SEARCH

SEARCH BY CITATION

Keywords:

  • breeding values;
  • costly mating preference;
  • cuticular hydrocarbons;
  • Drosophila;
  • natural selection;
  • selection gradients;
  • sexual selection

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Fundamental to many theories of sexual selection is the expectation that sexual traits, which males use in an attempt to increase mating success, confer costs as well as benefits to individual males. Although evolution of exaggerated male traits is predicted to be halted, by costs applied by natural selection, there is a lack of empirical work devoted to quantitatively establishing whether natural selection opposes sexual selection generated by the preferences of females. In this study, we quantified natural and sexual selection gradients on breeding values for cuticular hydrocarbon (CHC) components of male contact pheromones in Drosophila serrata. As male sexual traits may often be environmentally condition dependent, breeding values were used in the selection analysis to remove the possibility of environmental correlations between the measured trait and fitness biasing estimates of selection. The direction of natural selection was found to oppose sexual selection on a subset of CHCs examined. Opposing natural and sexual selection suggests that further evolution of the male pheromone may in part be limited by costs associated with attractive male CHC blends.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

In many species, males possess exaggerated display traits that are used to attract females (Andersson, 1994). When female preferences are first established in a population, it is anticipated that initial exaggeration of attractive male traits may occur exponentially, during which time costs of trait exaggeration are expected to rapidly accumulate (Fisher, 1930; Lande, 1981). In many populations, therefore, it is likely that sexual traits may be costly to produce (Kotiaho, 2001) and if exaggerated beyond their natural selection optimum, further evolution of male sexual traits may eventually be balanced and halted by costs (Fisher, 1930; Kirkpatrick, 1987). There are a number of ways natural selection could potentially oppose further exaggeration of sexual traits (Kotiaho, 2001), these include resource competition between the development of sexual traits and other morphological characters (Emlen, 2001) or life-history components, and may result in negative genetic correlations between attractive trait phenotypes and other components of fitness (Kokko et al., 2002). Furthermore, males may be able to differentially tolerate costs of trait exaggeration depending on individual condition (Kotiaho, 2001; Cotton et al., 2004).

A number of studies have demonstrated a cost to male traits (Emlen, 2001; Kotiaho, 2001), including negative genetic correlations between attractiveness of males and other fitness components (Brooks, 2000; Hine et al., 2002). The demonstration of costly male sexual traits provides support for the theoretical prediction of opposing natural and sexual selection. In addition, responses to sexual selection through male–male competition and natural selection on wing length were found to be opposing in Drosophila melanogaster providing support for equilibrium between natural and sexual selection (Wilkinson, 1987). Although elegant methods for measuring selection gradients based on response surface methodology have been formulated (Lande & Arnold, 1983), natural and sexual selection gradients on male display traits have not been quantified and directly compared in any population.

Selection gradients resulting from the multiple regression approach of Lande & Arnold (1983) can be biased if covariation among traits is the result of environmental rather than genetic variation (Rausher, 1992). Environmental covariances between traits and fitness may occur when environmental factors that influence the trait phenotype also affect fitness. This problem may be of particular concern for sexually selected traits, as condition dependence is expected to evolve once a trait becomes costly (Rowe & Houle, 1996), and sexual traits including insect pheromones (Clark et al., 1997; Rantala et al., 2003), often demonstrate condition-dependence (David et al., 2000; Kotiaho, 2001; Kruuk et al., 2002; Cotton et al., 2004). More accurate estimates of selection may be generated by the use of breeding values in place of phenotypic values in analysis (Stinchcombe et al., 2002). Breeding values allow selection to be measured directly on the additive genetic variation through which evolution of a trait occurs (Rausher, 1992), removing the problem of environmental bias by describing individuals as the mean deviation of their progeny from the population mean (Falconer & Mackay, 1996).

