Age-dependent trade-offs between immunity and male, but not female, reproduction

Authors


Summary

  1. Immune function is costly and must be traded off against other life-history traits, such as gamete production. Studies of immune trade-offs typically focus on adult individuals, yet the juvenile stage can be a highly protracted period when reproductive resources are acquired and immune challenges are ubiquitous. Trade-offs during development are likely to be important, yet no studies have considered changes in adult responses to immune challenges imposed at different stages of juvenile development.
  2. By manipulating the timing of a bacterial immune challenge to the larvae of the cotton bollworm moth, we examined potential trade-offs between investment into immunity at different stages of juvenile development (early or late) and subsequent adult reproductive investment into sperm or egg production.
  3. Our data reveal an age-dependent trade-off between juvenile immune function and adult male reproductive investment. Activation of the immune response during late development resulted in a reduced allocation of resources to eupyrene (fertilizing) sperm production. Immune activation from the injection procedure itself (irrespective of whether individuals were injected with an immune elicitor or a control solution) also caused reproductive trade-offs; males injected early in development produced fewer apyrene (nonfertilizing) sperm. Contrary to many other studies, our study demonstrates these immune trade-offs under ad libitum nutritional conditions. No trade-offs were observed between female immune activation and adult reproductive investment.
  4. We suggest the differences in trade-offs observed between male sperm types and the absence of reproductive trade-offs in females may be the result of ontogenetic differences in gamete production in this species. Our data reveal developmental windows when trade-offs between immune function and gametic investment are made, and highlight the importance of considering multiple developmental periods when making inferences regarding the fundamental trade-offs expected between immune function and reproduction.

Introduction

Studies of variation in immune response typically focus on adult individuals. However, for many species the juvenile stage can represent the majority of an individual's lifespan. For many insects, the juvenile phase is when individuals acquire most, if not all, of their resources for future reproduction, but when they are also vulnerable to attack from parasites and pathogens. Despite this, surprisingly little is known regarding how immune challenges inflicted at the juvenile stage influence the capacity of an individual to invest in reproductive traits (Sait, Gage & Cook 1998; Jacot et al. 2005a; Cotter, Beveridge & Simmons 2008), and particularly how they affect investment into postcopulatory traits such as gamete production.

Both the maintenance and up-regulation of immune system function is costly (Kraaijeveld & Godfray 1997), and trade-offs between investment in immunity and other life-history traits are predicted (Sheldon & Verhulst 1996; Lochmiller & Deerenberg 2000; Zuk & Stoehr 2002). Trade-offs between immunity and sexually selected traits have received considerable attention (for a review, see Lawniczak et al. 2007). Correlative studies of adult insects have revealed positive relationships between immune function and investment in sexual traits such as acoustic signals, pheromones, or male weaponry (Rantala et al. 2000, 2003; Simmons, Zuk & Rotenberry 2005; Pomfret & Knell 2006). These studies suggest that only adult individuals in good condition can afford high investment in both immunity and sexually selected traits, allowing females to maximize fitness by preferentially mating with the best quality males (Hamilton & Zuk 1982; Folstad & Karter 1992). Experimental manipulations have demonstrated that immune challenges induced at the final larval instar or adult stage of development reduce reproductive investment (Ahmed et al. 2002; Kerr, Gershman & Sakaluk 2010; Simmons 2012), and conversely, that increased investment into reproduction is associated with reduced immune function (Hosken 2001; Kerr, Gershman & Sakaluk 2010). Fitness trade-offs extend beyond copulation, however, postcopulatory trade-offs are comparatively poorly understood, especially the relationship between immunity and ejaculate quantity or quality (Simmons & Roberts 2005; Skau & Folstad 2005; Kerr, Gershman & Sakaluk 2010; Simmons 2012). If trade-offs between immune activity and sperm production occur, males that are forced to up-regulate their immune system may, as a consequence, produce fewer or poorer quality sperm. Such a relationship has been demonstrated in fish and humans, where there is a negative relationship between indices of immune function and ejaculate quality (Liljedal, Folstad & Skarstein 1999; Skau & Folstad 2005). A negative relationship between sperm viability and antibacterial immunity was observed in the field cricket, Teleogryllus oceanicus, but other common indices of invertebrate immune function were positively correlated with sperm viability (Simmons & Roberts 2005). Importantly, studies that have looked at postcopulatory trade-offs typically focus on challenges imposed at the final instar or adult stage. This ignores a vital component of life-history, the juvenile stage, when investment into reproductive tissues first commences. There is little information on how resource trade-offs at different stages of an individual's gonadic development may ultimately affect their reproductive investment.

