SEARCH

SEARCH BY CITATION

Keywords:

  • androgens;
  • galliformes;
  • mate choice;
  • maternal effects;
  • mating success;
  • major histocompatibility complex

Abstract

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

Females of several vertebrate species selectively mate with males on the basis of the major histocompatibility complex (MHC) genes. As androgen-mediated maternal effects have long-lasting consequences for the adult phenotype, both mating and reproductive success may depend on the combined effect of MHC genotype and exposure to androgens during early ontogeny. We studied how MHC-based mate choice in ring-necked pheasants (Phasianus colchicus) was influenced by an experimental in ovo testosterone (T) increase. There was no conclusive evidence of in ovo T treatment differentially affecting mate choice in relation to MHC genotype. However, females avoided mating with males with a wholly different MHC genotype compared with males sharing at least one MHC allele. Females also tended to avoid mating with MHC-identical males, though not significantly so. These findings suggest that female pheasants preferred males with intermediate MHC dissimilarity. Male MHC heterozygosity or diversity did not predict the expression of ornaments or male dominance rank. Thus, MHC-based mating preferences in the ring-necked pheasant do not seem to be mediated by ornaments’ expression and may have evolved mainly to reduce the costs of high heterozygosity at MHC loci for the progeny, such as increased risk of autoimmune diseases or disruption of coadapted gene pools.


Introduction

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

In sexually reproducing organisms, mate choice generates variance in individual fitness. Most theoretical and empirical work has focused on the mechanisms and consequences of sexual selection via preference for mates possessing genes for viability or sexual attractiveness (Kokko et al., 2002, 2003; Mead & Arnold, 2004; Puurtinen et al., 2009). Intersexual selection can thus lead to the evolution of exaggerated secondary sexual traits that may serve as indicators of mate genetic quality (Andersson, 1994). The honesty of such traits as quality indicators may be enforced by the differential viability costs for bearers of different genetic backgrounds (Andersson, 1994). Sexual selection models for ‘good genes’ predict additive genetic benefits from choosing high-quality individuals, implying that individuals of the choosy sex will tend to agree over mate choice, generating directional selection (Mead & Arnold, 2004). Conversely, compatibility models of sexual selection posit that fitness benefits from mate choice will arise only when specific paternal and maternal genotypes are combined in the progeny, thus generating nonadditive fitness variation (Zeh & Zeh, 1996, 1997; Penn et al., 2002; Agbali et al., 2010). Importantly, ‘good genes’ and ‘compatibility’ modes of sexual selection need not be mutually exclusive, and the interpretation of mate choice patterns will have to rest on the concomitant analysis of both mechanisms (Neff & Pitcher, 2005).

In the light of the Hamilton–Zuk hypothesis of parasite-mediated evolution of secondary sexual traits (Hamilton & Zuk, 1982), the role of the immune system in sexual selection has received increasing attention (Milinski, 2006; Piertney & Oliver, 2006). In the face of disease, males with a superior genetic background should be able to produce exaggerated secondary sexual traits, which may therefore act as ‘condition-dependent’ or ‘revealing’ handicaps (Zahavi, 1975, 1977; Andersson, 1994; but see Maynard Smith & Harper, 1995). Ornaments that honestly signal good genes may allow mates to accurately discriminate between low- and high-quality partners (Zahavi, 1975; Grafen, 1990).

The major histocompatibility complex (MHC) genes are extremely polymorphic and likely to play a pivotal role in parasite-mediated sexual selection because of their role in immune processes and antiparasite defence (Klein, 1986; Andersson, 1994; Danchin & Pontarotti, 2004; Kasahara et al., 2004; Trowsdale & Parham, 2004; Kelley et al., 2005). MHC heterozygosity confers a selective advantage in combating parasite infection (Penn et al., 2002; Bonneaud et al., 2006a; Wegner et al., 2008; Worley et al., 2010). MHC variation at particular loci can be the result of specific host–parasite coadaptational cycles (e.g. McCairns et al., 2011), making these genes focal candidate markers to study the genetic basis of mate choice. Moreover, the expression of showy male ornaments, as reliable indicators of condition, has been related to specific MHC genotypes (von Schantz et al., 1996, 1997; Ditchkoff et al., 2001; Buchholz, 2004; Jäger et al., 2007; Hale et al., 2009; Baratti et al., 2010), and females may choose males with specific MHC genotypes (Bernatchez & Landry, 2003; Milinski, 2006). Studies of MHC-based mate preferences revealed several instances of disassortative choice: according to the ‘heterozygote advantage hypothesis’, a preference for MHC-dissimilar mates may function to maximize offspring MHC heterozygosity (Potts & Wakeland, 1990; Apanius et al., 1997; Agbali et al., 2010). However, in natural populations and in outbred strains, female preference may not maximize genetic dissimilarity (Bos et al., 2009; Eizaguirre et al., 2009), because mating with the most genetically dissimilar males can involve the loss of local adaptations or breakage of coadapted gene complexes in the offspring (Bateson, 1983; Hendry et al., 2000). Hence, the balance between costs and benefits of choosing either MHC-identical or wholly different mates may result in adaptive preference for mates with intermediate levels of MHC dissimilarity (e.g. Forsberg et al., 2007; Eizaguirre et al., 2009). Moreover, other studies revealed that females may affect MHC diversity of their progeny either by choosing males possessing a high number of MHC alleles or by choosing males with a high number of MHC alleles when the choosy female has a low number of MHC alleles (and/or vice versa), according to a self-referent strategy of MHC-based female mate preference (‘allele counting’; Reusch et al., 2001; Aeschlimann et al., 2003; Griggio et al., 2011).

