Abstract
The expression of testosterone-dependent sexual traits might signal the ability of their bearers to cope with parasite infections. According to the immunocompetence handicap hypothesis (IHH), such signals would be honest because physiological costs of testosterone, such as a reduced ability to control parasite infections, would prevent cheating. We tested whether testosterone would affect the outcome of a standardized parasite challenge in red grouse, using a main parasite of the species, the nematode Trichostrongylus tenuis. We caught males in spring, removed their nematode parasites, and implanted them with testosterone or empty implants, as controls. After 1 month, they were reinfected with a standard dose of infective T. tenuis parasites. When challenged, testosterone males had relatively less globulin relative to albumin plasma proteins than control males, an indication that they had experienced increased physiological stress. Testosterone-treated males had significantly more T. tenuis parasites than controls in the next autumn and also had more coccidia and lost more weight than controls. Testosterone-treated males nevertheless benefited from their elevated spring testosterone: they had bigger sexual ornaments than controls both in spring and autumn, and they tended to have a higher pairing and breeding success than controls. Our results supported the IHH in showing that elevated testosterone impaired the ability of males to cope with a standardized challenge by a dominant parasite. Testosterone thus plays a key role in mediating trade-offs between reproductive activities and parasite defense, and testosterone-dependent comb size might honestly signal the ability of red grouse to control T. tenuis infection.
Many animals exhibit secondary sexual characters, the size or brightness of which function in intrasexual competition and mate choice (Andersson, 1994). In males, sexual ornament expression can signal various individual qualities, but one that has attracted researchers' interest for a long time is the possibility that sexual traits reliably advertise the inherent ability of individuals to resist parasite infections (Hamilton and Zuk, 1982). Females could then benefit from choosing males on the basis of these traits, by pairing with a mate that has fewer parasites and is better able to cope with parasites, hence passing “good genes” to its offspring (Andersson, 1994; Hamilton and Zuk, 1982; Møller, 1990; Møller et al., 1999; Zuk, 1992).
For sexual signaling to be reliable, the expression of these traits should be costly, in order to prevent cheating (Zahavi A and Zahavi A, 1997). The androgen testosterone plays an important role in the expression of many sexual traits, with elevated testosterone concentration being necessary for initiating their development or for increasing their size or elaboration (e.g., Folstad and Karter, 1992; Hillgarth and Wingfield, 1997). However, maintaining high levels of testosterone is often costly. Costs might include increased use of resources for the more intense displays and behaviors or interference with other physiological mechanisms (e.g., Grosmann, 1985; Wingfield et al., 2001). Because testosterone has multiple effects, it may create trade-offs between different fitness-related life-history traits or modulate trade-offs already existing between different characters due to resource allocation constraints. Folstad and Karter (1992) suggested that testosterone-dependent sexual traits would reliably signal health because of immunosuppressive effects of testosterone that would impair a male's ability to resist parasite infections (e.g., Folstad and Karter, 1992; Hillgarth and Wingfield, 1997; Roberts et al., 2004; Wingfield et al., 2001).
Because it is based on trade-offs, the immunocompetence handicap hypothesis (IHH) is best investigated using experimental studies (Getty, 2002; Roberts et al., 2004). Testosterone levels can be artificially increased by implants, so allowing workers to modify the intensity of sexual signaling by males and to investigate the effects on immunocompetence and parasite infections (e.g., Casto et al., 2001; Duckworth et al., 2001; Saino and Møller, 1994) or on fitness components such as viability, mating, and breeding success (e.g., Alatalo et al., 1996; Moss et al., 1994). In birds, direct experimental evidence for immunosuppression by testosterone is still limited. Some studies found evidence that testosterone is immunosuppressive (e.g., Buchanan et al., 2003; Casto et al., 2001; Lindström et al., 2001; Mougeot et al., 2004; Owen-Ashley et al., 2002, 2004; Peters, 2000), but others found little or no support for this effect (Hasselquist et al., 1999; Ros et al., 1997). A recent meta-analysis of both observational and experimental studies testing the IHH in birds provided limited support for the hypothesis (see Roberts et al., 2004). More experimental work is needed to better assess the relevance of this hypothesis. Moreover, the immune system is complex, and the lack of effect of testosterone on certain aspects of immune function (for example, cell-mediated versus humoral immunity) does not exclude the possibility that other aspects might be affected. A direct way of testing predictions of the IHH would thus be to modify testosterone levels while investigating individuals' responses to a challenge by a parasite that is known to adversely affect the study host. This would specifically test whether testosterone impairs a male's ability to control infection by a relevant parasite. We conducted such an experimental test, using the red grouse Lagopus lagopus scoticus as a study species.
