- Top of page
- Materials and methods
- Statistical analyses
- Data accessibility
By definition, parasites reduce the fitness of their hosts by diverting hosts' nutritional resources for their own growth and reproduction and by causing other fatal or debilitating effects (Schmid Hempel 2011). To counter this threat and to minimize the costs of parasitic infection, multicellular organisms have evolved an effective immune system to recognize and attack invading parasites. But immune defences are costly; they can cause self-harm when triggered (Sadd & Siva-Jothy 2006) and also demand nutritional resources that could otherwise be channelled into growth and reproduction (e.g. Moret & Schmid-Hempel 2000; Siva-Jothy & Thompson 2002; Cotter, Kruuk & Wilson 2004).
The nutritional state of the host can affect its ability to fight and resist an infection (Chandra 1996; Lochmiller & Deerenberg 2000) such that increasing an organism's access to resources can increase its resistance to parasites. For example, food-supplemented snowshoe hares (Lepus americanus) experienced reduced nematode prevalence compared to controls (Murray, Keith & Cary 1998), whilst experimental food restriction suppressed cell-mediated immunity in yellow-legged gulls (Larus cachinnans, Alonso-Alvarez & Tella 2001). Similarly, invertebrate studies have focused on the effect of nutrient deprivation or starvation on immune function and/or parasite resistance, with the consensus being that reduced resources compromise immunity (e.g. Moret & Schmid-Hempel 2000; Siva-Jothy & Thompson 2002; Ayres & Schneider 2009 but see Triggs & Knell 2012).
Often, energy is assumed to be the limiting resource that individuals must partition between traits, and indeed, mounting an immune response has been shown to increase the metabolic rate of both vertebrates (Demas et al. 1997) and invertebrates (Freitak et al. 2003). Despite the requirement for resources during an immune response, many animals display illness-induced anorexia, in which food intake is reduced immediately after an immune challenge (Kyriazakis, Tolkamp & Hutchings 1998; Adamo, Fidler & Forestell 2007). This may seem counter-intuitive but has been hypothesized to serve a number of possible functions, from reducing the risk of ingesting more parasites, to starving resident parasites of key macro- and micronutrients (see references in Kyriazakis, Tolkamp & Hutchings 1998; Adamo, Fidler & Forestell 2007). However, beyond the intake of energy, feeding comprises the ingestion of nutrients in particular ratios, which are allocated to different functions within the body, and there is good evidence that overingestion as well as underingestion of certain nutrients can be costly (Simpson et al. 2004; Raubenheimer, Lee & Simpson 2005; Cotter et al. 2011). Animals that would benefit from reducing the intake of a particular nutrient that favours parasite growth might be forced to decrease food consumption overall.
In lepidopteran larvae, resistance to parasites has been shown to depend on the relative amounts of macronutrients (protein and carbohydrate) in the diet and the diet that optimizes growth rates in uninfected individuals differs from the diet that optimizes the immune response (Lee et al. 2006; Povey et al. 2009; Cotter et al. 2011); thus, we might expect organisms to modify their intake based on their current nutritional requirements. This behaviour is known as self-medication, which Singer, Mace & Bernays (2009) define as ‘a specific therapeutic and adaptive change in behaviour in response to disease or parasitism’. It is generally recognized that verification of therapeutic self-medication must satisfy three criteria: (i) the behaviour should increase the fitness of infected individuals; (ii) it should decrease or have no effect on the fitness of uninfected individuals; and (iii) the behaviour should be specifically triggered by infection. There is evidence for therapeutic self-medication from several studies of vertebrates, most famously from chimpanzees that use plant-derived substances when infected with protozoan or helminth parasites (Huffman & Seifu 1989; Fowler, Koutsioni & Sommer 2007), and some experimental studies of livestock infected with gut nematodes using nitrogen-rich clover (Hutchings et al. 2003). There is also evidence from insect species for medicinal use of plant secondary compounds, such nicotine, pyrrolizidine alkaloids and iridoid glycosides (e.g. Krischik, Barbosa & Reichelderfer 1988; Christe et al. 2003; Castella et al. 2008; Singer, Mace & Bernays 2009). More recent studies have provided support for macronutrient self-medication in bacteria- or virus-challenged caterpillars (Lee et al. 2006; Povey et al. 2009). Although macronutrients are a ubiquitous part of the diet and their use is not restricted to self-medication, nearly all documented cases of self-medication involve increasing the amount of a nutrient or chemical that comprises some fraction of the normal diet (see Raubenheimer & Simpson 2009).
