SEARCH

SEARCH BY CITATION

Keywords:

  • good genes;
  • maternal effects;
  • sexual selection

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Peacocks are a classic example of sexual selection, where females preferentially mate with males who have longer, more elaborate trains. One of the central hypotheses of sexual selection theory is that large or elaborate male ‘ornaments’ may signal high genetic quality (good genes). Good genes are thought to be those associated with disease resistance and as diversity at the major histocompatibility complex (MHC) has been shown to equate to superior immune responses, we test whether the peacock’s train reveals genetic diversity at the MHC. We demonstrate via a captive breeding experiment that train length of adult males reflects genetic diversity at the MHC while controlling for genome-wide diversity and that peahens lay more, and larger, eggs for males with a more diverse MHC, but not for males with longer trains. Our results suggest that females are assessing and responding to male quality in terms of MHC diversity, but this assessment does not appear to be via train length, despite the fact that train length reflects MHC diversity.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Female choice is thought to have driven the evolution of elaborate plumage in sexually dimorphic species such as peafowl (Pavo cristatus), where males provide no paternal care (Petrie et al., 1991; Petrie & Halliday, 1994; Yasmin & Yahya, 1996; Loyau et al., 2005; but see Takahashi et al., 2008; Loyau et al., 2008). However, why females have mate preferences is still a mystery. Where males provide no direct benefits for offspring there is a prima facie case for the involvement of so-called ‘good genes’ or indirect benefits for offspring. However, although there is evidence (mainly from birds) that females may gain more surviving offspring from mating with preferred males (Petrie, 1994; Møller & Alatalo, 1999), it is less clear exactly what constitutes a genetically good quality male and how genetic diversity is maintained amongst males. W.D. Hamilton was the first person to suggest that secondary sexual traits might reveal an individual’s ability to resist disease, and that the constant co-evolutionary arms race between hosts and parasites would be sufficient to maintain genetic diversity amongst males even when subjected to strong directional sexual selection. Under this scenario the hypothetical benefit of mate choice is the advantage of parasite resistance accruing to the offspring of choosy females from mating with males with a superior genetic background (Hamilton & Zuk, 1982) in the face of disease, with only males of superior genetic constitution being able to produce exaggerated secondary sexual characters (Andersson, 1994; Jennions et al., 2001).

The major histocompatibility complex (MHC) is known to be one of the most polymorphic areas of the genome (Bernatchez & Landry, 2003; Piertney & Oliver, 2006). This high diversity is not surprising given the role the MHC plays in immune system function. MHC genes code for peptide-binding molecules that recognize and bind foreign peptides, initiating an immune response (Klein, 1986). Because each peptide-binding molecule recognizes a small number of foreign peptides, individuals with higher genetic diversity at the MHC will be capable of recognizing a greater number of pathogens and thus combating a wider range of diseases (Piertney & Oliver, 2006). This direct link between MHC diversity and immune system response allows us to extend Hamilton’s link between secondary sexual traits and an individual’s ability to resist disease to the proposal that secondary sexual traits might reveal an individual’s MHC diversity and that in species where females do gain more surviving offspring from their choice, they may well be assessing genetic quality of mates at the MHC.

There has been considerable debate about the role of the MHC in mate choice, with particular focus on MHC dissimilar mating (Penn, 2002) resulting in an optimal number of MHC alleles in the offspring (Milinski, 2006). Although the focus has been on the benefits of MHC diversity in offspring, mate choice based on MHC diversity in males (indicating mate quality) has rarely been addressed. Only two previous study systems have demonstrated that there may be a link between female choice and the underlying variation in genetic diversity at the MHC in mates (von Schantz et al., 1997; Reusch et al., 2001; Aeschlimann et al., 2003). In sticklebacks, Gasterosteus aculeatus, females utilize ‘allele counting’ to identify potential mates (Reusch et al., 2001), although this result is likely to reflect female choice for dissimilar mates producing offspring with an optimal number of alleles, rather than be a reflection of mate quality as such (Aeschlimann et al., 2003; Milinski, 2003) and, in pheasants, Phasianus colchicus, it has been shown that male viability and male spur length may be associated with a particular MHC genotype (von Schantz et al., 1997). However, no study to date has shown that elaborate sexually selected plumage in birds is associated with MHC diversity. The peacock’s train is a classic example of sexually selected plumage and, in this study, we test whether train length of blue peacocks is a revealing handicap reflecting underlying genetic quality at the MHC.