The red grouse is a medium size, territorial monogamous bird inhabiting heather moorland of the UK. Males compete for territories in autumn and defend them through the winter until spring, when breeding starts (Cramp and Simmons, 1980). Pairing starts in autumn but is not definitive until spring (Cramp and Simmons, 1980; Watson, 1985). An important parasite of red grouse is the cecal threadworm Trichostrongylus tenuis, a gut nematode well known for its pathological effects on red grouse (Delahay et al., 1995; Hudson, 1986a; Hudson et al., 1992; Shaw and Moss, 1990). The main sexual ornaments of grouse are their bright supraorbital combs, the size of which is testosterone dependent (Moss et al., 1979; Mougeot et al., 2005a) and which function in intra- and intersexual selection (Alatalo et al., 1996; Bart and Earnst, 1999; Moss et al., 1979; Rintamäki et al., 2000). Recent work showed that, in male red grouse, comb size indicates immunocompetence rather than T. tenuis infection intensity (Mougeot et al., 2004; Mougeot and Redpath, 2004). Males with bigger combs have greater cell-mediated immunity, despite cell-mediated immunity being reduced by testosterone (Mougeot et al., 2004). Males with bigger combs also have a lighter spleen than expected from their T. tenuis intensity, an indication that they might be better to cope with this parasite (see Mougeot and Redpath 2004). Previous experiments have shown that increased testosterone in autumn had no significant short-term effect on T. tenuis intensity (Mougeot et al., 2004). However, in autumn, parasite transmission is low, and ingested larvae can arrest their development until the next spring (Hudson and Dobson, 1997; Shaw, 1988a), so an effect of testosterone on this parasite might have been overlooked (see Mougeot et al., 2004). Indeed, this was supported by a further experiment, where increased autumn testosterone led to increased parasite densities in the following autumn (Seivwright 2004). In spring, however, T. tenuis larvae de-arrest, previously ingested larvae develop into egg-producing worms and parasite intensity can rapidly increase (Hudson and Dobson, 1997; Moss et al., 1993), and a short-term effect of testosterone on T. tenuis intensity might be more noticeable. Moreover, testosterone concentration naturally peaks in spring (see Mougeot et al., 2005a), when male sexual and displaying activities for mate choice are at their peak.
In this experiment, we caught males in spring, removed their nematode parasites, and implanted them with either testosterone or with empty implants (controls). One month later, we challenged males with a standard dose of 2000 T. tenuis infective larvae. If testosterone impairs a male's ability to cope with T. tenuis parasites, then we expected testosterone-implanted males to end up with more T. tenuis parasites than controls in the next autumn. We also investigated the effect of increased testosterone on infection intensity by another type of parasite (coccidia Isospora ssp.) on body mass and on the physiological condition of males using the relative concentration of albumin to globulin plasma proteins (A:G ratio) as an indicator of stress (Ots et al., 1998). We expected testosterone-implanted males to have a relatively greater A:G ratio, to lose more weight, to have more coccidia, and to have reduced survival probability than controls. We also investigated whether elevated testosterone benefited males, in terms of enhanced sexual ornamentation, increased pairing success, or breeding success.
METHODS
Experimental protocol
We conducted the experiment on a grouse moor located in Aberdeenshire, northeast Scotland (Tillypronie estate, 57° 09′ N–2° 59′ W). In spring 2004 (12–23 March), we caught 53 male grouse by dazzling and netting them at night. Each was individually ringed and fitted with a radio-collar (TW3-necklace radio-tags, Biotrack, Dorset, UK) to facilitate relocations and recaptures. Grouse were aged (young, i.e., born in previous summer versus old, i.e., >1 year old) from the shape and color of their second and third primaries (tips pointed and mottled in young, round and plain in old birds) and the texture of their claws (Parr, 1975).
Males were randomly assigned to one of two treatments: controls (n = 24) or implanted with testosterone (n = 29). All birds were implanted with one silastic tube (20 mm long, 0.62 mm of inner diameter, and 0.95 mm of outer diameter) sealed with glue at both ends. Implants were either empty tubes (control males) or filled with crystalline testosterone propionate (testosterone-treated males). The implants were inserted between skin and breast muscles on the flank, under local anesthesia. From previous trials on captive grouse, we estimated that the testosterone implant would be exhausted after 1.5 months (Redpath SM and Mougeot F, unpublished data).
On first capture, we weighed each male to the nearest 5 g using a Pesola balance. We also measured maximum wing length and the size (height and length) of a flattened supraorbital comb to the nearest millimeter, with a ruler. We calculated comb area (height × length, in square millimeter) as a measure of sexual ornament size (see Moss et al., 1979). We collected approximately 0.5 ml of blood from a sample of males from the brachial vein. Plasma samples were obtained by centrifuging blood for 10 min at 4000 rpm and were stored at −80°C until assayed for plasma protein (see below). We also collected fecal samples from each male to estimate T. tenuis and coccidia intensities. To do so, we kept males overnight in individual holding boxes, provided with heather Calluna vulgaris (their main food). Birds were released early the following morning, when we collected fecal samples. Samples were transported to the laboratory immediately, stored in a cold room at a constant temperature of 4°C to inhibit parasite egg development, and processed within 2 weeks of collection to ensure reliable parasite intensity estimates (Seivwright et al., 2004). Prior to release, we orally dosed all males with 1 ml of the anthelminthic levamisole hydrochloride (Nilvern Gold) to clear grouse of their helminth parasites and in particular of their T. tenuis worms (Hudson, 1986a). This treatment was used in recent experiments that demonstrated its effectiveness at clearing grouse from their T. tenuis parasites (e.g., Mougeot and Redpath, 2004; Mougeot et al., 2005b).