Here, we quantified natural and sexual selection on the breeding values of a set of male sexually selected traits comprising a contact pheromone system in D. serrata. Females of this species discriminate between males based on their multivariate cuticular hydrocarbon (CHC) phenotypes resulting in sexual selection on these male traits (Hine et al., 2002; Chenoweth & Blows, 2003). A negative genetic correlation between the attractiveness of males and the productivity their genes conferred to females (Hine et al., 2002) suggested that natural selection may oppose the evolution of CHCs through sexual selection in this system. We first estimated natural and sexual selection gradients (Lande & Arnold, 1983) on breeding values of the male CHCs. We then directly compared natural and sexual selection gradients on a subset of the male CHCs using a modelling approach developed for categorical variables in response surface designs (Draper & John, 1988).

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Experimental design

A standard half-sib breeding design was used to estimate breeding values for: (1) male CHCs (male display traits), (2) a mating success score to quantify sexual selection and (3) a productivity score to quantify natural selection (Fig. 1). Each of 140 5-day-old virgin males from the previously described Forster laboratory population of D. serrata (Higgie et al., 2000; Hine et al., 2002) were placed in a vial with three 5-day-old virgin females and allowed 24 h to mate before females were transferred to individual vials for 5 days to lay. Progeny from the resulting 420 families were anaesthetized using carbon dioxide and sorted by sex whilst virgins. All flies used in experiments were virgins, stored at a constant 25 °C, exposed to a 12 : 12 light : dark photoperiod, and reared under controlled laboratory conditions.

image

Figure 1. Experimental design to estimate breeding values for individual male CHCs, mating success and productivity. Sires were each mated to three females producing three families. Two sons from each family had each phenotypic measure quantified, resulting in a total of six sons used per family. R denotes a random individual from the Forster population of D. serrata.

Download figure to PowerPoint

Quantification of CHC blend

From each half-sib family group, two 3-day-old F1 males were chosen at random to have their CHCs assayed by gas chromatography as per the established technique of Blows & Allan, 1998. In previous studies of sexual selection on D. serrata (Hine et al., 2002; Blows et al., 2004; Chenoweth & Blows, 2005) nine CHCs were examined. In this study, seven of these CHCs were examined; Z,Z-5,9-C25:2, Z-9-C26:1, 2-Me-C26, Z,Z-5,9-C27:2, 2-Me-C28, Z,Z-5,9-C29:2 and 2-Me-C30, as contamination within the gas chromatograph obscured two CHC peaks. Relative proportions of CHCs on the cuticle of each individual were used to control for variability in sample concentration attributable to differences in CHC extraction and evaporation of the sample (Blows & Allan, 1998). To remove the unit-sum constraint caused by the relative proportions totalling one, CHC logcontrasts were calculated by dividing all the proportions by that of an arbitrarily chosen peak (in this case Z-9-C26:1) and finding the log10 of each new variable (Aitchison, 1986). Logcontrasts were standardized (mean = 0, SD = 1) prior to genetic analysis.

Mating success

To obtain breeding values for mating success, two F1 males were used from each family group in mate choice tests. During a test, a random virgin female was given the choice between the son and a random male of the same age. To distinguish between the two males, wing clipping was performed on opposite wings of sons and random males 2 days prior to commencement of the choice tests. Wing clipping of both males controlled for any bias in female preference for clipped or unclipped males. Females were observed until a male was successful in mating and each son was recorded as either the chosen or rejected male. This method of measuring sexual selection is inclusive of both female choice and male–male competition, however, male–male competition may not play a large role in the mating success of males, as D. serrata females exercise strong choice between conspecific males (Hine et al., 2002; Blows et al., 2004) and can utilize various mechanisms to control the success of males attempting copulation (Hoikkala & Crossley, 2000).

Productivity

To obtain breeding values for productivity, the remaining two 6-day-old F1 males from each family group were placed into separate vials, each containing two randomly chosen 6-day-old virgin females. Males were allowed 24 h to mate with the two females, after which time, females were transferred to separate laying vials for a period of 3 days. Productivity was measured on the fourth day of F2 progeny emergence. The live mass of progeny and number of progeny per vial were recorded. Males failing to produce offspring with either female were discarded, as it is not known whether mating took place or if males or females were infertile. The productivity of each son was expressed as two scores: (1) the average mass in grams of all the progeny produced by the two females as per Hine et al. (2002) and (2) the number of offspring averaged across both females. Productivity measured in this way encompassed both production and quality of sperm, as well as, offspring fitness measures such as larval survival and offspring viability.

