Sophie Armitage, Department of Animal and Plant Sciences, University of Sheffield, Sheffield, S10 2TN, UK. Tel.: +0114 2220080; fax: +0114 2220002; e-mail: firstname.lastname@example.org
Central to the conceptual basis of ecological immunity is the notion that immune effector systems are costly to produce, run, and/or maintain. Using the mealworm beetle, Tenebrio molitor, as a model we investigated two aspects of the costs of innate immunity. We conducted an experiment designed to identify the cost of an induced immune response, and the cost of constitutive investment in immunity, as well as potential interactions. The immune traits under consideration were the encapsulation response and prophylactic cuticular melanization, which are mechanistically linked by the melanin-producing phenoloxidase cascade. If immunity is costly, we predicted reduced longevity and/or fecundity as a consequence of investment in either immune trait. We found a measurable longevity cost associated with producing an inducible immune response (encapsulation). In contrast to other studies, this cost was expressed under ad libitum feeding conditions. We found no measurable costs for constitutive investment in immunity (prophylactic investment in cuticular colour).
Animal hosts usually defend themselves against attack from parasites and pathogens. Costs for such defences seem likely given that many tend to be inducible rather than constitutively maximally expressed (Harvell & Tollrian, 1999). Moreover, it is commonly assumed that effective immune defence is costly: the best working hypothesis for the maintenance of epigamic selection assumes resource-based trade-offs between immunity and the traits under selection (Sheldon & Verhulst, 1996; Siva-Jothy & Skarstein, 1998). Costs associated with immune defence fall into one of three broad categories. First the costs associated with maintaining/investing in constitutive immune function (Kraaijeveld & Godfray, 1997); secondly, the cost of producing an induced response to an immune insult (Moret & Schmid-Hempel, 2000; Hoang, 2001; Tiën et al., 2001); and thirdly, the cost of evolving immunity (Webster & Woolhouse, 1999). However, quantifying these costs has proven difficult, possibly because the extent to which a host invests resources in immune defence will be determined by the virulence of the pathogens it faces (Sasaki & Godfray, 1999; Fellowes & Travis, 2000).
A few studies have demonstrated costs for producing an induced response and evolving responses, but evidence showing a cost for constitutive defence is lacking. To date, most assessments of the costs of immunity have focused on finding trade-offs between immune function and life history traits. For example, selection experiments on Drosophila melanogaster (Kraaijeveld & Godfray, 1997; Fellowes et al., 1998; Kraaijeveld et al., 2001), the velvetbean caterpillar (Fuxa & Richter, 1998), and a snail (Webster & Woolhouse, 1999) have found traits such as reduced competitive ability, reduced egg viability and reduced fertility in response to selection for increased parasitoid or pathogen resistance. Likewise, field-based studies on bumblebees (König & Schmid-Hempel, 1995) and damselflies (Siva-Jothy et al., 1998) have shown reduction in immune function contingent on foraging activity and reproduction, respectively. However, Coustau et al. (2000) challenged the general assumption that investment in immunity always bears a cost and suggested that the costs may be restricted to some, but not all, resistance mechanisms. Additionally Ferrari et al. (2001) could not identify trade-offs between resistance against parasitoids and fungi in the pea aphid.
All the above empirical studies examined innate immunity [‘a set of disease-resistance mechanisms that are not specific to a particular pathogen’ (Goldsby et al., 2000)]: a shared feature of vertebrate and invertebrate immunity (Hultmark, 1993; Vilmos & Kurucz, 1998). One of the major effector systems the arthropod immune system has to deal with pathogens and parasites is the melanin-producing phenoloxidase (PO) cascade (Brey & Hultmark, 1998). This cascade is intimately involved in the encapsulation response towards parasites and pathogens (Shiao et al., 2001), as well as the recognition of nonself (Brey & Hultmark, 1998). Insects respond to large haemocoelic pathogens with the inducible encapsulation response (Gillespie et al., 1997). In most insects this consists of coordinated humoral and cellular responses that produce a layer of dead melanized haemocytes over the pathogen, resulting in its death and isolation (Götz & Boman, 1985). The PO enzyme cascade is also involved in the hardening and darkening of insect cuticle, an important barrier against pathogens (Ashida & Brey, 1995). Moreover, the insect cuticle is biochemically and immunologically active in expressing PO as well as antimicrobial peptides (Brey et al., 1993; Golkar et al., 1993) and so may represent more than just an inert barricade against pathogens.
