Nearly all animals in nature are infected by at least one parasite, and many of those parasites can significantly change the phenotype of their hosts, often in ways that increase the parasite’s likelihood of transmission. Hosts’ phenotypic changes are multidimensional, and manipulated traits include behavior, neurotransmission, coloration, morphology, and hormone levels. The field of parasitic manipulation of hosts’ phenotype has now accrued many examples of systems where parasites manipulate the phenotypes of their hosts and focus has shifted to answering three main questions. First, through what mechanisms do parasites manipulate the hosts’ phenotype? Parasites often induce changes in the hosts’ phenotypes that neuroscientists are unable to recreate under laboratory conditions, suggesting that parasites may have much to teach us about links between the brain, immune system, and the expression of phenotype. Second, what are the ecological implications of phenotypic manipulation? Manipulated hosts are often abundant, and changes in their phenotype may have important population, community, and ecosystem-level implications. Finally, how did parasitic manipulation of hosts’ phenotype evolve? The selective pressures faced by parasites are extremely complex, often with multiple hosts that are actively resisting infection, both in physiological and evolutionary time-scales. Here, we provide an overview of how the work presented in this special issue contributes to tackling these three main questions. Studies on parasites’ manipulation of their hosts’ phenotype are undertaken largely by parasitologists, and a major goal of this symposium is to recruit researchers from other fields to the study of these phenomena. Our ability to answer the three questions outlined above would be greatly enhanced by participation from individuals trained in the fields of, for example, neurobiology, physiology, immunology, ecology, evolutionary biology, and invertebrate biology. Conversely, because parasites that alter their hosts’ phenotype are widespread, these fields will benefit from such study.
Nearly all animals are infected by parasites, and many parasites exert profound influences on the phenotype of their hosts (Moore 2002; Hughes et al. 2012). However, the influence of parasites is often unnoticed by researchers. Ethologists and neuroscientists, for example, rarely look for, or consider how, infections might change the phenotype of the organisms they are studying. While not all parasites alter their hosts’ phenotype (e.g., Joseph and Faulkes 2014) nor are all traits of the host influenced by manipulative parasites, it is important to at least understand which parasites infect a focal host species and consider how the parasite may influence the trait in question. An ecologically and evolutionarily relevant understanding of hosts’ phenotype requires understanding how parasites influence it, and studying parasites may uncover previously unexplored links between the brain, immune system, and behavior.
Infection results in changes to the phenotype of the host through one of three main routes: (1) infection induces pathology that may or may not benefit the parasite; (2) the host induces adaptive changes in its phenotype in ways that reduce the costs of infection, or that eliminates the parasite; and (3) parasites may adaptively manipulate the phenotype of their host in ways that increase the probability of transmission (Poulin 2010). The category into which a parasite-associated change in phenotype should fall is not always obvious. For example, Paranjpe et al. (2014) found that side-blotched lizards (Uta stansburiana) infected with the apicomplexans Plasmodium mexicanum or Schellackia sp. thermoregulated at lower than preferred body temperatures and exhibited greater variability in body temperature relative to uninfected individuals, in contrast to some reptiles that respond to infections by increasing body temperatures via “behavioral fever” (Kluger 1979; Firth and Turner 1982; Monagas and Gatten 1983). This result could arise if pathology prohibits lizards from achieving their preferred thermal level, if the hosts adaptively combat the infection by reducing their body temperature, or because of the parasite’s manipulation if an inability to maintain optimal body temperatures makes the lizards more hospitable hosts for the parasites, or more likely to be bitten by the appropriate vector of the parasite. Alternatively, lizards that maintain lower or more variable body temperatures may be more susceptible to infection. Careful study and controlled infections are often required to confirm that the change in host phenotype arose post-infection, and to determine through which of the three routes, in which combinations, phenotypic changes were induced. While the manuscripts in this special issue deal mainly with cases in which parasites manipulate the host’s phenotype, ecologically and evolutionarily important phenotypic changes arise through the other routes as well (e.g., Boze and Moore 2014). Concerns about which category is associated with particular changes in the host’s phenotype may cause us to under-appreciate the population, community, and ecosystem-level implications of those changes. The focus should go beyond categorizing parasite-associated changes in phenotype of the host and should instead focus on how these categories differ in their effects on the evolution of phenotype, and what they can tell us about links between the brain, immune system, and behavior.
