Social Evolution Forum
FIND sef:
Why Coevolution is the Key to Understanding Biodiversity
Rust-infected flax leaves. By Carl Davies, CSIRO, via Wikimedia Commons.

Harold Henry Flor spent most of the 1930s in the greenhouse of the North Dakota Agricultural College at Fargo, making flax plants sick. Flor worked in crop pathology, a brutally practical field of study. Crop pathologists work to find out what bacteria or fungi or microorganisms make economically important plants sick, then try to find a way to kill those pathogens, or to breed plants to resist them. The process can resemble engineering more than natural history — infect plant A with fungus B, record the results, move on to fungus C. Yet with precisely that process, Flor prepared the way for a new branch of evolutionary biology.

Over years, Flor tested different strains of flax rust on different lines of flax, noting which rust strains could infect and damage each flax line. He systematically mated pairs of plants and pairs of rust strains and tested their offspring against each other. And slowly he built up a dataset to describe the genes that determine whether flax plants can resist rust infections, and whether rust can overcome the plants’ defenses to establish infection.

Sign up for our newsletters

I wish to receive updates from:

There wasn’t a clear hierarchy; gene variants that let plants resist one rust strain might not be much use against another strain. As Flor wrote in the first paper he published reporting his discoveries:

These facts suggest that the pathogenic range of each physiological race of the pathogen is conditioned by pairs of factors [gene variants] that are specific for each different resistant or immune factor possessed by the host variety.

Flor called this “gene-for-gene” infection. The statistical geneticist Charles Mode created a mathematical model of the evolutionary changes implied by these flax-rust genetic interactions: In a hypothetical population of flax with many different resistance variants, plants carrying the variant that protects against the most common rust strains would prosper, and leave more offspring, and in a few generations every plant in the population would be their descendents, carrying that resistance variant. This would create an advantage for a rust strain carrying a new gene variant that can overcome the successful plant resistance variant. Such a lucky mutant would quickly spread, creating natural selection favoring a new plant resistance variant, and the cycle would begin again. Mode called this evolutionary back-and-forth in host and pathogen “co-evolution.”

Half a century later, coevolution (the hyphen disappeared almost immediately) has been referenced or examined in thousands of studies, and merits a section to itself in the just-released Encyclopedia of Evolutionary Biology. I was lucky enough to be asked to write the introductory chapter [PDF] to that section, which gave me an excellent excuse to dig into the history of research on coevolution, and to think about its future.

The idea that a major source of the natural selection shaping the evolution of living things is other living things traces back to The Origin of Species, wherein Charles Darwin muses about clover flowers and pollen-carrying bees “becoming modified and adapted in the most perfect manner to each other.” But the modern, systematic study of coevolution really didn’t get underway until the middle of the Twentieth Century. Flor’s studies on the genetics of flax and rust and Mode’s model-building laid the groundwork for thinking about the short-term evolutionary dynamics of interacting species. At about the same time, Paul Ehrlich and Peter Raven compiled information about the relationships between large groups of butterfly species and the plants on which their larvae feed, showing that the associations ran deep into evolutionary history — which suggested that many groups of butterflies diversified into many different species after their ancestors evolved the capacity to penetrate the defenses of a previously inaccessible group of host plants. This suggested that coevolution could shape the formation and extinction of species over millions of years.

In other words, interactions between species — a fundamental consequence of biodiversity — may help to generate more biodiversity.

Those of us who study coevolution have spent much of the decades since those early studies trying to understand how, exactly, this happens. Studies of the reconstructed historical relationships between living species, have shown that groups of species engaged in highly specialized interactions tend to be more diverse than related groups of species that lack such interactions. Experimental and ecological studies of contemporary populations, however, have often failed to pinpoint the kind of selection that can help to create new species.

Consider one iconic example. Some orchids attract pollinating moths with nectar, offered at the bottom of a long, tubular spur, which runs back from an opening below the stamens and pistil. A moth must reach its proboscis down the spur to reach the nectar, bumping against the flower as it does so and picking up or depositing pollen. The spur, then, must be long enough that the moth has to stretch to reach the nectar, to make contact with the flower; but moths with longer probosces can collect more nectar with less effort. The result should be an “arms race” of coevolutionary selection favoring first longer spurs, then longer probosces, and so on — Darwin famously followed something like this logic to predict that an orchid with a dramatic, foot-long nectar spur must be matched by a moth with a similarly long proboscis, decades before anyone found such a moth.

