The Evolution of Cooperation

When and why individual organisms work together at the game of life, and what keeps cheaters in check

Jan 1, 2016
R. Ford Denison and Katherine Muller

ANT LADDER: Nestmates will climb on top of their nest mates to achieve collective goals, such as reaching food.© VOLKER MOHRKE/CORBIS

Evolution by natural selection, Darwin wrote, mainly depends on “success in leaving progeny.”1 He also recognized that such success may be achieved by “dependence of one being on another.” When are individuals most successful living on their own, and when can they benefit from working with others?

It’s not always an easy question to answer. For parasites living in or on other organisms, for example, maximizing reproduction is a tricky proposition. Using more host resources lets parasites produce more offspring, but overexploitation shortens host life span, reducing the amount of time the parasites have to reproduce. So it may make sense for parasites to avoid harming their hosts, and parasites that increase host life span may fare even better. As British evolutionary biologist and  geneticist John Maynard Smith noted more than 100 years after Darwin’s musings on reproduction and cooperation, you shouldn’t kill the goose that lays the golden eggs.2

But Maynard Smith recognized that this strategy is based on a critical assumption: that if you do not kill the golden goose, no one else will either. In other words, limiting host exploitation will only benefit a parasite if the host isn’t also inhabited by other, more virulent strains or species. If another parasite is using so many resources that it kills the host anyway, why should any organisms on the same host limit their own reproduction by using fewer host resources? This “tragedy of the commons” type of dilemma, in which individuals benefit from activities that undermine shared benefits, is a major reason why cooperation is not universal.


Whenever there are different genotypes in the mix, cheaters can arise.

Nevertheless, cooperation is found throughout the living world—from the cellular to the societal level. Our cells are descended from single-celled organisms that once competed with or preyed on one another, but now work together to function as a cohesive unit. Within our cells, the mitochondria that provide energy are descended from free-living bacteria that gave up their autonomy for a cooperative existence. Lichens, corals, and many plants host beneficial bacteria or fungi within their bodies and depend on them for vital nutrients; and different species of microorganisms living within a host may be interdependent on one another. Ants defend trees that house and feed them. Animals, from bees to lions, cooperate with close relatives, and human civilization depends on cooperation even among unrelated individuals.

What drives the evolution of these relationships, and why are they not more widespread? And can humans harness cooperative biology for their own benefit, for example, to increase crop yields?

Cooperating with kin

An early example of different species coming together to work as a team—one that changed the course of the evolution of life on Earth—is the origin of the eukaryotic cell. The cells of animals, plants, and fungi all contain mitochondria, which generate energy via respiration. Mitochondria are distant descendants of symbiotic bacteria, surrounded by their own membranes and containing their own DNA. Mitochondria presumably passed through a cooperative-but-potentially-independent stage before becoming completely integrated with their host cells. Today, mitochondria have lost enough genes that they cannot survive and reproduce outside of host cells, solidifying the cooperative relationship.

Interestingly, similarities among all mitochondria suggest that animals, plants, and fungi evolved from a one-time origin of this ancestral symbiosis between two microbial species. This is in stark contrast to the repeated evolution of multicellularity, which has appeared more than 20 times across the eukaryotic kingdom. (See “From Simple To Complex,” The Scientist, January 2011.) This discrepancy highlights how it is easier for cooperation to evolve among genetically identical cells. Indeed, multicellular organisms serve as the ultimate example of cooperation on a cellular level, with millions, billions, or even trillions of cells working together to form the tissues and organs of a complex individual. Such organisms cannot exist without cooperation among their cells, and the cells cannot exist outside of the cooperative system. The surprising ease with which these systems arose may be attributed to the fact that all the cells of a multicellular individual carry an identical (or nearly identical) genome.

In theory, multicellular organisms could have formed as individual cells banding together. For example, single-celled Dictyostelium amoebas and Myxobacteria both form multicellular structures that produce starvation-resistant spores, and bacteria can aggregate into seemingly cooperative multispecies biofilms, which show enhanced resistance to antibiotics. Alternatively, today’s multicellular organisms may have descended from cells that failed to go their own ways following cell division. This process is recapitulated by most modern multicellular organisms, which develop from single-celled embryos that divide repeatedly but remain in contact. This ensures that the cells have a shared genotype.

