Above I have provided an outline of the theoretical basis for understanding the evolution of parasite virulence. I have however also suggested that what will occur during actual virulence evolution is going to be dependent on specific circumstances. Though it is not within the scope of this book to provide numerous detailed examples of parasite virulence evolution, nonetheless in this section I delve into a handful of illustrative examples, beginning with relatively simple systems and progressing toward increasing complexity. These examples include, in order, the virulence displayed by phages against host bacteria, the virulence displayed by viruses against other viruses (i.e., evolution of defective interfering particles), the evolution of strategies of host dissolution (e.g., as seen with baculovirus), parasite movement between hosts that is mediated by other entities such as water or insect vectors, the complication of contact-mediated (between infected and not-yet infected host) parasite transmission strategies, and requirements for especially intimate contact (as seen with sexually transmitted diseases). Subsequently I consider pathogen evolution from a genomics perspective as well as parallels between cancer progression and parasite virulence evolution.
Phages illustrate a number of generalizations pertaining to parasite virulence, plus a few unique perspectives of their own. The most important of these perspectives may be in distinguishing the concept of virulence in terms of absolute host impact versus rapidity of host impact, plus in defining the concept of host itself (which here I will define as equivalent to a bacterial culture, which I justify in the following two paragraphs). In addition, in terms of phage virulence it is relatively simple to appreciate how bypassing a usual transmission mechanism can result in the evolution of greater parasite virulence. I begin, though, with a brief introduction to the history of phage virulence since otherwise my subsequent discussion will only confuse those who have greater familiarity with phages.
Phages have been known to science for approximately 100 years (Abedon et al., 2011) and in that time they have both informed and been informed by biology as a whole. In other instances phage biology seems to have diverged from the precepts adhered to by biologists in general. The concept of phage virulence is perhaps the best example of this latter tendency, though this was not always so. Phage virulence, originally, seems to have been a concept that was little different from the common idea that it may be measured in terms of host health. For early phage biologists, however, the bacterial culture, perhaps more so that the bacterium itself, was viewed as the phage host. Thus, phage infection can be viewed as a culture "infection", and the relative health of that culture a function of the inherent virulence of the added phages. Some phages when added had a negligible effect on a culture (e.g., no noticeable decline in culture turbidity nor decline in the potential of the culture to be propagated). These phages could be described effectively as avirulent. Other phages, by contrast, gave rise to dramatic drops in culture turbidity. Those latter phages could be described as virulent. In between there exist various phages that have negative impacts on cultures, though impacts that are less dramatic than culture-wide lysis. Such phages display an intermediate level of virulence.
An easy measure of phage virulence involved (and still involves) the addition of a certain quantity of phages to a certain quantity of bacteria, which is then followed for signs of virulence, especially culture-wide lysis. These assays actually have a time component since they involve, effectively, a race between bacterial growth to stationary phase (at which point most phages are no longer active) and phage growth to sufficient densities to infect and then lyse most the bacteria present. Those phages which could still lyse cultures starting from lower phage densities could be described as more virulent. This is because they have a greater potential to effect culture-wide lysis relative to phages displaying lower virulence. Certain phages also exist that do not lyse bacteria, i.e., the filamentous phages such as phage M13. Other phages are able to establish potentially mutualistic relationships with bacteria, particularly temperate phages such as phage λ. Especially in the latter case, the word virulent has come to describe phages that are obligately lytic, that is, unable to display either lysogenic or chronic infections, versus these other phage varieties. Within the context of potential to achieve culture-wide lysis, this description is reasonably accurate. Nonetheless, for those who are familiar with phage biology may be confused by the idea that the concept of phage virulence might have something to do with concepts of parasite virulence in general. It does, especially historically, and I apologize for that confusion.