Immune trade-offs in insects are often explored by assaying two key immune parameters: the phenoloxidase (PO) cascade and lytic activity. The enzyme PO is released in the haemolymph when a foreign body is detected, or a wound received. Its activation results in the melanization and death of the pathogen (Sugumaran 2002). PO activity is typically held as a correlate of an organism's ability to resist pathogens and parasites (Wilson et al. 2001; Rantala & Roff 2007). Lytic activity is the main defence against bacteria, and is measured as the ability to disrupt the structural integrity of invading pathogens. Lytic activity is measured in vitro by quantifying the ability of the haemolymph to destroy bacterial cells. These two indices can be negatively correlated, suggesting there may be trade-offs between components of the immune response itself (Cotter et al. 2004; Simmons & Roberts 2005).

Lipopolysaccharides (LPS), a component of the gram-negative bacterial cell wall, are nonpathogenic immune elicitors, which allow quantification of the costs associated with immune system up-regulation, rather than simply the physiological costs of infection. LPS elicits a persistent stimulation of the lysozyme-like immune response observed by an increase in lytic activity (Jacot et al. 2005b), yet may also result in reduced PO activity reflecting a trade-off between these two immune responses (Moret & Schmid-Hempel 2000; Simmons 2012). Studies that have elicited immune responses with LPS have demonstrated significant reproductive costs to immune system activation, suggesting important resource allocation trade-offs. Importantly, stimulation of the immune system with LPS is associated with decreases in the expression of secondary sexual traits, such as courtship activity (Jacot et al. 2005a; Fedorka & Mousseau 2007), and also decreases in gamete investment, such as ejaculate size (Kerr, Gershman & Sakaluk 2010), sperm viability (Simmons 2012) and fecundity (Ahmed et al. 2002).

We explored trade-offs between investment into immunity at different stages of juvenile development and subsequent adult reproduction using the cotton bollworm, Helicoverpa armigera (Hübner; formerly Heliothis armigera). The majority of the individual's life is spent as a juvenile or metamorphosing (approximately 30 of 42 days; K.B. McNamara unpublished data). The cotton bollworm is a polygamous species. For females, an increase in the number of matings results in higher fecundity, but a reduction in longevity (Hou & Sheng 1999). From the male perspective, investment in the quantity of sperm decreases with each mating event which results in a reduction in the number of eggs produced by successive females (Hou & Sheng 1999). This observation suggests that, in this species, males incur a significant fertility cost of mating. In common with all moths, cotton bollworm males transfer two types of sperm – eupyrene sperm, which are fertilizing sperm, and apyrene sperm, which are anucleate, nonfertilizing sperm. Apyrene sperm typically make up approximately 90 per cent of the sperm transferred. They are thought to act as a ‘cheap filler’, as the number of sperm transferred by males is a strong determinant of female postmating receptivity and future mating frequency and therefore sperm competition risk and intensity (Cook & Wedell 1999).

Here, by manipulating the timing and severity of an immune challenge to the larvae of H. armigera, we examined potential trade-offs between investment into immunity at both early and late stages during juvenile development and subsequent adult reproductive investment into sperm or egg production.

Materials and methods

Helicoverpa armigera were obtained as eggs from an existing culture of approximately 25 years (with occasional supplementation with new individuals) from the Commonwealth Scientific and Industrial Research Organisation in Narrabri, Australia in 2010. To ensure virginity and prevent cannibalism, larvae were reared individually until adult emergence on an artificial diet (described by Teakle & Jensen (1985)(except formalin was substituted with 0·08% propionic acid). Upon emergence, adult moths were provided with a feeder of sugar solution (ascorbic acid: honey: sugar: water; 1 : 10 : 10 : 100), and kept in 500-mL containers covered with a fine mesh.

Experimental up-regulation of immunity

To assess the impact of immune challenge at different developmental stages on adult reproductive traits, larvae were haphazardly assigned to one of six treatments that differed in the nature (treatment or control) and timing (3rd instar or 6th instar) of an administered immune challenge (individually housed larvae were selected from a large pool of individuals, and blindly allocated to a treatment). In the immune challenge treatment (+), individuals were injected with a nonpathogenic immune elicitor, a lipopolysaccharide (LPS; derived from Serratia marcascens; Sigma-Aldrich T3895), which was dissolved an isotonic ringer (Grace's insect medium; Sigma-Aldrich G81423). In the procedural control treatment, to account for the potential immune response to the injection itself, individuals were injected with the ringer only (−). Additional sham-treated controls (C) were conducted in which individuals were handled as for the other treatments, but not injected prior to immune testing.