Female birds have been shown to invest differentially in the offspring of males with superior genotypes, as would be expected if differential investment increased the viability of offspring with high reproductive value due to paternal genetic effects (Burley, 1986; Sheldon, 2000). So, if it is the case that MHC diversity is a genetic quality that females assess, we expect that females will differentially invest in offspring of those males with more diversity at the MHC. We report here on a captive breeding experiment on blue peafowl where a number of males were randomly mated with a number of females. The aims of the experiment were to determine separately the paternal and maternal genetic effects on the number and size of eggs produced, and thus to assess whether paternal reproductive success was associated with MHC diversity.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Captive breeding experiment

We conducted a captive breeding experiment in which a total of 46 adult males (fathers) each were randomly mated with four females, in total 189 adult females (mothers). Each mother was allocated to a single father for the entire breeding season, so each mother was mated with only one father. Eggs were collected daily, numbered and weighed on a digital top pan balance to the nearest 0.1 g. Any mother that died during the experiment was replaced and mothers and fathers were allocated to pens at random at the start of the breeding season. During the experiment only four females were replaced (2% of total females) and all replacements occurred during the first 4 weeks of the 20-week breeding season. Blood samples were taken from each individual for MHC and microsatellite genotyping, and parentage and sex of offspring were determined genetically. At the start of the breeding season, on the day adults were allocated to pens, body weight (g) and tarsus length (mm) were measured for all adults, and train length (length of the longest eye spot feather in mm), mean train eye spot size and number of train eye spots were measured for all males. Eye spot size was calculated as mean eye spot width (measured at the widest point) × mean eye spot height (measured at the greatest vertical diameter), with mean values being calculated from 10 randomly chosen eye spots for each father’s train. Eye spot number was the number of eye spots counted from photographs taken of each male. Because the count was taken from photographs, and it is difficult to get a photograph where the train is completely visible, this measure is an estimate of number of eye spots rather than an accurate count. The peafowl were fed a poultry layers pellet that did not contain any antibiotics. The breeding experiment was conducted over 2 years (in 1998 and 1999), with mothers and fathers randomly reassigned to mates between the 2 years. Here, we report only the results of the second year as many fathers were under 4 years of age in the first year, and the males’ train is not fully developed before the age of 4 years (Takahashi et al., 2008).

The random allocation of four mothers to each father was designed to reduce the impact of any maternal effects in the analyses of reproductive output, including any impact of previous mate history. However, because the breeding experiment was conducted over 2 years, we were able to determine whether the ‘quality’ of the mate in year one (measured as mate’s MHC diversity) had any impact on the reproductive output of mothers in year two. For both measures of reproductive output assessed (egg weight and egg number) we found no relationship between current reproductive output and the MHC diversity of the previous year’s mate (egg weight: F1,1342 = 0.011, = 0.918; number of eggs: F1,143 = 0.013, = 0.908).

Adults for the breeding experiment were from a well-established population that has been captive for at least 40 years, and to which there have been many introduction events from several parts of the UK throughout its existence. The captive adults, therefore, came from a relatively old, well-mixed population originating from a variety of sources; so, our results should not be biased by a small number of founders, or by recent population admixture causing linkage disequilibrium among traits.

MHC diversity

DNA was extracted from whole blood using BioRad Chelex®100 (Hillis et al., 1996). Exons 2 and 3 of the MHC class II β-chain were amplified for four individuals using primers C37 and C60 (Jacob et al., 2000) designed for the chicken (Gallus gallus). From the sequences of these four individuals, a new primer (PCMHC1: 5′-GCTCCTCTGCACCGTGA-3′) was designed to be used in conjunction with C37 to amplify 276 bp of the highly polymorphic exon 2. Exon 2 of the class II β-chain was amplified for each individual in 30 μL PCR reactions as follows: 1×Taq buffer [16 mm (NH4)2SO4, 67 mm Tris–HCl, 0.01% Tween-20], 1.0 mm MgCl2, 0.2 mm each dNTP, 1.0 μm each primer (C37 and PCMHC1), 1.0 U Taq (Bioline, London, UK) and 1.0 μL template DNA, with the following reaction cycle: 1 min at 95 °C, followed by 35 cycles of 95 °C for 1 min, 50 °C for 30 s, 72 °C for 1 min, with a final extension of 72 °C for 10 min. PCR products were then separated and detected using RSCA (reference strand-mediated conformational analysis) (Drake et al., 2004). Two alleles, named allele 1 and allele 4 with 20% sequence difference, were used as fluorescently labelled reference strands (FLRs). FLRs were created following a protocol described previously (Drake et al., 2004). Allele 1 was labelled with FAM(6-FAM) and allele 4 was labelled with JOE (ABI dyes, Foster City, CA, USA). Conditions for annealing FLRs to each individual’s PCR product were 95 °C for 10 min, 55 °C for 15 min, 50 °C for 15 min and 4 °C for 15 min, with 2 μL of PCR product and 2 μL of diluted FLR (diluted 1 in 20) per reaction. There were two separate annealing reactions per individual; one to allele1-FAM and one to allele4-JOE. The fluorescently labelled double-stranded products were detected on an ABI 310 Prism® Genetic Analyser, using 3% Genescan® nondenaturing polymer with ROX-500 internal size standard, and analysed and sized using Genescan® software (Applied Biosystems, Foster City, CA, USA) (Supporting Information Fig.  S1). Run conditions were 15 kV injection voltage, 15 s injection time, 13 kV run voltage, 32 °C run temperature and 15 min run time. All alleles present in each individual were identified by size (in bp) using both FLRs. All 12 alleles detected in the experimental population were isolated via cloning using Qiagen pDrive cloning vector (Qiagen, Crawley, UK) and QAComp-C01 competent cells (Qbiogene, Cambridge, UK) and then sequenced using BigDye Terminator Cycle Sequencing chemistry (Applied Biosystems), and sequences detected on an ABI 310 Prism® Genetic Analyser. Isolated alleles were then annealed to each of the two FLRs separately and run on the ABI 310 as above, 20 times each, to determine the range of the migration pattern for each allele. This was necessary to ensure correct allocation of alleles to samples run at different times.

