Parasite-altered feeding behavior in insects: integrating functional and mechanistic research frontiers

Parasites live at the expense of their hosts, and at least 50% of species on Earth are parasites (see Glossary; Price, 1980; Windsor, 1988; Poulin and Morand, 2004). The ubiquity of parasites and the intimacy of their interactions with their hosts generate enormous biological impacts. For example, infectious viruses, bacteria and fungi are famous for their influence on the population dynamics of their hosts (Lefèvre et al., 2009b; Lafferty and Kuris, 2012), the evolution of host resistance (e.g. Ebert, 1994; Woolhouse et al., 2002) and even the evolution of sex (Morran et al., 2011). Until recently, however, few studies considered the possible proximate influence of parasitism by micro- or macroparasites on the feeding behavior of their hosts. This topic is now gaining some deserved attention with the rapid development of fields such as ecoimmunology (see Glossary; e.g. Rolff and Siva-Jothy, 2003; Schmid-Hempel, 2011), medication behavior (e.g. de Roode et al., 2013; Abbott, 2014) and parasite manipulation of hosts (e.g. Moore, 2002; Thomas et al., 2005; Libersat et al., 2009; Hughes et al., 2012). The host's feeding behavior is likely to be an important factor in host–parasite interactions because it has strong effects on the fitness of both hosts and parasites, with possible conflicts between an optimal host diet versus a diet optimal for parasitic survival (Maure et al., 2016). In theory, parasite-induced changes in the feeding behavior of their hosts could benefit the fitness of the host, parasite, both or neither (Poulin, 1995; Hurd, 2001; Moore, 2012).

Here we discuss changes in feeding behavior of insects induced by infections from their parasites (including microbial pathogens) that are adaptive for either host or parasite. Most studies that show a fitness benefit of altering host feeding behavior (either to the benefit of the host or parasite) lack a mechanistic understanding of the initiation of the behavior (Table 1). Conversely, many mechanistic studies do not address how parasite-altered feeding behavior might be adaptive for either party (Table 1). Therefore, our aim here is to integrate both functional and mechanistic frontiers in this field of research in hopes of guiding future research.

We first use existing evolutionary theory as the basis for our functional perspective on parasite-altered feeding behavior, and we focus on insect hosts because of their suitability for experimental work in this field. Some earlier works (e.g. Poulin, 1995) focused on the question of whether these and other host behaviors are truly adaptations. Rather than revisiting this question, we move to more specific areas of evolutionary theory as guideposts for applying a functional approach, which tests fitness benefits of phenotypic traits. Coevolutionary theory (see Glossary) considers reciprocal adaptation between interacting species, and predicts the existence and potential escalation of adaptive antagonistic traits in hosts and their parasites (e.g. Kraaijeveld et al., 1998; Sasaki and Godfray, 1999). Even more pertinent to parasite-altered feeding behavior is an explicit consideration of adaptive plasticity theory (see Glossary), which predicts alternative adaptive phenotypic responses (e.g. feeding behavior and underlying physiology of the host) by a single genotype to different environmental conditions (e.g. imposed by parasites).

We identify four basic, functional types of parasite-altered feeding behavior of hosts (Fig. 1). Some of the existing terminology in the literature reflects the idiosyncrasies of particular sub-disciplines (e.g. anti-parasite medication behavior, de Roode et al., 2013; Abbott, 2014); thus, we advocate terms that can be applied more broadly: prophylaxis, therapy (see Glossary), compensation and manipulation. The first three are behavioral changes that are adaptive for the host. In this context, ‘prophylaxis’ refers to feeding changes prior to parasite infection, whereas ‘therapy’ refers to feeding responses after infection (de Roode et al., 2013). ‘Compensation’ refers to changes in feeding that redress physiological deficits imposed by either prophylaxis or therapy (or other physiological distress) (Singer et al., 2014). By contrast, ‘manipulation’ describes feeding changes in the host that are adaptive for the parasite (Moore, 2002; Thomas et al., 2005; Libersat et al., 2009; Hughes et al., 2012).