One month after implanting (10–15 April), we caught males, remeasured their comb size and body mass, took another blood plasma sample, and orally challenged them with 2000 T. tenuis parasite larvae that had been cultivated in the laboratory (see below). We recaught males in the same autumn (28 September–18 October), remeasured their comb size and body mass, collected a fecal sample for parasite egg counts (see below), removed the radio-tags, and released males into the wild.
Study parasites and fecal egg counts
One of the red grouse's dominant parasites is the cecal threadworm T. tenuis, a gut nematode that has a direct lifestyle with no alternative hosts within the same habitat (Hudson, 1986a). T. tenuis worms and infective larvae have well-established negative effects on the energetics, breeding success, and survival of grouse (Delahay et al., 1995; Hudson, 1986a; Hudson et al., 1992; Shaw and Moss, 1990). Red grouse are also often infected by coccidia Isospora ssp. (Fantham, 1911; Mougeot et al., 2004). Patterns of coccidia infection and effects are little known for red grouse, but coccidia can have important negative effects on the growth or condition of avian hosts, as reported in other species (e.g., Fehlberg and Pohlmeyer, 1991). For this study, we focused on these two parasites for which infection intensity is relatively easy to measure reliably from fecal samples of live birds.
On initial (spring) and final (autumn) captures, we estimated infection intensity by T. tenuis and coccidia using fecal egg counts. For each male, a subsample of approximately 0.2 g of faeces was diluted in 5 ml of saline water and mixed thoroughly. A subsample of this solution was placed in a MacMaster slide under a microscope in order to count eggs of each parasite species (see Seivwright, 2004, for further details on the method). T. tenuis fecal egg concentrations (eggs per gram) provide reliable estimates of worm burdens both in spring (Moss et al., 1990) and in autumn (Seivwright et al., 2004). For analyses, we used the egg count data, but we also calculated corresponding parasite intensities in terms of worms per grouse, which are more biologically meaningful than egg concentrations, using the equations provided by Seivwright et al. (2004). Infection rates by coccidia and the rate of oocyst shedding can show important diurnal variations (Brawner and Hill, 1999; Hudman et al., 2000), but this potential source of variation was minimized by collecting all fecal samples early in the morning.
Culture of T. tenuis infective larvae for parasite challenges
We used the remains of fecal samples collected on first capture to cultivate T. tenuis parasite infective larvae for the challenges, 1 month later. In outline, fecal samples were washed thoroughly over a 125-μm sieve (Endecotts Ltd, Moreden, London, UK) to remove coarse, fibrous material. The fecal residue containing the eggs was collected with a 25-μm sieve using distilled water, placed in petri dishes, and incubated at a constant temperature of 20°C. These fecal cultures were mixed and watered daily to maintain optimal moisture conditions for the hatching of T. tenuis eggs (Shaw, 1988b). After 2 weeks, the fecal contents were filtered and the larvae extracted and concentrated using a modified Baerman apparatus. Infective larvae were stored in distilled water for up to a week in a fridge at 4°C. On the day prior to challenges, the solution containing the infective larvae was mixed, and larval concentration was measured by counting live larvae in subsamples. Individual doses containing approximately 2000 T. tenuis infective larvae were then extracted and stored until given orally to males. According to previous work, about a third of these larvae should establish and develop into egg-producing worms (Shaw, 1988b); so we expected males to end up with 600–700 worms by the end of summer if the challenge had been successful. Further details on the methods for cultivating, storing, and counting T. tenuis larvae are given in Wilson GR and Wilson LP (1978), Shaw (1988b), and Seivwright (2004).
Plasma protein assays
Relative measures of gamma globulin and albumin concentrations were assayed by densitometric analysis after electrophoretic separation of plasma proteins on agarose gels (SAS-MX High Resolution Agarose Elecrophoresis Kit, Helena Biosystems Europe Ltd., Sunderland, UK). One μl of plasma was diluted 1:5 in barbital buffer (pH 8.6), and then 2 μl of this diluted sample was applied to the agarose gel. A standard of diluted chicken serum (Sigma Ltd., Poole, UK) was also applied in one lane of the gel. Electrophoreses were carried out at 250 V for 25 min. After electrophoresis, gels were stained, and densitometric analysis was performed using Gelworks 1D Advanced computer image analysis software (UVP Ltd., Cambridge, UK) Relative albumin and gamma globulin concentrations (A:G ratio) were measured as the proportion of the densitometric profile that they occupied. Plasma protein concentrations provide insights into individuals' physiological status and immunocompetence. Decreases in albumin concentrations are a common symptom of pathological states, and immunoglobulin levels typically increase in response to infections but decrease during stress and immunosuppression (Kawai, 1973; Ots et al., 1998).