Major histocompatibility complex diversity was calculated as the number of alleles present at exon 2 of the class II β-chain. This region exists as multiple copies in many species, and our results (up to five alleles detected per individual) suggest that it is triplicated in P. cristatus, and that all copies of the locus are polymorphic. In our sample of adults, we detected 12 different alleles (GenBank accession numbers: AY928093AY928104), with all individuals possessing between two and five alleles. All 46 fathers possessed either three or five alleles. Throughout the paper, fathers with three MHC alleles are referred to as ‘low MHC diversity’ fathers and those with five MHC alleles are referred to as ‘high MHC diversity’ fathers. The lack of fathers with two or four alleles is most likely a sampling effect due to the small number of fathers in the experiment. Among the mothers, the frequencies of the four genotypic classes were two alleles: 0.069, three alleles: 0.720, four alleles: 0.005 and five alleles: 0.206. It is not surprising, therefore, that among the much smaller sample of fathers the two rare genotypic classes were not sampled. Among the male offspring, we detected individuals with two, three, four and five alleles, at frequencies similar to those in the female offspring.

Parentage analysis

All adults were genotyped at 13 microsatellite loci (PC3, PC41, PC46, PC142, PC151, PC125, PC281, PC256, PC9, PC148, PC36, PC159 and PC243) using amplification and detection conditions described previously (Hale et al., 2004). Offspring were then genotyped for all loci at which their group of candidate mothers were polymorphic. For each offspring, the father was known and there were between four and six candidate mothers. Maternity was determined via exclusion – i.e. candidate mothers were excluded when their genotype (when combined with the known father’s genotype) could not possibly have produced the offspring genotype at at least one locus. Where maternity could not be conclusively determined via exclusion, offspring were removed from the analysis. Maternity of 1659 offspring (85% of total) was determined via exclusion. Maternity was determined by exclusion by hand, rather than using a computer program that assigns probability of parentage, due to the small number of candidate mothers per individual. The microsatellite genotypes were also used to estimate genome-wide genetic diversity, by calculating average heterozygosity across these 13 loci.

Sexing of offspring

Offspring were sexed via PCR amplification of the CHD genes using primers P2 and P8 (Griffiths et al., 1998) in 10-μL reactions containing 1×Taq buffer, 2 mm MgCl2, 200 μm each dNTP, 1.0 μm each primer, 0.3 U Taq and 0.5 μL of template DNA. The reaction cycle was 94 °C for 2 min, followed by 40 cycles of 94 °C for 15 s, 50 °C for 20 s and 72 °C for 25 s. PCR products were detected on 8% denaturing polyacrylamide gels. Females produced three bands and males two bands.

Immune response

Although our MHC genotyping method tells us which alleles were present in each individual across all copies of the locus, we cannot determine from our data how many copies of each allele are present, or which alleles are present at which copy of the locus. It is possible that one or more copies of the locus that we amplified are nonfunctional pseudogenes. Because we cannot distinguish functional from nonfunctional copies with our data, we measured T-cell-mediated immune response by measuring the degree of wing swelling after injection with phytohemagglutinin (PHA) in all adults and compared this with MHC diversity. A relationship between immune response and MHC diversity would suggest that our measure of diversity is likely to reflect the diversity present in functional MHC genes. Any relationship between MHC diversity and immune response would also provide additional evidence that immune response is affected by MHC diversity.

We estimated the response to PHA to obtain an in vivo response of T cells because PHA stimulates T-lymphocyte proliferation, followed by local recruitment of inflammatory cells and increased expression of MHC molecules at the site of injection (Goto et al., 1978; Abbas et al., 1994; Parmentier et al., 1998; Martin et al., 2006). This is a standard method to assess cell-mediated immunity in poultry (Stadecker et al., 1977; Goto et al., 1978; McCorkle et al., 1980; Cheng & Lamont, 1988) and experimental data have recently demonstrated that such a PHA test does reflect T-cell-mediated immune response in birds (Tella et al., 2008). The thickness of the left and right wing webs (patagium) of birds was measured at premarked sites with a pressure-sensitive caliper (cod SM112; Alpa SpA, Milano, Italy), to the nearest 0.01 mm. In order to avoid damage to the skin, we removed the spring from the pressure-sensitive caliper and replaced that by a weight of 16 g on top of the instrument. The right wing web was injected with 0.2 mg of PHA (L-8754; Sigma, Gillingham, UK) in 0.04 mL of phosphate-buffered saline. The left wing web was injected with 0.04 mL of phosphate-buffered saline only. Twenty-four hours later, we re-measured the thickness of both wing webs at the premarked injection sites in all individuals.