Our goal for this symposium was to expose a broad range of biologists to the field of phenotypic manipulation of hosts’ phenotypes, in the hope of recruiting a greater diversity of scientists to tackle emerging questions in this field. Our symposium focused on these emerging questions: (1) through what mechanisms do parasites manipulate hosts’ phenotypes? (2) What are the ecological implications of manipulation? (3) What are the evolutionary implications of infection? Decades of research, conducted mainly by parasitologists, reveal many cases of parasitic manipulation (Moore 2002, 2012; Hughes et al. 2012), and many questions about parasites will only be answered by multi-pronged investigations from scientists with a diverse set of skills. For example, understanding the mechanisms through which parasites induce changes in hosts’ phenotypes will require immunologists, neurobiologists, scientists familiar with “-omics” techniques, and bioinformaticians. Quantifying the ecological and evolutionary implications of parasite-induced changes in hosts’ phenotypes will be forwarded by collaboration among parasitologists, ecologists, evolutionary biologists, and neurobiologists. Below we discuss how participants in the symposium are addressing the three questions posed above, and discuss how an integrative approach to these questions will benefit the study of parasitic manipulation of phenotype while benefiting other fields as well.
Mechanisms of manipulation: How do parasites manipulate hosts’ phenotypes?
Parasites are capable of inducing phenotypic changes in their hosts that are both multidimensional and extremely specific (Cezilly and Perrot-Minnot 2010; Thomas et al. 2010; Cézilly et al. 2013). One of the best known examples is the protozoan parasite Toxoplasma gondii, which infects cats (Felidae) as definitive hosts (i.e., the hosts in which sexual reproduction occurs), but practically any species of warm-blooded animals can serve as an intermediate host (Dubey 2010). In rodents, infection by T. gondii is associated with decreased learning and memory, increased activity, increased time spent in the open, and reduced neophobia (Webster 2007), but the most dramatic consequence of infection is a change in the rats’ responses to the smell of cats’ urine. Uninfected rats avoid the smell of predators’ urine, including cats, but infected rats lose their aversion to the smell of their feline predators; in some cases, rats are attracted to this scent (Berdoy et al. 2000; Vyas et al. 2007; Lamberton et al. 2008). Infected rats still avoid the smell of urine from non-feline predators, showing the specificity of the change. The rats’ fearlessness likely increases the probability of contact with feline predators, which is predicted to increase the probability of T. gondii infecting its definitive host. Kaushik et al. (2014) found that rats infected with T. gondii prefer the smell of urine from wild cats (cheetahs and pumas) over that from domestic cats, which may be explained by wild cats being more suitable definitive hosts or by a longer coevolutionary history between wild cats and T. gondii. This alteration in rats’ behavior appears to involve long-lasting (Ingram et al. 2013) changes in the limbic system (House et al. 2011; Mitra et al. 2013), but neuroscientists cannot recreate this manipulated phenotype in the laboratory. Host-manipulating parasites can be thought of as “evolutionary neuroscientists” (Adamo 2013; Cézilly et al. 2013; Maure et al. 2013), and understanding how parasites manipulate their host’s phenotype will teach us about links between the brain, behavior, and immune system (Helluy 2013; Holland and Hamilton 2013; Libersat and Gal 2013; McCusker and Kelley 2013).
Neurobiologists often see animal behavior as the output of the nervous system, so it is a reasonable hypothesis that parasites would manipulate their hosts by affecting the nervous system. A secondary hypothesis is that the more precise the manipulation, the more likely it is to involve the nervous system. Nervous system output is usually described as being determined by electrical signals within and chemical signals between neurons, and neurobiologists have traditionally leaned more heavily on using electrophysiological techniques to control neurons, and thus behavior. In contrast, there are few documented cases of a parasite affecting its host by generating or altering electrical signals directly, but there are many cases of parasites controlling their host’s behavior by exploiting chemical signals (Lafferty and Shaw 2013).