To determine how the first multicellular organisms may have arisen, my (R.F.D.’s) former student Will Ratcliff and colleague Mike Travisano used artificial selection in 2012 to evolve simple multicellular forms from single-celled yeast.3 Once a day, they separated liquid cultures based on how fast they settled to the bottom of the test tube. Cluster-forming mutants settled faster, and only these faster settlers were transferred to fresh media. Simple multicellularity evolved within a few weeks, and closer analysis of the faster-settling mutants revealed that clusters were formed by cells staying together after division, not by independent cells aggregating together. The same outcome occurred when selecting for clumps formed by yeast strains—such as the flocculating yeasts used for brewing beer, which are known to aggregate under certain conditions4—and when conducting the experiments with algae.5 Collectively, these studies suggest that multicellularity likely arose from incomplete separation following cell division, and support the idea that genetic similarity matters in the evolution of cooperation.

Genetic similarity among multicellular individuals also plays a major role in the evolution of cooperation on the macro level. The great evolutionary theorist William Hamilton noted that a gene for cooperation can spread if cooperation helps others with that same gene to survive and reproduce.6 Close relatives are more likely to share genes, including genes for cooperation, so “kin selection” can favor cooperation. Evolutionary biologist J.B.S. Haldane joked that he would die for “two brothers or eight cousins,” the number of relatives who, on average, would replace his own contribution to the gene pool. Many animals follow this basic philosophy. A worker bee, for example, will give up her own reproductive capabilities and even die defending the hive to help the queen keep laying eggs—which contain the worker’s sisters and brothers. Most of these siblings will have the same hive-defense genes as the dying worker.

Kin selection is also apparent in animals that don’t have such extreme eusocial societal structures. Squirrels, for example, will occasionally adopt orphans, but only when they are so closely related that this is likely to increase representation of the foster mother’s genes in the gene pool.7 Closely related female lions in a pride will cooperate to defend their territory against intruders.

Such systems are not immune to cheating, however.  In the mid-1990s, Robert Heinsohn of Australian National University and Craig Packer of the University of Minnesota found that some lions, while no less closely related, are less likely to fight an intruder, thus reducing their own risk of injury.8 Bolder lions apparently resent this, but they don’t seem to retaliate against the “cheaters.” However, punishment of cheaters is often needed to maintain cooperation among unrelated individuals or between species.

Keeping cheaters in check

KEEPING CONTROL: Rhizobia bacteria live inside root nodules of plants, fixing nitrogen in exchange for energy-rich organic molecules. But if the rhizobia fail to fix nitrogen, the plant will induce senescence of the nodules (blue, far right). ALEX MAYWhenever there are different genotypes in the mix, cheaters can arise. For example, when Dictyostelium cells aggregate to form a fruiting body with spores supported by a stalk, only spore cells produce progeny. When a fruiting body forms from a mixture of two strains, one strain may contribute less to the stalk and more to the spores. For cooperation to evolve in the face of such competition, a system of checks and balances must be in place to guard against cheaters—strains that enhance their own Darwinian fitness at the expense of the others. One way is simply to exclude dissimilar strains from the cooperative group, a practice of at least some Dictyostelium strains.9 Similarly, Pseudomonas aeruginosa bacteria tend to kill nearby strains to create single-strain biofilms.10

Cooperation can also evolve when organisms become mutually dependent on one another, especially if the same individuals interact repeatedly. Among free-living bacteria, if some bacteria unavoidably “leak” expensive nutrients, nearby cells that have lost the ability to make those nutrients might be able to gather them from their neighbors. Without the cost of making expensive nutrients, these mutants might have greater fitness than their nutrient-making ancestors. Researchers at Michigan State and the University of Tennessee have suggested that this could lead to cooperation among species, with each species evolving to make only a subset of the nutrients they all need and getting the rest from their neighbors.11 But such systems are also prone to cheating. Planktonic bacteria floating around in oceans or lakes, for example, have only loose associations with one another, and selection would seem to favor species or strains that use, but do not make, any of these public goods. Why help neighbors who will soon leave? When pairs of bacterial species were mixed in liquid culture, selection favored the less-productive, not the more-productive, species.12 But interactions may not always be random, even among free-living bacteria. Christian Kost of the Max Planck Institute for Chemical Ecology and colleagues have shown that some bacteria connect to other cells, of the same and different species, via nanotubes through which they exchange amino acids.13 (See “Live Wires,” The Scientist, May 2013.) If such connections are common, that would allow cooperation based on reciprocity—trade rather than piracy.