I limit the rest of this section to considerations of virulence evolution as experienced by obligately lytic phages, that is, phages that upon successful infection display a lytic cycle and only a lytic cycle. Note that the action of these phages, when functioning well, can lead to the culture-wide lysis of susceptible bacteria. Therefore, differences in virulence can be measured in terms of likelihood of the occurrence of such lysis and/or the rapidity with which phage populations can grow to effect that lysis. Two other perspectives may be helpful in understanding these ideas. The first is that culture-wide lysis is definitively an example of a population exceeding an environment's carrying capacity. Thus, the more virulent a phage is then the greater its potential to exceed the carrying capacity of a bacterial culture (as a phage environment), i.e., as equivalent to defection in the context of a Tragedy of the Commons, or the destruction of the body as a parasite's growth environment. The second, additional perspective returns us to the ideas of economy versus expediency. That is, under certain circumstances phages can tradeoff shorter generation times, and thereby faster population growth, for lower overall yields. Indeed, it may always be possible for a phage population to display greater restraint, resulting in greater population fitness (as measured in terms of overall fecundity), but this restraint potentially comes at the expense of within-culture phage population growth rates.
Phage virulence evolution, in this regard, can be illustrated in terms of competition as it can occur between two otherwise identical phages. One displays a shorter latent period and smaller burst size (shorter generation time but lower fecundity) while the other displays a longer latent period and higher fecundity. I will call these phage variants expedient and economical, respectively. This system is affected by within-culture host density, where expedient phages will tend to display relatively faster population growth at higher host densities (relative to economical phages) whereas economical phages will display relatively faster population growth at lower host densities. For a given higher host density there will exist three expected outcomes (Abedon et al., 2003) : Economical phages growing alone will produce more phages overall, but at a slower rate; in terms of a Prisoner's Dilemma, this scenario represents a mutual cooperation (i.e., R) situation. Expedient phages, by contrast, will produce fewer phages overall, but at a faster rate than economical phages, which represents the mutual defection behavior (P). As one might expect, it is during mixed-population growth that the equivalent of unilateral defection and unilateral cooperation will occur, with the defectors (expedient phages at higher host densities) out competing the cooperators (T > S). This growth-rate advantage is the only competitive advantage for the expedient phages since their productivity is lower than that of the economical phage when both are growing alone (R > P). A representation of this experiment is presented as a previously viewed figure (see "Cooperation: Specific Circumstances: Cooperation Evolution in Batch Culture"), which I reprint below.
Note that this yield or fecundity advantage may be realized only to the extent that phage populations are provided with sufficient time to "mature" prior to transmission, which in this case is so long as transmission between cultures is delayed until after culture-wide lysis. This is because sooner transmission may select instead for more-rapid attainment of higher phage densities rather than higher end-point phage densities (Abedon, 2008) . It would be this pre-maturation transmission that can be viewed as equivalent to bypassing a usual transmission mechanism during serial passage. Note that the effect of pre-maturation transmission, for phages, is to create a growth environment that is effectively infinite in size, i.e., one that lacks limits on phage population growth and therefore within which selection acts primarily on within-culture phage population growth rates.
Another simple system relevant to parasite virulence evolution, and one that also can be seen with phages (Turner and Duffy, 2008) , is the evolution of defective interfering particles (DI particles). These are viruses that display reduced fitness (traditionally reduced all the way to zero) during individual infections of host cells or during coinfection with other DI particles. Alternatively, during mixed infections the fitness of DI particles is greater than that of the coinfecting wild-type virus. In terms of the Prisoner's Dilemma-type payoff values, these would be mutual-defection-type situations (DI particle only, P) versus mutual cooperation (wild-type only, R) versus unilateral cooperation (wild type but with DI particle coinfecting, or S) and unilateral defection (DI particle with wild type infection, or T). Note that in most instances this is not strictly a Prisoner's Dilemma since mutual defection (DI particle only, or P) has a lower fitness than unilateral cooperation (wild type coinfecting with DI particle, S). Instead, it can be more accurately described as a Game of Chicken (Nowak and Sigmund, 1999) .
In instances of mixed infection, the DI particle can viewed as the more virulent. This designation comes in part because it is certainly more expedient to tradeoff fecundity, absent coinfection with wild type, for greater competitive ability given such coinfection. Closer to the idea of virulence, however, rather than simply in terms of cooperation and defection, the DI particle can be viewed as an obligate parasite of wild-type infections. That is, the host of a DI particle is not the same host as the host of a wild-type virus, but instead DI particle hosts are limited to wild type-augmented infections (Turner and Duffy, 2008) . One in fact can view this greater virulence in terms of antagonistic pleiotropy. That is, in the course of evolving greater competitive ability in one circumstance (coinfection with wild type viruses), DI particles have declined in abilities in another circumstance (i.e., infection absent the wild type virus). Thus, DI particle evolution is ripe for interpretation as an example of evolution of greater virulence, i.e., as previously discussed.