We selected the LPS dose based on the literature for other invertebrates, controlling for body size (Adamo 1999; Ahmed et al. 2002; Schuhmann et al. 2003; Moret & Schmid-Hempel 2004; Jacot et al. 2005a). We administered 0·0898 μg of LPS per mg of larvae, using the average weight for each cohort of 3rd and 6th instars. Thus, at the 3rd instar, immune challenged larvae were injected with 2·38 μg of LPS in 0·25 μL of ringer (3+), while control larvae were injected with 0·25 μL of ringer (3−). At the 6th instar, immune challenged larvae were injected with 50 μg of LPS in 2·5 μL of ringer (6+), while control larvae were injected with 2·5 μL of ringer (6−). Individuals were injected using a 2·5-μL microsyringe (Hamilton: 7632-01) with a 33-gauge needle (Hamilton: 7803-05). On the day they moulted into their 3rd of 6th instar, larvae were weighed, an anterior proleg was cleaned with 100% ethanol, and for the treatments involving and injection, the needle was inserted into the proleg to transfer the LPS or control solution. Although the 3rd and 6th instar individuals were injected with the same amount of LPS, relative to instar body weight, the quantity of insect ringer in which it was dissolved did not also scale accordingly for logistical reasons associated with syringe capacity. The amount of insect ringer injected is unlikely to affect our results, given the large variation in body size, which results in an overlap in the amount of ringer administered (weight of ringer as a percentage of mean larval weight: 3rd instar = 0·70–2·98%, 6th instar = 0·39–2·05%).

To quantify the immune response to the LPS challenge, we assayed additional individuals from each of the four treatments (3C, 3−, 3+, 6C; 6−, 6+; methods as previously) for PO activity (n = 138) and antibacterial (lytic) activity (n = 80) 24 h after they were injected with either LPS or the control ringer solution. Prior to the assay, a larva was weighed and a haemolymph sample taken. Larvae were chilled on ice, and the tip of the first proleg was removed using fine dissection scissors. The exuded haemolymph was drawn into a pipette. A total of 2 μL from each individual was placed in a tube for lytic activity assays, and 3 μL was in placed in another tube containing 72 μL of chilled PBS (Amresco E404) and vortexed for PO activity assays. Both tubes were immediately frozen at −20 °C.

A lytic zone assay was used to determine the antibacterial response to the bacterium Micrococcus luteus (Sigma-Aldrich M3770). Agar plates were made with 10 mL of 1% agar in which 5 mg mL−1 of M. luteus was suspended (modified from Cotter et al. 2004). Using a sterilized Pasteur pipette, holes were punched into the agar. Into these wells, 1 μL of undiluted, thawed haemolymph was added, with two replicates made per haemolymph sample. The plates were incubated at 36 °C for 24 h and a digital image was recorded using a digital camera under ×10 magnification. The diameter of the area of the bacteria cleared by the haemolymph was measured using imagej (v1.43, National Institutes of Health, Bethesda, Maryland, USA).

Phenoloxidase activity was measured using a method modified from Hartzer, Zhu & Baker (2005). Two replicates of 25 μL of each sample (1 : 25; haemolymph : PBS) were added to a 96-well microtitre plate with 55 μL of PBS and was incubated at 33 °C for 20 min. After incubation, 20 μL of 10 mmol L−1 Dopamine hydrochloride (Sigma-Aldrich H8502) was added and the absorbance was measured at 492 nm every 5 min for 30 min (at 33 °C) using a FLUOstar Galaxy microplate reader (BMG LABTECH, Offenburg, Germany). This period was determined previously to be in the linear phase of the reaction.

Effect of instar at immune challenge on larval development

Five hundred and forty larvae were assigned to one of the four previously described experimental treatments [3−, 3+, 6−, 6+; no sham-treated control (C) treatments were conducted for these experiments, given that the immune responses of ringer-injected (−) and sham-treated controls did not differ ('Results')]. Following injection, individuals were returned to their feeding containers (no haemolymph was drawn from any of these larvae). Of the 540 tested, 174 died as a larva or pupa (preadult). A nominal logistic regression revealed that the likelihood of dying as a preadult increased for individuals injected at the 3rd instar (accounting for 74% of deaths; inline image = 20·62, P < 0·0001), but was not affected by whether individuals received LPS (inline image = 1·96, P = 0·16). A nonsignificant interaction between the instar at injection and whether the individual received LPS was removed from the model (inline image = 1·33, P = 0·25). For those individuals that survived and participated in mating trials, larval development time, adult body weight and sex were recorded.

Individuals that survived until adulthood were used to assess the impact of immune challenge on adult investment into sperm and egg production. For both assays, experimental animals were mated to stock-cultured individuals. All mating experiments were initiated at the onset of the scotophase and were conducted in 500-mL plastic containers with access to a honey solution feeder. Newly emerged adults take several days to reach reproductive maturity, and thus, all individuals were 2–3 days postimaginal eclosion at the time of first mating.