This language advantageously accommodates multiple functional mechanisms, such as alternative dietary mechanisms, e.g. medicinal versus nutritional therapy (see Glossary; Singer et al., 2014). Our suggested terminology also accommodates alternative life-history mechanisms of prophylaxis and therapy (as proposed by de Roode et al., 2013). For example, therapeutic changes in feeding behavior may have a social versus individual basis or may function via trans-generational versus cis-generational processes (Box 1). Because evidence for prophylaxis, therapy, compensation and manipulation as functional components of parasite-altered feeding behavior in insects is still somewhat new and patchy, we agree with Raubenheimer and Simpson (2009) that the time is right to adopt a lexicon that reflects and does not limit conceptual developments in this field of study.

Adaptive plasticity theory predicts the existence of prophylactic foraging changes in response to cues that predict the risk of parasitism in insects, and several recent ecoimmunological studies support this prediction. For example, upon seeing parasitoid wasps (see Glossary), female Drosophila melanogaster flies switch their oviposition preference from low- to high-ethanol substrates (Kacsoh et al., 2013), a behavioral change that pre-emptively confers resistance to their larval offspring against some of their lethal parasitoids (Milan et al., 2012). This case study is significant because it proves the existence of trans-generational prophylaxis (de Roode et al., 2013), thus raising the question of how widespread it is. In holometabolous insects, trans-generational prophylaxis might be especially common, because ovipositing adults often have better opportunities to deploy pre-emptive defenses than do larvae. We further expect trans-generational prophylaxis to evolve in holometabolous insects in response to the risk of attack by parasites, such as insect parasitoids, that are not transmitted from mother to offspring. Vertically transmitted parasites, in contrast, are expected to induce trans-generational therapy (rather than prophylaxis) because offspring would respond directly to infections acquired from parents (de Roode et al., 2013).

In addition, the growing evidence for density-dependent prophylaxis (see Glossary) as a physiological immunological defense against pathogens in insects (e.g. Reeson et al., 1998; Wilson et al., 2001; Triggs and Knell, 2012) raises the question of whether there might be associated prophylactic or compensatory feeding changes in these host species. The first case of density-dependent prophylaxis showed that Spodoptera exempta caterpillars reared at high densities were more resistant to nucleopolyhedravirus than those reared alone (Reeson et al., 1998). A recent study of nutritional immunology of S. littoralis caterpillars shows differences among the nutrient intake targets that optimally support alternative immune responses (e.g. phenoloxidase versus lysozyme activity), which were different from the optimal nutrient intake target for insect growth (Cotter et al., 2011). Although this study found no shifts in feeding behavior that directly support or compensate for the upregulation of the host's specific immunological defenses, this possibility needs further testing (Cotter et al., 2011). To our knowledge, density-dependent prophylactic feeding changes and associated compensatory feeding responses are strictly hypothetical at present.