What circumstances tend to give rise to evolution of DI particles? In fact they occur in response to high-multiplicity virus infection. That is, under circumstances where bottlenecking is not operating. In these instances infections can be non-clonal. Furthermore, under circumstances where the clonality of infections is intentionally repressed, DI particles are especially competitive, and this repression can be viewed as a diminishment of what otherwise could be seen as the bottlenecking life stage. Thus, DI particles evolve, as parasites to wild-type particles, especially when otherwise protective, virulence-reducing effects of clonality-generation bottlenecking are avoided.
Note that there exists a related phenomenon of viruses that require coinfection with other viruses in order to successfully infect. In these instances the parasitic virus (a hyperparasite) typically negatively impacts the fitness of the virus it is parasitizing. However, unlike the DI particle, this virulence is not an example of evolution of intra-specific (or intra-population) defection, but instead is equivalent to parasite-host interactions in general. This is because the parasitic virus is not related to the parasitized virus in the same manner that a DI particle is related to its wild-type parent. In fact, the hyperparasite may be little related to the parasitized virus at all.
Many viruses lyse their host cells to release progeny. Because of this, these viruses obligately reduce the fitness of free-living, single-celled hosts to zero (or, at least, have this effect given successful infection). Other viruses are released chronically so don't have as dramatic an impact on host-cell fitness. Nonetheless, the process of chronic release, in and of itself, can lead to reductions in host fitness. In both cases, with lytic as well as chronic release, we can describe the route of parasite transmission itself as having a virulent impact on single-celled hosts.
We can scale these ideas up to parasites of multicellular organisms where the processes involved in parasite exit from the host causes the host damage. In these cases, enhancement of transmission frequencies may be limited by the damage that transmission causes to the host. These limitations are either a consequence of pathogen restraint, i.e., such that fitness is optimized via some intermediate display of virulence or, alternatively, virulence is so extreme that hosts die. The former is analogous, presumably, to the chronic release of viruses from individual cells whereas the latter we can view as having an equivalence to lytic infections also of individual cells.
The classic examples of parasites of multicellular organisms that display this more "lytic" infection strategy are such things as baculovirus, anthrax infections, and perhaps to a more limited extent ebola virus. Baculovirus infects insects. Within insects, such as caterpillars, it replicates until high densities are found in insect tissues. Then it causes the insect's body dissolve, creating a puddle of high-density virus. This virus adheres to wherever the insect remained associated with at the time of its death. Since often this is in association with the insect's food source, simple eating by subsequent insects results in acquisition of the virus, and the cycle continues. Key to the success of the baculovirus strategy is the durability of the virus while it waits for another insect to infect. Note that fitness is presumably a function of rates of accumulation of virus numbers produced prior to insect killing, the duration of time until that killing, the potential for the insect to die in a location that is optimal for subsequent virus transmission, and the post-death durability of the virion. Tradeoffs presumably exist, such as between rapidity until virus release and total virus produced.
Anthrax, transmitted by Bacillus anthracis spores, provides a strategy that is equivalent to that of baculovirus. Here it is grazing animals that are most likely infected, typically by ingestion. In contrast to baculovirus, B. anthracis has a much wider host range: baculovirus is typically species specific while anthrax can affect various grazing animals, their predators, and, of course, also humans. Animal death is effected by various toxins produced by the vegetative bacterium. Once the host has died, vegetative cells bleed out of the animal, maturing into highly durable spores that resist inactivation. The animal's death, ideally for the parasite, is within an environment frequented by other potential hosts. Because the spores are so resistant to inactivation, the life cycle of the bacterium can span decades, though does not necessarily. As above, tradeoffs likely exist between the rapidity of animal death and total spore production. Note that Clostridium botulinum and indeed other otherwise saprophytic clostridia also can kill their hosts, continue replicating, and then produce spores.
