News & Views | Published:

Social evolution

Kinship is relative

Nature volume 430, pages 975976 (26 August 2004) | Download Citation


Kinship fosters the evolution of cooperation. However, a once-heretical theory and an unconventional social organism show that the cooperation-enhancing effect of kinship is sometimes negated.

As evolutionary biologist W. D. Hamilton showed more than 40 years ago, selfish genes can lead to cooperation and altruism1. A gene can spread by helping other individuals that carry copies of the gene. This extension of individual selection, called kin selection, works best with the recognition of close relatives, but Hamilton also thought that if the dispersal of individuals is limited, this might build up enough local genetic similarity to favour a less targeted kind of altruism towards neighbours in general. Because local dispersal is very common, this mechanism might greatly expand the range of cooperation in nature. However, later models2,3,4 suggest that this is not necessarily so: when limited dispersal makes neighbours close relatives, it also makes them close competitors, and this can negate the effect of kinship. On page 1024 of this issue5, Griffin and colleagues provide the first experimental test of this effect, confirming that both kinship and the scale of competition matter.

The study is part of a growing trend towards using microorganisms to study social evolution6. Sociobiologists have been slow to recognize that microorganisms have social interactions worthy of study. However, any reader of John Bonner's work on cellular slime moulds7 will realize that these single-celled amoebae interact in interesting ways; some give up their lives to form a stalk that furthers the dispersal of others. Analogous behaviour occurs in Myxobacteria8. And interesting behaviour is not limited to the exotica of the microbial world; the old laboratory workhorse Escherichia coli has strains that aid their own type by wielding poison–antidote systems to destroy other strains that lack the same system9. Perhaps the most common form of microbial cooperation is the secretion of substances outside the cell to do various kinds of work, such as digestion. The work provides a public good, in which the benefits are available not just to the secretor, but to anyone in the neighbourhood. This puts individuals into a social bind. Why work to help others when you can be lazy and freeload off your neighbours? The obvious answer is kin selection. If your effort helps others with the same genes, it is not wasted.

Once it is recognized that microbes have interesting social behaviour, two clear advantages can be exploited. First, many microbes are well understood biochemically and genetically. Second, it is easy to manipulate entire populations of microbes over space and to select them over time. Griffin et al.5 made full use of these advantages in their study of a freeloading mutant of the bacterium Pseudomonas aeruginosa. This ‘cheater’ mutant is deficient in production of a public benefit called siderophores. Siderophores are synthesized and excreted in response to iron deficiency. They then bind iron, making it available for uptake — either by the secretor or by its neighbours. The mutant saves on the cost of producing siderophores but can still obtain iron if it has secretor neighbours.

Griffin et al. selected populations, starting with equal numbers of cheaters and secretors, over six cycles of group inter-action, each lasting about seven generations (Fig. 1). In natural populations with limited dispersal, neighbours tend to be both relatives and competitors, but Griffin et al. were able to vary the relatedness and scale of competition independently. High or low relatedness was imposed by allowing bacteria to grow and interact in groups derived from a single bacterium (a cheater or a secretor) or from two bacteria (initially in a ratio of one cheater to one secretor), respectively. The scale of competition was manipulated during the transfers to form each next cycle of groups. Global competition was allowed by mixing the groups and then choosing random individuals to start the next cycle. The more productive groups thus contributed more to the next cycle. Local competition was imposed by taking an equal number of individuals from each group, without mixing between the cycles. Crossing the relatedness and competition treatments yields a simple two-by-two design that permits the study of both factors and their interaction.

Figure 1: Cheaters and secretors.
Figure 1

Cooperator Pseudomonas aeruginosa secrete a green siderophore, pyroverdin, which makes iron available for uptake by all cells. Cheaters save on the expense of producing pyroverdin but can exploit the siderophores of nearby secretors. Griffin et al.5 tested how the proportion of each varied over time in response to relatedness (high or low) and scale of competition (global or local). As these results show, relatedness is crucial but it is modulated by the scale of competition. (Simplified from ref. 5.)

As expected, at the end of the experiment the high-relatedness treatments yielded higher frequencies of the cooperative secretors than the low-relatedness treatments, demonstrating the importance of kinship in the evolution of cooperation. But the greater novelty of the study comes from another theory-confirming result: the local-competition treatments yielded more cheaters than the global-competition ones. Kinship is crucial, but it is modulated by the scale of competition. Local competition exactly cancelled out the effect of high relatedness, such that neither type was favoured (Fig. 1).

There is an interesting history behind the theories tested here. The first study to challenge Hamilton's ideas about limited dispersal was a computer simulation based on group-selection theory2. This theory divides selection into within-group and between-group components, with the latter favouring cooperation. Group selection has a shady past. At one time, some biologists saw group-selected cooperation everywhere in nature, even though the process was unsupported by rigorous models, and consequently the whole idea was brought into disrepute. But there has been growing agreement that group selection, properly applied, is a roughly equivalent way of understanding social selection — not so much an alternative theory as an alternative means of slicing up the way in which selection occurs. Thus, kin-selection theorists took the computer-simulation results2 seriously and quickly found two different ways of squaring them with their theory. First, one can add in the indirect effects for the cases where helping a neighbour takes away fitness from another, perhaps equally related, neighbour3. Alternatively, relatedness can be rescaled so that it measures genetic similarity relative to the locally competing individuals, rather than to the global population4.

It is the relativistic effect of kinship that Griffin et al. set out to show, and they have succeeded. Curiously, however, their experiment is perhaps more easily understood from a group-selection standpoint. The conditions of low and high relatedness correspond exactly to the presence and absence of within-group selection. The conditions of global and local competition correspond exactly to the presence and absence of between-group selection. The two-by-two crossing of these treatments therefore leads to the most basic group-selection experiment possible. The results confirm that cooperation is favoured by between-group selection and disfavoured by within-group selection.

Does all this mean that we should discard kin selection in favour of the simpler group-selection approach? Hardly. Kin selection has yielded far more insights into the complex behaviours of animals such as social insects. For example, elegant theories of sex-ratio conflict in social insects emerge naturally from kin-selection models, whereas the corresponding group-selection models are so complex that they have not been developed. But this history does suggest that group selection has matured enough for it to become a partner with kin selection in contributing insights, though the story is not without some irony. Once, group selectionists saw cooperation everywhere but were brought down to earth by individual selectionists. Now group selection is being used, not to show the ubiquity of cooperation but to rein in theories on an important form of cooperation envisaged by individual selectionists.


  1. 1.

    J. Theor. Biol. 7, 1–52 (1964).

  2. 2.

    , & Evol. Ecol. 6, 331–341 (1992).

  3. 3.

    Evol. Ecol. 6, 352–356 (1992).

  4. 4.

    Evol. Ecol. 8, 70–73 (1994).

  5. 5.

    , & Nature 430, 1024–1027 (2004).

  6. 6.

    Trends Ecol. Evol. 16, 178–183 (2001).

  7. 7.

    The Cellular Slime Molds (Princeton Univ. Press, 1967).

  8. 8.

    , & Proc. Natl Acad. Sci. USA 95, 12376–12380 (1998).

  9. 9.

    & Trends Microbiol. 7, 129–133 (1999).

Download references

Author information


  1. David C. Queller is in the Department of Ecology and Evolutionary Biology, Rice University, Houston, Texas 77005, USA. e-mail:

    • David C. Queller


  1. Search for David C. Queller in:

About this article

Publication history



Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing