Earlier I noted that Michael Lynch has doubts about whether natural selection was responsible for the evolution of multicelluarity:
These results, along with our theoretical work on network evolution, challenge the popular idea that modularity arises as a direct consequence of selection for morphological complexity, and by extension raise questions about the common assumption that natural selection was responsible for the emergence of multicellularity.
A couple of years ago, Lynch published a very interesting article entitled, “The frailty of adaptive hypotheses for the origins of organismal complexity” in PNAS (2007) vol. 104, pp. 8597–8604. You can read the article here.
A couple of excerpts relevant to the evolution of multicellularity are below the fold.