Survival, pairing, and breeding success
Paired grouse stay and roost together prior to laying, which usually take place in mid–late April (Cramp and Simmons, 1980). In mid April, during catching at night or during daytime radio tracking, we checked whether males were paired or not, by looking for the presence of a female roosting or resting in close proximity (less than 5 m). Males that were never seen with a female in April and July were classified as “unpaired.” To measure survival, we radio tracked males in April, July, and September and flushed them to see if they were alive. In July, we located males and flushed them together with their families to measure their breeding success (number of young at fledging).
Statistical analyses
We used SAS 8.01 for the statistical analyses (SAS, 2001). Dependent variables were fitted to generalized linear models (GLM) (GENMOD procedure) using the following error distributions and link functions—comb size, body mass, and (arcsine-transformed) A:G ratio: normal error distribution and identity link function; counts of T. tenuis or coccidia eggs and counts of young per brood: Poisson error distribution and log link function; and survival probability and pairing status (paired or not): binomial error distribution and logit link function. For parasite data analyses, the dependent variables were the number of eggs or oocystes counted, with (the loge of) weight of fecal sample used for the count included as an offset in the models (SAS, 2001). When testing for an effect of treatment on changes over time (time × treatment interaction) in study parameters, we used generalized linear mixed models (GLIMMIX procedure) that included individual males as a random effect, to account for repeated measures within individuals. Parasite data are typically overdispersed and are expressed as geometric means (back log-transformed average of log-transformed data). For power calculations, we used the power calculator provided on the University of California Los Angeles Department of Statistics web site (http://calculators.stat.ucla.edu/powercalc/). All data are expressed as means ± SD, and all tests are two tailed.
RESULTS
Effect of testosterone on comb size
Prior to treatment, comb size did not differ significantly between treatment groups (GLM; F1,51 = 0.47, p = .49; Figure 1a). Comb size variation during spring (prior to and 1 month after treatment) was not significantly explained by age or treatment and was not related to wing length (as an index of male size; Table 1). However, comb size increased significantly more in testosterone-treated males than in control males between March and April (Table 1: significant time × treatment interaction; Figure 1a). We further tested whether differences in comb size between treatment groups lasted until the next autumn. Comb size variation between April and September was explained by time and treatment effects and by the time × treatment interaction (Table 1). Males had smaller combs in autumn than in spring (Figure 1a), but comb size decreased significantly less between April and September in testosterone-treated than in control males. The relative differences in comb size caused by implants in spring were thus increased between spring and autumn, when testosterone-treated males still had bigger combs than control males (Figure 1a).
Changes over time in mean ± SE (a) comb size and (b) body mass of males according to treatment (○, controls; •, testosterone implanted). Sample size, above bars, refers to number of males.
Changes over time in mean ± SE (a) comb size and (b) body mass of males according to treatment (○, controls; •, testosterone implanted). Sample size, above bars, refers to number of males.
Effect of treatment on changes in comb size and body mass
| Dependent variablesa | Comb size | Body mass | |||||
|---|---|---|---|---|---|---|---|
| Time period | df | F-value | p-value | F-value | p-value | ||
| (1) March to April | |||||||
| Age | 1,42 | 2.05 | .17 | 0.86 | .36 | ||
| Wing length | 1,42 | 0.11 | .73 | 1.15 | .29 | ||
| Time | 1,42 | 4.03 | .05 | 9.46 | <.01 | ||
| Treatment | 1,42 | 0.17 | .67 | 2.24 | .14 | ||
| Time × treatment | 1,42 | 12.18 | <.001 | 7.65 | <.01 | ||
| (2) April to September | |||||||
| Age | 1,36 | 2.01 | .17 | 0.33 | .57 | ||
| Wing length | 1,36 | 0.18 | .67 | 2.55 | .12 | ||
| Time | 1,36 | 89.58 | <.0001 | 40.06 | <.0001 | ||
| Treatment | 1,36 | 9.98 | <.01 | 6.79 | <.05 | ||
| Time × treatment | 1,36 | 4.36 | <.05 | 0.02 | .88 | ||
| Dependent variablesa | Comb size | Body mass | |||||
|---|---|---|---|---|---|---|---|
| Time period | df | F-value | p-value | F-value | p-value | ||
| (1) March to April | |||||||
| Age | 1,42 | 2.05 | .17 | 0.86 | .36 | ||
| Wing length | 1,42 | 0.11 | .73 | 1.15 | .29 | ||
| Time | 1,42 | 4.03 | .05 | 9.46 | <.01 | ||
| Treatment | 1,42 | 0.17 | .67 | 2.24 | .14 | ||
| Time × treatment | 1,42 | 12.18 | <.001 | 7.65 | <.01 | ||
| (2) April to September | |||||||
| Age | 1,36 | 2.01 | .17 | 0.33 | .57 | ||
| Wing length | 1,36 | 0.18 | .67 | 2.55 | .12 | ||
| Time | 1,36 | 89.58 | <.0001 | 40.06 | <.0001 | ||
| Treatment | 1,36 | 9.98 | <.01 | 6.79 | <.05 | ||
| Time × treatment | 1,36 | 4.36 | <.05 | 0.02 | .88 | ||
Analyses were conducted separately for two time periods: (1) during spring, before and after treatment (March–April) and (2) between spring, after treatment, and autumn (April–September). Data are shown in Figure 1.
