Data from: Flower colour and visitation rates of Costus arabicus support the "bee avoidance" hypothesis for red-reflecting hummingbird-pollinated flowers

Floral colour mediates plant–pollinator interactions by often signalling floral resources. In this sense, hummingbird-pollinated flowers are frequently red-coloured, and there are two tentative hypotheses to explain this pattern: 1. hummingbirds are attracted to red due its easier detection and 2. bees are sensorially excluded from red flowers. The second hypothesis is based on bees’ red colour blindness, which lead them to be less frequent and less important than hummingbirds as pollinators of red-reflecting flowers. Here, we untangled the role of different flower traits mediating plant–pollinator interactions and empirically tested the above hypotheses. We chose Costus arabicus due to its synchronopatric white- and pink-flowered individuals and its bee and hummingbird pollination system. Although pink flowers are not totally achromatic as pure red ones, they show an achromaticity degree that could drive bee exclusion. Specifically, we tested whether differences on red reflectance work attracting hummingbirds or excluding bees and the consequent implications for the plant's reproduction.
Flower colour morphs of C. arabicus do differ neither in morphology nor in nectar sugar content. Moreover, white and pink flowers can be discriminated by the bees’ and hummingbirds’ colour vision system. Both groups are able to discriminate the red colour variation morph on the flower petals, the white flowers being more easily detected by bees and the pink flowers by hummingbirds. Bees preferentially visited the white flowers, whereas hummingbirds visited both colours at the same rate – both patterns corroborating the second hypothesis. Pollen loads deposited on stigmas did not differ between flower colour morphs, indicating that bees and hummingbirds play a similar role in the overall pollen deposition. However, bees are more likely to self-pollinate than hummingbirds. Self-pollination limits C. arabicus reproduction, and red-reflecting flowers may be better pollinated by discouraging bee visitation. Therefore, the intraspecific colour variation is driving flowers to show colour-related different levels of generalization. Our results support the ‘bee avoidance’ rather than the ‘hummingbird preference’ hypothesis. Sensory exclusion of bees seems to be the pressure for red-reflecting flowers evolution, driving specialization in hummingbird-pollinated flowers due to the costs of bee pollination on plant reproduction.

花色常通过传递花部奖励信号，调控植物与传粉者的互作关系。在此背景下，以蜂鸟为传粉者的花朵常呈红色，目前有两项试探性假说来解释这一规律：其一，红色更易被蜂鸟察觉，因此蜂鸟会被红色花朵吸引；其二，蜜蜂因感官限制无法识别红色，从而被排除在红色花朵的传粉互动之外。第二项假说基于蜜蜂的红色色盲特性：相较于蜂鸟，蜜蜂在红色反光花朵的传粉者中占比更低、作用也更弱。

本研究厘清了各类花部性状在调控植物-传粉者互作中的作用，并对上述假说开展了实证检验。我们选取阿拉伯闭鞘姜（Costus arabicus）作为研究对象，因其存在同域分布的白花植株与粉花植株，且兼具蜜蜂与蜂鸟的传粉繁育系统。尽管粉花并不像纯红色花朵那样完全呈非彩色，但它们具备一定的非彩色化程度，足以实现对蜜蜂的排除效应。具体而言，我们检验了红色反光率的差异是否会起到吸引蜂鸟或排斥蜜蜂的作用，以及这对植物繁育所产生的潜在影响。

阿拉伯闭鞘姜的花色变型在花部形态与花蜜糖含量上均无显著差异。此外，蜜蜂与蜂鸟的色觉系统均可区分白花与粉花：两类传粉者均能分辨花瓣上的红色反光变异，其中白花更易被蜜蜂察觉，而粉花则更易被蜂鸟识别。

蜜蜂优先访问白花，而蜂鸟对两种花色的访问频率无显著差异——这两种结果均验证了第二项假说。不同花色变型的柱头上的花粉沉积量并无显著差异，表明蜜蜂与蜂鸟在总体花粉沉积过程中发挥的作用相近。但相较于蜂鸟，蜜蜂更易引发自花授粉。自花授粉会限制阿拉伯闭鞘姜的繁育成功率，而红色反光花朵可通过减少蜜蜂访问来获得更优质的传粉服务。因此，种内花色变异正促使花朵呈现出与花色相关的不同传粉泛化水平。

本研究结果支持“蜜蜂规避”假说，而非“蜂鸟偏好”假说。蜜蜂的感官排除效应似乎是红色反光花朵演化的选择压力：鉴于蜜蜂传粉会对植物繁育造成成本，这一压力推动了以蜂鸟为专性传粉者的花朵类群的演化。