In contrast to the limited study of prophylactic changes in foraging behavior, therapeutic changes in foraging behavior in response to parasites have been investigated experimentally in several cases of medicinal and nutritional therapy (reviewed in de Roode et al., 2013; Abbott, 2014; see Box 1). Abbott (2014) has refined the set of criteria for defining therapeutic self-medication based on adaptive plasticity theory (sensu Singer et al., 2009), and here we extend them to other types of therapy. These criteria require demonstrating that the putatively therapeutic change in behavior: (1) increases the fitness of infected hosts, (2) decreases the fitness of uninfected hosts, (3) is detrimental to the parasite and (4) is preferentially enacted by the host in response to infection. A recent case study exemplifies this functional approach to medicinal therapy by showing explicit support for these criteria: wood ant workers (Formica fusca) infected with the fungus Beauveria bassiana preferentially ingested liquid diets containing the putative medicinal substance, H₂O₂ (4), which increased the survival of infected workers (1) but decreased the survival of uninfected workers (2), and reduced the spore viability of the fungal pathogen (3) (Bos et al., 2015). Further testing of these criteria can identify additional cases of therapeutic feeding behavior as well as distinguish therapy from compensation, which does not necessarily confer fitness costs to uninfected hosts (2) (Abbott, 2014). Information on physiological mechanisms underlying feeding behavior (Box 2) is also critical for discerning therapy and compensation. In light of the growing evidence for therapeutic changes in foraging, we expect the existence of associated compensatory foraging in insects. However, research in this field has tended to emphasize medicinal or nutritional therapy over compensation (Singer et al., 2014; Shikano and Cory, 2016). It is possible, however, that some cases reported as ‘macronutrient self-medication’ (e.g. Povey et al., 2009, 2013) might more accurately represent cases of nutritional compensation (Shikano and Cory, 2016). In contrast to cases of medicinal and nutritional therapy (Box 1), there is also increasing support for illness-induced anorexia as a type of therapeutic change in infected insect hosts (Adamo, 2006; Adamo et al., 2010), but a definitive test using the criteria above has yet to be conducted.

Parasite manipulation of host feeding behavior has received very limited study in insects. Hence, our discussion of functional aspects of manipulation is mostly hypothetical. We expect selection to favor parasite manipulation of host feeding behavior when the host's normal diet creates a suboptimal environment for the parasite's fitness. On the basis of coevolutionary theory (Hughes et al., 2012) and previous work (Adamo, 2002, 2013; Moore, 2002; Thomas et al., 2005; Escobedo et al., 2009; Helluy, 2013), we expect to see highly sophisticated manipulations of host feeding behavior by parasites with high host specificity. Host-specific parasitoid wasps, which infect a wide range of insects such as caterpillars, beetles, cockroaches and ants, are well known for manipulating the physiology of their insect hosts (Beckage and Gelman, 2004), and are likely candidates for subtle manipulations of host feeding behavior as well. To distinguish manipulation of host feeding behavior from mere disruption and dysfunction owing to parasitic infection, experiments will need to demonstrate that the parasite-altered behavior not only benefits the parasite, but also that the parasite-altered behavior is regulated or controlled physiologically by the parasite. For example, the wasp Cotesia congregata uses multiple mechanisms to suppress feeding by its caterpillar host, thus preventing the wasp larvae from being eaten by the host as they emerge and pupate on its integument (Adamo, 1997, 2005; Adamo et al., 2016). It is possible that host-specific pathogens might manipulate host feeding behavior in such a way as to promote transmission to new hosts, as hypothesized for insect vectors of plant viruses for example (Blanc and Michalakis, 2016), and this deserves further investigation.

Experimental tests of evolutionary theory are crucial for resolving questions about the functional significance of parasite-altered host feeding behavior (i.e. whether the host or the parasite benefits from the change). However, understanding the physiological (including genomic and epigenetic) mechanisms underlying changes in feeding behavior is crucial for informing function, e.g. distinguishing therapy from compensation, as well as for identifying potential applications to issues such as public and environmental health. Currently, little is known about the physiological bases of prophylaxis, therapy, compensation and manipulation in cases of parasite-altered feeding behavior of insects (Table 1). Regardless of the functional outcome, parasite-altered feeding behavior must act proximately via the host's nervous system (Schmid-Hempel, 2011; Simpson, 2013), which entails modulation of chemoreception and feedbacks on other physiological processes of food rejection and intake. We therefore see the study of modulation of the host's chemoreception as a major research frontier for explaining how parasite-infected hosts can adaptively modify the quality or quantity of food intake. Parasite manipulation of host feeding behavior warrants further study along several physiological frontiers: direct and indirect manipulations of host resources, and biochemical, genomic and mechanical mechanisms.

