Of flowers and pollinators – a case study in punctuated evolution

Darwin predicted that flowers and pollinators are engaged in an evolutionary arms race. But that’s not always the case. In a group called columbines, evolution happened in a stop-start ‘punctuated’ way, as the flowers encountered new pollinators with increasingly long tongues.

Charles Darwin was a visionary in more ways than one. In 1862, Darwin was studying a Malagasy orchid called Angraecum sesquipedale, whose nectar stores lie inaccessibly at the bottom of a 30cm long spur (tube). Darwin predicted that the flower was pollinated by a moth with tongue long enough to raid the spur.

Darwin’s moth, perfectly adapted to drink from a long-spurred orchid.Few people believed him, but in 1903, zoologists discovered Darwin’s predicted moth, Xanthopan morgani praedicta, and it did indeed have a very long tongue. Darwin’s accurately predicted the extraordinary but matching lengths of moth tongue and orchid spur, but his explanation for them is another story.

The arms race model

He suggested that the two species were locked in an ‘evolutionary arms race’. Orchids and pollinators gradually co-evolved over time, lengthening both tongues and spurs in response to each other.

Orchids with the longest spurs have an advantage. Their nectar stores are only just within reach of pollinators, so they are tempting but don’t sacrifice too much valuable nectar. For pollinators, the advantage belongs to those with the longest tongues because they have access to the most food.

The arms race model has become widespread and popular since Darwin’s time. It helps to explain relationships between predators and prey, parasites and hosts and even males and females. But its original function – to explain the relationship between flowers and pollinators – has just been called into question.


Justen Whittall and Scott Hodges from the University of California, Santa Barbara, tested the arms race theory by looking at another long-spurred flowering plants – the columbines (Aquilegia sp). In these flowers, every petal carries its own elongated nectar spur and the advent of these spurs coincided with the recent and rapid diversification of this group.

Columbines of North America have a great range of spur sizesWhittall and Hodges charted the evolutionary relationships between the 25 North American columbine species, whose spurs range form barely a centimetre in length, to just over twelve. They found that this great variety of lengths was driven by changes in pollinators, rather than gradual races against a single one.

The flowers with the shortest spurs were pollinated by short-tongued bumblebees. Hummingbirds, whose tongues are longer, pollinate columbines with longer spurs, while hawk-moths, with the longest tongues of all, carry the pollen of the longest-spurred flowers.

There is no overlap between these three groups and once a lineage switches pollinator it doesn’t go back. Over the course of their evolution, the columbine lineages went from bumblebees to hummingbirds, and then to hawkmoths, lengthening their spurs with every jump.

The ‘pollinator shift’ model

Based on these observations, Whittall and Hodges put forward an alternative to Darwin’s arms race model. They imagined a columbine ancestor that was well adapted to the tongue length of a specific pollinator (say, a bumblebee).

In part of its range, the flower started to be visited by a second pollinator (say, a hummingbird) with a much longer tongue. In this area, the plant rapidly evolved a longer spur in response to its new partner and over time, this led to two species with different spur lengths and different pollinators.

In this model, the columbines’ spurs evolved in a ‘punctuated’ stop-start way, very different to Darwin’s model of gradual change. Each pollinator shift triggered a large evolutionary rush, as the species lengthened their spurs in response to the longer tongues of their new partners. In between these shifts, the pace of evolution slowed down considerably.

A columbine flower - its long spurs are driven by evolutionary shifts between pollinatorsBut Darwin’s arms race idea isn’t out for the count yet. Whittall points out that columbines that are pollinated by hawkmoths have a great variety of spur lengths themselves that were most likely the result of an arms race. And the moths themselves evolved long before the columbines did, so the variations in their tongue lengths must have evolved in relationships with other plants.

An adaptive valley

The stop-start model also explains a difference between columbines around the world. Those in Europe and Asia have a much smaller range of spur lengths than their North American cousins, and none of them are pollinated by hawkmoths. Whittall and Hodges have an answer for this too – it’s because Eurasia has no hummingbirds.

Imagine if flowers tried to make the evolutionary leap from bumblebee to hawkmoth without the intermediate stepping stone of hummingbirds. At the intermediate spur length, the flower would have excluded its old pollinator, whose tongues would now be too short to reach any nectar. But it would have no advantage over its new pollinator, whose amply long tongues could drink the flowers dry.