Dependent variables were fitted to generalized linear mixed models using a normal error distribution. Models included “individual male” as a random effect to account for repeated measures.
Effect of testosterone on body mass
Body mass did not differ significantly between treatment groups prior to manipulation (GLM; F1,51 = 0.17, p = .58; Figure 1b). Variation in male body mass during spring was not explained by age or wing length (Table 1). However, within-individual changes in body mass between March and April differed between treatment groups: body mass varied little in control males but decreased in testosterone-treated males (Figure 1b). Body mass increased between April and September but did so similarly in both treatment groups (Table 1; nonsignificant time × treatment interaction). The difference in body mass caused by implants was thus maintained until the following autumn, when testosterone-treated males were significantly lighter than control males (Table 1; Figure 1b).
Effect of testosterone on relative albumin and globulin concentrations
Prior to treatment, albumin to globulin (A:G) ratios did not differ significantly between treatment groups (F1,40 = 2.22, p = .14), although ratios tended to be greater in controls than in treated males (Figure 2). However, changes over time (during spring) in albumin to globulin (A:G) ratio differed significantly between treatment groups (mixed model—time: F1,28 = 15.24, p < .01; treatment: F1,28 = 0.04, p = .84; time × treatment: F1,28 = 5.74, p < .05; Figure 2). A:G ratios increased between March and April in both treatment groups, but this increase was proportionally greater in testosterone-treated males than in control males (Figure 2).
Changes over time in albumin to globulin ratio (mean ± SE) according to treatment (○, controls; •, testosterone implanted).
Changes over time in albumin to globulin ratio (mean ± SE) according to treatment (○, controls; •, testosterone implanted).
Effect of testosterone on T. tenuis infection
In spring, prior to treatment, young males had fewer T. tenuis worms than old males, but infection intensity did not differ between control males and testosterone-treated males (GENMOD—age: F1,48 = 7.62, p < .01; treatment: F1,48 = 0.74, p = .39; age × treatment: F1,48 = 2.02, p = .16; Figure 3a). Average T. tenuis intensities (geometric means) were of 1213 and 893 eggs per gram in testosterone-treated and control males, respectively.
Changes over time in infection intensity by (a) Trichostrongylus tenuis and (b) coccidia according to treatment (white bars, controls; black bars, testosterone-implanted males). Parasite intensities are expressed as the mean ± SE of log-transformed concentration of parasite eggs in fecal sample. Sample size, above bars, refers to number of males. The experiment controlled for initial exposure to T. tenuis parasites: males were dosed with anthelminthic at the start of the experiment (nematode parasite removal) and challenged 1 month later with a standard dose of 2000 T. tenuis infective larvae.
Changes over time in infection intensity by (a) Trichostrongylus tenuis and (b) coccidia according to treatment (white bars, controls; black bars, testosterone-implanted males). Parasite intensities are expressed as the mean ± SE of log-transformed concentration of parasite eggs in fecal sample. Sample size, above bars, refers to number of males. The experiment controlled for initial exposure to T. tenuis parasites: males were dosed with anthelminthic at the start of the experiment (nematode parasite removal) and challenged 1 month later with a standard dose of 2000 T. tenuis infective larvae.
Five months later, in autumn (after parasite removal and challenge with 2000 T. tenuis infective larvae), young and old males had similar parasite intensity, but testosterone-treated males had significantly more worms than control males, in both age groups (GENMOD—age: F1,33 = 0.29, p = .59; treatment: F1,33 = 6.78, p < .01; age × treatment: F1,33 = 2.12, p = .15; Figure 3a). Testosterone-implanted males then had about twice as many T. tenuis parasites as controls (geometric means of 3370 and 1100 eggs per gram [equivalent to geometric means of 381 and 781 worms per grouse] in testosterone-treated and control males, respectively).
Because testosterone males tended to have more T. tenuis parasites than controls prior to experiment (see Figure 3a), we also tested whether changes over time in T. tenuis intensity between March (baseline) and September differed between treatment groups. This was the case (GLIMMIX model with individual as a random effect; significant time × treatment interaction: F1,35 = 4.27, p < .05), indicating that the autumn difference between treatment groups in T. tenuis intensity was not due to one group of males being, by chance, more susceptible to the parasite, or on areas with more natural infective larvae.
We further tested whether differences in T. tenuis intensity between treatment groups at the end of the experiments depended on initial comb size, as an index of initial testosterone concentration. Autumn T. tenuis intensity was significantly explained by treatment after controlling for initial comb size (GLM for log-transformed parasite intensity; treatment effect: F1,34 = 7, p < .05) but was not explained by initial comb size, although there was a tendency for higher parasite intensities to be associated with bigger initial comb size (F1,34 = 2.50, p = .13), in both treatment groups (comb size × treatment interaction: F1,34 = 0.62, p = .44).
