A common circumstance in which payoff values may be interdependent in a Prisoner's Dilemma-like situation can be seen between parasites and their hosts. From the perspective of a parasite, taking resources from their host is beneficial. That is, defection against the host provides (presumably) a bigger payoff for the parasite than cooperation with the host. I say "presumably" because otherwise one would expect the relationship to evolve towards a form of commensalism, or even mutualism, were cooperation with the host to provide a larger fitness payoff to the symbiont than a defection behavior against the host. At the same time, too much taking on the part of the parasite can result in declines in the potential for the host to continue to support the parasite. Indeed, the host might even die, thereby, depending on the parasite, eliminating the potential for the host to continue giving at all. Contrast predation where host death does not interfere with the fitness of the antagonist.
One can view the potential of a host to continue to support a parasite as a function of the host's health, where the ability to continue to support the parasite, to the same degree, would decline as a host's health declines. The host, for example, might be more susceptible to predation, or otherwise less able to obtain resources, thereby degrading the host as a resource for the parasite. Host health also can be measured in terms of host fitness. Parasite-induced disease, for example, might degrade a host's ability to survive, again such as by reducing the host's ability to run away from predators, or (also again) degrade a host's ability to subsequently convert gathered resources into host progeny. The resulting host fitness, with versus without parasite impact, can represent the host's payoff in an evolutionary Prisoner's Dilemma (especially S with parasite impact versus something better represented by R without parasite impact).
Thus, by defecting too greatly against its host, a parasite may gain in the short term and do so explicitly at the expense of its host, thereby receiving the equivalent of a Prisoner's Dilemma payoff of T for unilateral defection. In the longer term, to the extent that a parasite's fitness is tied to that of its host, and so long as the parasite's taking behavior results in declines in host health, then the parasite by overly defecting ultimately might see declines in its own fitness. These postulated declines, it is important to point out, are a result of declines in host health rather than a result of the host for example immunologically fighting back against the parasite. That is, these are reductions in the payoff associated with T rather a transition from a payoff of T to one of P, though the latter is possible as well. Thus, the parasite gains by taking from the host, essentially by cheating in what otherwise could be a potentially cooperative interaction of symbiont with host, but the gains associated with this cheating can decline as the cheating takes its toll on the host as a resource to the symbiont.
Commonly, the impact of a parasite on its host's health or fitness is described as virulence. Virulence tends to vary depending on circumstances, including in terms of host genotype as well as host phenotype. The only thing completely under parasite control, from the perspective of evolution, however, is the parasite's own genotype. Thus, one can speak of an evolution of virulence in which a parasite's propensity to cooperate or defect with its host is optimized in such a way so as to enhance parasite fitness. The resulting virulence will reflect the parasite's infection modes where, especially, parasites that can benefit from host health will be expected to evolve a lower virulence – perhaps resulting instead in the evolution of commensalism rather than parasitism – whereas those parasites that do not benefit from host health will not be expected to evolve lower levels of virulence. Instead, the parasite might benefit particularly from some optimal level of disease. Note the implicit assumption that parasite evolution may be ongoing. Indeed, parasite evolutionary tendencies towards virulence optimization can respond to changes in environmental circumstances (e.g., host density in the environment) plus, of course, to host genotype or host physiological responses to the parasite, particularly anti-parasite immunity.
In addition to optimizing the relationship between parasite and host, the parasite often must also contend with a second Prisoner's Dilemma-like situation, which stems from the behaviors of coinfecting parasites. This is particularly where short-term defection by parasites against the host may provide a competitive advantage to the defecting parasite at the expense of the long-term fitness of the infecting parasite population. This then is the Parasite's Dilemma: To display increased virulence at the expense of long-term, between-host fitness versus reduced fitness at the expense of within-host intraspecific competitiveness.