Parasite-altered feeding behavior can manifest as changes in food intake quality or quantity. Altered dietary choices based on food quality, which indicates the effects of food on the growth and development of the consumer, are based on physiological responses to a combination of several food components, such as macro- and micronutrients, toxins and non-digestible material (Simpson and Raubenheimer, 2012). It is important to note that a food's quality for a consumer is not absolutely determined by properties of the food; rather food quality is also a product of the consumer's physiological needs. In host–parasite interactions, the infection state of the host can determine the effect of any given food on the host's fitness (i.e. food quality). For example, infected hosts might experience increased fitness from consuming a higher ratio of protein to carbohydrate than the optimal ratio for uninfected hosts. Mechanisms responsible for the adaptive modification of the quality or quantity of food intake are therefore based on physiological feedbacks from post-ingestive processes to the host's sensory system (Simpson and Raubenheimer, 2012).

Uncovering the physiological feedbacks that change the chemosensory response in parasitized hosts is a research priority. Changes in gustatory response are likely to mediate feeding changes associated with functional mechanisms discussed above. The finding of parasite-induced changes in gustation in the woolly bear caterpillar, Grammia incorrupta (formerly G. geneura) (Bernays and Singer, 2005), as part of the physiological mechanism underlying medicinal therapy (Singer et al., 2009) opens the door for further study along these lines. Although there are other examples of parasitized insects changing their feeding choices with respect to nutrients and plant secondary metabolites, studies investigating changes in gustation are limited. We therefore turn to the more extensive literature on physiological regulation of feeding in uninfected insects (Box 2). Based on the current understanding of feeding regulation in unparasitized insects (Behmer, 2009; Simpson and Raubenheimer, 2012), i­t would follow that shifting demands for nutrients and changes in blood metabolite levels could feed back on chemoreception as the basis for nutritional therapy and compensation in parasitized hosts.

Hormones and neuropeptides (see Glossary) warrant attention as potential mediators of prophylactic or therapeutic changes in feeding by insects in response to parasites. Research on other aspects of parasite-altered behavior of insects has shown the important role of hormones and neuropeptides in escape (Lefèvre et al., 2009a), reproductive (Roy et al., 2006) and grooming behaviors (Libersat et al., 2009). The increased foraging behavior in bees infected with Nosema ceranae is linked to changes in the vitellogenin–juvenile hormone regulatory network, which affects the timing of a bee's switch from nursing to foraging activities (Nelson et al., 2007). Infection results in an accelerated shift from nursing to foraging behavior, and can benefit the colony because infected bees spend less time with the brood and young workers, and more time with foragers and older workers (Seeley, 1982). In addition, infected bees show upregulation of octopamine pathways, and this neurohormone is linked to changes in foraging behavior (Mayack et al., 2015). Although the adaptive significance of the behavioral change remains untested, the authors suggest that this may be a parasite-induced change in behavior that enables the parasite to obtain energy for reproduction.

Neuropeptides are important neuromodulators that can accumulate in parasitized insects (Zitnan et al., 1995) and influence feeding behavior in unparasitized insects (reviewed in Audsley and Weaver, 2009; Caers et al., 2012). Therefore, they are likely candidates for mediating parasite-induced changes in insect feeding behavior. Neuropeptides have been shown to regulate the quantity of ingested food (Wei et al., 2000; Maestro et al., 2001; Downer et al., 2007; Al-Anzi et al., 2010), exhibit myoactivity on the visceral muscles of the gut, and may act elsewhere, such as the central nervous system (CNS) or gut stretch receptors (Audsley and Weaver, 2009). Neuropeptides are involved in the case of trans-generational prophylaxis by D. melanogaster flies discussed above (Kacsoh et al., 2013). While a visual cue initiates the behavioral change in D. melanogaster oviposition, neuropeptide F mediates the lasting effects of the oviposition behavior for 4 days after parasitoid exposure (Kacsoh et al., 2013). However, examples of neuropeptides as mediators of prophylactic or therapeutic changes in feeding are lacking. This is likely to be a case in which absence of evidence is not evidence of absence, because the functions of neuropeptides are complex and their mechanisms of action, interactions with each other, and secretion are still poorly understood.