As an extension of the Hamilton–Zuk hypothesis, the ‘immunocompetence handicap hypothesis’ predicts that testosterone-dependent ornaments signal the ability to cope with the immunosuppressive effects of testosterone (Folstad & Karter, 1992). In avian species, maternal effects via sex steroids in the eggs, besides affecting offspring behaviour soon after hatching, also entail long-term effects for the adult phenotype and may function to enhance the mating success of the offspring (Groothuis et al., 2005; Groothuis & Schwabl, 2008). However, the effects of egg steroids on the offspring may depend on their genetically based resistance to parasites and thus on their MHC genotype. Ultimately, the epigenetic effects exerted by maternal transfer of steroids to the eggs should depend on the developmental programme specified by the biparentally inherited genes (Badyaev, 2008).

We investigated patterns of MHC-based mate choice in the ring-necked pheasant (Phasianus colchicus). In the sample of individuals used in this study, we previously demonstrated that an experimental increase in egg testosterone (T) concentration enhanced copulation success of males, particularly with females hatched from control eggs (Bonisoli-Alquati et al., 2011a). This effect was apparently independent of the expression of male secondary sexual traits (Bonisoli-Alquati et al., 2011a), although a T-induced change in the covariation pattern of different male ornaments could be involved (Bonisoli-Alquati et al., 2011b). Here, for the first time in any species, we analysed the combined effects of MHC genotype of potential mates and of an experimental increase in egg T concentration on female mate choice, by performing a correlative evaluation of patterns of MHC-based mate choice and analysing their interaction with the experimental manipulation of T in ovo.

Specifically, we first studied mate choice in relation to the MHC genotype of both partners. Females may prefer to mate with genetically similar, dissimilar or intermediately different males, depending on the specific MHC combination that maximizes offspring fitness (Penn & Potts, 1999; Huchard et al., 2010). We investigated these possible strategies of MHC-based mate preferences by analysing variation in the number of copulations in relation to indices of MHC differences between mating partners and male MHC heterozygosity, as well as in relation to the number of MHC alleles of both mating partners, to shed light on the different potential mechanisms by which females can affect offspring MHC genotype. We assumed that the number of copulations of a female with a particular male reflected the strength of female preference (see Materials and Methods and Bonisoli-Alquati et al., 2011a). We then assessed whether any pattern of MHC-based female choice was influenced by experimental in ovo T manipulation, as expected if maternal effects via egg androgens differentially affect adult phenotype depending on MHC genotype. Finally, we investigated whether the expression of male secondary sexual traits and social rank was associated with the combined effects of male MHC genotype and egg T treatment, thus serving as a mechanistic basis for an interaction effect of MHC genotype and maternal effects on mate choice.

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

Study species

The ring-necked pheasant is a highly sexually dimorphic species, in which males are larger than females and characterized by conspicuous ornaments: elongated tail feathers, ear tufts, long spurs and wattles, which have an important role during male courtship (Johnsgard, 1999). Wattle size is positively correlated with viability (Papeschi & Dessì-Fulgheri, 2003) and circulating T levels (Briganti et al., 1999; Papeschi et al., 2000) and linked to early nutritional conditions (Ohlsson et al., 2002). Spur length is also a condition-dependent trait that positively covaries with age, viability and body size (von Schantz et al., 1997), although the role of spur length in female mate choice is still controversial (Hillgarth, 1990a). This ground-nesting species is exposed to a variety of virulent parasites (e.g. coccidia of the genus Eimeria and the roundworm Heterakis gallinarum), and more heavily parasitized individuals court less vigorously (Hillgarth, 1990b). Ring-necked pheasants show heritable parasite resistance (Hillgarth, 1990b). Males are polygamous, defend mating territories and do not contribute to parental care (Ridley & Hill, 1987). Female mate choice is based on costly territorial displays related to social rank (Mateos & Carranza, 1997), courtship displays and ornaments (von Schantz et al., 1997; Mateos, 1998). For the purposes of this study, we assumed that the frequency of copulations of a female with a male reflects the strength of female preference, as all main models of sexual selection in polygynous species posit that females are the choosy sex, whereas males should be prone to opportunistically copulate with any receptive partner, although we do not exclude that male choice may play a minor role (see Bonisoli-Alquati et al., 2011a).