The estimate of immune response is the difference in wing web thickness between day 2 and day 1 for the PHA-inoculated wing minus the difference in wing web thickness between days 2 and 1 for the phosphate-buffered saline-inoculated wing (see Saino et al., 1997 for details of the methods used here). Estimates of T-cell response during repeat measurements of the same individuals were highly repeatable, with repeatability values exceeding 0.88.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Major histocompatibility complex diversity (measured as the number of alleles present) was positively related to fathers’ train length, after controlling for fathers’ body condition and neutral genetic variation via heterozygosity at 13 microsatellite loci (train length for fathers with low MHC diversity: 147.13 ± 1.17 cm, high MHC diversity: 155.25 ± 2.58 cm; Table 1). Fathers with higher MHC diversity (five alleles present) had trains approximately 8 cm longer (5.5% of mean total length) than fathers with relatively low MHC diversity (three alleles). Train length was not related to diversity at the selectively neutral microsatellite loci (Table 1), so is reflecting diversity specifically at the MHC, not diversity of the overall genome. Thus, adult females could potentially use train length as an indication of male MHC diversity. Train length was also not significantly related to body condition (Table 1), but instead reflected underlying MHC diversity. MHC diversity was not associated with either eye spot size (F1,41 = 1.334, = 0.255) or eye spot number (F1,41 = 0.705, = 0.406); so, these potential measures of the ‘attractiveness’ of the males’ train cannot be signals of the males’ MHC diversity, at least in our population.

Table 1.   General linear model for train length (mm) (dependent variable) in relation to MHC diversity [number of alleles (factor)], weight and tarsus length (representing body condition) and heterozygosity (arcsine square root transformed) calculated from 13 microsatellite loci to give a measure of genome-wide genetic diversity (three covariates).
VariableSSd.f.FPPartial η2
Body weight9950.712.1050.1540.049
Tarsus length7079.311.4980.2280.035
Heterozygosity5441.411.1510.2900.027
MHC diversity36 091.817.6360.0080.157
Error193 786.541   

T-cell response was positively associated with the number of MHC alleles both in fathers and mothers (Table 2, Fig. 1). Fathers with higher MHC diversity had a stronger immune response (mean ± SEM: 1.37 ± 0.13) than those with fewer MHC alleles (1.00 ± 0.08). All three MHC diversity categories differed significantly in T-cell response when both sexes were combined (< 0.05, Fisher LSD post hoc tests; sex was not a significant factor), but within females there was no significant difference in T-cell response between those with three MHC alleles and those with five MHC alleles (= 0.211, Fisher LSD). T-cell response was not body condition dependent in either sex (not associated with weight and tarsus length in Table 2).

Table 2.   General linear model for T-cell response of adults measured as the degree of wing swelling after injection with PHA (dependent variable) in relation to body weight and tarsus length (covariates) and sex and MHC diversity (factors) (= 42 for males and = 141 for females).
VariableSSd.f.FPPartial η2
Body weight0.56311.6490.2010.009
Tarsus length0.27010.7910.3750.004
MHC diversity3.52925.1720.0060.056
Sex0.26310.7720.3810.004
Error60.048176   
image

Figure 1.  Relationship between T-cell response and MHC diversity for both mothers (open circles, = 141) and fathers (solid circles, = 42). Error bars are SEM. There is a clear positive relationship with T-cell response increasing with increasing MHC diversity in both sexes (Table 2).

Download figure to PowerPoint

If MHC diversity is a genetic quality that females assess, we expect that females mated to males with high MHC diversity should invest more in their offspring than those mated to males with low MHC diversity. In addition, if females are using train length as the major ‘signal’ of male MHC diversity, we expect that females should invest more in their offspring from males with longer trains. To determine whether this was the case, we first examined the relationship between both egg weight and the number of offspring (fertilized eggs) produced by females mated to fathers with high MHC diversity compared with females mated to males with low MHC diversity. We then added train length to these analyses, to determine whether train length is the signal of MHC diversity that females are responding to. If it is the signal of MHC diversity being used, then train length should be a significant predictor of egg weight and number, and adding it to the analyses should result in the correlated variable of MHC diversity having less explanatory power than when train length was not included in the analyses. A subset of the eggs laid was hatched, and chick weight measured upon hatching, and at 22 days after hatching. Egg weight positively correlated with the weight of chicks upon hatching (r = 0.738, < 0.001, = 911) and at 22 days of age (r = 0.129, < 0.001, = 836). Therefore, egg weight provides an estimate of chick weight at an early age.

