While most scientists agree that selection can occur at the individual level, defining what constitutes an "individual can be a difficult endeavor. Within individual organisms, groups of genes essentially "cooperate" that results in a self-perpetuating coherence between genetic individuals that allows selection to occur at the organismic level. However, this does not mean that conflicts do not occur within the genome and sometimes "selfish genes" become both the interactor and replicator. Nonetheless, organisms are usually a cohesive enough unit to interact with its environment as a whole.
Many scientists agree that "individuality" can occur on many levels and its evolutionary importance is dependent upon the degree of coherence exhibited by the interactor. Similarly, Wilson and Sober believed that this type of individuality could be elevated to the group level. They argued that if the genes can "cooperate" to form individuals, than if groups are able to achieve similar genetic coherence they can also become the vehicles of selection. (Sober and Wilson 1998) Likewise, Marguilis (1992) has termed this phenomenon as "individuality at a more complex level" and as such we must now identify groups that exhibit "individuality" (interacts as a cohesive whole) and whether conditions permit selection to occur at levels higher than the individual.
Brandon argues that groups (and communities) can be the level at which selection is occurring provided that the three conditions (hertiability, variability, and differential reproduction) required for individual selection can be translated to that level. Further, Brandon's views of the units and levels of selection may help resolve the scaling dilemma that scientists encounter when attempting to explain evolution of traits. In his discussion, Brandon uses "units" of selection to refer to replicators and "levels" of selection to refer to interactors. He argues that the unit of selection must be inferred from the level in which selection is occurring. To determine the level at which selection is operating, Brandon utilizes the principle of "screening off". Simply put, the principle of "screening off" states that the level of organization that is most directly (most proximate) responsible for particular effects in a system screens off more distal properties at lower levels of selection from those effects. Therefore, the unit of selection is the level of organization that most directly explains the traits of the interactor (Brandon 1990). This view is contrary to Williams's parsimonious view, that group selection should never be invoked if the phenomenon can be explained by individual selection (Williams 1996).
Interdemic selection occurs in subdivided populations, where there is little or no gene flow between subdivisions (demes). In turn, this genetic isolation increases genetic variance between demes, which can provide the variance (between demes) necessary for selection to occur. The changes in allele frequencies of the population are due to the differential extinction and proliferation of demes. Since little mixing occurs between demes, Brandon likened this process of population fission to asexual reproduction in individuals. As a consequence of this mode of group reproduction, the selected traits exhibit a high heritablity between parent and offspring groups. Further, there is some evidence that suggest that multiple levels of selection should be considered simultaneously. For example, Wade (1978) found a greater rate of change in allele frequency occurred when both group and individual selection were operating in the same direction. If this case is common, than "screening off" lower levels of interactors may limit the understanding of the selective processes. However, Maynard Smith and others suggest that interdemic selection is unlikely to occur in a natural setting. They argue that the low to non-existent gene flow required to form genetically divergent groups would not only increase the extinction rate of demes but also reduce the overall recolonization rate achieved by the selected groups traits (Maynard Smith, 1976; Harrison and Hastings 1996).
Intrademic group selection occurs when the population is subdivided into trait groups during one stage of the individuals' lifetime. After reproduction within their separate trait groups, individuals disperse and mix with the global population. The changes in allele frequencies of the population are due to the global mixing of individuals after the differential reproduction of trait groups. While many have argued that this not group selection because the groups disappear after globing mixing, Brandon (1990), pointed out that this process is similar to sexual reproduction in organism in that daughter groups are a result of blending inheritance which half the between group variance. This reduction in variance is problematic because it can potentially reduce the efficiency of selection. However, if individuals with the same selected trait can positively assort themselves prior to recolonization or there is incomplete mixing within the migrant pool, variability between populations can be maintained. In addition, the degree of mixing of colonizing individuals can fall anywhere along a selection continuum from strictly interdemic (no mixing) to intrademic (complete random mixing).
Community level selection can potentially alter gene frequencies of all species within a community, by selection acting directly upon species interactions. Again the communities must vary in the selected trait, community reproduction must be differential with respect to alternative community traits, and the selected traits must be at least partially heritable. One way community selection may occur is through the extended phenotype of a keystone and dominant species. Whitham's study demonstrates that the extended phenotype of a dominant species can act as the common selective force of an entire community. In his study, hybrids of two cottonwood species were created to form individuals that varied in their ability to produce leaf tannins. This heritable trait explained a large proportion of the differences in litter decomposition rates and nitrogen mineralization rates of the community and thus had a significant impact on the shared community environment. In another study, the genes for tannin production and the genes for aphid resistance interacted to affect other community features such as species richness and species abundances (Whitham 2003). These studies suggest that extended phenotypes of cottonwood (tannin production) have community-wide consequences and as such community level selection seems plausible in a natural setting.
Brandon, R. 1990. Adaptation and Environment. Princeton: Princeton University Press.
Harrison, S. and Hastings, A. 1996. Genetic and evolutionary consequences of metapopulation
structure. Trends Ecol. Evol., 11: 180183.
Low, T. 2002. The New Nature. Penguin Books Australia
Margulis L, 1992, Symbiosis in Cell Evolution: Microbial Communities in the Archean and Ptroterozoic Eons. 2d ed., W.H. Freeman, New York.
Maynard Smith, J. 1976. Group selection. Q. Rev. Biol., 51: 277283.
Sober, E., Wilson, D.S. 1998. Unto Others: The Evolution and Psychology
of Unselfish Behavior. Cambridge: Harvard University Press
Wade, M.J. 1978. A critical review of the models of group selection. Quarterly Review of Biology 53: 101-114.
Whitham,T.G., et al, 2003. Community and Ecosystem Genetics: A Consequence of the Extended Phenotype. Ecology, 84:3 p 559573