A similar example of interspecies trade can be found in just about every soil ecosystem, where most plant species depend on symbiotic fungi that help them acquire soil phosphorus, and a smaller number of plant species (including legumes) depend on symbiotic bacteria such as rhizobia to convert atmospheric nitrogen into compounds that plants use to make essential proteins. Legumes are unlikely to cheat their bacterial symbionts because the rhizobia bacteria can’t fix nitrogen without the energy-rich organic molecules provided by plants. The nitrogen the rhizobia provide can allow greater host-plant photosynthesis, potentially generating more organic molecules for the rhizobia.

That said, each plant typically hosts several different strains of rhizobia. And just like multiple parasites inhabiting a single host, strains that divert resources to their own reproduction would tend to outcompete strains that put all their energy into the “public good” of host-plant health. One widespread form of rhizobial cheating is hoarding more plant resources for future reproduction, rather than using those resources only to power nitrogen fixation. Even if cheaters supply some nitrogen, they reduce a host plant’s overall health by occupying root nodules that would otherwise be occupied by more-beneficial strains.

But plants have evolved ways to prevent a two-way trade from degenerating into a one-way resource grab. If the bacteria inside one root nodule stop fixing nitrogen, the plant can shut off the oxygen supply to that nodule, limiting rhizobial reproduction. The best evidence that plants respond to rhizobial behavior comes from experiments by my (R.F.D.’s) group, led by former students Toby Kiers and Ryoko Oono. Comparing soybean and alfalfa root nodules in normal air to nodules on the same plant in an atmosphere with only traces of nitrogen, we found that rhizobia reproduced less frequently when they could only fix enough nitrogen for their own needs, with no surplus for the plant.14 Soybean plants reduced oxygen supply to rhizobia that didn’t supply them with nitrogen. This presumably limits rhizobial metabolism so they waste fewer plant resources and may also explain their decreased reproduction. Similarly, plants supplied less energy to mycorrhizal fungi that provided them with less phosphorus.15  Without such sanctions by the plant host, strains that diverted resources to their own reproduction would displace more-cooperative strains over the course of evolution.

Some hosts manipulate their partners in ways that enhance current cooperation. Alfalfa and some other legume species cause rhizobia in their root nodules to swell to two or more times their usual size. Swollen rhizobia can no longer reproduce, but we (Oono and R.F.D.) found that they fix more nitrogen, relative to their cost to the plant.16 Similarly, researchers in Mexico and Germany found that Acacia cornigera trees protected by Pseudomyrmex ferrugineus ants manipulate the ants to keep them loyal. The nectar they give the ants contains chemicals that prevent the ants from digesting nectar from other plants. Individual ants apparently learn to stay on their host plant.17 This sort of manipulation can ensure that partners continue to cooperate with their current hosts. But cooperation based on manipulation may lapse whenever manipulation does, and thus does not necessarily favor the evolution of cooperation over generations. Sanctions that reduce the frequency of cheaters in future generations may have longer-lasting benefits.

Obligatory cooperation

MAINTAINING COOPERATION: For cooperation between species to withstand the inherently selfish nature of evolution, individuals that fail to cooperate must have fewer descendants than cooperators, on average. This could result from fitness-reducing sanctions against cheaters or strict dependence of each partner on the other for survival. Partners may also manipulate each other in ways that enhance cooperation in the short term, without necessarily favoring evolution of cooperation over generations. Among related individuals, kin selection favors cooperation with related individuals that are likely to also carry the same genes for cooperation. These mechanisms for enhancing cooperation are not always foolproof, however.
See full infographic: WEB | PDF
Another way to reduce cheating in interspecies relationships is to increase mutual dependence. When symbionts lose genes needed for survival outside their host, they cannot escape and may evolve to be even more beneficial, especially if their next host is their current host’s offspring.

Aphids, for example, rely on symbiotic bacteria contained in specialized cells for essential amino acids lacking in their diet of sugary plant sap. In return, bacteria gain access to their host’s offspring by entering aphid egg cells, being ingested by the offspring, or other mechanisms of transmission. Such symbiont inheritance, known as vertical transmission, means that bacterial strains benefit from helping their host lay as many eggs as possible. Thus, the most beneficial symbionts become the most frequent in the host population.  

Even in these systems, however, cheating can arise. When a host carries different strains or species of vertically transmitted bacteria, they may compete with each other to reach the host’s offspring. The winners in such within-host competitions will not necessarily be those that are most beneficial to the host, unless the host has specific mechanisms for favoring more-beneficial strains. The problem of competition between symbionts is somewhat alleviated by the fact that a very small fraction of symbiotic bacteria reaches the next generation of hosts. This bottleneck means that a hypothetical mutation that allows a strain to gain a slight advantage over a competitor by exploiting the host does not greatly improve its chances in reaching the next generation. To accomplish this advantage, a strain would have to exploit its host enough to reduce total egg production—with negative consequences that could outweigh the benefits of occupying a larger fraction of those eggs.

