Experimental Demonstration of the Prisoner’s Dilemma

Turner and Chao (1999) illustrated the Prisoner’s Dilemma using bacteriophage φ6, a viral parasite of E. coli. Their ancestral stock of φ6 had been propagated at low density, such that usually only a single phage infected each host. By propagating φ6 for 250 generations at higher density, so that approximately five phage infected each cell, they derived a strain, φH2, which had evolved higher competitive ability at the expense of a lower efficiency of transmission. The competitive advantage of this strain as a function of its frequency was determined to have a roughly twofold advantage over its ancestor when rare and a smaller advantage when common (Fig. WN20.8A). This can be interpreted as a stronger ability to compete for resources produced by the ancestral strain but poorer ability to produce those resources itself. Despite the negative frequency dependence, however, φH2 was at an advantage even when common and hence represents an ESS. Moreover, a pure derived population was less fit than a pure ancestral population. This shows that the full payoff matrix is indeed an example of the Prisoner’s Dilemma.

The fitness of a pure derived population must be shown to be lower than that of a pure ancestral population. This measurement was made using an ingenious trick: Turner and Chao allowed phage in pure culture to bind to host cells and then mixed these infected cells in equal proportions to give a competitive assay in which each virus met only its own type within cells.