Ebola is a very different story, at least upon infecting humans. Here, especially as with baculovirus, the portal of exit involves release within various bodily fluids. Transmission of the virus, however, requires contact with fresh tissues. As a consequence, cultural practices can exacerbate transmission likelihood, while isolation can limit transmission. Because of its high virulence but relatively limited potential to move between hosts, ebola virus epidemics tend to be spectacular but nonetheless short lived. Indeed, the virus reservoir does not even appear to be humans. Instead, this virus seems to be limited as a human pathogen by its mechanism of transmission, which effects behavioral changes among potential hosts that tend to place limits on transmission. Unlike baculovirus and B. anthracis, as well as many viruses of single-celled organisms, ties between transmission and host death does not necessarily make for a highly successful parasite, at least in this case of humans infected by ebola virus during outbreaks prior to the West African-centered epidemic or indeed pandemic of 2014.
Chronic parasite shedding giving rise to host morbidity seems to be more common than mechanisms that tend to result more directly in host death. Examples include diarrhea, or even ebola virus-mediated bleeding out (which is not always lethal and which also is associated with diarrhea). Parasites that are spread by coughing also, usually, are not lethal (though as with diarrhea, of course, some respiratory parasites fairly often do give rise to host mortality). The cough itself, however, has the effect of decreasing host health and therefore represents a direct tie between pathogen transmission and pathogen virulence. We can view these less lethal, more chronically released infectious diseases in terms of the parasite generally deriving a benefit in terms of transmission ability from ongoing host viability, that is, such that potential new hosts, for example, may be coughed upon. An important exception is cholera, where rates of mortality are fairly high, plus rapidly occur. In this instance, however, the mode of transmission may be enhanced by this relatively high level of virulence, not just because diarrhea gives rise to increased pathogen exit from infected hosts, but also because the mechanism of Vibrio cholerae transmission involves water contamination rather than contact between hosts. Note that many of the ideas discussed in this and the following section are adapted from those summarized by Ewald (2004) ; see also Ewald (1994) .
Greater parasite durability, during transmission, can result in tolerance for greater virulence, and this is particularly so for sit-and-wait parasites, which can lie where their host died until another host comes along. Durability thus can be viewed as an extra-host (that is, outside of the host) enhancement of transmissibility, one with potential virulence consequences. Assisted movement outside of the host is another means of extra-host enhancement of parasite transmissibility. Such enhancement can be mediated by arthropod vectors or by vehicle transmission (such as via water or air). By allowing parasites to more readily reach hosts, either spatially or over longer periods, one would predict that greater virulence may be tolerated by the parasites, e.g., as seen with baculovirus. In addition, the medium through which extra-host movement occurs itself may be viewed as a means of durability enhancement, e.g., such as in evolved associations with arthropod vectors, or because water prevents damage due to desiccation.
An additional mechanism may operate given vector- or vehicle-mediated transmission, and that is that an infected host may no longer need to be ambulatory to effect transmission. While in some instances sick organisms may attract conspecifics, e.g., such as one observes in clinical settings (indeed, as potentially resulting in ebola transmission), in most cases that mechanism does not appear to operate. Similarly, while sit-and-wait pathogens can bide their time until a host's conspecifics accidentally stumble upon the now dead infected host, that presumably is not the most efficient nor effective means of transmission for most parasites. By contrast, what vectors and vehicles provide are means by which parasites can move away from infected hosts, thereby disseminating to other hosts mostly independent of infected-host health. This means that virulence, especially short of host death, may not greatly interfere with parasite transmission potential, resulting in a prospect for the evolution of greater parasite virulence.
In some instances, greater immobility might even result in greater transmissibility, e.g., such as if an individual is more susceptibility to biting insects such as mosquito vectors when sick versus if healthy. Another mechanism can involve greater susceptibility to predation for parasites that can be spread through that route, e.g., such as various parasitic worms that can weaken their would-be-prey hosts. Yet another mechanism, as can be observed in the laboratory, is a bypassing of normal routes of transmission. Ultimately, though, while these mechanisms of enhanced transmissibility may allow for tolerance of a greater virulence on the part of the parasite, they don't guarantee an evolution of high levels of virulence. More generally, virulence evolution is multifactorial, and while general principles likely operate, so too can they conflict, resulting in a utility to better understanding parasite characteristics before speculating excessively on what factors might impact the optimization of their virulence.