Experimental design

Five hundred ring-necked pheasant eggs were randomly chosen from a large stock of several thousands of eggs purchased from a commercial breeder (L’Envol de Retz, Machecoul, France; http://www.envol-de-retz.com/). As detailed in the study of Bonisoli-Alquati et al. (2011a,b), half of the eggs (T eggs) were inoculated with a solution of T in ethyl alcohol dissolved in distilled water, whereas the other half (control eggs) were inoculated with a solution of ethyl alcohol in distilled water. Egg T injections increased the egg T-albumen concentration by 2 standard deviations (see Bonisoli-Alquati et al., 2011a,b). All eggs were incubated in the same professional incubator. All hatched chicks (201/250 control eggs, 80.4%; 205/250 T-injected eggs, 82.0%; difference in hatchability between the two groups: χ2 = 0.21, d.f. = 1, = 0.65) were reared in captivity in the same aviary (12 × 8 × 3.5 m indoor; ca. 20 × 10 × 2.5 m outdoor). We separated males and females after 200 days.

Once the birds were 270 days old, we measured a suite of ‘ordinary’ and secondary sexual traits on 92 males (43 control and 49 from T-injected eggs, T males hereafter) and 108 females (55 control and 53 from T-injected eggs, T females hereafter), randomly selected from the original pool of individuals. We measured body mass, length of spurs, tarsus and right ear tuft, wattle size and colour (both in the visible and in the UV range), and cell-mediated immunity (CMI hereafter), as described by Bonisoli-Alquati et al. (2011a). Finally, we collected blood samples from the wing vein for MHC genotyping.

We randomly assigned males to three groups (= 14 per group, including seven control and seven T males). We sequentially transferred each group into the aviary holding the 108 females and left them there for 1 week. The adopted 1 : 8 male-to-female ratio was similar to that observed in territories maintained by dominant males in wild populations (Johnsgard, 1999). We conducted behavioural observations for five consecutive days, starting 2 days after we transferred the males to the aviary. We used these observations to estimate male dominance rank and total number of copulations obtained (copulation success). This latter was used as an index of female mate preference (see Bonisoli-Alquati et al., 2011a). Each day, we conducted observations for 6 h, from 7:00 a.m. to 1:00 p.m. We observed 1767 male–male interactions, which we used to calculate male social rank in its group as indexed by the David’s score (David, 1987). This index is relatively insensitive to minor deviations from linearity in hierarchy rank, and it is considered more reliable than other dominance indices (see Gammell et al., 2003).

The eggs were purchased from a company holding a very large breeding stock (see company website). The 500 T- or sham-inoculated eggs were randomly sampled from a very large sample of eggs, and the individuals included in the present study were then randomly chosen among those that reached sexual maturity. We thus regard the possibility that closely related males and females occurred in our sample and thus that relatedness between mating partners could affect our results to any extent, as a remote one.

Genetic analyses and MHC polymorphism screening

We collected blood samples from each individual into tubes containing EDTA and stored them at −80 °C until DNA extraction. Genomic DNA was isolated from whole blood using a Puregene D-5000A isolation kit (Gentra Systems, Minneapolis, MN, USA).

We performed amplification of a section of the MHC class II β second exon by PCR as described in the study of Baratti et al. (2010). In birds, the MHC of the fowl (Gallus gallus domesticus) is the only well-characterized avian MHC (Guillemot et al., 1988; Kroemer et al., 1990), and a similar MHC organization seems to characterize the ring-necked pheasant (Jarvi & Briles, 1992; Wittzell et al., 1994, 1998, 1999; Jarvi et al., 1996; von Schantz et al., 1996). The MHC of pheasants (MHCPhco) appears to be highly polymorphic and characterized by two unlinked gene clusters, corresponding to those found in the fowl (BF1, BF2 and B-LB1, B-LB2) and in the black grouse (Tetrao tetrix) BLB genes (Strand et al., 2007), both comprising class I and class IIB genes (3–4 genes). Besides, the BLB MHC loci are expressed both in the chicken and in the black grouse (Jacob et al., 2000; Strand et al., 2007). The exon 2 of the B complex coding for the β1 chain is the most variable part of MHC class IIB genes and is commonly used for population genetic analyses (Ditchkoff et al., 2001; Oliver & Piertney, 2006; Zhang et al., 2006; van Oosterhout, 2009).

We used the capillary electrophoresis–single strand conformation polymorphism technique (CE-SSCP) for screening genetic polymorphism (Kourkine et al., 2002). This method allows the detection of mutations in sequences of identical length by the electrophoresis variance of the molecules. Samples were prepared as follows: 1 μL PCR product was added to 10.5 μL formamide and 0.75 μL GeneScan Standard Rox 500, then denatured for 3′ at 94 °C and cooled on ice. Samples were then loaded onto a 47-cm-length, 50-μm-diameter capillary filled with a polymer containing 10% glycerol and 1X Applied Biosystems (ABI, Foster City, CA, USA) Buffer 310 and the concentration was tested at 3% and 4%. The electrophoresis conditions showed that a 15-kV EP voltage and the 4% polymer concentration at 30 °C heat plate temperature (25 min) combination was best for the separation of peaks with run conditions at optimal levels. We used 5′-end-labelled primers for both DNA strands, with different dyes (HEX for the forward and 6-FAM for the reverse) to allow the differentiation of the two single strands.