Jewel wasps’ manipulation of their cockroach host, detailed by Libersat and Gal (2014), is one of the best examples of how a nervous system can be hijacked by another animal using a cocktail of neuroactive chemicals. “Parasite” usually conjures the image of small animals living partly or entirely within the body of a host for most of its life cycle, but parasitoids, like the jewel wasp, are free-living as adults, but lay eggs within a host. Adult jewel wasps manipulate the host in ways that benefit their offspring. Like many parasites, the parasitoid jewel wasp uses neuroactive chemicals that alters the host’s behavior and delivers them by precise and sudden injection of venom. The venom specifically impairs the initiation of voluntary walking in the stung roach, but not other locomotor behaviors, such as escape responses. The cocktails of venom achieve this through the dopaminergic system in the subesophageal ganglion, and almost certainly the brain (supraesophageal ganglion), although research on the latter is not as advanced. Other venoms can be extremely variable (Sunagar et al. 2014), so studying other parasitoid systems is likely to yield entirely new mechanisms of manipulation by venom. Undoubtedly, the success of the jewel-wasp research program lies in part with the fact that cockroaches were intensively studied as subjects for neuroethology, e.g., walking (Ritzmann 1993; Zill 2010), escape (Camhi 1984; Camhi and Levy 1988), and memory (Mizunami et al. 1998), including the locomotor behaviors manipulated by the jewel wasp. This provided a foundation upon which the investigation of manipulation could be investigated.
Like cockroaches, decapod crustaceans have been stalwart model organisms in neuroethology for decades (Wiese 2002a, b), which makes crustacean-parasite systems promising models of understanding the mechanisms of parasitic manipulation. Carreon et al. (2011) and Carreon and Faulkes (2014) examined larval tapeworms infecting the nervous system of shrimp. Because the parasites infect the nervous system, they are in an excellent position to manipulate the host, but this risks disrupting the fine connections between neurons needed for basic functioning. Carreon and Faulkes (2014) hypothesized that parasites living in the nervous system will be distributed non-randomly, perhaps due to the trade-off between control and damage. The larval tapeworms in shrimp appear to fall somewhere between the two extremes. They hypothesize that the position of parasites in the nervous system can suggest whether parasites use targeted (scalpel) or non-specific (shotgun) tactics to affect their hosts’ behavior.
Parasites’ ability to infect, and presumably manipulate, the nervous system, has a trade-off over evolutionary timeframes as well as physiological or organismal scales, as Fredensborg (2014) shows. Helminth parasites that infect the nervous system are limited to a much narrower range of hosts than those helminths that infect other types of tissue, but have larger, more specific effects. Those that infect tissues such as muscle can infect a wider variety of hosts, but the level of manipulation is much less specific. That is, parasites that infect nervous systems might alter higher level behaviors, such as microhabitat choice, while those infecting muscle might induce general lethargy. Across a wider range of parasitic taxa, the location of infection, and the type of host correlates with the mechanisms of manipulation (Lafferty and Shaw 2013).
Another way in which parasites manipulate host’s phenotypes relies less on altering neurons in the nervous system, and more on interfering with connections between the host’s immune and nervous systems (e.g., Helluy 2013; Holland and Hamilton 2013; McCusker and Kelley 2013). Adamo (2014) discusses ways in which sexually transmitted infections (STIs) may exploit immune/neural communication signals that increase the probability of transmission to another host. Infected hosts frequently are avoided by conspecifics, considered less attractive mates, and initiate sickness behaviors that divert energy away from reproduction and toward immune-system functioning. All of these scenarios result in fewer opportunities for transmission for the STI, and thus in some cases selection should act on STIs to mask signs of infection in their host and reduce hosts’ sickness behavior. A final behavioral response to infection by hosts is terminal reproductive investment, putting as much energy as possible into reproduction when future reproductive prospects are dim (e.g., because infection will eventually result in the host’s death). This final behavioral defense may benefit both host and parasite, as it increases fitness of the host given current circumstances and exposes parasites to more potential hosts. These behavioral changes are likely achieved through interactions with the immune and nervous systems and may have important implications for theories of sexual selection, provide important insights for the field of psychoneuroimmunology, and provide practical benefits (e.g., allowing us to suppress sickness behavior and sexual dysfunction).