Harnessing cooperation

Our own research focuses on the problem of mediocre rhizobia strains that provide soybeans or alfalfa with some nitrogen, but much less than the best strains. While host sanctions keep root symbionts that provide little or no nitrogen or phosphorus in check, rhizobia may not trigger sanctions until they reduce their nitrogen fixation rate by more than 50 percent of their potential.18 Breeding soybeans and other legume crops for stricter sanctions could increase yields significantly, while still relying on symbiosis rather than fertilizer. Within a few years, we should have enough data to tell whether this approach will work.

Whenever there are different genotypes in the mix, cheaters can arise.

In other systems, it may be beneficial to reduce cooperation. While research into the interspecies relationships of the bacteria, fungi, and protozoans living in and on the human body is still in its infancy, recent theoretical work has suggested that microbial cooperation causes instability of species networks, and that competition reduces cooperation and promotes network stability.19

The key to harnessing such cooperative relationships is to understand them at the most basic biological levels. Continued research on within- and between-species cooperation will be necessary to make the most of our social world. 

R. Ford Denison is an adjunct professor at the University of Minnesota. Katherine Muller is a PhD student in his lab.


  1. C.R. Darwin, On the origin of species by means of natural selection, or the preservation of favoured races in the struggle for life, 1st ed., London: John Murray, 1859, page 62.
  2. J. Maynard Smith, “Generating novelty by symbiosis,” Nature, 341:284-85, 1989.
  3. W.C. Ratcliff et al., “Experimental evolution of multicellularity,” PNAS, 109:1595-1600, 2012.
  4. J.T. Pentz et al., “Clonal development is evolutionarily superior to aggregation in wild-collected Saccharomyces cerevisiae,” in H. Sayama et al., eds., Artificial Life 14: Proceedings of the Fourteenth International Conference on the Simulation and Synthesis of Living Systems, (Cambridge, MA: The MIT Press, 2014), 550549-554, 2014.
  5. W.C. Ratcliff et al., “Experimental evolution of an alternating uni- and multicellular life cycle in Chlamydomonas reinhardtii, ” Nat Commun, 4:2742, 2013.
  6. W.D. Hamilton, “The evolution of altruistic behavior,” Am Nat, 97:354-56, 1963.
  7. J.C. Gorrell et al., “Adopting kin enhances inclusive fitness in asocial red squirrels,”  Nat Commun, 1:22, 2010.
  8. R. Heinsohn, C. Packer, “Complex cooperative strategies in group-territorial African lions,” Science, 269:1260-62, 1995.
  9. E.A. Ostrowski et al., “Kin discrimination increases with genetic distance
  10. in a social amoeba,” PLOS Biol, 6:e287, 2008.
  11. N.M. Oliveira et al., “Biofilm formation as a response to ecological competition,” PLOS Biol, 13:e1002191, 2015.
  12. J.J. Morris et al., “The black queen hypothesis: Evolution of dependencies through adaptive gene loss,” mBio, 3:e00036-12, 2012.
  13. K.R. Foster, T. Bell, “Competition, not cooperation, dominates interactions among culturable microbial species,” Curr Biol, 22:1845-50, 2012.
  14. S. Pande et al., “Metabolic cross-feeding via intercellular nanotubes among bacteria,” Nat Commun, 6:6238, 2015.
  15. E.T. Kiers et al., “Host sanctions and the legume-rhizobium mutualism,” Nature, 425:78-81, 2003.
  16. E.T. Kiers et al., “Reciprocal rewards stabilize cooperation in the mycorrhizal symbiosis,” Science, 333:880-82, 2011.
  17. R. Oono, R.F. Denison, “Comparing symbiotic efficiency between swollen versus nonswollen rhizobial bacteroids,” Plant Physiol, 154:1541-48, 2010.
  18. M. Heil et al., “Partner manipulation stabilises a horizontally transmitted mutualism,” Ecol Lett, 17:185-92, 2014.
  19. E.T. Kiers et al., “Measured sanctions: Legume hosts detect quantitative variation in rhizobium cooperation and punish accordingly,” Evol Ecol Res, 8:1077-86. 2006.
  20. 19.    K.Z. Coyte et al., “The ecology of the microbiome: Networks, competition, and stability,” Science, 350:663-66, 2015.