A combination of vectors, parasite, and afflicted-host complexity can create a very complicated selective scenario, as the following quote suggests for the malarial parasite, Plasmodium spp. (Mackinnon and Marsh, 2010) , p. 869:
Ultimately, selection acts on whole-organism phenotypes (traits), such as asexual replication rate, transmissibility, persistence, and virulence. Combined, these traits determine the parasite's fitness, that is, the total number of successful transmissions to a new host that the parasite achieves within the duration of the infection of its current host, roughly calculated as the rate of transmission (transmissibility) multiplied by the duration of infection (persistence). Virulence, defined [in the quoted article] as the propensity to cause host death, shortens the duration of infection, and so acts negatively on parasite fitness. Studies in a laboratory model of malaria, P. chabaudi in mice, have shown that the intrinsic asexual replication rate of a parasite strain is the primary determinant of transmissibility, persistence, and virulence, and therefore underpins parasite fitness… Similarly, in natural infections of P. falciparum, asexual parasite density is related to transmissibility and persistence…, although the contributions of parasite strain and host immunity cannot be separated in this case. … Among the many sources of selection acting on the malaria parasite, five dominate: immunity, vector availability, host death, drugs, and coinfection. If the host dies (which occurs in about 1% of cases), it usually does so during the acute phase of the illness before the transmissible forms (gametocytes) appear in the bloodstream, thus causing the parasite to lose all of its transmission potential… Some drugs, if used effectively, have the same effect. Host immunity reduces transmission potential by curtailing asexual replication, thereby reducing both gametocyte production (transmissibility) and the duration of infection (persistence). Ineffectively used drugs can do the same. Immunity and drugs, when effective, can prevent host death and rescue the parasite [the latter by preventing host death]. Thus, there are both fitness benefits (avoiding host death) and costs (reduced transmission potential) of immunity and drug use. By contrast, a loss of vector availability is always costly to the parasite: When the population density of the mosquito vector is reduced by seasonal factors, the fitness cost is from curtailed infections, whereas in nonseasonal transmission areas, the fitness cost is from reduced transmissibility.
Direct-contact transmission involves the actual touching of an infected host by a second, susceptible host. In addition, there exists droplet transmission which, like direct-contact transmission, requires relatively close interaction between individuals, with droplet transmission not requiring actual contact so much as proximity. Indirect-contact transmission, like direct-contact transmission, requires close spatial proximity, but not temporal proximity, since such transmission is mediated by inanimate objects, e.g., such as doorknobs. What all of these mechanisms require is that the infected host be present in the vicinity of potential hosts over the period that the infected host is shedding parasites. This means either that potential hosts must move to come into contact with the shedding host, or the shedding host must move to come into contact with potential hosts.
The idea of individuals coming to the infected individual I have already considered above under the headings of "Bleeding (or Squirting) Out" and "Vectors and Vehicles". For example, baculovirus or anthrax spores sit and wait for potential hosts to come to them, while in the clinic nosocomial infections can be passed from patient to patient as medicated by healthcare worker "vectors" whose job it is to actively seek out sick individuals. By contrast, in this section I walk through the logic of virulence evolution in the face of requirements that parasite transmission is enhanced by infected hosts mingling with others. In the following section I take this idea one step further in considering not just mingling but a requirement for highly intimate interactions (i.e. sex) for parasite transmission.
Mobility typically requires health, and for parasites that are transmitted intraspecifically, mobility may be required for sufficient mingling between conspecifics to assure parasite transmission. Such a requirement can set up conflicts between parasite virulence and parasite transmission. Similar conflicts can arise if infected individuals come to be actively shunned by others (which, in principle, may be viewed as having the effect of increasing the distances that must be traveled to effect adequate mingling). My favorite way of illustrating this point is to describe an illness that impacts health to the point that one feels obliged to avoid attending work or school. By contrast, consider an illness that has less of an effect on health. With greater virulence there may be a greater parasite load and potentially greater parasite shedding, but less access by the host to conspecifics. On the other hand, with less virulence, there may be lower parasite loads and less parasite shedding, but, by allowing infecting individuals to socialize, greater opportunities for transmission.