Effect of instar at immune challenge on the number of sperm transferred by males

To assess the effect of age at immune challenge on adult male reproductive investment, 124 males from the four treatments (3−, 3+, 6−, 6+) were provided with a virgin, stock-cultured female. Couples were allowed 5 h in which to mate. Couples that did not mate within 5 h were separated, and at the following scotophase, unmated males were provided with another novel, stock-cultured virgin female with which to mate. This was repeated for up to 3 days at which point unmated males were discarded. Mating couples were observed until the end of copulation, after which females were immediately frozen for later sperm analysis.

To calculate the number of eupyrene and apyrene sperm transferred to a female, the ampulla (containing sperm) was placed in a small drop of insect ringer (1·049 g NaCl, 0·032 g KCl, 0·048 g CaCl2, and 0·032 g NaHCO3 in 100 mL distilled H2O) on a cavity slide and broken open to release the sperm mass into the solution. The sperm mass was teased apart gently using fine dissecting needles and the solution stirred to ensure an even dispersal of sperm throughout the solution. To estimate the number of apyrene and eupyrene sperm, the sperm solution was washed from the well slide into a container using approximately 30 mL of distilled water (which was weighed) and homogenized. Six 10 μL sub-samples were pipetted from the diluted solution onto glass slides and air-dried. The number of apyrene and eupyrene sperm in each sub-sample was counted using dark-field phase contrast microscopy at ×100 magnification. Under this magnification, the two types of sperm are easily differentiated, with eupyrene sperm being considerably longer and thicker in diameter. Using the mean of the sub-samples, the total number of both types of sperm in the solution was calculated by multiplying the sperm number by the solution's dilution factor.

Effect of instar at immune challenge on female fecundity

To assess the effect of age at immune challenge on adult female reproductive investment, 139 females from each of the four treatments (3−, 3+, 6−, 6+) were provided with a virgin, stock-cultured male. Couples were allowed 5 h in which to mate. Couples that did not mate within 5 h were separated, and at the following scotophase, unmated females were provided with another novel, virgin stock-cultured male with which to mate. This was repeated for up to 3 days at which point unmated females were discarded.

Immediately following copulation, couples were separated. Males were killed and females were placed in 500 mL containers with access to honey solution. Eggs were collected from the container each day. Lifetime fecundity and adult longevity were recorded.

Statistics

All analyses were conducted in jmp (version 9.0.0, SAS Institute Inc., Cary, NC, USA). Data were transformed to normality before analysis, where appropriate. All data were modelled with ancovas. Nonsignificant interactions were removed from final models (Engqvist 2005).

Results

Lytic activity

In total, 80 larvae were injected and assayed for lytic activity (3C, n = 12; 3−, n = 15; 3+, n = 13; 6C, n = 12; 6−, n = 15; 6+, n = 13). Individuals that received the immune elicitor LPS (+) had a significantly higher lytic activity than individuals that received a ringer solution (−) or that were not injected (C; F2,75 = 34·74, P < 0·0001; Fig. 1a). Lytic activity also increased with larval instar (F1,75 = 140·83, P < 0·0001; Fig. 1a), and larval weight (F1,75 = 6·84, P = 0·01) {We standardized larval weight [(actual weight − mean instar weight)/standard deviation of instar weight], as instar and larval weight are necessarily collinear}. Nonsignificant interactions between larval instar and injection treatment (F2,79 = 1·40, P = 0·12), and between injection treatment and standardized larval weight (F2,129 = 0·05, P = 0·94) were removed from the model.

Figure 1.

(a) Means ± standard error lytic activity in response to instar and LPS status. Individuals that received LPS (+) had a greater lytic activity than those that received the control ringer solution (−) and those that were only handled (C). Sixth instar larvae also had a greater lytic activity than 3rd instar individuals, regardless of whether they were given LPS (+), ringer (−), or were only handled (C). Different letters denote significantly different comparisons. (b) Means ± standard error PO activity in response to instar and LPS status. Sixth instar larvae had a greater PO activity than 3rd instars, irrespective of whether they were given LPS (+), ringer (−), or were only handled (C). (c) There is a positive relationship between investment in lytic activity and PO. Within each treatment, this was significant for 6th instar larvae injected with control ringer solution only. Solid fit lines refer to LPS-treated trials (3+, 6+), dashed lines refer to ringer-treated trials (3−, 6−).

