Hosts have evolved a diverse range of defence mechanisms in response to challenge by infectious organisms (parasites and pathogens). Whether defence is through avoidance of infection, control of the growth of the parasite once infected, clearance of the infection, tolerance to the disease caused by infection or innate and/or acquired immunity, it will have important implications for the population ecology (epidemiology) of the host-parasite interaction. As a consequence, it is important to understand the evolutionary dynamics of defence in the light of the ecological feedbacks that are intrinsic to the interaction. Here, we review the theoretical models that examine how these feedbacks influence the nature and extent of the defence that will evolve. We begin by briefly comparing different evolutionary modelling approaches and discuss in detail the modern game theoretical approach (adaptive dynamics) that allows ecological feedbacks to be taken into account. Next, we discuss a number of models of host defence in detail and, in particular, make a distinction between 'resistance' and 'tolerance'. Finally, we discuss coevolutionary models and the potential use of models that include genetic and game theoretical approaches. Our aim is to review theoretical approaches that investigate the evolution of defence and to explain how the type of defence and the costs associated with its acquisition are important in determining the level of defence that evolves.

Abstract. Tolerance to parasites reduces the harm that infection causes the host (virulence). Here we investigate the evolution of parasites in response to host tolerance. We show that parasites may evolve either higher or lower withinhost growth rates depending on the nature of the tolerance mechanism. If tolerance reduces virulence by a constant factor, the parasite is always selected to increase its growth rate. Alternatively, if tolerance reduces virulence in a nonlinear manner such that it is less effective at reducing the damage caused by higher growth rates, this may select for faster or slower replicating parasites. If the host is able to completely tolerate pathogen damage up to a certain replication rate, this may result in apparent commensalism, whereby infection causes no apparent virulence but the original evolution of tolerance has been costly. Tolerance tends to increase disease prevalence and may therefore lead to more, rather than less, disease-induced mortality. If the parasite is selected, even a highly efficient tolerance mechanism may result in more individuals in total dying from disease. However, the evolution of tolerance often, although not always, reduces the individual risk of dying from infection.

There is a wide variety of resistance mechanisms that hosts may evolve in response to their parasites. These can be functionally classified as avoidance (lower probability of becoming infected), recovery (faster rate of clearance), tolerance (reduced death rate when infected), or acquired immunity. It is commonly thought that longer lived organisms should invest more in costly resistance.We show that due to epidemiological feedbacks the situation is often more complex. Using evolutionary theory we examine how the optimal investment in costly resistance varies with life span in a broad range of scenarios. In the absence of acquired immunity, longer lived populations do generally invest more in resistance. If hosts have acquired immunity, the optimal resistance may either increase or decrease with increasing life span. In addition, there may be evolutionary bistability with high and low investments in avoidance or tolerance. The optimal investment in the duration of acquired immunity always increases with life span, and due to bistability, shorter lived hosts may commonly not evolve any immunity. In contrast, the optimal investment in the probability of acquiring immunity initially increases and then decreases with life span. Our results have important implications for the evolution of invertebrate and vertebrate immunity, and for the evolution of acquired immunity itself.

No abstract

Many host‐parasite models assume that transmission increases linearly with host population density (‘density‐dependent transmission’), but various alternative transmission functions have been proposed in an effort to capture the complexity of real biological systems. The most common alternative (usually applied to sexually transmitted parasites) assumes instead that the rate at which hosts contact one another is independent of population density, leading to ‘frequency‐dependent’ transmission. This straight‐forward distinction generates fundamentally different dynamics (e.g. deterministic, parasite‐driven extinction with frequency‐ but not density‐dependence). Here, we consider the situation where transmission occurs through two different types of contact, one of which is density‐dependent (e.g. social contacts), the other density‐independent (e.g. sexual contacts). Drawing on a range of biological examples, we propose that this type of contact structure may be widespread in natural populations. When our model is characterized mainly by density‐dependent transmission, we find that allowing even small amounts of transmission to occur through density‐independent contacts leads to the possibility of deterministic, parasite‐driven extinction (and lowers the threshold for parasite persistence). Contrastingly, allowing some density‐dependent transmission to occur in a model characterized mainly by density‐independent contacts (i.e. by frequency‐dependent transmission) does not affect the extinction threshold, but does increase the likelihood of parasite persistence. The idea that directly transmitted parasites exploit different types of host contact is not new, but here we show that the impact on dynamics can be fundamental even in the simplest cases. For example, in systems where density‐dependent transmission is normally assumed de facto, we show that parasite‐driven extinction can occur if a small amount of transmission occurs through density‐independent contacts. Many empirical studies are still guided by the traditional density/frequency dichotomy, but our combined transmission function may provide a better model for systems in which both types of transmission occur.