Out of screened profiles, we selected individuals that were homozygous and we directly sequenced them using a sequencing kit (Big Dye Terminator Sequencing version 1.1-ABI PRISM; PE Biosystems, Foster City, CA, USA). We grouped heterozygotes based on similar SSCP profile and then cloned one to three samples for each group using a pGEM®-T Easy Vector System (Promega, Madison, WI, USA). All sequences were corrected, aligned and compared with those previously described for the same species (Wittzell et al., 1994; von Schantz et al., 1996; Baratti et al., 2010). All alleles were deposited in GenBank (http://www.ncbi.nlm.nih.gov/; see accession numbers in Table S1).

Due to several problems in generating PCR and cloning artefacts in the analyses of polymorphic gene complexes, we decreased the number of cycles until a reliable amplification fragment was visible (30 cycles), and established criteria to avoid misinterpretation of alleles. An allele was considered to be real only when it occurred in more than one clone, from the same or from different individuals, and when it was supported by two independent PCRs. As a control, we also ran clones in CE-SSCP analysis. All CE-SSCP analyses and sequencing were performed with a 310 ABI® automated sequencer (PE Applied Biosystems, Foster City, CA, USA). GeneScan software version 3.7 (PE Applied Biosystems) was used to visualize CE-SSCP profiles.

Statistical methods

Analysis of genetic data

We used chromas version 2.01 (Technelesyum Pty Ltd, Brisbane, QLD, Australia) to correct chromatograms. We translated the sequences, previously aligned with clustal x version 1.83 (Thompson et al., 1997), into amino acids using mega version 4.1 (Tamura et al., 2007). We calculated the ratio of nonsynonymous (dN) to synonymous (dS) substitutions in mega, using the Nei–Gojobori method with the Jukes–Cantor correction for multiple substitutions (Nei & Gojobori, 1986) with 1000 bootstrap replicates to obtain standard errors. We inferred the location of the putative protein binding region (PBR) codons from published MHC sequences described in Westerdahl et al. (2000). The PBR codons are expected to experience positive Darwinian selection (Nei & Kumar, 2000), with dN/dS ratios in excess of unity. By contrast, purifying selection is generally found to govern sequence variation outside the PBR. We performed the z-test, as implemented in mega, to test for positive selection acting on the MHC fragment. We calculated genetic distance among all MHC-Phco alleles based on Kimura’s (Kimura, 1980) two-parameter model.

Analyses of the association between phenotype, mating success and MHC genotype

Because of high genetic diversity within our study population (24 different MHC genotypes in 37 males, mean number of individuals per genotype = 1.5), we could not analyse the associations between specific MHC genotypes and the phenotypic traits of males, as done in previous studies of the same species (von Schantz et al., 1996; Baratti et al., 2010). However, we checked whether male traits or social rank differed between homo- and heterozygote birds (homozygote birds carried only a single allele, whereas all other birds were classified as heterozygotes; see Lehmann et al., 2007) at MHC loci in linear models where we entered hormonal treatment, heterozygosity and their interaction as predictors (sample size, males: homozygous = 10, heterozygous = 27). In addition, we tested whether male traits or social rank covaried with male MHC diversity (number of alleles; males had either one, two or three alleles; see Results) by means of correlation analyses.

For the analysis of mate choice, we calculated the number of copulations in each male–female pair, considering also those pairs where no copulation occurred (i.e. including 0 as an informative response value, see Bonisoli-Alquati et al., 2011a). We assessed several possible strategies of MHC-based mate choice, using three main different metrics. First, to test whether females preferred to mate with males with similar MHC, dissimilar MHC, or having intermediate levels of MHC dissimilarity, we investigated whether differences in MHC genotype between mating partners predicted female mate choice. Secondly, we tested whether females preferred males with a high MHC diversity or whether females chose males according to both their own and male’s number of MHC alleles (allele counting). Thirdly, we tested whether females preferred heterozygote over homozygote partners at MHC loci.

We used three indices to quantify MHC differences between partners, considering all the possible pairings of males and females (= 2997 pairings): (i) MHC similarity, expressed as variant-sharing values (calculated according to Wetton et al., 1987), whereby the proportion of shared MHC alleles in a pair was calculated as twice the number of the alleles shared by the two partners divided by the sum of the alleles of the two individuals. Thus, MHC similarity was 0 for pairs that shared no alleles (= 2077, or 69.3% of all possible pairings), 1 for those pairs with identical MHC genotype (= 58, 1.9%) and had intermediate values for all the other pairs sharing at least one allele (= 862, 28.8%); (ii) overall genotypic distance between the two partners (MF distance), calculated as the mean of all the pairwise amino acid distances of all alleles carried by the two partners (modified from Landry et al., 2001); (iii) overall genotypic distance between the two partners based on putative sites involved in the peptide binding region (MF distance PBR); calculations were similar to those used to obtain MF distance, but distances between alleles were based on possible PBR sites only (Landry et al., 2001).