Phenoloxidase activity

In total, 138 larvae were assayed for PO activity (3C, n = 21; 3−, n = 26; 3+, n = 17; 6C, n = 25; 6−, n = 23; 6+, n = 26). PO activity was higher for 6th instar larvae compared with 3rd instar larvae (F1,133 = 69·18, P < 0·0001; Fig. 1b), irrespective of their injection treatment (C, − or +; F2,133 = 0·52, P = 0·59; Fig. 1b). PO was also not affected by standardized (described previously) larval weight (F1,133 = 1·73, P = 0·19). A nonsignificant interaction between larval instar and injection treatment (F2,129 = 2·17, P = 0·12), and injection treatment and standardized larval weight (F2,129 = 1·29, P = 0·28) were removed from the model.

Trade-offs between immune parameters

We examined the relationship between PO and lytic activity for those individuals for which both immune assays were conducted (3−, n = 15; 3+, n = 11; 6−, n = 15; 6+, n = 13). PO and lytic activity were positively correlated for 6th instar individuals injected with ringer solution (6−, r = 0·66, P = 0·007), but were not correlated for the remaining treatments (3−, = −0·05, P = 0·87; 3+, r = −0·06, P = 0·87; 6+, r = 0·10, P = 0·74). Across all immune treatments, there was also a positive relationship between lytic activity and PO (r = 0·55, n = 54, P < 0·0001; Fig. 1c).

Developmental duration

We examined larval and pupal duration for those males and females that survived to adulthood and participated in mating trials. Data were not recorded for one 3+ male.

Larval duration (days) was affected by an interaction between the instar at injection and whether the individual received LPS: individuals that received LPS at the 6th instar had a longer larval duration than the other treatments (F1,256 = 4·02, P = 0·046; Table 1). This significant interaction also drove two significant main effects: larval duration was longer for individuals that received LPS (F1,256 = 12·87, P = 0·0004), and for 6th instar-injected individuals (F1,256 = 16·71, P < 0·0001). Larval duration, however, was not affected by the sex (F1,256 = 1·69, P = 0·19), or weight of the individual (F1,256 = 2·78, P = 0·09). Several nonsignificant interactions were removed from the final model: instar at injection and the individual's sex (F1,253 = 0·40, P = 0·53), whether the individual received LPS and the individual's sex (F1,253 = 1·85, P = 0·17), and instar at injection and whether the individual received LPS and the individual's sex (F1,253 = 0·00, P = 0·95).

Table 1. Means ± standard errors for developmental duration (days) according to instar at injection (3rd or 6th) and whether the individual received LPS (− or +)
 Treatment
3−3+6−6+
N 64706265
Larval duration17·52 ± 0·1817·82 ± 0·1917·98 ± 0·1819·03 ± 0·22
Pupal duration14·17 ± 0·1314·15 ± 0·1313·97 ± 0·1513·51 ± 0·15

Pupal duration (days) was also affected by an interaction between the instar at injection and whether the individual received LPS: pupal duration was shorter for individuals that received LPS at the 6th instar than for all other treatments (F1,256 = 4·78, P = 0·03; Table 1). Females also had a shorter pupal duration (females = 13·31 ± 0·09 (n = 139); males = 14·67 ± 0·08 (n = 123); F1,256 = 128·58, P < 0·001). Pupal duration was shorter for individuals injected at the 6th instar (F1,256 = 10·69, P = 0·0012), and for those individuals that received LPS (F1,256 = 5·19, P = 0·02). Pupal duration, however, was not affected by the weight of the individual (F1,256 = 2·78, P = 0·09). Several nonsignificant interactions were removed from the final model: instar at injection and the individual's sex (F1,253 = 1·46, P = 0·23), whether the individual received LPS and the individual's sex (F1,253 = 0·56, P = 0·45), and instar at injection and whether the individual received LPS and the individual's sex (F1,253 = 0·35, = 0·56).

Sperm quantity

Eupyrene and apyrene sperm were collected for 70 males (3−, n = 18; 3+, n = 21; 6−, n = 17; 6+, n = 14). The number of eupyrene sperm number (square root transformed) was not affected by whether the male received LPS per se (F1,65 = 1·35, P = 0·25). However, there was a significant interaction effect between the instar at injection and whether the individual received LPS (F1,65 = 4·19, P = 0·04; Fig. 2a): 6th instar males injected with LPS transferred 60·56% fewer sperm than 6th instar males that were given control ringer solution. Males also transferred more eupyrene sperm when mated to heavier females (F1,65 = 4·49, β = 485·07 ± 229·04, P = 0·04). We previously determined that male adult weight was significantly affected by whether they received LPS [adults are heavier when injected with LPS as larvae (F1,68 = 5·94, P = 0·02)]. However, because male weight did not affect the number of eupyrene sperm transferred (F1,64 = 0·44, P = 0·51), this potentially collinear variable was removed from our initial model.

