SPECIAL ISSUE REVIEW 915 The adaptive evolution of virulence: a review of theoretical predictions and empirical tests CLAYTON E. CRESSLER1,2, DAVID V. MCLEOD1, CARLY ROZINS1,JOSÉEVANDEN HOOGEN1 and TROY DAY1,2,3* 1 Department of Mathematics & Statistics, Queen’s University, Kingston, ON K7L 3N6, Canada 2 Department of Biology, Queen’s University, Kingston, ON K7L 3N6, Canada 3 Fogarty International Center, National Institutes of Health, Bethesda, Maryland 20892, USA (Received 12 May 2015; revised 29 June 2015; accepted 7 July 2015; first published online 25 August 2015) SUMMARY Why is it that some parasites cause high levels of host damage (i.e. virulence) whereas others are relatively benign? There are now numerous reviews of virulence evolution in the literature but it is nevertheless still difficult to find a comprehen- sive treatment of the theory and data on the subject that is easily accessible to non-specialists. Here we attempt to do so by distilling the vast theoretical literature on the topic into a set of relatively few robust predictions. We then provide a com- prehensive assessment of the available empirical literature that tests these predictions. Our results show that there have been some notable successes in integrating theory and data but also that theory and empiricism in this field do not ‘speak’ to each other very well. We offer a few suggestions for how the connection between the two might be improved. Key words: Tradeoff hypothesis, evolutionary medicine, infectious disease. INTRODUCTION permutations and extensions of mathematical models that it has become difficult to see the forest The virulence of parasites is shaped by evolutionary for the trees. While it would be impossible to com- trade-offsatdifferent biological scales. At the host prehensively review all of this theory, we endeavor population scale, parasites that are able to best monop- to highlight some of the key conceptual questions olize susceptible individuals tend to be most success- that have been tackled using theory. As we note ful. However, the ability of a parasite to monopolize throughout, comprehensive reviews of the theoretic- susceptible hosts is intimately tied to its ability to al literature have been written for each of these persist within hosts and to spread effectively between questions. However, we feel that there is much to them. Therefore success at the host population scale be gained from distilling this theoretical literature, is also tied to success at the scale of physiological in particular, to identify broad-scale, robust, interactions within a host (van Baalen and Sabelis, predictions. 1995a;Alizon,2008b; Schmid-Hempel, 2011). Likewise there is an ever-growing empirical litera- At the same time, a parasite’s success at the ture that aims to test the predictions of theory. Some within-host scale is also shaped by interactions of these tests are tied very directly to mathematical with other, co-infecting, parasite strains and models, but more often the connections to theory species. These interactions may be antagonistic, are loose. To highlight the successes and shortcom- when parasites compete for resources or provoke a ings of the match between previous theoretical and cross-reactive immune response, or they may be fa- empirical research, we therefore also discuss the cilitative, when closely related strains cooperate or available evidence for each of the broad-scale theor- cotransmit (Pedersen and Fenton, 2007; Lion, etical predictions that we review. 2013; Alizon, 2013b). Because parasites must ultim- The structure of our review roughly mirrors the ately be successful at both the within- and between- two biological scales of the problem. We begin by host scales, virulence is expected to evolve to balance considering theory based on trade-offs at the these potentially conflicting selection pressures between-host scale. We focus primarily on the (Mideo et al. 2008). trade-off between parasite virulence and transmis- There is now a vast theoretical literature that sion, reflecting the theoretical attention paid here explores virulence evolution and makes predictions (Alizon et al. 2009). In this section, we ignore any about how we expect virulence to evolve under theory that includes either multiple infections different conditions. Indeed there are now so many (e.g. Gandon et al. 2001a) or an explicit dynamical consideration of within-host processes (e.g. André * Corresponding author. Department of Mathematics, Statistics & Biology, Queen’s University, Kingston, ON et al. 2003). We then consider how multiple K7L 3N6, Canada. E-mail: [email protected] infections influence the evolution of virulence, Parasitology (2016), 143, 915–930. © Cambridge University Press 2015. This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribu- tion, and reproduction in any medium, provided the original work is properly cited. Downloaded fromdoi:10.1017/S003118201500092X https://www.cambridge.org/core. IP address: 170.106.202.226, on 