Assessment of the relationship between egg weight and both mothers’ and fathers’ MHC diversity was accomplished using a two-stage analysis. First, variance in egg weight due to date when the egg was laid, maternal body condition, genome-wide genetic diversity and genetic similarity among mates was accounted for with multiple regression. The dependent variable was egg weight, the independent variables were date laid (to account for seasonal variation in egg weight), mothers’ weight and mothers’ tarsus length (to account for maternal body condition), mothers’ heterozygosity and fathers’ heterozygosity calculated from 13 microsatellite loci (to account for genome-wide genetic diversity), relatedness and MHC similarity (to account for genetic similarity of mates). Relatedness was calculated from the 13 microsatellite loci using the formula developed by Queller & Goodnight (1989). MHC similarity was calculated as the proportion of MHC alleles present that were shared among mates. All proportions were arcsine square root transformed to ensure that these data had an underlying distribution that was nearly normal.

Second, the residual egg weight from this regression was used as the dependent variable in a nested anova, with fathers’ MHC diversity, mothers’ MHC diversity, offspring sex and father’s ID (nested within fathers’ MHC diversity) as factors. Use of residual egg weight from the regression analysis ensured that any relationship between egg weight and MHC diversity detected in the analysis of variance was not driven by differences in maternal body condition among pens, or any relationship between MHC diversity and genome-wide genetic diversity or genetic similarity among mates. Father’s ID was nested within fathers’ MHC diversity as each father had only one value for MHC diversity, and father’s ID was included to account for pseudoreplication due to multiple eggs being fathered by the same individual. Offspring with missing data due to unknown maternity or sex were excluded from the analysis [offspring excluded = 426 (21.7%); total offspring included = 1535].

Maternal body condition, paternal weight and laying date were strongly correlated with egg weight (Table 3, whole model r = 0.242, adjusted r= 0.053). Mother’s microsatellite heterozygosity was a marginally significant predictor of egg weight, although not father’s heterozygosity, parental relatedness or MHC similarity. The residuals from this model explained 94.7% of the variance in actual egg weight; so, the significant explanatory variables of variance in residual egg weight from the nested anova (see below) should also be important predictors of actual egg weights.

Table 3.   Multiple regression of egg weight (dependent variable) with date laid, maternal body condition (weight and tarsus), paternal body condition (weight and tarsus), genome-wide genetic diversity (mothers’ heterozygosity and fathers’ heterozygosity) and genetic similarity among mates (relatedness and MHC similarity) as the independent variables (= 1535). The resulting residuals were then used as the dependent variable in an anova to assess whether egg weight was associated with MHC diversity of either parent after removal of variance due to the variables above (see Table 4).
VariableβSE of βtP
Date laid−0.0890.025−3.538< 0.001
Mothers’ weight0.1220.0294.293< 0.001
Mothers’ tarsus length0.0790.0292.7440.006
Mothers’ heterozygosity−0.0560.026−2.1240.034
Fathers’ weight0.1060.0293.621< 0.001
Fathers’ tarsus−0.0120.030−0.3810.703
Fathers’ heterozygosity−0.0270.029−0.9310.352
Relatedness0.0480.0281.7570.079
MHC similarity0.0170.0260.6480.517

Offspring sex was not a significant predictor of residual egg weight (Table 4). However, both mothers’ and fathers’ MHC diversity were significant predictors of residual egg weight, and there was a significant interaction between mothers’ and fathers’ MHC diversity (Table 4; Fig. 2). Females with high MHC diversity laid larger eggs for the more MHC diverse fathers compared with the less diverse fathers, whereas mothers with low MHC diversity had a much smaller increase in investment (if any) for fathers of high compared with low MHC diversity. Although the interaction between mothers’ and fathers’ MHC diversity is clear (Table 4; Fig. 2), there were also significant differences in reproductive investment when MHC diversity in one parental sex was held constant. For example, when mothers’ MHC diversity was held constant (analysed for each MHC class of mothers separately), investment increased significantly with increasing fathers’ MHC diversity for mothers’ with three MHC alleles (fathers’ MHC diversity F1,1123 = 4.74, = 0.029), as well as for mothers’ with five MHC alleles (fathers’ MHC diversity F1,229 = 11.85, < 0.001). There was, however, no increase in investment for mothers with only two MHC alleles (fathers’ MHC diversity F1,91 = 0.28, = 0.597). If we hold fathers’ MHC diversity constant (analysed for each MHC class of fathers separately) investment increased significantly with increasing mothers’ MHC diversity for fathers with three MHC alleles (mothers’ MHC diversity F1,1075 = 3.31, = 0.020), as well as for fathers with five MHC alleles (mothers’ MHC diversity F1,378 = 3.65, = 0.027), suggesting that mothers with higher MHC diversity are capable of producing larger eggs, irrespective of the quality of the male.

