Natural Selection of Memory-one Strategies for the Iterated Prisoner's Dilemma