We analysed the effect of MHC differences between partners on the number of copulations in mixed models that included male treatment, female treatment, their interaction and the MHC difference indices as fixed effects. Male and female identities were included in the models as crossed random effects (see also Bonisoli-Alquati et al., 2011a). We first fitted a model including male treatment, female treatment and their interaction as a basis for testing the effect of MHC genotype on female choice, because we have previously shown on a sample of birds including the same individuals used in the present study that the combined effect of male and female treatment significantly predicted the number of copulations (see Introduction and Bonisoli-Alquati et al., 2011a). To this ‘base’ model, we then added one of the indices of MHC differences between partners (MHC similarity, MF distance or MF distance PBR) as a covariate. We ran separate models for each index of MHC differences between partners. We checked for nonlinearity in the relationship between female choice and indices of MHC differences by testing the quadratic terms of each index or by recoding MHC similarity (that ranges between 0 and 1) as a three-level fixed factor (shared MHC alleles), accounting for MHC genotypes being different (no allele shared), identical (sharing of all alleles) or similar (sharing of some, but not all, alleles). We investigated whether MHC differences between partners affected mate choice differently in T-treated and control individuals by testing the interaction terms between male treatment, female treatment and indices of MHC differences between partners up to three-way interactions (i.e. male treatment × female treatment  × MHC difference index). To test for differential nonlinear variations among male treatment × female treatment groups, we also added to the model the three-way interaction with the squared terms of the MHC similarity indices.

Similar mixed models were built to test for ‘allele counting’ mating strategy. First, directional preference of females for males with high MHC diversity was investigated in a mixed model including male MHC diversity as a continuous predictor. Differential effects of in ovo T injection were investigated by adding to the models the interaction terms between male MHC diversity, male treatment and female treatment, up to the three-way interaction male MHC diversity × male treatment × female treatment. Secondly, we analysed the possibility of mate choice via self-referent ‘allele counting’ by running a mixed model including male treatment, female treatment, their interaction and the number of MHC alleles of males and of females as fixed factors. The number of alleles of males and females was coded as 3-level and 4-level fixed factors, respectively (males had up to three alleles, whereas females up to 4; see Results). A statistically significant interaction whereby females with few alleles preferentially mated with males possessing many alleles, and/or females with many alleles mated with males possessing few, would lend support to such a strategy (e.g. Griggio et al., 2011). Differential effects of in ovo T injection were tested by additionally including in the mixed model the interaction terms between male and female MHC allele numbers and male and female treatment, up to the four-way interaction (male number of alleles × female number of alleles × male treatment  × female treatment).

Finally, we tested directional preference of females for male MHC heterozygosity in a mixed model including male heterozygosity as a covariate. Differential effects of in ovo T injection were investigated by adding to the model the interaction terms between male MHC heterozygosity, male treatment and female treatment, up to the three-way interaction (male MHC heterozygosity × male treatment × female treatment).

Mixed models were run assuming a Poisson error distribution, which is appropriate to model count data. Akaike information criterion (AIC) values were used to compare model fit and the relative support for different strategies of MHC-based mate choice (Zuur et al., 2009). As the models were highly underdispersed because of zero inflation in the number of copulations, we checked the robustness of the results by means of a randomization procedure (see Results and Bonisoli-Alquati et al., 2011a). For randomizations involving MHC difference indices, male and female genotypes were randomized 1000 times. This means that the procedure substituted, for example, the genotype of male i with that of male j in all data lines referring to male i. Genotypes were randomized within each sex only, that is, female genotypes were not assigned to males in the randomization procedure and vice versa. At each run, indices of MHC differences were recalculated, and models were refitted while holding other random and fixed effects constant (see Anderson & ter Braak, 2003; Bonisoli-Alquati et al., 2011a). For randomizations involving male MHC heterozygosity and diversity, only male genotypes were randomized. Mixed models were fitted by means of the lmer procedure of the lme4 package of the software r 2.8.1 (R Development Core Team, 2010).

Results

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

Genetic characterization

The 222-bp fragment of the MHC class II B gene, translating into 74 amino acids, contained 21 putative PBR sites (Table S2). We used CE-SSCP to analyse 118 individuals, including 81 females (41 control and 40 T females) and 37 males (18 control and 17 T males). The analyses revealed 37 homozygotes (10 males and 27 females), which were directly sequenced, and 81 heterozygotes (27 males and 54 females).

We found 15 different sequences in the 118 individuals (Table S2), four of which had been previously characterized (Wittzell et al., 1994; von Schantz et al., 1996; Baratti et al., 2010), whereas 11 are described here for the first time. A phylogenetic analysis was carried out on these alleles (see details in Supporting information and Fig. S1). As the assignment of alleles to either of the different MHC loci was not possible using our methods, an individual carrying two different MHC sequences was considered as heterozygote. Different alleles combined to form a total of 47 MHC genotypes (24 males and 34 females; 11 genotypes where shared between the sexes). Individuals showed 1–4 alleles, supporting the existence of two paralogous class IIB genes in this species (Wittzell et al., 1999). Females had up to four different alleles (27 females had one allele; 30 females had two alleles; 20 females had three alleles; four females had four alleles). Females with four alleles all had the same genotype. We detected 1–3 alleles in males (10 males had one allele; 18 males had two alleles; nine males had three alleles). The frequency of individuals with different numbers of MHC alleles did not differ between males and females (χ2 = 2.99, d.f. = 3, = 0.39).

The proportion of nonsynonymous (dN) and synonymous (dS) substitutions was estimated in both PBR and non-PBR regions. In PBR codons, dN was significantly higher than dS (= 2.35, = 0.01), whereas this was not the case in the non-PBR regions (= 0.94, = 0.35).