Table 4.   Analysis of variance of residual egg weight from the multiple regression in Table 3, to assess the relationship between egg weight and MHC diversity after removal of any variance in egg weight due to date laid, maternal body condition, genome-wide genetic diversity and genetic similarity among mates (see Table 3) (= 1535).
VariableSSd.f.FPPartial η2
Fathers’ MHC diversity791.68123.520< 0.0010.016
Mothers’ MHC diversity1444.64314.306< 0.0010.028
Offspring sex37.5811.1160.2910.001
Fathers’ ID6946.05444.690< 0.0010.122
Mothers’ MHC ×  fathers’ MHC1121.57216.661< 0.0010.022
Mothers’ MHC ×  offspring sex25.1330.2490.8620.001
Fathers’ MHC ×  offspring sex5.9610.1770.6740.000
Error49 782.101479   
image

Figure 2.  Relationship between residual egg weight (residuals from multiple regression described in Table 3) and both mothers’ and fathers’ MHC diversity (= 1535 offspring). Open circles (dashed line) represent fathers with low MHC diversity (three alleles) and solid circles represent fathers with high MHC diversity (five alleles). Error bars are SEM. Females laid larger eggs for fathers with a high MHC diversity compared with fathers with low MHC diversity after controlling for the mother’s body condition, date of egg laying, neutral genetic diversity and both neutral genetic relatedness and MHC similarity between the parents (Tables 3 and 4).

Download figure to PowerPoint

Egg weight was not related to genetic similarity of the parents at the MHC (Table 3); so, our results exclude the possibility that the observed relationship between MHC diversity and egg size is a reflection of an underlying relationship between egg weight and genetic similarity of mates (Roberts et al., 2006). Egg weight was also unrelated to the offsprings’ own MHC diversity (one-way anova: F3,217 = 1.620, = 0.186, calculated from 221 randomly chosen offspring genotyped at the MHC), excluding the possibility that the observed relationship is driven by optimization of diversity in the offspring. Thus, MHC diversity of both parents, rather than MHC similarity between parents or offspring MHC diversity, influenced egg weight.

If females are using train length as a signal of fathers’ MHC diversity, then we expect that train length would be a significant predictor of egg weight, and the correlated variable of fathers’ MHC diversity should be a less important predictor of egg weight when train length is included in the analysis. However, when we added train length as an independent variable to the multiple regression described above, it was not a significant predictor of egg weight (β = 0.010, = 0.354, = 0.724; Supporting information Table S1). There was also no impact on the importance of fathers’ MHC diversity as an explanatory variable in the nested anova calculated using residual egg weight (from the multiple regression including train length as an independent variable) as the dependent variable (F1,1479 = 23.20, < 0.001, partial η2 = 0.015; Supporting information Table S2). Therefore, it appears that although females are investing in eggs differentially based on fathers’ MHC diversity, they are not using the correlated variable of train length as the signal of that MHC diversity.

Females not only laid larger eggs for fathers with high MHC diversity, they also laid more eggs (Table 5, Fig. 3, whole model r = 0.588, adjusted r2 = 0.158). We conducted a GLM with the number of fertilized eggs produced per father as the dependent variable, fathers’ MHC diversity as a factor, and maternal and paternal body condition (weight and tarsus), maternal and paternal heterozygosity, mothers’ MHC diversity, relatedness and MHC similarity as covariates (Table 5). In this analysis, the unit of measurement was the father and there were multiple mothers mated to each father. Therefore, mothers’ variables are weighted averages across mothers mated to each father, weighted by the number of eggs that each mother contributed to the total for that father. As with the analyses for egg weight, we subsequently added train length as an independent variable to the above analysis to see whether females could be using train length as the signal of fathers’ MHC diversity when investing differentially in the number of eggs based on MHC diversity. However, as with egg weight, train length was not a significant predictor of the number of eggs laid (F1,34 = 0.207, = 0.652, partial η2 = 0.006; Supporting information Table S3). Adding train length to the analysis resulted in a small decrease in the impact of fathers’ MHC diversity (F1,34 = 3.78, = 0.060, partial η2 = 0.100). Neither egg weight nor egg number was significantly associated with any one particular MHC allele, only with the total number of alleles present.

Table 5.   General linear model assessing the impact of fathers’ MHC diversity on maternal reproductive output with number of eggs per father as the dependent variable.
VariableSSd.f.FPPartial η2
  1. Fathers’ MHC is a factor, whereas all other variables are covariates (= 46). Father is the unit of measurement. As there were multiple mothers for each father, mothers’ characteristics are weighted averages across mothers for each father, weighted by the number of eggs contributed to the total for that father.