Implicit in the notions of self-medication and illness-induced anorexia is that changes in feeding behaviour should be dynamic, with the magnitude of the response depending on the stage of infection and the host's capacity to resist or tolerate infection. To capture this dynamic, studies must control for differences in feeding behaviour prior to and during infection, that is, dietary preferences should be compared longitudinally within groups pre- and post-challenge. In addition, studies must consider the possibility that the capacity to self-medicate could have a significant genetic component, such that the magnitude, direction or timing of behavioural changes differs between families or genotypes (Lefevre et al. 2010).
Here, we assess the effects of dietary protein and carbohydrate balance on the outcome of infection with nucleopolyhedrovirus (NPV) in larvae of the African armyworm, Spodoptera exempta, and on the associated immune response. This is a natural host–pathogen interaction in sub-Saharan Africa (Graham et al. 2012), and S. exempta larvae feed on a wide range of graminaceous crops and pasture grasses that vary in their nutritional composition (Yarro 1984; Rose, Dewhurst & Page 2000). Using artificial diets to control macronutrient composition precisely, we measured the diet choice of individuals from different full-sibling families both before and after challenge with NPV, thus providing the strongest test yet for dynamical self-medication using dietary macronutrients. In so doing, we also examined the absolute amount of each macronutrient consumed to test whether sickness-induced anorexia, and/or selective intake of specific nutrients, occurred in response to infection. Our study tested the following specific predictions: (i) resistance to NPV will decline as the relative protein content of the diet is reduced, (ii) diet-related resistance to NPV will be associated with diet-related changes in immune function, providing a potential mechanism for changes in resistance, (iii) virus-challenged insects will prefer a diet rich in the macronutrient that favours NPV resistance in the short term and revert to diets similar to non-challenged individuals when the infection is under control, (iv) infection with NPV will trigger a short-term anorexic response, limiting the potential for further exposure to the virus or starving it of resources, and finally, (v) the degree of plasticity in the self-medication response will vary among full-sibling families, consistent with genetic variation in the trait.
- Top of page
- Materials and methods
- Statistical analyses
- Data accessibility
Here, we provide the clearest evidence to date for therapeutic self-medication, sensu Singer, Mace & Bernays (2009), using dietary macronutrients. Consistent with this phenomenon, S. exempta larvae challenged with a high (LD50) dose of NPV chose a diet that was rich in protein (containing c. 50% more P than C) compared to that of uninfected control larvae, which chose a diet that was carbohydrate-biased (c. 50% more C than P). By choosing a relatively protein-rich diet, NPV-challenged insects improved their survival prospects from <40% on foods containing the most carbohydrates (P : C = 7 : 35 and 14 : 28) to around 80% on the most protein-rich foods (P : C = 28 : 14 and 35 : 7). In this and previous studies, the survival of non-infected larvae was high and independent of P : C ratio, but larval growth rate and overall performance (survival × larval growth rate) peaked on a diet that was slightly carbohydrate-rich and dropped off dramatically on diets with an excess of protein (Lee, Simpson & Raubenheimer 2004). Thus, the main criteria for self-medication are satisfied.
Comparison of overall feeding patterns of virus-challenged and control insects in both experiments suggests that challenged individuals self-medicate on protein, but closer analysis of the feeding dynamics supports a plastic response in which feeding behaviour changes as the viral infection progresses. Among caterpillars that had been given a high (LD50) dose of virus, those which survived viral challenge behaved very differently from those that died. The first day post-inoculation was characterized by a sharp increase in P consumption and an elevated P : C ratio in survivors relative to controls and casualties. P and C consumption then declined in survivors over the course of experiment, resulting in a decrease in total food consumption. Note that this dynamic is masked if survivors and casualties are lumped together.