Relationships between male traits, and MHC heterozygosity and diversity

Male MHC heterozygosity did not predict variation in any male trait or social rank (Table 1), and there was no significant covariation between male phenotype and male MHC diversity (all |r| < 0.27, all > 0.10, details not shown for brevity). In linear models of male traits, the two-way interaction between male treatment and heterozygosity or MHC diversity was always nonsignificant (all > 0.07, details not shown for brevity).

Table 1.   Descriptive statistics (mean and SE) of male traits and social rank in relation to major histocompatibility complex heterozygosity. No trait differed significantly between homozygote and heterozygote birds (t-tests, all > 0.14).
Male traitsHomozygotes (= 10)Heterozygotes (= 27)
Tarsus length (mm)73.41 (1.28)72.39 (0.63)
Body mass (g)1287.0 (4.55)1296. 3 (2.65)
CMI (mm × 100)35.90 (13.16)32.07 (12.70)
Ear tuft (mm)20.60 (0.60)21.00 (0.41)
Spur length (mm)20.91 (0.33)21.21 (0.42)
Wattle area (cm2)6.96 (0.30)6.90 (0.13)
Wattle hue (visible)0.36 (0.01)0.35 (0.01)
Wattle chroma (visible)0.53 (0.01)0.54 (0.01)
Wattle brightness (visible)0.14 (0.01)0.14 (0.01)
Wattle hue (UV)−0.46 (0.03)−0.45 (0.02)
Wattle chroma (UV)0.08 (0.01)0.08 (0.01)
Wattle brightness (UV)0.10 (0.01)0.09 (0.01)
Social dominance rank6.90 (1.16)7.70 (0.84)

Mate choice and MHC genotype

The mixed models testing the effects of MHC genotype on the number of copulations are summarized in Table S3. The best-supported models included MHC similarity, entered either as a linear continuous variable (AIC = 1852) or as shared MHC alleles (AIC = 1850). All the other models were less supported (ΔAIC > 2 from the shared MHC allele model, Table S3). Inclusion of the squared terms of MHC similarity and distance indices did not improve model fit (Table S3).

Parameter estimates for the best-fitting model, with MHC similarity coded as shared MHC alleles, revealed that females copulated more with males that were at intermediate levels of MHC similarity than with males that had a wholly different or identical MHC genotype (Table 2, Fig. 1). Post hoc tests showed a statistically significant difference in the mean number of copulations between partners sharing some and partners sharing no MHC alleles. The differences in the mean number of copulations between the shared MHC allele categories ‘some’ vs. ‘all’ and ‘none’ vs. ‘all’ were both statistically nonsignificant (Table 2). Results from the model including MHC similarity as a continuous variable were similar, showing a positive, statistically significant effect of MHC similarity on the number of copulations (Table 2). The squared term of MHC similarity was not statistically significant [estimate: −0.919 (0.809 SE), = −1.14, = 0.26] and did not improve model fit (Table S3). However, the lack of a statistically significant difference between the ‘all’ and the other shared MHC allele categories was possibly due to the large variance in number of copulations and the small frequency of male–female pairs that had identical MHC genotype, because the estimate for MHC-identical males and females was based on only four copulations out of 58 pairings. We thus repeated the analyses while removing MHC-identical pairings, and the results were robust to exclusion of these data (Table 2).

Table 2.   Parameter estimates for best-fitting Poisson mixed models analysing the effects of major histocompatibility complex (MHC) differences between mating partners in terms of MHC similarity (either as a continuous predictor or as shared MHC alleles) on the number of copulations (see Table S3 for a comprehensive list of all tested models and their AIC values). All models included also male treatment, female treatment and their interaction as fixed effects (see Materials and Methods). The interaction term male treatment  × female treatment was statistically significant in all models (all < 0.008; see also Statistical Methods and Bonisoli-Alquati et al., 2011a).
PredictorsEstimate (SE)zP
  1. 2 statistic from likelihood ratio test with reduced model, with d.f. = 2.

  2. Letters (a,b) denote statistically significant (< 0.05) differences at post hoc tests.

All data (N = 2997 pairings)
 MHC similarity (AIC = 1852)0.415 (0.198)2.100.036
 MHC similarity (shared MHC alleles) (AIC = 1850)
  None−2.637 (0.267)a7.65*0.021
  Some−2.338 (0.274)b  
  All−3.098 (0.572)a,b  
Excluding pairs with MHC-identical partners (N = 2939 pairings)
 MHC similarity (AIC = 1830)0.620 (0.218)2.840.005
 MHC similarity (shared MHC alleles) (AIC = 1831)
  None−2.630 (0.266)a2.470.013
  Some−2.335 (0.272)b  
image

Figure 1.  Mean (with 95% confidence limits) number of copulations in relation to major histocompatibility complex (MHC) similarity of mates, expressed as shared MHC alleles (none: male and female did not share any allele; all: male and female had the same genotype; some: male and female shared at least one allele) and male treatment. Means are estimated on 463 copulation events. Numbers above bars indicate sample size (number of possible male–female pairs, = 2997 pairings).