Figure 2.

(a) Means ± standard error eupyrene sperm transferred in response to instar and LPS status. Males injected at the 6th instar transferred less eupyrene sperm when given LPS compared with control ringer solution. This effect of LPS was absent in 3rd instar males. (b) Means ± standard error apyrene sperm transferred in response to instar and LPS status. Males injected at the 3rd instar transferred less apyrene sperm than those injected at the 6th instar, irrespective of whether they received LPS.

Males injected at the 6th instar transferred more apyrene sperm than those injected at the 3rd (F1,65 = 10·55, P = 0·002; Fig. 2b), irrespective of whether they received LPS (F1,65 = 0·79, P = 0·38), and of the weight of the female with which they mated (F1,65 = 1·23, P = 0·27). Male weight was marginally associated with apyrene sperm number and was therefore retained in the model (F1,65 = 3·95, P = 0·051). However, a nonsignificant interaction between instar at injection and whether the individual received LPS was removed from the model (F1,64 = 0·16, P = 0·69).

Female fecundity

Of the 139 females provided with a potential mate, 99 mated (3−, n = 27; 3+, n = 23; 6−, n = 24; 6+, n = 25). Of these, one 3+ female failed to lay eggs. Also, five females (3−, n = 1; 3+, n = 2; 6+, n = 2) laid <25 eggs, significantly fewer than the population mean of 955 ± 45 eggs (n = 93). These eggs were uniformly unfertilized, and we were unable exclude the possibility that this was owing to infertile pairings. Thus, we excluded these cases in which all eggs were unfertilized from analyses.

The total number of eggs produced (fecundity) increased with female adult longevity (F1,87 = 23·97, β = 55·97 ± 11·43, P < 0·001), but was not affected by the instar at injection (F1,87 = 0·61, P = 0·43), whether the individual received LPS (F1,87 = 0·18, P = 0·64), female weight (F1,87 = 0·26, P = 0·61), or the weight of the male with which she mated (F1,87 = 1·32, P = 0·25; mean fecundity ± standard error = 3− = 922·61 ± 91·07; 3+ = 1006·05 ±  95·22; 6− = 977·67 ± 87·82; 6+ = 924·35 ± 96·53). A nonsignificant interaction between instar at injection and whether the individual received LPS was removed from the model (F1,86 = 1·51, P = 0·22).

Discussion

Our data reveal an age-dependent trade-off between juvenile immune function and adult male reproductive investment. Up-regulation of the immune response during the 6th instar delayed male juvenile development and resulted in a reduced allocation of resources to eupyrene (fertilizing) sperm production. The immune challenge associated with the act of injection (irrespective of whether LPS was transferred) also elicited reproductive trade-offs; males injected at the 3rd instar produced fewer apyrene sperm than males injected later in their larval development. Interestingly, contrary to many other studies (Moret & Schmid-Hempel 2000; Simmons 2012), our study demonstrates these immune trade-offs under ad libitum nutritional conditions. No trade-offs, however, were observed between female immune activation and adult reproductive investment. Our data suggest that there are developmental windows when trade-offs between immune function and gametic investment are made. Furthermore, although there is evidence in other species of increased investment into reproduction following an immune challenge, in response to the potential increased risk of mortality (terminal investment; for a review, see Reaney & Knell 2010), our results are not consistent with this pattern.

Effects of immune challenge on sperm production

Why investment into eupyrene, but not apyrene, sperm numbers decreased under experimental immune activation may be explained by the ontogenetic timetable of spermatogenesis. Apyrene and eupyrene sperm are derived from the same bipotential spermatocytes, yet there are temporal differences in the commitment of these spermatocytes to either eupyrene or apyrene spermatogenesis. Although there is variation among species, eupyrene spermatogenesis typically commences during the later larval instars and ceases at pupation, while apyrene spermatogenesis begins later in the developmental cycle, starting just before or after pupation, and ending at or during adulthood (Friedlander 1997). Sixth instar challenged caterpillars were immune challenged at the point in time when eupyrene sperm are typically produced; the resource allocation shift at this critical time-point may thus generate a reduction in this specific sperm type. However, by the time apyrene spermatogenesis occurs during the pupal stage, immune up-regulation from the LPS may have ceased or at least be reduced (Simmons 2012), and the impetus for reproductive trade-offs may have diminished.

