Australasian Science: Australia's authority on science since 1938

Flower Evolution from the Birds to the Bees

 Credit: BirdImages/iStockphoto

Credit: BirdImages/iStockphoto

By Mani Shrestha, Adrian G. Dyer and Martin Burd

Walking around in the Australian bush we can see a dazzling array of different flower colours, but have you ever wondered how and why these evolved?

Plants face a challenging courtship problem. They can’t walk, talk... or the other pleasant parts that might “go with a date”. So to enable sexual reproduction to enhance the genetic variability of their offspring, plants must often rely on other mechanisms to enable pollen transfer.

Some plants like grasses use wind to blow pollen onto the flower’s stigma, which captures pollen and starts the process of transmitting sperm in the pollen toward the egg of an ovule. But wind pollination is not very precise. Pollen can be blown onto the stigmas of flowers of the wrong species, or into the faces of animals – as the millions of us who suffer from hay fever can testify.

Enter the birds and the bees as important collaborators. Better than walk, these guys can fly very quickly between flowers to deliver pollen.

But, as Charles Darwin postulated, birds and bees will not evolve to do this unless there is a fitness benefit to them. A nutritional reward like nectar or the nitrogen-rich pollen itself is the reward that encourages birds and bees to fly between flowers searching for more rewards, and incidentally transferring the pollen while doing so.


Now this all might appear very straightforward, but the complication is that birds and bees see colours very differently to humans, and to each other. While humans typically see blue, green and red light, bees see ultraviolet, blue and green light while many birds see four types of colours including violet (or ultraviolet), blue, green and red light (Fig. 1). Hence flowers pollinated by bees may need to present very different colours than flowers pollinated by birds so that the colour signals of the flower can best target the perceptual channels of the most efficient pollinator in an environment.

This is analogous to basic shopping theory. If a shop requires loyal customers to keep visiting, it needs to have displays that clearly communicate to targeted customers. So how do we get to understand what is important for pollinators to see?

To solve this interesting question in the Australian context, we teamed up to combine our backgrounds in botany, vision and ecology. First we measured the colour properties of flowers with a spectrophotometer, which plots the amount of reflected radiation against the wavelength of the radiation (Fig. 2). The wavelength of radiation is a convenient way of describing the visual part of the electromagnetic spectrum, such that red light is at wavelengths longer than 600nm, blue light is at wavelengths of about 400–500 nm, and UV light (which humans do not normally see) is at wavelengths shorter than 400 nm.

We measured the flowers of dozens of species with this device. To fully understand how these plants are related required detailed botanical knowledge and a consideration of the evolutionary relationships among species and families of plants in our sample.

We next identified which type of pollinator visited each particular species of flowering plant. It was then possible to interpret the complex statistical relationships between bees, birds and flowering plants. This revealed that bee-visited flowers frequently had spectral signals at the wavelengths that bees discriminate or detect best (about 400 nm and 500 nm), while bird-pollinated flowers had spectral signals frequently occurring at the longer red wavelengths of the spectrum, which birds discriminate better than bees.

Indeed, the evidence suggested that bird-pollinated flowers had often changed in evolutionary history from being bee-visited flowers to producing signals that now communicate best to birds.


This raised the question of how different environments might suit certain pollinators and affect what flower colours, or even types of plant, might be most efficient at surviving in certain climates. To approach this interesting question we considered different flowering plants in both sub-tropical and sub-alpine regions of Nepal. In particular, the Himalayas of Nepal is typified by very steep terrain that can act as a natural experiment for testing how flowers may evolve in different climates (Fig. 3).

It had been believed that flies, which mainly use scent to find flowers, were the most important pollinator in many alpine regions, so it was expected that bee flower colours would be comparatively rare at high altitude. However, by collecting a large amount of flower spectral data from the sub-tropical and sub-alpine regions, and comparing their spectra, it was possible to show that bees have been the most dominant influence on flowering plant colour evolution in both regions of Nepal.

Interestingly, the phylogenetic relationship between the respective regions suggested that flowering plants have independently evolved such signals, revealing that bees are the most important pollinator in radically different environments, and that plant flowers can change their characteristics to suit the most efficient pollinator.

This finding was a surprise as flies are thought to be the main pollinator in many mountain regions, but it appears that in the Himalayas several bee species have evolved to be active at high altitude, and these insects have been such effective pollinators that they have led to the evolution of distinctive bee-friendly colours.

However, while bee colours were prevalent at all elevations, flower colours in high altitude zones were more diverse and had more often undergone larger steps of evolutionary change than those at lower elevation, suggesting that more work on this question would be valuable.

This line of inquiry takes us back to Australia as a case study. It appears that the reason why some plants evolve red flowers may be that by advertising to a certain group of “loyal” and constant pollinators that prefer a certain spectral cue, some plants reproduce better and have evolved such specific cues. Indeed, when our team plotted the relative spectral position of Australian flowers we could clearly see that some bird-pollinated flowers occupied very specific regions of a colour space.

The colour spaces we used are three-dimensional computer models of how flowers would be seen by honeyeaters, a large family of nectar-feeding birds that are the dominant avian pollinators in Australia, or by sunbirds and some other bird families with a different visual system. While about half of the honeyeater-pollinated flowers we sampled had similar positions in a colour space to insect-pollinated flowers, the other half of the sample flowers had evolved specific reflectance colours that were confined to a small region of a colour space that honeyeaters see much better than bees.

In some other parts of the world, sunbirds are the dominant bird pollinators of flowers, such as the nectar-feeding sunbirds of southern Asia and Africa. Sunbirds have ultraviolet peak sensitivity at about 375 nm, and thus see colours differently to many Australian birds, so potentially flowers may have evolved different colours in environments where sunbirds are common.

These computer models of bird vision allowed us to test the classic hypothesis that certain flowers have evolved colour signals to suit particular pollination syndromes. In the future it will be possible to further test such ideas by evaluating how flower colour evolved in other regions like Asia or the Americas, which will provide vital information about how pollination networks may operate under changing climatic conditions in some environments.

Mani Shrestha is a postdoctoral researcher at Monash University, where he is working on the potential effects of climate change on complex plant–pollinator interactions. Adrian G. Dyer is a vision scientist at RMIT University. Martin Burd is an evolutionary biologist at Monash University.