Download figure to PowerPoint

The interaction terms between male treatment, female treatment and MHC genotype never improved model fit when models were tested on the entire data set (details in Table S3). Interestingly, when we excluded MHC-identical pairings from the models of MHC similarity, the models with the addition of the male treatment × MHC similarity and female treatment × MHC similarity interactions resulted in a slight improvement in model fit (MHC similarity as a continuous variable, AIC = 1828 vs. 1830; shared MHC alleles: AIC = 1830 vs. 1831). Parameter estimates revealed a marginally nonsignificant interaction between male treatment and MHC similarity of partners (continuous variable: = −1.90, = 0.058; shared MHC alleles: z = −1.95, = 0.051; the interaction term between female treatment and MHC similarity was nonsignificant in both models; P-values always > 0.10). Hence, these results at best suggest that the difference in the number of copulations obtained by T and control males varied according to MHC similarity with their female partners (Fig. 1).

Randomization tests confirmed all statistically significant effects reported in Table 2 (details not shown for brevity).

Discussion

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

We investigated whether mate choice, male secondary sexual characters and social dominance rank varied in relation to the independent and combined effects of MHC-Phco IIB gene diversity and experimental in ovo T increase, simulating higher maternal transfer of androgens, in the ring-necked pheasant. We found that females avoided mating with males with a wholly different MHC genotype and tended to prefer males sharing some MHC alleles over MHC-identical ones. Moreover, male secondary sexual characters and social rank were not predicted by male MHC heterozygosity or diversity. At the same time, increased egg T levels did not markedly affect MHC-based mate preferences, although they affected female mate preference per se (see Bonisoli-Alquati et al., 2011a). A tendency towards T males realizing more copulations than controls emerged in pairs sharing no MHC alleles, suggesting the intriguing possibility that steroid-mediated maternal effects interact with MHC-based female mate choice. However, this effect was weak and not conclusively supported by statistical tests and thus will not be discussed further.

Female mate choice in relation to male MHC genotype

Female mate choice was not affected by MHC allele amino acid distance of mating partners, nor by male MHC heterozygosity. In addition, an ‘allele-counting’ strategy of MHC-based mate preference was not supported. However, copulations took place more frequently between males and females with intermediate levels of MHC similarity and less so between individuals that had either a wholly different or identical MHC genotype (in terms of shared alleles), although the difference was statistically significant only between the ‘no shared alleles’ and ‘some shared alleles’ pairing categories.

Mating preferences may have evolved along a continuum between avoidance of the outbreeding costs of mating with genetically incompatible mates and avoidance of inbreeding costs (Bateson, 1983; Thornhill, 1993). In the case of MHC-based mate preferences, the choice for intermediate levels of diversity could be the outcome of a trade-off between maximizing offspring immunocompetence and increasing the risk of developing autoimmune diseases (Kalbe et al., 2009). Although the choice for partners with similar genotypes can lead to inbreeding depression and higher susceptibility to pathogens (e.g. Potts & Wakeland, 1990), mating with highly dissimilar partners may result in too many MHC alleles in the offspring, possibly increasing the risk of developing autoimmune diseases (Wegner et al., 2003). Individuals with more MHC alleles present more antigens to the immune system, triggering a thymic selection and a consequent depletion of T cells (Woelfing et al., 2009). In addition, whereas mating with a highly dissimilar partner may provide a ‘moving target’ to pathogens escaping immune recognition (Penn & Potts, 1999), MHC-disassortative mating may also result in disruption of coadapted gene pools and loss of defence to local pathogen strains (Dionne et al., 2007; Eizaguirre & Lenz, 2010).

Our findings are therefore compatible with a strategy of MHC-based female mate choice whereby females affect offspring MHC genotype by choosing males with intermediate MHC dissimilarity, rather than by ‘allele counting’ (Piertney & Oliver, 2006). We could not statistically demonstrate avoidance of MHC-identical males, likely because of the low power of statistical tests due to the very small number of MHC-identical pairings (58 of 2997 pairings, and four of 463 copulations). However, we emphasize that, especially under natural conditions, the odds of a female pheasant encountering a wholly identical male at MHC loci are expected to be very small. In fact, both the high MHC polymorphism observed in this species and the pheasant mating system make the encounter between MHC-identical partners an unlikely event (see also Bonneaud et al., 2006b). Differently from other galliform birds, male pheasants do not gather at leks, but rather attract females to well-spaced territories. Thus, females are unlikely to visit a large number of males before making their mate choice. Selection for avoidance of MHC-identical partners may therefore be very weak because of the low encounter frequency between MHC-identical individuals.