Fathers’ weight221.63711.1190.2970.031
Fathers’ tarsus length81.65810.4120.5250.012
Fathers’ heterozygosity594.31213.0020.0920.079
Mothers’ weight1603.83118.1010.0070.188
Mothers’ tarsus length99.05210.5000.4840.014
Mothers’ heterozygosity76.89510.3880.5370.011
Mothers’ MHC diversity295.17611.4910.2300.041
Relatedness38.87110.1960.6600.006
MHC similarity0.08610.0000.9830.000
Fathers’ MHC diversity1031.30915.2090.0280.130
Error6929.55635   
image

Figure 3.  Number of fertilized eggs produced for fathers with high MHC diversity compared with for fathers with low MHC diversity. Mothers produced significantly more eggs for fathers with high diversity (Table 5, = 46).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Mothers mated to fathers with high MHC diversity produced larger, and more eggs than mothers mated to fathers with low MHC diversity, and the increase in egg weight was greater when females had high MHC diversity themselves. An increase in investment is expected if fathers’ MHC diversity affects offspring fitness, because females should invest differentially in reproduction when the reproductive value of their offspring is elevated due to greater genetic quality (Burley, 1986), and differential maternal investment has previously been demonstrated in peafowl (Petrie & Williams, 1993; Petrie et al., 2001; Loyau et al., 2007). Our results suggest that MHC diversity is a quality of potential mates that females assess and respond to via increased reproductive investment. This is not the first study to suggest a link between a visual signal of attractiveness and MHC diversity. Interestingly, human females rate photographs of males that are more heterozygous at the MHC as visually more attractive (Roberts et al., 2005).

Major histocompatibility complex diversity was positively associated with T-cell response, demonstrating that immune response in this population was associated with our measure of MHC diversity. Although our measure of immune response (T-cell response to PHA injection) does not necessarily replicate the range of pathogens and parasites to which these birds would naturally be exposed, it does provide a relative measure of T-cell-mediated immune response among individuals, and demonstrates that the loci amplified in this study represent functional MHC genes.

Major histocompatibility complex diversity of fathers was also positively associated with train length, with MHC diversity explaining nearly 16% of the variance in train length among the adult males in our population. This strong association between MHC diversity and train length suggests that train length does provide an honest signal of the MHC diversity of males. However, are females making use of this signal (train length) to assess the MHC diversity of potential mates? Previous work has shown that females produce more eggs when mated to males with larger trains (Petrie & Williams, 1993), and larger eggs for males with a higher density of eye spots in their trains (Loyau et al., 2007) making train characteristics prime candidates for the signal females are using to assess mate quality. However, in the current study, train length was not a significant predictor of either egg weight or the number of eggs laid. The lack of concordance between the current and past studies could be due to different measures of the train being used [train mass (g) in Petrie & Williams, 1993 and eye spot density in Loyau et al., 2007 compared with train length (mm) in the current study] or differences between studies in the among-individual variance in train measures. Our results suggest that although train length does provide information about the MHC diversity of males, and females do invest differentially in offspring of fathers with differing MHC diversity, they are not using train length as the primary signal to assess MHC diversity of mates.

Train length was not condition dependent, although it was associated with MHC variation. This is somewhat surprising as secondary sexual characteristics are expected to be condition dependent as they are expensive to produce (Hamilton & Zuk, 1982; Andersson, 1986, 1994). However, it is possible that the variance in condition in our small population of fathers was not great enough to detect a relationship between train length and condition, or representative of the variance in condition expected in a wild population.

Females must be assessing MHC diversity of mates in order for us to detect significant differential reproductive investment based on mates’ MHC diversity. Although we have found that MHC diversity is reflected in the length of peacocks’ trains, it cannot be assumed that this is the only part of the genome reflected in the peacock’s train nor can it be assumed that it is the only part of the phenotype that reflects MHC diversity. Other parts of the genome could well relate to other aspects of the train and these may not necessarily be linked to parasite resistance. It may be that other parts of the plumage, vocal or odour signals also reflect MHC diversity, and females are using one or more of these characteristics to assess MHC diversity. It may also be that females are differentially investing in the offspring of mates based on train length, but not in terms of the two measures of reproductive output we have measured in this study (egg weight and number). The fact that the major explanatory variable for egg weight was father’s ID (partial η2 = 0.122, Table 4) suggests that there are many other aspects of the fathers ‘quality’ that mothers are assessing and responding to that we have not measured.

Although the interaction between mothers’ and fathers’ MHC diversity clearly has a significant impact on egg weight, there was also an independent effect of mothers’ MHC diversity on egg weight as well as fathers’ MHC diversity. This was demonstrated by analysing the impact of mothers’ MHC diversity on egg weight within each category of fathers’ MHC diversity and vice versa. There was a steadily increasing response in investment as mothers’ MHC diversity increased among mothers mated to fathers with low MHC diversity, but this was not the case among mothers mated to fathers with high MHC diversity (see Fig. 2). Among mothers mated to fathers with high MHC diversity, there was a slight increase in egg weight between mothers with three alleles compared with mothers with two alleles, but then a steep increase in egg weight between mothers with three alleles and mothers with five alleles. This suggests that mothers with three MHC alleles either choose to invest substantially less in the offspring of fathers with high MHC diversity than mothers with five alleles (i.e. the benefits of mating with a male with high MHC diversity are greatly reduced when the mother’s MHC diversity is low) or they are not able to substantially increase their investment for fathers with high MHC diversity, or a combination of both.