Immune activation, however, did not alter sperm production in 3rd instar challenged males. Unlike the 6th instar individuals, these larvae had several weeks to mitigate or overcome the costs of this immune challenge before allocation to apyrene of eupyrene spermatogenesis, and thus, any reproductive costs might be predicted to be on spermatogonia development (which typically commences from the first instar; Friedlander 1997) and thus reflected in a reduction in both types of sperm. This was clearly not the case. Rather, 3rd instar larvae that were injected (with either LPS or a control ringer) transferred fewer apyrene sperm as adults. The observed reduction in apyrene, but not eupyrene sperm for 3rd instar individuals is puzzling: eupyrene spermatogenesis is more typically sensitive to experimental manipulations than apyrene spermatogenesis (Friedlander 1997). Why 3rd instar-injected individuals transferred fewer apyrene sperm, regardless of their immune challenge is thus unclear. Our results suggest that the injection itself may have incurred significant costs for the smaller 3rd instar larvae, indeed mortality associated with the injection was high in 3rd instar larvae. However, it does not explain why only apyrene sperm production was affected, given the timetable of spermatogenesis in this species.

The effect of immune activation on male reproductive traits in insects generally provides evidence for significant reproductive trade-offs. Male Drosophila melanogaster that encapsulated a parasitoid egg showed lower fertilization success compared with noninfected males (Fellowes, Kraaijeveld & Godfray 1999), while male Drosophila nigrospiracula infected with a mite ectoparasites had smaller testes compared with controls (Polak 1998). Also, immune challenged crickets Teleogryllus oceanicus have reduced sperm viability, but only when nutrient deprived (Simmons 2012). Nonetheless, such trade-offs are not always found: ectoparasites do not alter sperm production in the dragonfly, Coenagrion puella (Rolff, Braune & Siva-Jothy 2001). In the Indian meal moth, Plodia interpunctella, immune activation at the final larval instar using a baculovirus reduced the number of both apyrene and eupyrene sperm (although this effect was marginally nonsignificant; Sait, Gage & Cook 1998). While the total numbers of apyrene and eupyrene sperm in an ejaculate are resource-dependent (Gage & Cook 1994), the proportion of each sperm type in an ejaculate is frequently resilient to manipulations of resources, such as through nutritional stress (Gage & Cook 1994). In this context, our finding that apyrene and eupyrene sperm were affected independently according to both immune challenge and age is unusual. Our study highlights the importance of considering different periods of development when examining resource trade-offs.

It should be noted that, because we assessed the number of sperm transferred, and not the number of sperm produced, the reduced eupyrene sperm transfer of 6th instar males may in fact reflect an altered ejaculation strategy by immune challenged males, rather than a resource-limited reduction in sperm production. It is, however, unclear why males would facultatively reduce the number of fertilizing sperm in an ejaculate following an immune challenge if sperm production itself was not affected, indeed, the terminal investment hypothesis would predict the opposite pattern (Reaney & Knell 2010). We suggest the more parsimonious explanation for the patterns observed is a reduced production of sperm owing to resource limitation derived from an immune function trade-off.

Effects of immune challenge on female fecundity

Female egg production was not affected by immune activation with LPS, unlike other studies, which have reported both increases (Adamo 1999) and decreases (Ahmed et al. 2002) in fecundity. Differences in male and female immune investment are expected (Adamo, Jensen & Younger 2001; Rolff 2002), with females showing increased immune parameters (Adamo, Jensen & Younger 2001; Rolff & Siva-Jothy 2002). Such differences in immune response may reflect the fundamental differences in male and female reproductive strategies (Rolff 2002). In species with conventional sex roles, male fitness is thought to be limited by the number of females he can fertilize, while female fitness is limited by the number of offspring produced (Bateman 1948) and in association, her longevity. The sex differences we observed here in reproductive trade-offs may not necessarily be indicative of the higher residual immune function of females, but may also reflect temporal differences between males and females in their investment into gamete production. Helicoverpa armigera oocytes are not developed at emergence, and adult feeding is required for ovary and egg maturation (Song et al. 2007). Females emerge earlier than males, and are not immediately receptive; pheromone production (which signals receptivity) initiates at least 2 days postemergence (Hou & Sheng 2000). The absence of an effect of immune activation on fecundity here could indicate that females mitigate such costs by altering their larval or adult feeding behaviour. It remains to be tested as to whether an immune challenge at this adult premating period, when females are accumulating further resources for oogenesis, would affect female reproductive output.