While many studies have focused on measuring one or a few potential mechanisms at a time, the rise of “-omics” techniques allows us to focus on candidate mechanisms while casting a wide net to capture other mechanisms as well (Biron and Loxdale 2013; Hughes 2013; Hébert and Aubin-Horth 2014). de Bekker et al. (2013) use metabolomics to understand manipulation of ants by fungal parasites, and de Bekker et al. (2014) discuss the importance of controlled infections and controlling for heterogeneity when employing “-omics” techniques. These authors highlight the sensitivity of the transcriptome, metabolome, and proteome to abiotic factors such as time of day, temperature, and humidity and biotic factors (e.g., parasite dose, interactions with conspecifics), and stress the importance of laboratory infections to control for these factors. Removing noise associated with these factors will be crucial for acquiring high-quality data and honing in on mechanisms of manipulation. As these papers show, parasites can alter neural, hormonal, immune, and/or molecular physiology through a wide array of mechanisms. There are surprisingly few systems in which parasitic infection (or, in the case of parasitoids, attack) on hosts can be controlled rigorously, and in which the necessary baseline knowledge of the hosts’ physiology and behavior is present. Thus, understanding the mechanisms of parasitic manipulation remains a challenging field, but one ripe for further study and begging for an integrative approach.
What are the ecological implications of parasites changing their host’s phenotype?
While individual parasites may be small and difficult to see, the combined biomass of parasites in an ecosystem can exceed the biomass of top predators (Kuris et al. 2008). In addition to accounting for an important proportion of the biomass in an ecosystem, the biomass of hijacked hosts can sometimes exceed the biomass of uninfected hosts within a species (Kuris et al. 2008). Phenotype-manipulating parasites can have important impacts on, for example, the dynamics of food webs, apparent competition, and biodiversity of ecosystems, yet ecological implications of manipulating parasites have been studied in few systems and effects are often hypothesized but not quantified (Léfevre et al. 2008; Lafferty and Kuris 2012; Mescher 2012). Not all parasites manipulate their hosts in the strict sense (i.e., changes in the host increase the fitness of the parasite, and are therefore adaptive for the parasite), and both adaptive and non-adaptive changes have important biological consequences.
Parasites that manipulate their host’s phenotype are likely to have important ecological implications when they exert dramatic changes, infect a large proportion of the host population, or when they infect abundant or ecologically important hosts (Lafferty and Kuris 2012). The nematode Physocephalus sexalatus infects an ecosystem engineer, Phanaeus dung beetles, and impacts behaviors that define this insect as an ecosystem engineer (Boze and Moore 2014). Infected dung beetles eat half as much feces as uninfected conspecifics, which lowers the rate at which nutrients are returned to the soil and increases the risk of parasitic infection for other organisms in the ecosystem. Parasites produce eggs that pass with the host’s feces back into the environment, and many of these eggs are killed and thus removed from the environment when ingested by dung beetles (Miller 1961). Additionally, infected dung beetles dig shallower tunnels in which to bury feces, and so infected individuals contribute less to the aeration and fertility of the soil. However, infection is not associated with changes in the behavior classically associated with increased risk of transmission (i.e., increased risk of predation). Instead, the anorexia observed in infected dung beetles may well be part of sickness behavior and may enhance the dung beetle’s defense against parasites. While this example likely does not count as “parasitic manipulation” as the behavioral changes do not clearly benefit the parasite, the effects of the modification of the host’s phenotype on the ecosystem are clear.