A sphinx moth, Xanthopan morgani, with a very long proboscis, as predicted from an orchid. By Esculapio, via Wikimedia Commons.

A sphinx moth, Xanthopan morgani, with a very long proboscis, as predicted from an orchid. By Esculapio, via Wikimedia Commons.

While that plant-moth arms race might be easy to envision, it’s less easy to see how it might lead to the formation of new species of orchids or moths — both species should evolve in one direction until one or the other reaches some physiological limit, then remain closely matched.

Well, imagine not one arms race, but many. These different races occur in semi-isolated populations scattered across the landscape. Constraints against longer spurs or longer probosces could differ from place to place — maybe orchids grow in some places with poor soil nutrients, which limit their growth, or maybe long probosces make for awkward flight in open, windy habitats. Or maybe some sites have different, more efficient pollinators with shorter tongues, like hummingbirds. Those differences, then, could lead the various populations of orchids and moths to spin off on their own coevolutionary journeys — and, eventually, to form different species with very different nectar spurs and probosces.

This idea, that coevolution may promote biodiversity not in and of itself, but in concert with other ecological and evolutionary forces, is at the heart of John Thompson’s “geographic theory mosaic of coevolution” — meaning mosaics of populations, scattered across patchy, geographically variable environments. Some sites in a mosaic may not even support true coevolution — may not even have both species present — but over the whole system, coevolution interacts with different environmental conditions to promote and maintain new genetic diversity. Thompson, and scientists inspired by his thinking, have started to document how these much more complex processes operate in a variety of living communities, including at least one plant-pollinator interaction quite like Darwin’s orchid.

Fully describing and understanding geographic mosaics of coevolution is intensive work, potentially involving long-term field surveys, carefully designed experiments in natural and controlled settings, and population and molecular genetic studies. Some elements of these have become much, much easier since Flor’s day — in particular, modern DNA sequencing methods allow us to identify and describe specific genes that might be involved in coevolutionary intractions much faster than his laborious controlled crosses. But, as with Flor’s foundational work, modern coevolutionary studies need long hours spent in careful observation spent of the interacting species. To those of us who specialize in species interactions, ferreting out the natural history and evolutionary dynamics of our favorite organisms is all part of the fun.


Darwin C. 1862. The Various Contrivances by Which Orchids are Fertilised by Insects New York: D. Appleton and Company.

Ehrlich, P. and P. Raven. 1964. Butterflies and plants: A study in coevolution. Evolution 18:586–608. doi: 10.2307/2406212

Flor, H. H. 1956. The complementary genic systems in flax and flax rust. Adv. Genet. 8:29–52. doi: 10.1016/s0065-2660(08)60498-8

Grant, V. 1949. Pollination systems as isolating mechanisms in angiosperms. Evolution 3:82–97. doi: 10.2307/2405454

Loegering, W. Q., and A. H. Ellingboe. 1987. H.H. Flor: Pioneer in phytopathology. Annu. Rev. Phytopathol. 25:59–66. doi: 10.1146/

Mode, C. J. 1958. A mathematical model for the co-evolution of obligate parasites and their hosts. Evolution 12:158–165. JSTOR. doi:

Thompson, JN. 2005. The Geographic Mosaic of Coevolution Chicago: University of Chicago Press.

Yoder JB. 2016. Coevolution, Introduction to. In: Kliman RM (ed.), Encyclopedia of Evolutionary Biology. Vol. 1: 314–21. Oxford: Academic Press. doi: 10.1016/B978-0-12-800049-6.00185-2

Zhang F., C. Hui, A. Pauw. 2012. Adaptive divergence in Darwin’s race: How coevolution can generate trait diversity in a pollination system. Evolution 67(2):548–60. doi: 10.1111/j.1558-5646.2012.01796.x

1 Comment

Join the discussion

One Comment

  1. Farida Jalaludin says:

    I did not read the article to the end when this question popped into my head and I had to ask.
    Is this the reason monkey’s still exist?