Effect of testosterone on coccidia infection
Prior to treatment, coccidia intensity was higher in young than in old males (geometric means of 13,170 and 2452 eggs per gram of faeces for young and old, respectively) and tended to be higher in control males than in testosterone-treated males (GENMOD—age: F1,48 = 4.92, p < .05; treatment: F1,48 = 3.47, p = .06; age × treatment: F1,48 = 0.63, p = .43; Figure 3b). Average coccidia intensities (geometric means) were of 4571 and 7324 eggs per gram in testosterone-treated and control males, respectively.
In autumn, coccidia intensity still differed between age groups, although not significantly, and was higher in testosterone-treated males than in control males, in both age groups (GENMOD—age: F1,33 = 3.39, p = .07; treatment: F1,33 = 6.80, p < .01; age × treatment: F1,33 = 2.44, p = .12; Figure 3b). Average coccidia intensities were of 1287 and 701 eggs per gram in testosterone-treated and control males, respectively.
Because of the initial differences between treatment groups, we further tested for a treatment effect on changes over time (from spring to autumn) in coccidia intensity. Coccidia intensity decreased between spring and autumn but did so less in testosterone-treated males than in control males (GLIMMIX—age: F1,60 = 7.07, p < .01; time: F1,54 = 6.67, p < .05; treatment: F1,72 = 1.09, p = .30; time × treatment: F1,54 = 13.31, p < .001; Figure 3b).
Effect of testosterone on pairing success, breeding success, and survival
One month after implanting, the proportion of males that were paired did not differ between age groups (GENMOD:
Breeding success (number of young fledged per male in July) did not differ between age groups (GENMOD:
The percentage survival of males did not differ between age groups or between treatment groups in April, July, or September (GENMOD—treatment effect: April,
Percentage survival between April and September in control males (○) and in testosterone-implanted males (•). Sample size included 24 control males and 29 testosterone-implanted males.
Percentage survival between April and September in control males (○) and in testosterone-implanted males (•). Sample size included 24 control males and 29 testosterone-implanted males.
When accounting for the differences in mortality (by considering a breeding success of zero if a male died and failed to reproduce), males treated with testosterone in spring still tended to raise more young (1.03 ± 1.76, n = 29) than control males (0.42 ± 1.14, n = 24; GENMOD:
DISCUSSION
Our initial hypothesis was that elevated spring testosterone would be costly, in terms of increased susceptibility to parasites and reduced condition. This was supported by the data, which showed that testosterone led to increased infection by T. tenuis after a standardized challenge and also led to increased infection by coccidia and resulted in a greater loss of weight. We also found some benefits of elevated spring testosterone, in terms of enhanced sexual ornamentation and a tendency for increased pairing and breeding success. We discuss the role that this hormone might play in trade-offs between reproductive activities and parasite defense.
Effects of testosterone on T. tenuis parasite intensities
After experimental treatment, we found that T. tenuis parasite intensities differed significantly between treatment groups in autumn, with testosterone-implanted males having almost twice as many T. tenuis parasites as the controls. This could be explained in terms of either increased host susceptibility or increased exposure to infective stages. Grouse ingest parasite infective larvae through their main food, heather (Hudson, 1986b). Male red grouse implanted with testosterone are typically more aggressive, interact more often with neighboring males, and expand their territory to the detriment of other males (Moss et al., 1979, 1994; Mougeot et al., 2005a). The greater increase in parasite intensity observed in testosterone-treated than in control males might have been because testosterone increased exposure to infective larvae, for instance, by increasing interactions with other infected individuals or by increasing territory size or home range (Clayton and Moore, 1997). An alternative explanation is that the testosterone treatment affected host susceptibility and the outcome of the parasite challenge.
In wild grouse, parasite intensity can increase rapidly in spring, when larvae that have arrested their development in the previous autumn develop into egg-producing worms (Hudson and Dobson, 1997; Moss et al., 1993; Shaw, 1988a). This reemergence of arrested larvae accounts for the increased recruitment into the adult worm population usually observed in grouse during spring, but the timing of de-arrestment can differ between sites: February–March in a study by Hudson and Dobson (1997) and mid-April in another study by Moss et al. (1993). By removing and challenging with T. tenuis larvae, our experiment aimed at mimicking this sudden increase in infective larvae in spring, the effect of testosterone on this process, and standardizing initial exposure to infective larvae. When we challenged males with the parasite larvae, 1 month after implanting, testosterone-treated males had bigger combs, which underlined higher plasma testosterone concentration, than control males (testosterone concentration and comb size are positively correlated; see Mougeot et al., 2005a).
Differences between treatment groups in relative albumin and globulin concentration (A:G ratios) at the time of the parasite challenge also indicated that testosterone-treated males were physiologically stressed and immunologically more susceptible. At the start of the experiment, males were dosed with an anthelminthic, which cleared them of their T. tenuis and other nematode parasites. In red grouse, globulin concentration increases with T. tenuis intensity (Wilson, 1983; Wilson GR and Wilson LP, 1978). Thus, the overall increase in A:G ratio observed in control males during spring (Figure 1) could be explained by the removal of these parasites. One month after implanting, testosterone-treated males had relatively higher A:G ratio than control males, which suggests that the testosterone treatment caused increased physiological stress and a relative decrease in gamma globulin concentration (Kawai, 1973; Ots et al., 1998) and might have increased susceptibility to T. tenuis infective larvae.