The importance of the cuticle in insect defence has been highlighted by recent work on density dependent prophylaxis (DDP) (Reeson et al., 1998). In systems displaying DDP individuals raised at higher population densities invest more in prophylactic immune function presumably because of the increased chance of encountering pathogens at high conspecific densities (Reeson et al., 1998; Wilson et al., 2001). Such individuals show higher resistance to pathogens as well as increased cuticular melanization (Reeson et al., 1998). In the mealworm beetle, Tenebrio molitor L. (Coleoptera: Tenebrionidae), resistance to an entomopathogenic fungus (Metarhizium anisopliae) was positively correlated with the degree of cuticular melanization (Barnes & Siva-Jothy, 2000) independent of rearing density. Beetles that produced darker cuticle were more resistant to the fungus.
In this paper we conducted an experiment with two aims: first to assess the costs of producing an induced immune response to an immune insult: we predicted that there would be a cost to utilizing the immune system that resulted in reduced longevity and/or fecundity in immune challenged beetles. Secondly we assessed the costs of investing in cuticular melanization: given that dark (black) T. molitor have invested more in cuticular immunity compared with lighter (tan) conspecifics we predicted that any costs of this investment will result in black beetles having reduced longevity/fecundity compared with tan beetles. Both aspects of immune function that we examined (the encapsulation response and cuticular melanization) depend on the production of melanin via the PO cascade.
Materials and methods
Experimental beetles came from 15 isofemale lines reared at the University of Sheffield for at least seven generations of brother–sister matings. Utilizing isofemale lines meant that individuals within each group of four (two tan and two black beetles) were genetically very similar to one another. Source of all isofemale lines were from different stock cultures and so probably represented the genetic responses of more than one population. The beetles were maintained in an insectory at 22 ± 3 °C, with an ad libitum diet of rat chow [special diets services: 77% cereal (wheat, maize, barley, wheatfeed) 15% vegetable proteins (soya bean meal), 5% animal protein (fish meal) and 3% vitamins (major and trace) and amino acids], and water and apple added every 2–3 weeks.
Cultures were examined every other day for pupae. Pupae were collected, sexed, weighed, and then kept individually in grid box containers. All beetles used in this experiment fell within the overall weight range of 0.115–0.150 g whilst females derived from specific isofemale lines were all within 0.010 g of each others' fresh pupal weight. Larval density was maintained constant in all stock cultures.
Measuring cuticular colour
Experimental beetles from each isofemale line were the two darkest and two lightest female offspring collected over a random (in time) 7-day period. These offspring were derived from a single mother producing eggs over at least a 4-week period. Consequently, the allocation of beetles to colour categories represents a random sample of black and tan offspring from that isofemale line.
When adult females reached 4 days old, their cuticular colour was analysed using the protocol of Thompson et al. (2002). In short, beetles were anaesthetized on ice, and a digital image of the elytra was captured. The degree of cuticular darkness was analysed from the captured image using Optimas 6® software (Bothell, WA, USA) giving a weighted average luminance (WAL) on a greyscale between 0 and 255 (0 darkest, 255 lightest). Within the field of DDP it is common to use the term ‘melanization’ in relation to cuticular darkness (e.g. Reeson et al., 1998; Barnes & Siva-Jothy, 2000; Wilson et al., 2001), and we will adhere to this convention in this paper; however, it is important to note that cuticular colour is actually the result of the processes of both melanization and sclerotization (Sugumaran, 1991).
