Control of Cheaters in the Transition to Multicellularity

Explores the multilevel selection framework, the role of model organisms, and philosophy of science.


We often take for granted the fact that we are superorganisms composed of trillions of cells that happen to be working together. Framed in this way, our existence and the existence of all multicellular organisms seems like an evolutionary miracle. What causes all these cells to cooperate with one another? While this question is easy to answer—cooperation can be advantageous for cooperators—how such cooperation arose and how it has been maintained are much more difficult questions. I will argue that adopting a multi-level selection framework is necessary to explain the transition to multicellularity, using the amoebae D. discoideum as a case study; in doing so, I hope to illustrate the importance of defining the scope of our models in evolutionary biology.

The Major Transitions in Evolution

The transition from unicellularity to multicellularity is an example of what Maynard Smith and Szathmáry termed “the major transitions in evolution” (Okasha, 1013). The challenge of the major transitions is to explain how lower level units can cooperate and form a higher-level unit which is “evolutionarily stable” and able to reproduce (Okasha, 1014). However, because all organisms are self-interested and designed to maximize their fitness, it seems that any attempts at multicellularity would descend into utter chaos and that the major transitions would not be possible. But this need not be the case if we adopt a multi-level selection framework.

Multi-level selection theory is the idea that natural selection can operate on multiple levels—the individual, the group, the species—and that these selective forces often interact with one another. If we want to explain cooperation among organisms, we must look not only at natural selection’s effects on individuals but also at its effect on the group. Sober illustrates this point well, showing that while “selfish” types can be more successful at the level of individual-selection, at the group-level cooperative individuals can actually be fitter on average given the right set of conditions. This is “Simpson’s paradox”, and an understanding of it is critical to multi-level selection theory (Sober, 103).

Intuition Pump: Teams vs. Players

To illustrate this idea, imagine that two teams are playing a basketball game. Team A is composed of selfish individuals who all want to score points, whereas Team B is a cohesive unit in which some players are fine not scoring a lot of points if it means that their team will win. While Team A may have a lot of high scorers, the average score per player on Team B may be higher because they worked together, and thus Team B may win the game given the right set of conditions.

At first glance, how Team B could have evolved to cooperate is puzzling. Organisms are self-interested, so it doesn’t make much sense that they would sacrifice their own fitness (scoring points) for the betterment of the group. This is because we typically think of fitness in terms of direct fitness—e.g. the number of points an individual scores or the number of offspring they have. However, as Hamilton (1964) explained, an individual’s fitness can be measured not only through direct fitness but also through inclusive fitness—the indirect propagation of one’s genes via individuals whom one shares genes with, or in the case of the basketball example, helping your teammates score points. This idea can be used to explain Team B’s cooperation, for individuals sacrifice their own success not just because they care about the team but because they have some stock in the success of the team as a whole.

Okasha’s idea of MLS1 vs. MLS2 is useful in explaining how this type of cooperation could have evolved. In MLS1, the fitness of the collective (the team of individuals) is defined simply as the average fitness of all the individuals (the number of points each person scores) in the collective (Okasha, 1018). This can explain why individuals on Team B win, for being on Team B is of benefit to the individual because the average number of points scored on that team is higher. However, MLS1 can’t explain why teams like Team B tend to win more often than those like Team A, or how this cooperative behavior evolved.

MLS2, on the other hand, is defined by the fitness of the collective itself—i.e. how many other collectives like itself that it produces—not by the average fitness of the individuals that make up the collective. MLS2 is concerned with the frequency of different types of groups. However, for this to be considered natural selection, these group level traits must be heritable, vary, and have a direct effect on fitness. Furthermore, these traits cannot be simply reducible to the individual level, otherwise this would be MLS1 (Okasha, 1017).

Preconditions for Multi-Level Selection

Thus we have reached the basic problem with the major transitions. Making the the transition from unicellularity to multicellularity requires that a group of individuals essentially becomes a superindividual, a “new higher-level evolutionary unit”; to do this, fitness must be transferred from the level of the individual to the level of the superindividual (Michod 2003, 64). Essentially a transition must be made from MLS1 to MLS2, because for this group-level trait to flourish, it must be heritable, vary, and affect the fitness of the collective. Michod and Nedelcu argue that creation of new levels of fitness occurs only through cooperation, and with cooperation arises the opportunity for conflict. The effective management of this conflict “is fundamental to the emergence of individuality at a higher level” (66).

As can be seen, the multi-level selection framework’s importance in the major transitions is largely accepted (Okasha, 1014). However, how such transitions actually occurred has been the subject of much debate. Evolutionary biologists disagree about the relative importance of factors such as control of cheaters and kin discrimination. Using the amoeba D. discoideum as a case study, I will argue that control of cheaters is the major issue that any multicellular organism deals with, and high relatedness among cooperators and evolution in response to these cheaters are the main ways which cooperation is maintained.

Case Study: D. discoideum

D. discoideum lifecycle (source).

The amoeba D. discoideum is typically solitary; however, when food is scarce, many of these amoeba will aggregate, forming a slime mold in order to move to areas with more food (Gilbert 2007). This behavior is an interesting case study in evolutionary biology because these slime molds can be composed of many different amoebas that are not genetically identical. Heterogeneous multicellular organisms like this are referred to as chimeras (Foster 2002). However, most multicellular organisms go through a single-cell bottleneck, meaning that all the cells that make up the organism originate from a single cell, and therefore all these cells are (essentially) genetically identical before undergoing differentiation. This is how our body works: all our cells originate from one cell, the zygote, and therefore relatedness among these cells is 1. Multicellularity with high genetic relatedness among cells is so common and chimeric multicellularity so uncommon because it prevents most intercellular conflict, as cooperation is easy when everyone has essentially the same interest.

Are Slime Molds Organisms?

Could these multicellular slime molds be considered organisms? According to Strassman and Queller’s definition of organismality, yes. They argue that organisms are simply groups of cooperating elements that have minimal conflict between them. D. discoideum, even though it doesn’t go through a single-cell bottleneck, still can be considered a multicellular organism when many of these amoebae group together, given that conflict is at a minimum (Strassman and Queller, 2010, 608).

A field of slime molds. Credit: Owen Gilbert (source).

Reproduction and Cheating (of a Different Sort)

Before we discuss how conflict is managed in chimeric multicellular organisms, we must first define the specific type of conflict we will be discussing: cheating. After D. discoideum have aggregated and formed a motile slime mold, in order for the organism to reproduce, a fruiting body must be formed. This fruiting body has a stalk-spore structure in which around 20% of cells must sacrifice themselves to become part of a stalk on which the reproductive bodies of the organism, the spores, are released. The stalk is important because it allows the spores to be spread farther, and thus gives the fruiting body a better chance of successfully reproducing. While the cells which form the stalk receive an inclusive fitness benefit from the cells that form the spores, the sporus (where the spores are located) is prime real estate, for as a spore you get to reproduce and receive direct fitness benefits.

In general, cheating has to do with the expectation of fairness, which is fundamental to any sort of cooperation (Strassman and Queller 2011A). In the case of D. discoideum, fairness means that any specific genetic strain of D. discoideum has a representation in the sporus (the reproductive unit of the fruiting body) which is proportional to its representation in the fruiting body overall (2011A). Cheating in this case would mean being disproportionately represented in the sporus, and not contributing anything to the stalk—essentially receiving all the benefits of cooperation without contributing anything yourself. Certain cheater mutants are better at getting into the spore than the wild type cells, and though these mutants receive an individual fitness benefit, they reduce the cohesiveness of the group and thus decrease the group’s overall success, threatening cooperation (Strassman and Queller 2011A).

What is Cooperation?

But we need to get a bit more specific about what cheating and cooperation are. Ghoul et al. define cooperation as “a behavior that is favored by selection, at least in part, from its beneficial effect on the recipient” (Ghoul 2013, 320). Ghoul et al. also draw a distinction between facultative and obligate cheaters; the former will only cheat in certain situations, depending on its social partners, whereas the latter will always cheat. The type of cheater present is important to the success of a cooperative group, because “cooperation can be maintained with facultative cheats, but will be lost if obligate cheats spread to fixation” (322). However, when obligate cheaters do spread to fixation, these groups cannot reproduce because none of the individuals are willing to sacrifice themselves to be the spore; thus, from a group-level perspective, obligate cheating is not advantageous in certain populations because it is negatively frequency dependent (326). In D. discoideum, facultative cheaters have been found to cheat in chimeric populations but contribute to the group and produce a normal fruiting body in clonal populations (324).

Game Theory of Slime Molds and Empirical Evidence

High relatedness is one way of defending against cheaters, because as mentioned before, sharing a common interest is a good way of ensuring cooperation. Within D. discoideum populations, “high relatedness means that cheaters and cooperators will tend to be in different groups, which both limits opportunities for cheaters to exploit cooperators, and exposes any groups-level defects of cheaters to selection” (Gilbert 2007, 8913). In other words, the negative frequency dependence of certain types of cheaters (obligate) is exposed by group-level selection, for even if obligate cheating were to sweep to fixation, the groups they would form would be unable to reproduce because no individual would be willing to be part of the stalk. This is not to say that cheaters never survive; when still in a non-social state, cheaters do just as well as cooperative individuals. However, when conditions become stressful and D. discoideum become social and aggregate, a group of obligate cheaters will be unable to, over the long run, maintain cooperation and successfully reproduce as a group.

Gilbert et al. found that in D. discoideum a cheater mutant “devastates cooperation at low relatedness but does not spread at high relatedness” (8913). In fact, when relatedness is .25 or greater in a chimeric D. discoideum group, cheater mutants cannot invade and spread (8914). This supports the idea that high relatedness is one defense against cheating, and this intuitively makes sense; as Hamilton argued, the degree of relatedness among individuals in a population is critical for kin selection to be maintained, and the same is true in these fruiting bodies.

Experiments done by Levin et al. have suggested that cheaters are also dealt with by evolving defenses in response to them, creating a sort of co-evolutionary arms race among cooperators and cheaters. Levin et al. found that “resistance [to cheaters] evolved after obligate cheaters emerged but before they swept through the population,” suggesting that the cooperators must have evolved in response to these cheaters (2015, 758). They competed three populations against one another: an obligative cheating strain, the cooperative ancestors of the cheating strain (who were cheated by the cheating strain), and a novel strain that had never encountered the cheating strain. They found that while the cheaters were able to cheat—i.e. increase their representation in the sporus of the fruiting body—over multiple generations of interacting with their ancestors, no such successful cheating occurred when the cheating strain was mixed with the new population. This suggests that “fruiting clones [i.e. cooperators] that have evolved in the presence of nonfruiters [i.e. obligate cheaters] are resistant to the evolved nonfruiters’ cheating” (762). Furthermore, the authors found that in a number of these tests, not only did the evolved fruiters resist cheating, but they also cheated the cheaters (764). As the authors write, these findings provide “experimental evidence for the maintenance of at least a simple form of multicellularity by means other than high relatedness” (756).


D. discoideum has been explored as a case study of how high relatedness and evolution in response to cheaters can facilitate cooperation. However, even if a simple form of multicellularity can be maintained via these mechanisms, this doesn’t necessarily mean that these mechanisms can, or did, drive a major evolutionary transition. West et al. draw a distinction between obligate multicellularity—essentially multicellularity that results from a single-cell bottleneck—and facultative multicellularity—the type we see in chimeric D. discoideum slime molds. They argue that the former is the only type of multicellularity which can result in a major transition, for while control of cheaters is important to the success of facultative multicellular organisms such as D. discoideum, these features really don’t matter in obligate multicellular organisms because they already have high relatedness, and therefore low levels of conflict. Thus, West et al. argue that control of cheaters has played only a minor role in the major transition to multicellularity (2015, 10118).

This seems to make sense, given that facultative multicellular organisms such as D. discoideum have only evolved to the point of being slime molds, whereas obligate multicellular organisms have been evolutionarily quite successful. West et al. argue that this is because “it is hard to evolve the complete repression of competition that would be required for a major transition because the marginal benefits of repressing competition will often plateau, such that an intermediate level of repression will be favored” (10118). In other words, no matter how successfully cheaters are suppressed in facultative multicellular organisms, there is always going to be some degree of competition which prevents a major transition from being evolutionarily stable. According to West et al., any real major transition to multicellularity must be a transition to obligate multicellularity, and this can only occur in a clonal population.

However, Michod and Nedelcu (2003) disagree, arguing that conflict meditation is key to major transitions: “Through the evolution of conflict modifiers, developmental programs evolve so that heritability of fitness at the group level may increase leading ultimately to the creation of a new evolutionary individual” (72). But maybe these two views are not mutually exclusive. As we have seen, the high relatedness of clonal populations is actually a means of eliminating conflict; thus, in clonal populations the “conflict modifiers” which Michod and Nedelcu discuss are already built in. Furthermore, Michod and Nedelcu aren’t explicitly claiming that development of “conflict modifiers” will result in a major transition; rather, they argue that they are integral in the transfer of fitness from a lower to a higher level organism. Therefore, these two views may actually support rather than oppose each other.

The D. discoideum case study is illuminating because it shows that our models can work in certain situations but not others. The cheater suppression model is useful in explaining the challenges associated with multicellularity via aggregation, but not necessarily multicellularity via fragmentation or single-cell bottleneck (Michod and Roze 2001, 4). Thus we must understand the scope and usefulness of our models. While control of cheaters and kin discrimination seem necessary to facilitate cooperation in multicellular organisms which arise via aggregation, these factors may not be as important in other types of multicellularity, and the D. discoideum model may not be of use in those cases. That being said, there are certain elements of our theory, such as taking a multi-level selection approach, which most evolutionary biologists seem to agree upon, and which are useful in all our models. The exploration of the major transitions illustrates how evolutionary explanations are often retrospective in nature, and thus speculative. While we can get experimental data on model organisms such as D. discoideum, we will never be able to go back in time and witness the major transitions, and thus we are ultimately speculating and theorizing. This means that we will probably be debating the relative importance of different factors in the major transitions for quite a long time.


Foster, Kevin R., Angelo Fortunato, Joan E. Strassmann, and David C. Queller, “The Costs and Benefits of Being a Chimera,” Proc. R. Soc. Lond. B 269 (2002): 2357ꟷ2362.

Ghoul, Melanie, Ashleigh S. Griffin, and Stuart A. West, “Toward an Evolutionary Definition of Cheating,” Evolution 68 (2013): 318ꟷ331.

Gilbert, Owen M., Kevin R. Foster, Natasha J. Mehdiabadi, Joan E. Strassmann, and David C. Queller, “High Relatedness Maintains Multicellular Cooperation in a Social Amoeba by Controlling Cheater Mutants,” PNAS 104 (2007): 8913ꟷ8917.

Hamilton, W. D., “The Genetical Evolution of Social Behavior I,” Journal of Theoretical Biology 7 (1964): 1ꟷ16.

Levin, Samuel R., Debra A. Brock, David C. Queller, and Joan E. Strassmann, “Concurrent Co-Evolution of Intra-Organismal Cheaters and Resisters,” Journal of Evolutionary Biology 28 (2015): 756ꟷ65.

Michod, Richard E., and Aurora M. Nedelcu, “On the Reorganization of Fitness During Evolutionary Transitions in Individuality,” Integrative and Comparative Biology 43 (2003): 64ꟷ73.

Michod, Richard E., and Denis Roze, “Cooperation and Conflict in the Evolution of Multicellularity,” Heredity 86 (2001): 1ꟷ7.

Okasha, Samir, “Multilevel Selection and the Major Transitions in Evolution,” Philosophy of Science 72 (2005): 1013ꟷ1025.

Sober, Elliott. 2000. Philosophy of Biology. Boulder, CO: Westview Press.

Strassmann, Joan E., and David C. Queller, “The Social Organism: Congresses, Parties, and Committees,” Evolution (2010): 605ꟷ616.

Strassmann, Joan E., and David C. Queller, “How Social Evolution Theory Impacts Our Understanding of Development in the Social Amoeba Dictyostelium,” Development, Growth and Differentiation 53 (2011A): 597ꟷ607.

Strassmann, Joan E., and David C. Queller. “Evolution of Cooperation and Control of Cheating in a Social Microbe,” PNAS 108 (2011B): 10855ꟷ10862.

West, Stuart A., Roberta M. Fisher, Andy Gardner, and E. Toby Kiers, “Major Evolutionary Transitions in Individuality,” PNAS 112 (2015): 10112ꟷ10119.

(Image credit: Nat Geo)