Rather than increasing food intake, a host may adaptively decrease its intake of some or all food in response to parasitism. This anorexic response could be achieved through changes in blood nutrient composition (imposed by the parasite) to reduce the quantity of food intake by progressively reducing phagostimulatory input during a meal and desensitizing gustatory receptors (Simpson and Simpson, 1992). The mechanism by which food intake is reduced could also involve increased blood osmolality, prolonged intervals between feeding or decreased duration of feeding bouts (Simpson, 1995). This anorexia could serve a therapeutic function by decreasing mobility and thus conserving host energy stores to be used by other host processes. Alternatively, reduced feeding during infection may be a means to ameliorate the trade-off in physiological allocation between immune function and other physiological processes (Smith and Holt, 1996), such as digestion (Adamo et al., 2008, 2010). Crickets (Gryllus texensis) experimentally infected with a bacterial pathogen (Serratia marcescens) reduce their food consumption, specifically their intake of a high-lipid food (Adamo et al., 2010). After consuming a diet high in lipid (34% fat), which increases the hemolymph lipid concentration in uninfected crickets, infected crickets show reduced resistance against the pathogen compared with their counterparts on the control diet (10% fat). This relationship is mediated by monomeric apolipophorin III, a lipid transport protein in the hemolymph that increases in concentration on a high-lipid diet (Adamo et al., 2010). Further mechanistic study of illness-induced anorexia in other systems is needed to determine whether it plays a general role in ameliorating trade-offs between digestion and immunological defense.

Compensatory changes in feeding by infected hosts are especially likely to involve physiological mechanisms modifying the quantity of food intake. For example, increased food intake by infected hosts can compensate for extra energetic or nutrient costs imposed by the growing parasites (Ponton et al., 2011a,b). Although reduced fecundity often accompanies infection (Hurd, 1993, 2001), Ponton et al. (2011a) showed that the increase in nutrient intake in infected beetles was able to compensate for parasite-induced reduction in fecundity. In this case, the mechanism is currently unknown, but the authors hypothesize that the dual increase in protein and carbohydrate consumption results in part from blood-borne nutrient feedback. It is also unknown whether volumetric feedbacks are altered with infection in insects to benefit the host. Perhaps the threshold for stretch receptors in infected hosts could change to modify the volume of intake, possibly combating infection by alleviating trade-offs in resource allocation (Box 2). Hypothetically, parasite-induced alteration of stretch receptor sensitivity could thus prolong a meal to increase food intake in service of compensation or therapy.

Parasites and insect hosts may compete for existing and ingested resources (Schaub, 1992; Adamo, 1997), which may facilitate changes in host feeding behavior. Hosts often respond to the physical presence of parasitic infection by economizing on their resources, e.g. reducing egg production or increasing foraging activity. Parasites could exploit this standard response and use it to gain increased transmission (e.g. increased foraging increases exposure to predation for hosts of parasites with complex life cycles involving multiple hosts) or to prolong survival (the host invests more into maintenance). Pre-existing compensatory mechanisms of hosts can therefore be readily hijacked by parasites.

The depletion of a host's internal resources by the parasite may lead to compensatory feeding, further increasing resources for the parasite (Raubenheimer and Simpson, 1993; Lefèvre et al., 2008a). Apparent manipulation of a host's quantity of food intake could be a byproduct of the parasite's ingestion of host resources, resulting in volumetric feedbacks and increased host feeding rate or duration. Alternatively, parasites could increase their hosts' feeding on a particular resource when the nutritional requirements of hosts and parasites differ. Parasite manipulation of resources (Polak, 1996; Ponton et al., 2011b) can change host hemolymph composition and alter chemoreception through blood-borne feedbacks. Thompson and colleagues (Thompson and Redak, 2005; Thompson et al., 2005) have demonstrated the effects of parasitism by Cotesia congregata wasps on blood metabolite levels of their Manduca sexta caterpillar hosts, concluding that the observed changes create an optimal environment for the parasite and cause physiological dysfunction of nutrient regulation for the host. This example raises the question of whether this type of breakdown in nutrient regulation in parasitized hosts may act in similar systems to curb host defenses, such as nutritional therapy or compensation, which could otherwise be deployed against the parasite.