Note the similarity of the Parasite's Dilemma, when viewed solely from the parasite's perspective, to the Tragedy of the Commons (see "Games of Cooperation and Defection: Tragedy of the Commons"), where the commons is the host's body and virulence is a measure of the degree that parasites exploit the body environment toward their individual gain. Note too that we have already seen essentially the equivalent conflict when discussing the evolution of cooperation in biofilms, that is, with tradeoffs between microcolony growth rates and bacterial ability to found new microcolonies (see "Specific Circumstances: Biofilms and Cooperation"). Furthermore, in terms of antagonistic pleiotropy, this Parasite's Dilemma perspective on the evolution of virulence is very similar to previously considered potential conflicts between growth and transmission as may be observed, for example, during serial passage experiments (see "Serial Passage").
For most of the 20th century, the prevailing dogma was that disease organisms eventually should evolve toward benign coexistence with their hosts; harmful diseases were interpreted as a transitory state of maladaptation… This view arose more from assumptions about the harmony of nature than from rigorous application of evolutionary principles. Specifically, it failed to cast the problem in the context of natural selection. Rather than asking whether harmful or mild variants would win out in competition with each other over the short run, the focus was on what was stable over the long run. Natural selection [however] is powerless to favor long-term stability if the variants that win in the short term destabilize the system. Natural selection may favor the evolution of extreme harmfulness if the host exploitation that causes this harm enhances the competitive success of the harmful variants over benign variants in the short run. If predator-like variants of a pathogen population out-produce and out-transmit benign variants, benign coexistence may be precluded. — Paul W. Ewald (2004)
The concept of virulence, historically, has come from medicine. Because the emphasis of medicine is on patient health, virulence, understandably, has been viewed from the patient's perspective rather than that of the parasite. That is, the question of what's in it for the parasite would seem to be obvious – it gets a home, it gets food – and thereby under most circumstances the issue need not be pondered further. Besides having a patient-centered rather than parasite-centered perspective, medicine also is an applied discipline. As such, there has been a tendency, at least historically, for identifying what works rather than why it works, and when why questions were asked it was often proximate rather than ultimate explanations that were sought. The result has been less of an emphasis on evolutionary theory, in medicine, and much less evolutionary theorizing as it pertains to parasites as opposed to the patient herself. Early musings on virulence evolution consequently took more of a seat-of-the-pants approach, with less rigor than one typically would prefer. Furthermore, even evolutionary theory itself was perhaps not fully up to the task of explaining parasite virulence evolution, even as late as 100 years after the publication by Darwin (in 1859) of "The Origin of Species".
As a result of these otherwise reasonable circumstances, ideas on virulence evolution got off to a rocky start. The chief concern was a failure to view the relative strength of individual versus group selection. In modern evolutionary biology, a default assumption is that individual selection is stronger than group selection. That is, what is good for the individual often will prevail over what is good for the group (hence the "dilemma" in the Prisoner's Dilemma). It is especially when behaviors seem to run counter to this rule that evolutionary ecologists take note. In taking an inherently group selectionist perspective, an implicit assumption was made by early musers on virulence evolution that what is good for the species must be good for the individual. From that assumption, it is clear that an obligate parasite that drove its host (or hosts) to extinction would itself go extinct. Therefore, a "smart" parasite would not do that, implying a tendency for evolution to reduce parasite virulence towards commensalism. Why then do highly virulent parasites exist? They must be newly acquired parasites by a given host species, parasites that have not yet evolved a reduced virulence, right?
The above musings do have a certain logic to them. Unfortunately, they also are very unhelpful towards gaining an understanding of just how parasite virulence evolves. This is because the important question is not what is good for the (parasite) species, but instead what is good for the individual (parasite). What is good for the parasite? In a word: transmission (Ewald, 1994) . Since all hosts eventually die, a successful parasite must be able to move from infected host to (especially) not-yet-infected hosts. Transmission thus can be viewed as the measure of parasite fitness. Virulence optimization, too, can be viewed in terms of a larger consideration of optimization of transmission.