The experimental treatment
Each isofemale line contributed one treatment group to the experiment. Each treatment group contained two tan and two black females that had all undergone imaginal eclosion within 7 days of each other. Colour variation within and between isofemale lines varied: our criterion for allocating a female to the tan or black group was that they had to differ by at least 10 points (approximately 50% of the total range) on our colour scale. These individuals represent the darkest and lightest individuals from a random subsample of each mother's offspring: they therefore represent a random subsample of black and tan. One tan and one black female were assigned at random to the experimental treatment and one tan and one black assigned at random to the control group. At 4 days after adult eclosion, all individuals in the treatment group were chilled, the cuticle swabbed with 70% ethanol (to ensure sterility) and three short (1 mm) lengths of sterile nylon monofilament (Sunline, fil classe I.G.F.A., Siglon V transparent; 0.128 mm diameter; Tokyo, Japan) were inserted through the pleural membrane between the third and fourth sternite using sterile instruments [examination of nylon removed after 24 h from identically treated beetles (n = 5) revealed cellular encapsulation and melanization on all filaments from each beetle: individuals therefore responded to all nylon inserts (S. Armitage, unpublished results)]. Control individuals were cold anaesthetized and swabbed with ethanol. We did not conduct a treatment control for logistic reasons and because the main physiological response to the sham-treatment (wounding) is the aggregation of haemocytes in the wound and their subsequent melanization (i.e. the same response as occurs towards the nylon inserts which are magnitudes larger than the hole they were inserted through). However, differences between the treatment group and control group may be partly explained by uncontrolled effects of the treatment (e.g. wound-healing and stress).
Focal females were mated to males derived from genetically diverse stock populations. Stud males were removed from the stock populations upon pupation. Upon imaginal eclosion, they were fed ad libitum rat chow and apple and were mated to females once they were 7 days old. Once adult females were 7 days old (the age of sexual maturity; Happ, 1970), they were allowed to mate once with a 7-day-old virgin male. Matings were carried out in a Petri dish lined with filter paper and continuously observed. Once copulation had been achieved, the pair were immediately separated and isolated. Females were mated once every 14 days (again with a 7-day-old virgin male) to ensure sperm availability was constant across all treatment groups.
Once mated, each female was placed in a Petri dish containing approximately 10 g of freeze-sterilized plain flour and water in a 0.5-mL centrifuge tube with a cotton wool bung. Water was replenished every 14 days. The flour in the Petri dish was sieved for eggs every day for the first 7 days after mating, then once every 7 days after this period, and egg numbers recorded. Petri dishes were inspected daily for deaths whereupon the flour was sieved for eggs. Our experiments are therefore conducted in a relatively ‘clean’ environment compared with stock cultures, where beetles are in contact with frass as well as live and dead conspecifics.
We used a mixed-model anova with isofemale line as random factor and colour and treatment as fixed factors. To test the influence of isofemale line we fitted all two-way interactions with isofemale line and kept them if P < 0.25. This is a conservative estimate. The significance levels of main effects were therefore tested over the interaction term with isofemale line if appropriate. We have not reported nonsignificant interactions. In order to meet the assumptions of parametric ancova, longevity data were log-transformed and fecundity data were square root transformed (Sokal & Rohlf, 1995).
Effect of treatment on longevity
There was a significant effect of treatment upon longevity, with a significant reduction in adult lifespan in individuals receiving an immune challenge compared with controls (Fig. 1, Table 1). The interaction between colour and isofemale line had a significant effect on longevity (Table 1) but cuticular colour had no effect on longevity (Fig. 2, Table 1). Fecundity covaried significantly with longevity (Table 1). Furthermore, there was a significant positive relationship between longevity and fecundity across isofemale lines (Fig. 3).
Table 1. anova table for the effect of immune challenge and cuticular colour on the dependent variable ln(longevity).
Sum of squares
Treatment has two levels: control (no immune challenge) and experimental (immune challenge). Colour has two levels: tan and black. Sqrt(fecundity) is included as a covariate; isofemale line is included as a random factor. Superscript symbols denote the error term used for the test of significance.
Error term used for the test of significance: sum of squares = 1.706, d.f. = 13.709.
Error term used for the test of significance: sum of squares = 1.731, d.f. = 14.149.
Error term used for the test of significance: sum of squares = 1.656, d.f. = 28.
Fecundity was not affected by immune challenge or by cuticular colour (Fig. 4, Table 2). Cuticular colour did, however, show a significant interaction with isofemale line, indicating that colour was important in the context of some isofemale lines but not others (Table 2). There was also a significant interaction between longevity and isofemale line indicating variation in longevity between isofemale lines (Table 2). Isofemale line had a significant effect upon fecundity, indicating that some lines were comparatively more fecund (Table 2). Longevity also covaried significantly with fecundity (Table 2).