Experiment 2 showed that this change in behaviour was not simply caused by families which naturally choose higher levels of protein being more likely to survive infection. We also tested diet preference before infection so that we could be sure that any differences in feeding behaviour were a response to the virus challenge. Prior to inoculation, the digestible component of the diet comprised around two-thirds carbohydrate and one-third protein. In the non-challenged controls, the amount of protein in the diet remained low but gradually increased as pupation approached. In contrast, sublethally infected larvae radically changed their feeding behaviour on a daily basis (Fig. 5), and this is likely to have coincided with temporal changes in the viral infection process (Keddie, Aponte & Volkman 1989; Washburn, Kirkpatrick & Volkman 1996; Cory & Myers 2003). On day 1, there was a dramatic reduction in the amount of carbohydrate consumed by the virus-challenged larvae and a decline in the overall feeding rate (Fig. 5b, d). This change in feeding behaviour coincided with the period when virus released from the ingested OBs invades the larval midgut epithelial cells and replicates in their nuclei. Importantly, the amount of protein eaten by inoculated larvae was maintained at pre-infection levels, such that the percentage of protein in the diet increased from <40% to c. 60% in all of the families we tested. By day 2, carbohydrate intake returned to pre-infection levels in the sublethally infected insects, such that total food consumption increased and the overall P : C ratio declined towards 1 : 1. This change in feeding behaviour coincided with a period when many infected midgut cells are likely to have become melanized, encapsulated and/or sloughed into the gut lumen to be replaced by healthy cells, and in some larvae, virus will have migrated into the insect haemocoel to infect haemocytes and other tissues. By day 3, the total food intake of virus-challenged larvae continued to increase, perhaps to offset the reduced food consumption earlier in the infection. Finally, by day 4, the dietary P : C ratio and total food intake of virus-challenged caterpillars became comparable to that of non-infected control larvae, presumably as the infection has been controlled and is no longer imposing a nutritional demand on its host.
Although we detected genetic variation for nutrient consumption, this explained a relatively small amount of the variation in feeding behaviour and was independent of treatment or time post-infection. Rather, diet choice showed a high degree of phenotypic plasticity and different families demonstrated the capacity to respond to infection by self-medicating. Of particular note is that the immediate response following inoculation with a sublethal dose of virus is that the larvae limit their consumption of carbohydrate and food intake overall, but maintain a constant level of protein ingested. This behaviour is consistent with a form of illness-induced anorexia (Kyriazakis, Tolkamp & Hutchings 1998; Adamo, Fidler & Forestell 2007). Specifically, the anorexic response could limit the ingestion of further virus OBs with contaminated food, or it could be a mechanism by the host to reduce calorie intake overall (or carbohydrate intake specifically) without sacrificing protein consumption. Another explanation is that this is the most efficient mechanism by which the host can alter the blend of ingested food to bias it towards proteins; this would be an adaptive response if a protein-rich diet enhances resistance to the virus or limits the virus replication rate.
To explore the impact of macronutrients on possible viral resistance mechanisms, we assayed several aspects of immune function. In both virus-challenged and control larvae, the haemolymph protein pool increased linearly with the amount of protein in the diet. Thus, short-term changes in larval feeding behaviour are reflected in rapid changes in the nutritional composition of their blood (see also Povey et al. 2009). The P : C composition of the diet was also reflected in constitutive levels of PO activity, antimicrobial activity and haemocyte density, all three of which increased (linearly or nonlinearly) with increasing protein content of the diet, though unlike the other haemolymph properties, peak antimicrobial activity was not achieved on the most protein-rich diet. This suggests that larvae that switch from a carbohydrate-biased diet onto a diet that is relatively protein-rich will generally have more haemocytes and higher levels of PO with which to melanize and encapsulate virus-infected cells (Washburn, Kirkpatrick & Volkman 1996; Trudeau, Washburn & Volkman 2001), as well as a greater capacity to combat concomitant microbial infections. However, only PO activity and haemocyte density were significantly modulated by viral infection, with virus-challenged larvae having marginally more haemocytes and lower PO activity. Haemocytes are involved in the encapsulation of virus-infected tissues, and so, their greater density in infected larvae may reflect their increased production following infection. The reduction in PO activity in virus-infected larvae is counter-intuitive, but is consistent with previous studies, suggesting phenotypic and genetic trade-offs between immune traits (Cotter, Kruuk & Wilson 2004; Cotter et al. 2004; Povey et al. 2009; Rao, Ling & Yu 2010). Thus, whilst pre-ingestive behavioural plasticity allows infected individuals to capture the resources required to mount an effective immune response, post-ingestive internal trade-offs may constrain immune expression (Cotter et al. 2011). It is also worth noting, however, that other important viral resistance mechanisms have not been quantified in this study, such as the sloughing and replacement of infected midgut epithelial cells, and the resource implications of these processes are not easily quantified.