One can view this situation simplistically in terms of parasite diffusion (outside of the host), distance between hosts, and rate of parasite inactivation (also outside of the host), i.e., as can equivalently be applied to the evolution of phage virulence (see Phage Virulence). The farther parasites must diffuse to find new hosts then the lower their density since diffusion has the effect of reducing density while increasing the volume occupied. As a consequence, even large parasite numbers as found at their site of origin (the infected host from which they are being released), can be greatly reduced in terms of density given a requirement for substantial movement. Even directed movement will not be perfect, and time delays will result in parasite losses. Thus, intermingling has an effect that is equivalent to increasing host density, resulting in less travel time and travel distance, so that greater parasite densities may be received by susceptible hosts. With contact transmission, these distances can be reduced to zero, or nearly to zero, thus the incentive for some parasites to limit virulence to the point that such contact may be occur, even at the expense of overall levels of parasite shedding.
One can make social policy decisions based on these ideas, such as by urging individuals who are even a little bit sick to stay at home, thereby reducing transmission opportunities. Such policies might even lead to the evolution of reductions in the virulence displayed by a parasite, e.g., if only people who don't feel sick (but nevertheless are shedding) are able to effect parasite transmission then evolution will favor those parasites that have an only minimal impact on host health. Note, however, the requirement by such a mechanism that individuals actively become infected with this more benign strain for natural selection to actually work to reduce parasite virulence. That is, practices that serve to eliminate transmission entirely would constitute a stronger form of hard selection, potentially driving a parasite to extinction rather than serving to reduce pathogen virulence.
Sexual transmission is a special case of direct-contact transmission where, of course, that contact is particularly intimate. For parasites whose transmission is mostly or entirely limited to sexual means, there exist limitations on virulence that may be viewed as both equivalent to and extensions of those seen simply with direct-contact transmission. That is, not only must the infected host remain ambulatory, but in fact must remain sexually motivated (and/or receptive). Furthermore, not only must infected individuals present themselves as sufficiently disease free that others are willing to be in their presence, but also that others are willing, indeed interested in having sex with them. That is perhaps a fairly tall order for an infectious disease. It is very illustrative, however, of the power of virulence evolution. In fact, a number of diseases exist that are sufficiently asymptomatic, but nonetheless infectious, that they are not just persistently present within populations, but in some cases, in human populations, can be growing in prevalence. Of course, it doesn't hurt the cause of these parasites that sex drives can be sufficiently compelling, and important evolutionarily, that protection against sexually transmitted diseases often is of only secondary concern. (Or that humans both wear clothing and have a propensity to engage in sexual relations in the dark!)
Were sexually transmitted diseases to retain these properties lower virulence throughout their course, then they presumably would be of less concern. In displaying lower virulence than many other diseases, however, sexually transmitted diseases also tend to display extended spans over which they may be transmitted. Thus, part of the challenge to these parasites is that they also must persist despite immune systems. Their success in persistence, combined with the window of especially promiscuous sexuality being fairly short in human societies (multiple years rather than multiple decades), means that eventually individuals can end up with diseases that no longer have much stake in maintaining an ability to be transmitted, but nonetheless are still well equipped to evade elimination by immune systems. These diseases can be relatively benign over relatively long periods, but nonetheless eventually give rise to substantial morbidity and even mortality years after the start of the infection (e.g., tertiary syphilis or AIDS).
Ewald (2004) speculates that the virulence of sexually transmitted diseases may be modified, toward lower virulence, by decreasing sexual promiscuity, since that would demand of the sexually transmitted parasites longer spans of benign infection before changes in partners occurred. Of course, as with any scheme for the evolution of reduced parasite virulence, there must exist a host population that is infected with the reduced-virulence parasite for selection to favor those individuals over less-benign variants. Thus, in the case of sexually transmitted diseases, the etiology would have to thrive among individuals who practice relatively long-term but nonetheless serial monogamy. This population would have to strike a perhaps difficult balance between becoming infected and passing it on to others, while at the same time not being too likely to become infected nor too likely to pass the disease on to others, since the latter would imply levels of promiscuity that might not achieve an evolution of reduced pathogen virulence. In other words, for many infectious diseases of humans, it often appears that parasite extinction and parasite virulence modification may be achieved not only with similar potentials but, indeed, in the course of employing similar approaches, e.g., such as by reducing especially unprotected sexual promiscuity in the case of sexually transmitted diseases.