24 Sep 2021 at 17:14:56, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S003118201500092X Clayton E. Cressler and others 916 highlighting especially how and why multiple infec- the empirical literature, we will explicitly state how tion can alter predictions based on trade-off theory. virulence was quantified. As will be seen, the Specifically, we review models of superinfection, match between the theoretical and empirical defini- coinfection and kin selection. tions of virulence is often only approximate. Each section of the review is relatively self-con- tained to provide readers with bite-sized chunks of the literature to digest. As will be seen, one of the TRADE-OFFS AND THE EVOLUTION OF take-away messages of this review is that, to a large VIRULENCE extent, theory and empiricism in this field do not Trade-offs between different components of parasite ‘speak’ to each other very well. Experiments fitness provide the dominant conceptual framework seldom measure the traits explored by the theory, for understanding the adaptive evolution of viru- and theory seldom models the traits measured by lence (Alizon et al. 2009). This idea was first intro- empiricists. Although neither empiricists nor theo- duced in Anderson and May (1982) to help explain reticians are particularly to blame for this miscom- patterns in myxomatosis data, and by Ewald (1983) munication, what is sorely needed at this point is a to explain the severity of vector-transmitted tighter integration of mathematical modeling with disease. It arises quite naturally from a consideration empirical research. We close our review by high- of the classic expression of parasite fitness, R0. The lighting recent work in this vein (Mideo et al. R0 expression for a simple SIR epidemiological 2011; Berngruber et al. 2013, 2015). model is illustrative: β ¼ S DEFINING VIRULENCE R0 μ þ n þ γ Before we begin, it is necessary to define what we mean by virulence. The most general definition of Here, parasite fitness is the product of the rate at virulence is the reduction in host fitness caused by which new infections are caused by an infected − infection (Read, 1994). This definition, ironically, host (βS) and the duration of infection (μ + ν + γ) 1 may be a primary reason for the lack of integration (Bremmerman and Thieme, 1989). In this formula- between theory and data in this field: whereas tion, β is the transmission rate, S is the density of fitness can be quantified precisely in a mathematical susceptibles, γ is the rate at which an infected host model, it is exceedingly difficult to measure empiric- clears the disease, μ is the background mortality ally (Metcalf et al. 2015). In mathematical models, rate and ν is the mortality rate due to infection, virulence is quantified as an infection-induced in- often referred to as ‘virulence’. More precisely, β is crease in host mortality rate or reduction in host re- the rate of contact between susceptible and infec- productive rate. In experimental or observational tious individuals multiplied by the probability of data, virulence is often quantified by measures of transmission per contact; it has units of individ- − − ‘harm’ done by the parasite, such as host anemia, ual 1 × time 1. Parasite fitness is increased by in- weight loss, or morbidity, assuming that this harm creasing transmission (i.e. increasing the value of β) is correlated with negative impacts on host fitness. and/or by prolonging the infection (e.g. decreasing However, the relationship between these metrics mortality ν or clearance γ). Trade-off theory and host mortality is not typically straightforward. assumes that a parasite cannot simultaneously in- Even more troubling, common empirical measures crease transmission and prolong infection, and so of infection-induced mortality (such as case mortal- parasites are attempting to maximize R0 subject to ity rate or lethal dose) do not have a simple relation- these constraints. Although within-host processes ship with the theoretical measure (instantaneous may not be explicitly represented in the models, epi- mortality rate); as such, the evolutionary response demiological trade-offs are thought to emerge from of these empirical measures can be opposite in direc- the dynamics of the within-host interactions tion to that of the theoretical measure (Day, 2002a). between the immune system and parasite. Throughout our review of theory, ‘virulence’ will By far, the most widely studied trade-off involves typically refer to the instantaneous mortality rate transmission and virulence (Anderson and May, caused by infection, unless otherwise noted. 1982;Frank,1996; Alizon et al. 2009). Transmission Although parasites often affect host reproduction, and virulence are linked by within-host replication: in- the overwhelming majority of modeling work has creasing parasite abundance increases the likelihood of studied the evolution of virulence as host mortality transmission, but also increases the likelihood of host rate.