Parasites may adaptively manipulate the feeding behavior of the insect host by introducing molecules that directly alter host physiology (i.e. chemical manipulation; Moore, 2002; Thomas et al., 2005). Hymenopteran parasitoids inject their hosts with venom and hormones that act as neuromodulators (Beckage, 1997). These parasite-derived factors have the ability to alter host development (Jones et al., 1986; Dover et al., 1995; Steiner et al., 1999; Cole et al., 2002), induce immunosuppression (Beckage and Gelman, 2004) and drastically alter host behavior (Thomas et al., 2002, 2003; Libersat et al., 2009; Hughes et al., 2008; de Bekker et al., 2014; Takasuka et al., 2015). Although none has yet been identified, these same factors may play a role in altering host feeding behavior. Identifying parasite-derived factors that control host behavior via the CNS (neuropharmacological factors; Adamo, 2013) is key to elucidating the mechanisms involved in manipulation.

Alternatively, manipulation of host feeding behavior may result from parasites inducing the host to make factors that benefit the parasite (i.e. genomic manipulation). Such factors are known to influence other aspects of host behavior; for example, the baculovirus Lymantria dispar nucleopolyhedrovirus induces its caterpillar host to secrete a virally encoded enzyme into the hemolymph, causing the host to climb to an elevated position prior to death, which increases viral transmission once the host disintegrates or liquefies (O'Reilly et al., 1992). In this example, foraging behavior is indirectly targeted because the virally encoded enzyme inactivates one of the host's major hormones, 20-hydroxyecdysone, a hormone that is thought to induce many caterpillars to climb down from their foraging positions in the treetops to hide in bark or leaf litter (Hoover et al., 2011). However, to our knowledge, there are no known examples of such factors directly influencing feeding behavior by the host. Proteomics approaches to studying the genome-wide changes in protein expression of the insect host CNS caused by parasites (Biron et al., 2005, 2006; Ponton et al., 2006; Lefèvre et al., 2008b) provide a unique opportunity to explore host-derived factors responsible for host manipulation, and we expect this approach to increasingly yield insight into physiological mechanisms of host manipulation by parasites.

The host's immune response is another prime target for manipulation by parasites because host immune responses can influence neural function and reliably change the physiology and behavior of the host (i.e. psychoneuroimmunological mechanisms sensu Adamo, 2012, 2013). The link between the immune and nervous systems could be a major axis relating resource manipulation, chemical manipulation and genomic manipulation (Adamo, 2002, 2013; Schmid-Hempel, 2011; Adamo et al., 2016). For example, the parasitoid C. congregata suppresses host feeding by inducing a currently unidentified host-derived molecule to manipulate immune–neural connections of its caterpillar host, M. sexta (Adamo, 2005). Wasp larvae inhibit the breakdown of octopamine (released during immune and stress responses in other insects; Adamo, 2010), which decreases the host caterpillar's feeding (Adamo, 1998, 2005) by disrupting the neural output of the frontal ganglion, which in turn disrupts host swallowing (Miles and Booker, 2000). However, while increased octopamine levels are necessary to arrest host feeding, they are not sufficient. Most recently, Adamo et al. (2016) found that a surge of insect cytokines (which is potentially induced by the parasitoid) as well as changes in the expression of host immune genes [i.e. plasmatocyte spreading peptide (PSP) and spätzle] play an important role in suppressing feeding behavior of the host after parasitoid larvae emerge and pupate on the host's body. This body of work demonstrates the redundancy and multi-functionality of physiological mechanisms responsible for parasite manipulation of host feeding behavior (Adamo et al., 2016).

