The answer is that the origins and evolution of multicellularity do not violate evolutionary theory, but they do require us to approach evolutionary processes more broadly. The first new idea we need to integrate into our theoretical framework is that of inclusive fitness, which is sometimes referred to as kin selection. For the moment, let us think about traits that favor the formation of a multicellular organism - later we will consider traits that have a favorable effect on other, related organisms, whether or not they directly benefit the cell or organism that expresses that trait. Finally, we will consider social situations in which behaviors have become fixed to various extents, and are extended to strangers (humans can, but do not always, display such behaviors). The importance of mutual aid in evolutionary thinking, that is the roles of cooperation, empathy, and altruism in social populations, was a point emphasize by the early evolutionary biologist (and anarchist) (Prince) Peter Kropotkin (1842–1921).
All traits can be considered from a cost-benefit perspective. There are costs (let us call that term “c”) in terms of energy needed to produce a trait and risks associated with expressing the trait, and benefits (“b”) in terms of the trait’s effects on reproductive success. To be evolutionarily preferred (or selected), the benefit b must be greater than the cost c, that is b > c. Previously we had tacitly assumed that both cost and benefit applied to a single organism, but in cooperative behaviors and traits, this is not the case. We can therefore extend our thinking as follows: assume that an organism displays a trait. That trait has a cost to produce and yet may have little or no direct benefit to the organism and may even harm it, but let us assume further that this same trait benefits neighboring organisms. This is like (but not exactly the same as) the fireman who risks his life to save an unrelated child in a burning building. How is it possible for a biological system (the fireman), the product of evolutionary processes, to display this type of self-sacrificing behavior?
Let us consider an examples of this type of behavior, provided by social amoebae of the genus Dictyostelium114. These organisms have a complex life style that includes a stage in which unicellular amoeba-like organisms crawl around in the soil eating bacteria, growing, and dividing. In this phase of their life cycle, the cells divide asexually in what is known as a vegetative cycle (as if vegetables don’t have sex, but we will come back to that!)[→]. If the environment turns hostile, the isolated amoeba begin to secrete a small molecule that influences their own and their neighbor’s behaviors. They begin to migrate toward one another, forming aggregates of thousands of cells. Now something rather amazing happens: these aggregates begin to act as coordinated entities, they migrate around as multicellular “slugs” for a number of hours. Within the soil they respond to environmental signals, for example moving toward light, and then settle down and undergo a rather spectacular process of differentiation115. All through the cellular aggregation and slug migration stages, the original amoeboid cells remain distinct. Upon differentiation ~20% of the cells in the slug differentiate into stalk cells, which can no longer divide, in fact they die. Before they die the stalk cells act together, through changes in shape, to lift the non-stalk cells above the soil, where they go on to form spores. The stalk cells sacrificed themselves so that other cells can form spores. These spores are specialized cells that can survive harsh conditions; they can be transported by the wind and other mechanisms into new environments. Once these spore cells land in a new environment, they convert back into unicellular amoeba that begin to feed and reproduce vegetatively. The available evidence indicates that within the slug the “decision” on whether a cell will form a stalk or a spore cell is stochastic rather than innate. By stochastic we mean that the decision is controlled by underlying random processes, processes that we will consider in greater detail later on. What is important at this point is that this stochastic process is not based on genetic (genotypic) differences between the cells within a slug - two genotypically identical cells may both form spores, both stalk cells, or one might become a stalk and one a spore cell116.
114 Molecular phylogeny and evolution of morphology in the social amoebas: http://www.ncbi.nlm.nih.gov/pmc/arti...1/#!po=38.8889 and A Simple Mechanism for Complex Social Behavior: http://youtu.be/vjRPla0BONA A nice video here: http://youtu.be/bkVhLJLG7ug
115 Behavior of cellular slime molds in the soil: http://www.mycologia.org/content/97/1/178.full
116 This type of behavior occurs in a number of organisms, including the bacteria: see From cell differentiation to cell collectives: Bacillus subtilis uses division of labor to migrate: http://www.ncbi.nlm.nih.gov/pubmed/25894589