In red grouse, there is little evidence of acquired immunity to T. tenuis infection (Hudson and Dobson, 1997; Shaw and Moss, 1989), but innate immunity is likely to be important. In captivity, grouse show a wide variation in innate susceptibility to the same dose of T. tenuis larvae, and in wild grouse, relative differences in parasite intensity among individuals within years tend to persist across years, so that relatively high or low parasite intensities appear to be characteristics of individual birds (Moss et al., 1993; Shaw and Moss, 1989). Our observations are thus consistent with the hypothesis that elevated spring testosterone was associated with a reduced ability of males to cope with the T. tenuis challenge. If parasite transmission was low, as suggested by the low intensities of T. tenuis measured in grouse in spring, prior to parasite removal, then the observed difference in T. tenuis intensities between treatment groups might have been be mainly due to differences in susceptibility.
With this experiment, we cannot clearly distinguish between the relative importance of susceptibility and exposure mechanisms leading to higher T. tenuis intensity in testosterone-implanted males. However, recent experiments conducted on red grouse showed that elevated testosterone was associated with reduced T-cell–mediated immunity (Mougeot et al., 2004) and that increased parasite intensity following experimentally elevated testosterone was due to effects on host susceptibility rather than exposure (Mougeot et al., 2005d). Whether the actions of testosterone on immune function were direct, as suggested originally by Folstad and Karter (1992), or indirect, via cascading effects on other hormones like corticosterone (the stress-mediated IHH; Buchanan, 2000; Evans et al., 2000; Owen-Ashley et al., 2004), is still unclear, but recent works pointed toward an indirect pathway (Mougeot et al., 2005d; Owen-Ashley et al., 2004). In that respect, our observation that males implanted with testosterone experienced increased physiological stress (as highlighted by the differences in A:G ratios between treated and control males) is consistent with the stress-mediated IHH.
Effect of testosterone on coccidia infection
We also found that testosterone-implanted males had greater coccidia intensities than controls in the autumn after implanting, while they tended to have fewer coccidia than controls at the onset of the experiment. The experiment did not control or standardize treatment groups for infection by coccidia, and it is not known what effect the anthelminthic treatment might have had on this parasite. The overall decrease in coccidia intensity between spring and autumn might have been due to the anthelminthic treatment or might reflect natural seasonal variations, but more work is needed on patterns of coccidia infection in natural red grouse populations to reach a conclusion. Nevertheless, the results suggested that testosterone affected the ability of males to control infection by coccidia. These results are also consistent with those of another study showing that testosterone implants in autumn increased coccidia infection in male red grouse (Mougeot et al., 2004). The findings further agree with other studies on birds showing that experimentally increased testosterone results in greater parasite intensities (Duckworth et al., 1999; Hughes and Randolph, 2001; Saino et al., 1995).
Effects of testosterone on changes in body mass
Testosterone-implanted males lost more weight than controls during spring, when the testosterone implants were active, and the birds remained lighter than controls in the next autumn. This is consistent with previous findings showing that elevated testosterone results in a loss of weight and condition (e.g., Duckworth et al., 2001; Mougeot et al., 2004; Zuk, 1996). This might be due to the behavioral and physiological actions of testosterone, with more frequent displays and interactions, greater energy expenditure, and possibly also reduced feeding rate (Wingfield et al., 2001). The greater mass loss might also have been due to the greater parasite intensities and their detrimental effects on host condition.
Effects of testosterone on survival and breeding success
The effect of testosterone on parasite intensities and body mass (condition) suggests that testosterone was costly and could impair survival. We found that testosterone-implanted males tended to suffer higher mortality than controls in early spring, although the difference was not statistically significant. Nevertheless, a 12% difference in survival might be a substantial cost. Post hoc analyses indicated that, for survival rates of 71% and 83%, as observed between March and September in treatment and control groups respectively, a sample size of 167 in each treatment group would be required to achieve a statistical power of 80% with a 5% significance level. A far greater sample size would thus be required to ascertain this finding, but it is nevertheless consistent with other studies conducted on the same species (Redpath et al., in press) and on other species (e.g., Wingfield et al., 2001).