From the point of view of a small parasite, the internal milieu of their host is a complete ecosystem in its own right, containing diverse habitats (differing tissues) and potentially several other parasitic species. Thus, understanding interactions between parasites in their hosts is crucial for understanding the parasite’s population and community ecology. For example, competition between parasites within a host can constrain parasites’ sizes and reduce the parasite’s fecundity (e.g., Brown et al. 2003; Fredensborg and Poulin 2005). Exploring the host’s dynamics is particularly interesting when the interests of hosts and manipulating parasites align, as manipulation is believed to be costly and parasites may share the cost of infection (Poulin 1994, 2010). Weinersmith et al. (2014) looked for signs of density dependence in two manipulating trematode parasites of the California killifish (Fundulus parvipinnis), the most abundant fish species in southern Californian and Baja Californian estuaries (Allen et al. 2006; Hechinger et al. 2007). Euhaplorchis californiensis and Renicola buchanani infect killifish and induce conspicuous behaviors that increase the risk of infection by the parasites’ shared definitive host, predatory birds (Lafferty and Morris 1996). While other studies have observed negative density dependence in behavior-manipulating parasites (i.e., a decrease in the parasite’s size as its intensity or density increased in the host) (Brown et al. 2003; Dianne et al. 2012), Weinersmith and colleagues observed no density dependence for R. buchanani and there was positive density dependence for E. californiensis. Two major differences between previous studies and this study of killifish are that the parasites infecting killifish are small relative to their host (and are therefore less likely to experience limitation of resources); the magnitude of the expression of the manipulated phenotype increases with increasing density of parasites in killifish. Euhaplorchis californiensis may have been able to achieve a larger size in the presence of conspecifics by investing less in manipulation at high densities. Controlled infections are needed, but this study suggests that within-host dynamics do not limit the population dynamics of E. californiensis and R. buchanani, and that positive density dependence may help to explain why E. californiensis is the most abundant trematode parasite in the estuaries where it is found (Hechinger et al. 2007). While many studies in this field end with speculations about the implications of parasitic infection, direct or indirect ecological effects are typically not quantified. Indeed, as Hughes (2014) notes, even though there are cases that now appear to be dramatic, textbook examples of parasitic manipulation were often unseen or not verified for long periods of time. We have much yet to learn about this phenomenon.
How did manipulation of the host’s phenotype evolve?
Parasitic manipulations of hosts are among the most complex adaptive traits in the organic world (Hughes 2014). Teasing apart the origin of complex traits, like parasitic alterations of host’s phenotype, is still difficult for evolutionary biologists, and is typically considered on a case-by-case basis. Parasitic manipulation is apparently rare across parasites, perhaps because the transition from one peak in the fitness landscape to another is challenging. The study of the evolution of parasitism is made even more complicated because it often requires understanding the evolutionary history and phylogeny both of parasites and their hosts, as well as a thorough quantification of parasites’ extended phenotype. Controlling the phenotype of another organism requires that parasites adapt to, and thus specialize in, particular host species. Fredensborg (2014) explores how the type of manipulation a parasite uses constrains the diversity of hosts it is able to infect. A previous analysis found the location of infection and the type of host to be associated with the type of manipulation a parasite induced (Lafferty and Shaw 2013). Parasites infecting invertebrates or the host’s body cavity tend to manipulate the host in ways that increase contact rates of the host with its predators, while parasites infecting vertebrates or hosts’ muscles, central nervous system, and some other sites impair the host’s response to predators (Lafferty and Shaw 2013). Fredensborg (2014) built upon this analysis and found that parasites that infect the central nervous system and/or manipulate the host’s neuromodulation are more constrained in the diversity of hosts they can infect (i.e., they are constrained to infect hosts that are more closely related to each other) relative to parasites that reside elsewhere or rely on debilitation of the host to increase transmission. Residing in the central nervous system or manipulating neuromodulation forces manipulating parasites to be specialists, while residing elsewhere or depending upon debilitation to increase transmission rates allows parasites to adopt a more generalist strategy (Fredensborg 2014).