Finally, this study builds on two previous investigations of the impact of macronutrients on insect resistance to pathogens and the dietary choices insects make when faced with a pathogen challenge (Lee et al. 2006; Povey et al. 2009). Each study used different host–pathogen combinations, but broadly similar protocols in the same research laboratory, providing the opportunity to explore the generality of their key findings. Lee et al. (2006) found that S. littoralis larvae challenged with an LD50 dose of S. littoralis NPV had highest survival on the diet with the highest relative protein content, as also observed here for S. exempta and its specific NPV, so demonstrating the importance of protein for resisting baculovirus across different host–virus combinations. Povey et al. (2009) conducted a similar experiment using S. exempta challenged with the bacterium, Bacillus subtilis, suggesting that protein is perhaps ubiquitously important for resisting entomopathogens. This comparison is particularly revealing since the baculovirus infects orally, whereas the bacterium was injected into the haemocoel, suggesting that dietary protein may benefit multiple defence mechanisms in the gut, haemocoel and elsewhere. In diet-choice experiments, S. littoralis larvae that were challenged with an LD30 dose of baculovirus ate significantly less food post-infection than did the control larvae (Lee et al. 2006), so demonstrating a similar anorexic response to that shown by the S. exempta larvae receiving an LD50 dose of virus in the present study (Experiment 1). Moreover, in both these experiments, larvae that subsequently survived a potentially lethal dose of virus chose a P : C ratio that was significantly more protein-rich than those that succumbed. However, because of the high levels of virus-induced mortality in prior experiments, and the fact that dietary preferences before viral challenge were not quantified, we could not exclude the possibility that these results depended on genetic or other intrinsic differences in dietary preferences of larvae that predisposed them to dying of NPV (Lee et al. 2006). Both of these deficiencies were remedied in Experiment 2 of the present study by challenging S. exempta larvae with a low dose of virus and by quantifying feeding preferences prior to virus challenge, so that we could monitor shifts in feeding behaviour from pre- to post-infection. These clearly revealed that individuals from different families all switched to a relatively protein-rich diet immediately following infection before returning to a diet that resembled that of control larvae over the following days. It is also worth noting that in none of these experiments did we quantify viral loads in dead or surviving larvae, and so we cannot rule out the possibility that protein-biased diets either alter host tolerance or trigger the virus to switch to a vertically transmitted mode. These possibilities would make interesting avenues for further study.
In conclusion, as predicted, we showed that (i) survival following virus challenge declined as the relative protein content of the diet was reduced; (ii) increasing dietary P : C ratio resulted in higher levels of all immune traits, so providing a potential mechanism for changes in resistance; (iii) when given a choice between complementary diets, virus-challenged insects temporarily increased the relative protein content of their diet, but in insects challenged with a low viral dose, this was achieved by reducing the intake of carbohydrates whilst maintaining protein intake; (iv) infection with a low dose of NPV triggered a short-term anorexic response, so limiting the potential for further exposure to the virus or starving it of key resources. In contrast, we found little evidence for prediction (v), that the degree of plasticity in the ‘self-medication’ response would vary between full-sibling families. Whilst the total amounts of each macronutrient consumed varied between families, the P : C ratio achieved did not, suggesting that this choice is not genetically determined but is a form of phenotypic plasticity common to all genotypes. Our results have clear implications for the foraging behaviour of S. exempta larvae in the wild and may help explain the diverse range of graminaceous plant species included in their diet (Yarro 1984; Rose, Dewhurst & Page 2000).