Table 2. anova table for the effect of immune challenge and cuticular colour on the dependent variable fecundity.
Sum of squares
Treatment has two levels: control (no immune challenge) and experimental (immune challenge). Colour has two levels: tan and black. ln(longevity) is included as a covariate; isofemale line is included as a random factor. Superscript symbols denote the error term used for the test of significance.
Error term used for the test of significance: sum of squares = 34.992, d.f. = 15.167.
Error term used for the test of significance: sum of squares = 10.946, d.f. = 14.641.
Error term used for the test of significance: sum of squares = 10.301, d.f. = 14.
There is a measurable cost to our experimental treatment in T. molitor: longevity was significantly reduced in immune challenged beetles compared with controls even when living under ad libitum laboratory conditions, isolated from conspecifics, and in ‘clean’ conditions (see Methods). Previous work with insects has found a survival cost when using the immune system, but under starved conditions (Moret & Schmid-Hempel, 2000); by contrast, our result shows this cost can also be expressed under nonstressed conditions. Two recent studies examining the effect of successful immune defence against a natural parasite reported a decrease in survivorship of parasitized hosts (Hoang, 2001; Tiën et al., 2001) and support our finding. Further indirect support for a cost of the encapsulation response comes from studies showing that foraging activity (König & Schmid-Hempel, 1995) and mating behaviour (Siva-Jothy et al., 1998) reduce the magnitude of this immune response.
Given that an immune insult results in reduced longevity, and longevity is positively related to fecundity, one may expect a corresponding reduction in fecundity with reduced longevity. However, immune challenged beetles showed no reduction in fecundity. An explanation is that immune challenged beetles invested more per unit time into egg production than control beetles, and hence suffered decreased longevity whilst maintaining fecundity. Despite the lack of a fecundity cost, it is important to note that (a) there are important fitness consequences of survivorship in comparison with fecundity (Crone, 2001) and (b) that natural selection on longevity is common in the wild (Kingsolver et al., 2001).
A further cost to immune challenged beetles might be a compromised immune system because haemocytes are utilized in the encapsulation response (Götz & Boman, 1985) and as haemocytes age they may also lose their function (Kurtz, 2002). The combination of these two effects may mean that older, challenged beetles are less able to maintain cell-based immunity compared with younger and/or unchallenged beetles. Moreover, we suggest that under food-limited conditions, which are more likely to be the case in the field, longevity differences between immune-challenged and unchallenged individuals might be more pronounced.
Cuticular colour and the cost of prophylactic responses
Investment in prophylactic immune defence, here measured as cuticular colour, showed no measurable fitness cost under the conditions of our experiment. We found no difference in fecundity or longevity between black and tan beetles. Furthermore, colour did not appear to contribute to the interaction with the treatment, indicating that colour may not be influential in determining the costs of the induced encapsulation response discussed above. Two additional considerations need to be taken into account when interpreting these data. First we did not use replicates from the same population: Consequently, it is difficult to interpret negative results. Secondly, the isofemale lines are inbred, so we can expect low levels of genetic variation within them, a factor that may also account for the lack of interactions between colour and treatment.
The cuticle is the first physical line of defence against pathogens, and it is known that darker beetles are more resistant to an entomopathogenic fungus (Barnes & Siva-Jothy, 2000). Thus, we predicted that increased investment in this aspect of constitutive immune defence should produce a fitness cost for black beetles. Our results suggest there may be no such cost under the conditions of our experiment. This is in accord with the findings of Ferrari et al. (2001) who found no trade-off between resistance mechanisms against three different natural enemies in the pea aphid [but see Boots & Bowers (1999) who question the assumption that different resistance mechanisms should be traded off against each other].
In summary, we looked specifically at two aspects of immunity that are linked by their codependence on the melanin-producing pathway. Despite this mechanistic link, our study was only able to identify a cost of utilizing the inducible immune response. This cost exists although the animals were reared in ad libitum feeding conditions and in a relatively ‘clean’ laboratory environment.
The sclab massive would like to big up Nick Colegrave. Andrew MacColl and two anonymous referees made comments that greatly improved the manuscript. SAOA was supported by a NERC studentship, JJWT and MS-J were supported by NERC grant no. GR3/12121, and JR was supported by a Marie Curie Fellowship.