Mauck et al. (2014) explores questions about local adaptation of Cucumber mosiac virus (CMV), a pathogen of plants that is transmitted between hosts by aphids. Two virus isolates displayed increased infectivity and higher titers of virus when infecting a native host relative to a novel one. They also found that one virus isolate manipulated the phenotype of the host plant and vector (aphids) in ways likely to increase transmission when infecting a native host plant, while transfer to a novel host plant resulted in changes in the host’s phenotype that likely reduced transmission of the virus. The second virus isolate showed non-significant trends toward manipulating its native host and vector in ways that increased transmission, but the virus’s ability to manipulate a non-native host could not be explored because this isolate was rarely able to infect a novel host plant. Mauck et al. (2014) highlight the importance of considering genetic differences in the ability of a parasite’s or pathogen’s (in this case, isolate’s) ability to manipulate the host’s phenotype, and of the importance of local adaptation in the success of manipulation. This is not only important for our understanding of the evolution of manipulation, but also for applied purposes and for laboratory studies of manipulation. For example, understanding genetic differences and the importance of local adaptation would be important when considering the efficacy of using a parasite or pathogen to control an invasive plant or pest. Additionally, in laboratory studies, it is often more feasible to collect and infect wild hosts from populations that do not co-occur with the parasite rather than to establish breeding colonies using hosts with an evolutionary history with the parasite of interest. However, if the parasite is locally adapted to the host population, then laboratory-breeding may be necessary in order to accurately quantify the parasite’s manipulation of the host’s phenotype and to understand the ecological implications of infection.
Interest in manipulative parasites has been rising over the past decades (Moore 2012), and many documented cases of parasites’ manipulations of hosts’ phenotype now await study by integrative teams of scientists to elucidate how those manipulations occur, their ecological importance, and how manipulation evolved in that system. These integrative collaborations would be mutually beneficial. Through natural selection manipulative parasites have “learned” how to exploit the nervous and immune systems of their hosts to induce precise changes in the host’s phenotype. Neurobiologists, physiologists, and immunologists can hasten their understanding of links between these systems and the expression of phenotype in the host, and may discover previously unexplored connections fruitful for further study. For behavioral ecologists, studying behavior-manipulating parasites can help explain behavioral variation, average values of traits, expression of suboptimal behaviors, and correlations between behavioral traits. For ecologists, incorporating manipulative parasites will help to explain the flow of energy through food webs (e.g., if parasites make their hosts easier for predators to catch) and the distribution of animals in an environment (e.g., if manipulated hosts exhibit different microhabitat choice). Parasites that manipulate their host’s phenotype increase or decrease the variability in phenotypic traits on which natural selection can act, and may be of interest to evolutionary biologists tracking the trajectory of phenotypic traits across evolutionary time. We hope that in showcasing the host–parasite systems discussed above, we can encourage a wider array of researchers to examine the effects of parasites in organisms with which they are familiar, but for which they previously have not considered the impact of parasitic infections.
In this article, we have pointed out several desiderata for the study of parasitic manipulation. First, that the prevalence and abundance of parasitic infection in the wild are characterized (e.g., Joseph and Faulkes 2014; Paranjpe et al. 2014). Second, that the physiology (Carreon and Faulkes 2014; Libersat and Gal 2014), phylogeny (Fredensborg 2014; Hughes 2014; Mauck et al. 2014), and ecology (Boze and Moore 2014) of the host are well understood. Third, that hosts and parasites can be housed and infected under controlled conditions (de Bekker et al. 2014; Libersat and Gal 2014; Mauck et al. 2014). Currently, there are very few host–parasite systems that check even half of these items off the list. This is not surprising, given the wide range of skills needed to accomplish even a few of these. We hope that in showcasing these host–parasite systems, we can encourage a wider array of researchers to examine the effects of parasites in organisms with which they may not be familiar, or for which they may not have previously considered the impact of parasitic infections.
We wish to thank Jon Harrison, Lori Strong, the rest of the organizers of the 2014 SICB conference, and the ICB editorial team for their help organizing the symposium and this special issue. We thank Janice Moore and Brian Fredensborg for helpful comments on an earlier draft of this manuscript. We dedicate this paper to Ada Marie Weinersmith, who was delivered the same week as this paper.