behaviour, linalool scent, canopy access, Yunnan rainforest, Pollinator, Visitation, Within, Single