Pollination syndromes point to species interactions present and past

Ruby-throated Hummingbird at Cardinal Flower

Want hummingbirds? Paint the town red. Photo by U.S. Fish and Wildlife Service Northeast Region.

In my part of North America, spring is finally underway after a long slog of a winter. The trees lining the streets of my Minneapolis neighborhood are lacy-green with budding leaves, flowerbeds all over the University of Minnesota campus are yellow and red and pink with daffodils and tulips, and violets are popping up in the edges of lawns everywhere I look.

Of course, all of this colorful display isn’t for my benefit. Showy flowers are an adaptation to attract animal pollinators. Some flowers are quite precisely matched to a single species of pollinator, but most flowers have lots of visitors. These less specialized flowers are still adapted for their attractive function, though—and this is the origin of pollination syndromes.

Pollination syndromes are suites of traits that are common to flowers using the same group of pollinators, or pollination method. You can find many different lists of syndromes out there, but some of the most commonly cited are:

  • Bee-pollinated flowers are usually blue or yellow, often with contrasting “guides” that point towards nectar rewards, and they usually have some sort of scent.
  • Bird-pollinated flowers tend to be red and tubular, and often open downwards. They produce lots of relatively weak nectar, and generally don’t have very strong scents.
  • Butterfly-pollinated flowers come in many different colors, but are often grouped into larger displays, and may have tubular shapes that admit only a butterfly’s long proboscis.
  • Moth-pollinated flowers are usually white, opening in the evenings, and strongly scented.
  • Fly-pollinated flowers may not be very colorful, but they often have strong and unpleasant scents, like rotting meat.

Pollination syndromes were first proposed in the 1870s, and from the start they’ve been controversial in much the same way that many attempts to classify living things often are.

On the one hand, there are apparent (and sometimes carefully verified) functions to certain features of pollination syndromes—pollinating birds don’t have particularly strong senses of smell, so bird-pollinated plants aren’t well scented, bees are particularly sensitive to colors at the blue end of the spectrum, and tubular floral shapes can exclude all pollinators but the more efficient target species. But just reading my summary of selected syndromes (how many times did I say “usually?”) should make you think that they’re more a set of rules of thumb than a hard-and-fast classification scheme, and that’s often the best way to understand them.

2012.09.03 - Hylephila phyleus I

Butterflies like big displays, and sweet scents. Photo by jby.

However, there’s some new quantitative corroboration for those natural historical rules of thumb. Victor Rosas-Guerrero and his coauthors conducted a statistical survey—a meta-analysis—of published pollination efficiency studies in 417 different plant species. Across all those plant species, they found that the pollinator species you’d expect based on each species’s rule-of-thumb traits was more efficent than other species that visited the flowers. This held whether “efficiency” was judged based on how much pollen was found on the pollinators’ bodies, how much pollen they deposited, or how many fruits or seeds resulted from their visits.

What’s more interesting is what Rosas-Guerrero et al. found in the blurry boundaries between pollination syndromes. Many of the pollination studies they collected identified secondary pollinators—species that moved pollen, but not quite as efficiently as the syndrome-appropriate pollinator. In many cases, these secondary pollinators were from groups with older evolutionary origins than the syndrome-appropriate pollinator, and they often reflected prior pollination associations deduced from analyses that reconstructed the evolutionary past of the plant species.

Finally, the authors note that when they examined plant species whose most efficient pollinators weren’t the ones appropriate to the plants’ pollination syndromes, the most efficient pollinator often turned out to be from a group that was the secondary, possibly ancestral, pollinator of other plants with the same syndrome.

So, on the one hand, pollination syndromes reflect the collective effects of many different plant species all adapting, over time, to attract similar pollinators. But on the other hand, even flowers that mostly fit these rule-of-thumb classifications still carry the marks of adaptation to past pollinators. Evolution is an ongoing process—in this case, from the origins of modern plants to the flowers by your front stoop.


Ollerton J., Alarcon R., Waser N.M., Price M.V., Watts S., Cranmer L., Hingston A., Peter C.I. & Rotenberry J. (2009). A global test of the pollination syndrome hypothesis, Annals of Botany, 103 (9) 1471-1480. DOI:

Rosas-Guerrero V., Aguilar R., Martén-Rodríguez S., Ashworth L., Lopezaraiza-Mikel M., Bastida J.M., Quesada M. & Irwin R. (2014). A quantitative review of pollination syndromes: do floral traits predict effective pollinators?, Ecology Letters, 17 (3) 388-400. DOI:

2 comments on “Pollination syndromes point to species interactions present and past

  1. jeffollerton says:

    Hi Jeremy – thanks for posting this and for highlighting such an interesting topic, and our work on it. Just a few of points to add to what you say:

    The traditional pollination syndromes are “controversial” with respect to their relative frequency within the flowering plants, not to whether they exist, which is how I’ve seen our 2009 paper described. Our results suggest that about 30% of angiosperms fit into these traditional classifications, which we argue is not good grounds for their use as a classificatory framework in pollination biology.

    The other area of controversy which is only rarely touched upon relates to the biological assumptions underpinning interpretations of the patterns we see, some of which are either incorrect or have hardly been tested. For example, “bird pollinated flowers are not scented because birds have a poor sense of smell”. Actually that’s not true, though there’s a limited literature on the topic:

    Click to access p0237-p0238.pdf

    The Rosas-Guerrero et al. paper that you discuss is an interesting addition to the subject of syndromes, but there are problems with their approach which we’ve pointed out in a response that Ecology Letters has seen fit not to publish….. This is not the place for a technical rebuttal, but I will mention two aspects that I think alter an interpretation of their results:

    1. The studies that they used in their analyses are not a random sample of the flowering plants (contra Ollerton et al. 2009) – they are a sample of the flowering plants that pollination ecologists have chosen to study. The biases inherent within this should be obvious.

    2. You say that “Many of the pollination studies they collected identified secondary pollinators—species that moved pollen, but not quite as efficiently as the syndrome-appropriate pollinator.” Actually, if you look carefully at their figures, the majority (almost 60%) of the plants they included had “secondary” pollinators which (judging from the widths of connecting bars in their Fig 2) are often just as effective as the “expected” syndrome pollinators. But because they don’t fit the expectations of the syndrome they are relegated to being “secondary” pollinators. Doesn’t that strike you as circular reasoning?

    We’re currently preparing a discussion paper about this which we hope will be published later this year (but clearly not by Ecology Letters….)



  2. Jeremy Yoder says:

    Hey, Jeff—thanks for commenting! (That experimental test of hummingbird olfaction is pretty cool.)

    I’d absolutely concede your first point; that’s inherent in any meta-analysis. And I guess (among other biases) we might expect that plants with a more limited pollinator community get more attention, since they’re easier to track and manipulate? I do notice that bird- and bee-sydrome plants are the two biggest groups in the sample.

    On the second point, I guess I’m not sure what to think. It’s certainly clear from Figure 1 that there’s a lot of variation in predictive power of syndromes, and close-second or just-as-good performance by pollinators that don’t match the syndrome is consistent with that. I’d say that Figure 1 is evidence in support of the (average) predictive power of syndromes, and Figure 2 is presented in support of the evolutionary argument—which isn’t exactly circular. But if, in the core effect size calculation, they’re averaging just-as-good non-syndrome pollinators in with a bunch of much less effective pollinators—and now that I take another look at the Methods section, it looks like that’s what they might have done—that would inflate the apparent “fit” of the syndromes.

    Anyway, I’ll definitely be interested in seeing your followup paper. Drop me a line when it’s out!

Comments are closed.