Previous studies, conducted in diverse taxa, although often providing genetic evidence for offspring cohorts having an intermediate number of MHC alleles, have led to varied conclusions with respect to the specific mechanisms of MHC-based mate choice (e.g. Reusch et al., 2001; Aeschlimann et al., 2003; Olsson et al., 2003; Forsberg et al., 2007; Eizaguirre et al., 2009; Roberts, 2009; Griggio et al., 2011; Juola & Dearborn, 2012). Our study, by analysing the effects of several metrics of MHC genotype on mating patterns, adds a further piece to the complex mosaic of studies investigating MHC-based mate choice in birds. These studies vary from showing that females avoided mating with MHC-similar mates (Freeman-Gallant et al., 2003), avoided males without shared alleles and those with a low MHC diversity (Bonneaud et al., 2006b), preferred males with high MHC diversity as extra-pair mates (Richardson et al., 2005) or chose males via self-referent ‘allele counting’ (Griggio et al., 2011). In addition, MHC genotype may be involved in cryptic mate choice (Gillingham et al., 2009; Hale et al., 2009). Thus, mechanisms of MHC-based mate choice appear rather species-specific, and the general picture is further complicated by the high variability in MHC polymorphism among avian species, as well as by heterogeneity in metrics of genotypic variation at MHC loci among studies.

The mechanisms that mediate MHC-based mate choice in birds are still poorly understood (Zelano & Edwards, 2002). The lack of evidence of female mate preference based on male ornamentation from the same group of birds used in this study (Bonisoli-Alquati et al., 2011a) is compatible with the observed patterns of MHC-based mate choice, because mate selection for nonadditive genetic benefits may not be mediated by expression of male ornamental traits (Neff & Pitcher, 2005). MHC genes that cause nonadditive genetic variance in offspring quality (i.e. ‘compatible genes’; Neff & Pitcher, 2005) but are not revealed by male sexual ornamentation could result in variation in female mate choice unrelated to male ornaments. On the other hand, the overall expression of different male traits and their covariation, rather than individual sexual ornaments, may provide females with a reliable cue to the MHC genotype of individual males.

Some studies have suggested the involvement of the olfactory channel in MHC-based mate choice (Reusch et al., 2001; Aeschlimann et al., 2003; Milinski et al., 2005). Despite the traditional view that birds have a poorly developed sense of smell, olfaction may play a largely underestimated role in bird behaviour (Balthazart & Taziaux, 2009). A number of olfactory receptors in the chicken have recently been identified, implying that olfactory signals may be of some importance in mate recognition among galliform birds (Gomez & Celii, 2008).

Male secondary sexual traits, social rank and MHC genotype

In the present study, male ornament expression and social rank were not associated with MHC heterozygosity or diversity. Previous studies of ring-necked pheasants reported some associations between specific MHC genotypes and male morphology (von Schantz et al., 1996, 1997; Baratti et al., 2010). For instance, Baratti et al. (2010) found a significant difference in MHC genotypes between males with large and those with small wattles, whereas another study demonstrated that MHC genotype was associated with male spur length (von Schantz et al., 1996). Unfortunately, the high polymorphism observed in our captive sample of males (24 genotypes in 37 individuals) prevented any analysis of the association between male traits and specific MHC genotypes, thus making previous pheasant studies not directly comparable to the present one, as well as suggesting that ring-necked pheasant strains may retain different levels of MHC diversity. Moreover, genotype-by-environment interactions may result in variable relationships between male traits and MHC genotype: for instance, the lack of association between male traits and MHC genotype in our study might be due to the disruption of the natural association between ornament expression and MHC genotype under captive conditions (with ad libitum food and routine antiparasite pharmacological prophylaxis).

In conclusion, our study showed that female ring-necked pheasants avoid mating with the most MHC-dissimilar males, suggesting mating preferences for males with intermediate levels of MHC dissimilarity: although we could not statistically demonstrate avoidance of MHC-identical partners, we emphasize that the frequency of encounters with MHC-identical males was very small and may be an extremely unlikely event in natural pheasant populations. MHC-based mate preferences were apparently not mediated by male ornament expression and were only weakly affected by androgen-mediated maternal effects. Our findings suggest that MHC-based female mate choice in the ring-necked pheasant has evolved mainly to reduce the costs of high diversity at MHC loci, such as a higher risk of offspring autoimmune disorders or disruption of coadapted gene complexes.

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 thank G.P. Oldani for inestimable assistance and help. B. Leoni, S. Bocchi, A.T. Gerevini, U. Oldani and P. Usorini contributed to data collection, and we thankfully acknowledge their contribution. We also thank Dr. J.L. Waldron for comments, three referees and the Editor for constructive criticism. We thank R. Martinelli, A. Cavalleri and E. Venturelli for their help with hormone assays. A.B.-A. was funded by an MIUR PhD grant. Research was funded by a MIUR PRIN grant to N.S. (grant no. 20082X5YZC).

References

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

Dryad deposited at the Dryad: doi: 10.5061/dryad.5fc56

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

Table S1 Avian taxa included in the phylogenetic analysis with GenBank Accession Numbers.

Table S2 Alignment of partial MHC II β-class amino acid sequences of the 15 alleles characterized (see also Table S1).

Table S3 Summary of Poisson mixed models analysing the effects of MHC differences between mating partners (MHC similarity, shared MHC alleles, MF distance, MF distance PBR), male MHC diversity, the combined effects of male and female number of alleles (nallM × nallF), as well as male MHC heterozygosity, on the number of copulations.

Figure S1 Consensus tree of exon2 of MHC class IIB gene sequences, with the posterior probability (PP) values at the nodes (only PP of major nodes are shown).

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

FilenameFormatSizeDescription
JEB_2534_sm_TableS1-S3-FigS1.doc133KSupporting 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.