Individuals can benefit from having high levels of circulating testosterone in a variety of ways, such as increased competitive ability for access to breeding resources such as a territory or mates, and thus benefit in terms of increased offspring production (e.g., Moss et al., 1994; Raouf et al., 1997; Veiga et al., 2001; Wingfield et al., 1990). It is well established that the size of red grouse combs is testosterone dependent (Moss et al., 1979; Mougeot et al., 2004, 2005a). Accordingly, we found that testosterone-implanted males had bigger combs than controls 1 month after implanting. Comb size plays an important role in male-male competition, as males with bigger combs advertise a dominant status, are more aggressive (Moss et al., 1979, 1994), and are more likely to obtain a territory during autumn contests than others (MacColl et al., 2000). Comb size also plays a role in mate choice in birds of the grouse family, with females preferring males with bigger combs (Bart and Earnst, 1999; Brodsky, 1988; Rintamäki et al., 2000). Male grouse treated with testosterone could thus have benefited from their bigger combs in terms of sexual attractiveness and pairing success in addition to increased territory size, and elevated testosterone could thus lead to higher reproductive success (Moss et al., 1994). The effects of increased spring testosterone on breeding success depend crucially on the timing and duration of implants because testosterone can interfere with paternal care (e.g., Hunt et al., 1999). Implants that maintain high testosterone into the chick-rearing period can lead to decreased parental care and reduced breeding success. In our study, implants were designed to be exhausted in late April, that is, before hatching, so we did not expect any negative effect of the testosterone implants on paternal care.
We found that breeding success tended to be higher (+0.7 chick per male on average) in testosterone-implanted males than in control males. Breeding success was particularly poor for the species in the year the study was conducted, probably because of bad weather in early June, so we might have lacked statistical power to detect significant differences in breeding success between treatment groups. Post hoc analyses indicated that, for the observed breeding success of control and testosterone-treated males, a sample size of 55 males in each group would be required to achieve a statistical power of 80% with a 5% significance level. Nevertheless, the finding is consistent with previous studies on red grouse showing that testosterone-implanted male red grouse achieved a better breeding success and thus benefited from their elevated testosterone levels (Moss et al., 1994; Redpath et al., in press).
Overall, testosterone-treated males tended to survive less but bred better than control males. The net balance was still for testosterone-treated males to raise more offspring (about two times more) than control males over the breeding season that followed treatment. The quality of the offspring produced could also have differed. The differential costs and benefits of elevated testosterone might last longer than one breeding season. For instance, the higher parasite intensities and the poorer condition found in the testosterone-treated males 6 months after treatment could impact on their future survival and breeding success. So although males treated with testosterone tended to benefit on the short term, in terms of breeding success, the long-terms costs and benefits should be investigated.
Delayed effects of elevated spring testosterone on autumn comb size
We also found that the difference in comb size between treatments was maintained until the following autumn, that is, after the testosterone implants were exhausted. This observation suggests further benefits of elevated testosterone and is consistent with previous studies showing long-lasting effects of autumn testosterone implants on aggressiveness in grouse (Mougeot et al., 2003a,b, 2005c). This observation however differs from that of Moss et al. (1994), who gave male red grouse testosterone pellets and found no long-lasting effect of the brief increase in aggressiveness on subsequent territorial behavior or density. The differences might be due to the type of implants used (pellets causing a sharp but brief increase in testosterone concentration versus a silastic tube causing a more regular and prolonged increase). In our study, the difference in comb size between testosterone-treated and control males might have lasted until autumn because the implants allowed males to acquire a dominant status in spring and maintained this status afterward. Because testosterone-implanted males continued to have bigger combs than controls in the autumn, they might further benefit during the autumn territorial contests, by being more likely to maintain a territory for the next breeding season (see MacColl et al., 2000).
In conclusion, our experiment showed that elevated spring testosterone was costly, in terms of increased parasite intensities. The IHH predicts a physiological cost of elevated testosterone leading to a reduced ability to control parasite infections (Casto et al., 2001; Folstad and Karter, 1992; Lochmiller, 1995). Direct experimental evidence for immunosuppression by testosterone is still limited in birds (Roberts et al., 2004). In red grouse, previous work conducted showed that testosterone was immunosuppressive in red grouse (Mougeot et al. 2004). The results of this experiment showed that elevated spring testosterone caused increased physiological stress and led to greater intensities by two main parasites, T. tenuis and coccidia. Our results are thus consistent with the key predictions of the IHH in showing that males can benefit from elevated spring testosterone and that comb size might reliably indicate a male's healthiness because testosterone also impairs a male's ability to control infection by a main parasite.
We are grateful to the landowner and keepers of the Tillypronie estate for allowing us to conduct the work on their grouse moor. Particular thanks are due to the headkeeper S. McConnachie for his help with organizing the fieldwork. We also thank F. Leckie, S. Evans, and G. Jones for their help with the fieldwork and laboratory analyses. M. Evans and P. Sharp kindly provided advice and assistance on implanting procedures. R. Moss and three anonymous reviewers provided helpful comments on the manuscript. This work was funded by a NERC grant (NER/A/S/1999/00074) and was carried out under a U.K. Home Office license (PPL80/1438).
References
Author notes
aCentre for Ecology and Hydrology, Hill of Brathens, Banchory, AB31 4BW, Scotland, UK
bInstituto de Investigaciones en Recursos Cinegeticos (CSIC-UCLM-JCCM), Ronda de Toledo s/n, 13005 Ciudad Real, Spain
cNatural Environment Research Council Molecular Genetics in Ecology Initiative, School of Biological Sciences, University of Aberdeen, Aberdeen AB24 2TZ, UK