Effect of immune challenge on larval development

Immune activation also resulted in age-dependent developmental costs: individuals immune challenged at the 6th instar delayed their pupation, yet also demonstrated a reduced pupal duration, such that their emergence date was not affected. Investment into inducible immunity (following an immune insult) or constitutive immunity (such as prophylactic cuticular melanization) frequently results in developmental delays. For example, melanic mutant H. armigera demonstrated slower development at the larval and pupal stage, and a lower mating rate and fecundity (Ma et al. 2008). The development of resistance is also associated with developmental costs. Lines of P. interpunctella that were selected for increased viral resistance showed an associated developmental delay, reduction in egg viability and increase in pupal weight (Boots & Begon 1993). Induced responses to bacterial infection in invertebrates are often associated with ‘malaise syndrome’. Here, feeding and growth may be reduced, and metamorphosis delayed, while excretion may increase. These may act as a potential mechanism to clear the gut (a major site of bacterial infection in insects) of infected matter (Dunn et al. 1985). The observed prepupal developmental delay for 6th instar immune challenged males may have also contributed to the lack of response in apyrene sperm production, which undergoes spermatogenesis at the pupal stage.

Ontogenetic changes in immune response

Our data also revealed an increased investment into immune function (both PO and lytic activity) with larval instar. A potential caveat is that the haemolymph samples were stored at −20 °C. Samples were analysed at the same time, meaning that 3rd instar samples were typically stored for a longer period than 6th instar samples. If PO and lytic activity degrade at −20 °C, then this may potentially affect our finding that 3rd instars have a lower immune function that 6th instars. However, it would not affect our key finding that within instars LPS increases lytic activity, as samples within instars were treated identically. These results are consistent with studies of viral resistance in invertebrates, which demonstrate increasing resistance with age, a pattern that can be accounted for by increasing larval weight. Yet, age-related resistance, in some species, is above that predicted by body weight alone, suggesting the importance of developmental period on immunocompetence (Briese 1986). Resistance in Harmigera, however, is determined largely by larval weight, rather than instar or age per se (Hochberg 1991). Lysozyme and other antibacterial peptides are synthesized by the fat body within hours of an immune challenge in other Lepidoptera (Dunn et al. 1985). The larger fat reserves of larger (older) instars might allow for a greater potential for synthesis and thus a greater immune response. Importantly, because the effect of relative weight on lytic activity was identical in LPS- and ringer-only-treated groups, any effects of LPS appear to be qualitative, rather than dependent on the eventual concentration of LPS in the haemolymph. This idea is supported by statistically similar results in lytic response when individuals were given a lower quantity of LPS (unpublished data). This finding further confirms that the small proportional difference in administered ringer between 3rd and 6th instars did not confound results.

Resource trade-offs may also occur between components of the immune system itself, in both immune and nonimmune challenged individuals. Investment in PO and lytic activity are negatively correlated in a number of invertebrate species (Moret & Schmid-Hempel 2001; Cotter, Kruuk & Wilson 2004; Cotter et al. 2004, 2008; Povey et al. 2009). As expected, the bacterial immune challenge in our study elicited an up-regulation of the antibacterial (lytic) response, yet unlike patterns observed for other species, this did not come at the expense of PO activity. Rather, our data revealed a positive correlation between PO and lytic activity, particularly for nonimmune challenged individuals at the 6th instar. Trade-offs within the immune system may become more obvious when resources, such as dietary protein are limiting. In our study, individuals were provided with a high-protein ad libitum diet, which is positively correlated with indices of immune function (Siva-Jothy & Thompson 2002; Povey et al. 2009; Triggs & Knell 2012). Immune challenged individuals are also known to actively select foods that can fuel increased immune function, potentially mitigating the resource costs elicited from its up-regulation (Povey et al. 2009). The fact that individuals in this study were maintained on a high-protein ad libitum diet may have reduced the necessity for trade-offs between components of immune function. However, the fact that phenotypic trade-offs in male sperm quantity occurred under these same conditions suggests that the absence of any additional trade-offs between immune components here is real.

Concluding remarks

Immune function has clear costs, and life-history trade-offs are made to balance these costs to maximize fitness over the lifetime of the individual. Our study demonstrates the complexity of potential trade-offs between gamete investment and immune function. These trade-offs differed not only according to sex, but were also dependent on the developmental window when the immune challenge occurred. The fact that the nature of these trade-offs differed over the juvenile period highlights the importance of considering the entire lifespan of an individual when making inferences about the type and degree of trade-offs between immune function and reproduction.

Acknowledgements

We would like to thanks CSIRO Narrabri for our Helicoverpa armigera culture and Dr Sharon Downes for the culturing advice. We also thank Dr Sheena Cotter, Maxine Beveridge and Kylie Wilson for their advice on immune assays. K.B.M. was funded by the Margaret Clayton Fellowship (Monash University) and by an ARC Australian Postdoctoral Fellowship (DP110101163). T.M.J. was supported by a University of Melbourne Research Fellowship. L.W.S. was supported by an ARC Australian Professorial Fellowship (DP110104594).

Ancillary