Between bumblebee and hawkmoth lies an ‘adaptive valley’, where intermediate-length flowers have no advantage and are ignored by natural selection. In Eurasia, there is not enough impetus for a species to cross it. But North America, the hummingbirds act as a stepping stone that allowed the columbines to ford this gap and evolve even more extreme flowers.

Reference: Whittall & Hodges. 2007. Pollinator shifts drive increasingly long nectar spurs in columbine flowers. Nature 447: 706-709.

Technorati Tags: , , , , , , , , , ,

Images: Moth from MSN Encarta, columbine montage from Justen Whittall’s website.

Related stories about evolution:
Natural selection does a handbrake turn – quick evolution at work
Salamander robot walks, swims and sheds light on evolutionary step from sea to land
Human cone cell lets mice see in new colours
Living optic fibres bypass the retina’s back-to-front structure
Viruses evolve to be more infectious in a well-connected population
The evolution of animal personalities – they’re a fact of life


5 Responses

  1. Do you know what the term “rapidly” represents in this study? I did some cursory looking for the information, but couldn’t find a time frame to put “rapidly” into context. 10 years, 100 years…?

  2. Not sure, although it’s a good question. I’ll go to the original paper tonight and see if I can suss this out. Otherwise, I’ll take the word ‘rapidly’ out because it could be misleading.

  3. I wasn’t meaning that I didn’t believe it, or that you should change your article.

    I searched though a dozen sites to find out more details, including the NSF’s own site, and could never find a reference to how they discovered this information….meaning by only examining plants, doing contolled studies of new plants, etc. I also could never find a refernce to time. Article after article would simply say the same thing…rapidly or “very rapidly.” Some context would be nice.

    I don’t subscribe to Nature so I wasn’t sure if the article in the magazine had more details than what I could scrape up online.

  4. Ed & liza,

    Thanks for writing about and your interest in our recent findings published in Nature. I happened upon your blog and thought I could shed some light on the question at hand. I hope that you don’t mind a long-winded response. As you probably know, getting a scientist to talk about their work can produce LOTS of verbiage!

    I certainly agree that the question of what we mean by rapid is important and I often get asked this question.

    I first published about evolution in columbines in 1994 when I sequenced a region of DNA called the ITS (internal transcribed spacers) for a number of species as well as a portion of the chloroplast genome. These regions of DNA, especially the ITS, are often used to assess the relationships among closely related species because they accumulate mutations quickly. What I found however was that there were hardly any differences among Aquilegia species even though I sampled over a wide geographic area (Europe, Asia and North America). This suggested that either Aquilegia mutates much more slowly than any other species that had been studied or that the species were very closely related and had speciated very recently (the more time since speciation, the more differences we should find). At this point, all we could say was RELATIVELY recently because we couldn’t put a date on it.

    Recently, Kathleen Kay, Justen Whittall and I published a survey of studies that had tried to put an absolute time rate on changes in the ITS region.

    Kay, KM, JB Whittall and SA Hodges. 2006. A survey of nrITS substitution rates across angiosperms supports an approximate molecular clock with life history effects. BMC Evolutionary Biology 6:36

    We surveyed studies that had independent (e.g., fossil) evidence of the age since species had diverged. These species were also sequenced for the ITS and a rate of change established. From these studies, we found the average rate of change across plant studies. Using the average rate, assuming that Aquilegia is not exceptionally different from other plants, we can estimate the age of divergence of the North American species (the focus of our recent study) from the Eurasian species. This gives an estimate of 1.2 million years for the formation of about 25 species. Even if we use the slowest rate estimate we found in our survey we would estimate a time of 2.8 million years. Both of these are quite fast compared to estimates of speciation in other species. The very fastest speciation rate estimated in animals is for a genus of crickets on the Hawaiian islands with 6 species arising in just 0.43 million years.

    One of the remarkable things about Aquilegia is that even though they are very different morphologically (especially their flowers) and even though they are found throughout the northern hemisphere in a variety of habitats, we have had a very hard time finding differences in their DNA. We have sequenced many other DNA regions besides the ITS and found similar patterns. The paper in Nature was possible because we used a method that scans the entire genome in order to identify differences.

    We are starting a genome project now to identify the specific genes involved in speciation and adaptation.

    Thanks again for your interest and I hope this helps clarify things.

  5. Yeah, I was just going to say the same thing 😉

    Seriously though, thanks Scott for the great clarification.

Comments are closed.

%d bloggers like this: