Levels of Selection

So far we have presented a brief look at systematics, highlighting some issues of what systematics studies and what issues systematists face when mak­ing inferences about the distant past.

Traditionally, Darwinians have understood selection as acting primarily at the level of the organism. For them, it is the differential survival and reproduction of individual organisms that drives the evolutionary process. There are alternative views, however. Advocates of group selection argue that groups of organisms, rather than individual organisms, may some­times function as levels of selection; “genic selectionists” such as Richard Dawkins (1976) argue that the true level of selection is in fact the gene; while proponents of multi-level selection (Wilson 1975, Wilson and Wilson 2007) argue that natural selection can occur simultaneously at more than one hierarchical level. Does natural selection act on organisms, genes, groups, colonies, demes, species, or some combination of these?

The levels-of-selection question arises because, in principle, the process of natural selection can operate on any population of entities that satisfies three fundamental requirements, first articulated by Darwin in On the Origin of Species and later elaborated more formally by others, including Richard Lewontin (1983) (see §4 above). The first is that the entities should vary in their traits - they must not all be alike. The second is that the variants should enjoy differential reproductive success - some must have more offspring than others.

4.1 The problem of altruism

Debates over the empirical facts and what they might mean has been intimately bound up with the problem of altruism, because altruism is a very clear case in which the level of selection really matters for understand­ing and explaining the biological world and for evaluating the quality of present evolutionary theory. In evolutionary biology, “altruism” refers to any behavior that is costly to the individual performing the behavior, but benefits others, where the costs and benefits are measured in number of offspring, the units of reproductive fitness. Altruism in this sense is com­mon in nature, particularly among animals living in social groups, but prima facie, it is hard to see how it could have evolved by natural selection acting on organisms. By definition, an animal that behaves altruistically will secure fewer resources and have fewer offspring than its selfish counterparts, and so will be selected against.

One solution to this puzzle, first suggested by Darwin (1871) himself, is that altruism can evolve by selection at the group level. It is possible that groups containing many altruists will out-reproduce groups containing mainly selfish organisms, even though within any group, altruists do worse. In principle, altruism and other group-beneficial behaviors might evolve by natural selection acting on groups, rather than organisms.

Cogent though this argument is, it has been regarded with skepticism by biologists, especially since the publication of G.C. Williams' Adaptation and Natural Selection (1966), in which Williams was very critical of what now is sometimes referred to as “naive group selection” thinking. One reason for doubt about the fact and importance of group selection is that many have argued that the puzzle of altruism can be solved in ways that need not invoke group-level phenomena. According to one influential view, the inclusive fitness or kin selection approach first developed by W.D. Hamilton (1964) provides a superior explanation of how altruism evolved.

The basic idea behind kin selection is straightforward. Consider an animal that behaves altruistically, for example by sharing food with others. This behavior is individually disadvantageous, so cannot easily evolve by selection. However, if the animal shares food mainly with its close relatives, rather than with unrelated members of the population, then the behavior can evolve because relatives share genes. In this scenario, there is a certain probability that the recipient of the food will also possess the gene that “causes” the sharing behavior. In other words, if altruistic actions are directed toward kin, the beneficiaries of the actions will them­selves be altruists with greater than random chance, and so the altruistic behavior will spread.

Hamilton described these relationships formally in what has come to be known as “Hamilton’s rule.” The simplest statement of this is b > c/r, where b is benefit conferred by the altruist, c is the cost incurred to the altruist, and r is the coefficient of relatedness between the entities.

Despite Hamilton’s important contributions, the issue of group selec­tion has not been fully settled. Some theorists hold that kin selection, far from being an alternative to group selection, is in fact a version of group selection, expressed in different language and using different mathematical models. This issue is partially (though only partially) semantic (cf. Sober and Wilson 1998). Moreover, kin selection can only explain the existence of altruism directed towards relatives, but there are well studied cases of unrelated organisms (those in which r is very low) forming cooperative groups of varying degrees of integration.^[110] Finally, some recent theorists have stressed that individual organisms are themselves groups of cooperating cells, while each cell is a group of cooperating sub-units, including organelles, chromo­somes and genes (Michod 1999). Since cells and multi-celled organisms clearly have evolved, with sub-units that work for the good of the group, it cannot be true that group selection is of negligible importance in the history of life, according to this argument.

From what has been said so far, the levels question may sound purely empirical. Given that natural selection can operate at many different levels of the biological hierarchy, surely it is just a matter of finding out the levels(s) at which it does act? Surely this is a matter of ordinary empirical enquiry? In fact matters are not quite so simple.

As an example, consider that there is a good deal of debate over how to understand certain biological entities. Are eusocial colonies of insects best understood as groups or as individuals? If they are individuals made up of organisms (individual bees, ants, termites, or wasps) that are parts, it would seem that the individual-group distinction is misleading, and perhaps that models of selection use ontologies that do not properly reflect the organization of the biological world (Wheeler 1911, Wilson and Sober 1989). This comes to a puzzle about colony concepts, akin to the puzzle about species concepts in §3.2 above, but is not peculiar to either of these entities. Similar conceptual problems have come to light in debates about entities at other levels of organization, including groups of related species (Haber and Hamilton 2005, Hamilton and Haber 2006, Okasha 2006) and theoretical early systems of interacting molecules called “hyper­cycles” (Eigen and Schuster 1979, Maynard Smith and Szathmary 1998). The debate in the latter case is precisely about whether hypercycles are best understood as single entities or as collections of interacting individual biological molecules.

This is not to say that there has not been empirical work to test group, kin, and multi-level selection theory. On the contrary, tests of these theories have been conducted (Wade 1976, 1977, Craig and Muir 1995, Muir 1995, Goodnight and Stevens 1997). However, it is often not clear what these studies mean for the debate. As Dawkins demonstrated in The Selfish Gene (1976), it is often possible to recast what appears to be altruism at the level of the organism as selfishness at the level of the gene, and it is hard to see what facts might establish that one or the other interpretation is correct.

5.