Mothers with high MHC diversity are expected to be healthier individuals than those with low MHC diversity and so should have more energy available to invest in their offspring. Although mothers’ body condition was accounted for in the multiple regression of egg weight by including mothers’ weight and tarsus length, our measure of mothers’ body condition is unlikely to be an exhaustive measure of overall health, and therefore ability to invest in offspring. The increase in egg weight associated with increasing mothers’ MHC diversity, when fathers’ MHC diversity was held constant, suggests that MHC diversity may be influencing the ability of mothers to invest in their offspring in ways other than simply through improved body condition. This may in part explain the finding that the increase in egg weight when mated to high MHC diversity fathers compared with low MHC diversity fathers is much greater for mothers with high MHC diversity compared with mothers with low MHC diversity. It is possible that mothers with high MHC diversity are more likely to be able to respond to high quality males with more and larger eggs, than mothers with low MHC diversity, because they are healthier individuals overall.

Why are females investing more in eggs fathered by males with high MHC diversity? Fathers with high MHC diversity do not necessarily produce offspring with high MHC diversity, because offspring MHC diversity will rather be dependent upon the genetic similarity of the parents. In our data set, there was no significant relationship between offspring MHC diversity and father’s MHC diversity (one-way anova: F1,220 = 2.713, = 0.101). One possibility is that females are investing more in the eggs of males with high MHC diversity, because these males are more likely to carry rare alleles than males with low diversity. Because rare alleles occur at low frequency, they generally only occur in heterozygous form, and so the higher diversity an individual has, the greater the chance that such an individual will carry rare alleles. The MHC is one of the most variable coding regions in the genome, and this variation is likely to be maintained at least in part by negative frequency-dependent selection (Hedrick, 2002). That is, rare alleles are at a selective advantage because pathogens will not have adapted to rare new alleles (Takahata & Nei, 1990) or may have lost adaptations to rare old alleles that were once common (Slade & McCallum, 1992). Thus, individuals with rare alleles at the MHC should have a better immune response, be resistant to a wider range of diseases than individuals with common alleles and so should be more ‘fit’ in a pathogen-rich environment.

One of the unresolved questions in evolutionary biology is what maintains genetic diversity at the MHC. Overdominance, negative frequency-dependent selection, mutation and selection varying over time and space have all been proposed as mechanisms that may work independently or in concert to maintain MHC diversity (Hedrick, 2002; Bernatchez & Landry, 2003). While there is clear evidence that MHC diversity will aid in the battle against parasites (McClelland et al., 2003), there is also the suggestion in both mice and humans that MHC diversity is a preferred characteristic in mates (Potts et al., 1991; Roberts et al., 2005). Although it has been suggested that the selective forces diversifying MHC Class II genes in birds are likely to be similar to those in mammals (Potts & Wakeland, 1993; Edwards et al., 1995), there is very little research addressing this question. The results presented here suggest that sexual selection could act together with other forms of natural selection in promoting MHC diversity.

The idea that sexual selection can promote genetic diversity has had a chequered history (Kirkpatrick & Ryan, 1991; Pomiankowski & Møller, 1995; Tomkins et al., 2004), largely because it has been thought that any variance in fitness-related traits should rapidly spread to fixation, and most genetic mechanisms for maintaining variation (such as an elevated mutation rate or recombination) are costly (but see Møller & Cuervo, 2003 who demonstrate higher mutation rates in sexually selected species). However, recent theoretical considerations have shown that even where genetic mechanisms that maintain genomic variation are costly, sexual selection can allow for the maintenance of genetic mechanisms which promote and maintain genetic variation (Petrie & Roberts, 2007). Our findings suggest that mate choice may contribute to maintenance of genetic variation at the MHC, through selection for mates with high MHC diversity.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We are particularly grateful to Quinton Spratt for allowing us to work on his farm and thank Doug Maisie, James Futter, John Howe, Jaqui Shykoff, L. Morris Gosling, Candy Rowe, Ellen Vale, Bill Sutherland, Nicky Crockford, John Reynolds, Jim Kaufman, Terry Burke and Isabelle Cote for help. We also thank one anonymous reviewer for comments that greatly improved the manuscript. This work was supported by NERC (NER/A/S/2002/00959, M.P. and K.W.) and conducted under Home Office License (UK).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Figure S1 RSCA MHC genotyping chromatograms for three individuals, including all peaks produced by annealing MHC PCR products with both labelled FLRs (allele1–FAM in blue and allele4–JOE in green).

Table S1 Multiple regression of egg weight (dependent variable) as per Table  3 in the manuscript, with the addition of train length as an independent variable.

Table S2 Analysis of variance of residual egg weight from the multiple regression in Table  S1 (i.e. regression including train length as an independent variable), as per the analysis in Table  4 in the manuscript.

Table S3 General linear model assessing the impact of father’s MHC diversity on maternal reproductive output with number of eggs per father as the dependent variable (as per the analysis in Table  5 in the manuscript), with the addition of train length as an independent variable.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
JEB_1746_sm_Figure_Tables.doc176KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.