Finally, we note the possibility that parasite manipulation of host feeding behavior may occur through mechanical or physical means (Molyneux and Jefferies, 1986; Hurd, 2003). Such manipulations include obscuring host phagoreceptors, blocking the host foregut and damaging host tissue. The protist parasite Leishmania impairs the sandfly vector's ability to fully engorge during a blood meal and therefore induces the vector to increase feeding time and frequency, increasing parasite fitness by enhancing transmission efficiency (Rogers and Bates, 2007; Jenni et al., 1980). The parasite causes a mechanical blockage of the host's anterior midgut with promastigote secretory gel. In another example, increased transmission of the malaria parasite is partially due to mechanical damage of mosquito salivary glands by parasite sporozoites. Mosquitoes with damaged salivary glands have reduced apyrase activity, which increases mosquito feeding attempts and transmission of the malaria parasite (Ribeiro et al., 1984, 1985; Koella and Packer, 1996; Koella et al., 1998). Therefore, where the parasite resides in the host and the tissue with which it interacts inform the mechanism of manipulation.

The field of parasite-altered feeding behavior in insects is burgeoning with new studies. We have argued that therapy, compensation, prophylaxis and manipulation are adaptive types of parasite-altered feeding behavior that will be most profitably studied by integrating functional and mechanistic approaches. To better understand and distinguish prophylaxis, therapy and compensation, further study of physiological feedbacks affecting host sensory systems is especially needed. For host manipulation in particular, research on mechanisms by which parasites control host feedbacks will be important to integrate with functional approaches. Although few empirical examples exist, parasite manipulation of host feeding behavior may be subtle and thus more common than indicated in the literature.

To expand the study of this field, we propose some new research questions that stem from the above discussion. The functional genomics revolution offers great promise for identifying proteomic pathways underlying changes in foraging by infected hosts. How do these pathways compare among cases of prophylaxis, therapy, compensation and manipulation? Are some physiological feedbacks more commonly modified as anti-parasite defenses or easier to manipulate than others? To what extent might they be predicted by the phylogenetic relationships of parasites and hosts? How can informatics and other advanced approaches shed new light on the critical issue of modulation of neural pathways underlying changes in host foraging? Lastly, given recent findings of the surprisingly potent role of the host microbiome in modifying host physiology and behavior, including anti-parasite defenses in insects (Oliver et al., 2014), it will be of interest to investigate how microbial mutualists might mediate parasite-altered feeding behavior in insects (Ponton et al., 2013).

We close by discussing the potential relevance of parasite-altered feeding in insects to human and environmental health. Several observations of dietary effects on insect host or parasite fitness are quite suggestive (Moore, 2002; McArt et al., 2014). In a recent example, Hien et al. (2016) showed that variation in the plant source of nectar meals consumed by malaria-carrying Anopheles coluzzi mosquitoes modified the infection prevalence and intensity as well as fitness components of the mosquito host and Plasmodium parasite. This observation raises the question of whether parasite-altered nectar feeding behavior occurs in this system, which would putatively link variation in ecological communities to variation in public health. Turning to the critical ecosystem service of pollination, recent evidence shows medicinal effects of floral nectar chemistry on bumblebees infected with an intestinal parasite (Crithidia bombi) (Richardson et al., 2015), as well as changes in bumblebee pollination behavior in response to floral nectar chemistry (Richardson et al., 2016). To the extent that bumblebees engage in medicinal therapy (Barrachi et al., 2015), parasite-altered foraging behavior may play a role in the dynamics of plant pollination in both agricultural and non-agricultural ecosystems. In summary, parasite-altered feeding behavior in insects is a topic of both basic and applied interest, offering great opportunities for experimental research in comparative physiology to inform ecology, evolution, and human and environmental health.

The authors thank Charlotte Rutledge for the invitation to write this Commentary, and two anonymous reviewers for helping to improve the manuscript.