Genetic analysis

Breeding values for the six CHC logcontrasts, mating success and productivity were calculated as best linear unbiased predictors (BLUPs) in a mixed model with the use of restricted maximum likelihood (REML) for the estimation of variance components. The 84 sires included in the genetic analysis were those with data collected for each of CHC, mating success and productivity from at least two of the possible three families.

The use of breeding values to measure selection provided two further benefits in addition to the removal of environmental correlations among traits. First, CHC and fitness phenotypes could be measured independently using separate males from a family group. As male pheromone concentrations can change with exposure to females in some insect species (Birch & Haynes, 1982), measuring CHCs, mating success and productivity on separate males removed the possibility of CHC phenotypes changing between mate choice tests and quantification. Second, allowing productivity to be measured on separate individuals to mating success removed the opportunity for the experience of either success or rejection in mate choice tests to bias productivity measures, as all males used were virgins.

As the male CHC and productivity variables were normally distributed, breeding values were estimated using Proc Mixed in SAS. For each CHC, BLUPs were generated separately from the standardized logcontrasts of the six CHCs using the standard half-sib nested design (Falconer & Mackay, 1996). For productivity, mass and number of offspring was found to be highly correlated (r2 = 0.9103). To allow the most efficient use of the two productivity measures, a multivariate analysis was performed (Lynch & Walsh, 1998, p. 792) to generate breeding values for number of offspring encompassing information on mass of offspring. Due to the binomial distribution of mating success scores, a different method to estimate breeding values was required. The Glimmix macro implemented by Proc Mixed (Littell et al., 1996) was used to generate BLUPs for the binomial mating success measure and males from each dam were treated as separate observations. The predictors of mating success were computed on a logit scale.

Selection analysis

The Lande & Arnold (1983) multiple regression approach for estimating selection gradients is primarily applied to phenotypic data (Kruuk et al., 2003). Here, we employed this approach to estimate linear natural and sexual selection gradients (βn and βs) on breeding values. The selection analysis was conducted in two parts. First, linear natural and sexual selection gradients were estimated as partial regression coefficients of productivity and mating success BLUPs on standardized CHC BLUPs for all six CHCs (Table 1). Utilizing breeding values, however, decreases sample size (Stinchcombe et al., 2002) and subsequently reduced the degrees of freedom available to test for difference in overall direction of natural and sexual selection acting on CHCs. Therefore, to conserve d.f., regression coefficients were estimated for a second time after employment of variable selection via best subsets (Draper & Smith, 1981).

Table 1.  Standardized multivariate natural (βn) and sexual selection (βs) gradients on the CHCs of male D. serrata.
CHCh2Natural selectionSexual selection
βnPt (d.f. = 1)βsPt (d.f. = 1)
  1. CHC, cuticular hydrocarbon.

Z,Z-5,9-C25:20.240.0020.9470.660.0140.6610.441
2-Me-C260.46−0.0100.773−0.2890.0950.0152.484
Z,Z-5,9-C27:20.310.0250.3660.909−0.0570.078−1.785
2-Me-C280.230.0010.9800.025−0.0720.174−1.371
Z,Z-5,9-C29:20.53−0.0460.038−2.1140.0140.5780.558
2-Me-C300.060.0030.9080.1160.0020.9390.077

To perform variable selection, all the possible subsets of the six CHC BLUPs were evaluated using Proc Reg in SAS. The rankings of adjusted r2 were examined to identify separate best subsets important in explaining genetic variation in mating success (Table 2) and productivity (Table 3). The overall best subset was selected as the combination of CHCs identified in the mating success and productivity best subsets. Linear natural and sexual selection gradients were estimated as the partial regression coefficients of productivity and mating success BLUPs on the overall best subset CHC BLUPs. To allow regression coefficients to be compared within each selection type CHC–BLUPs were standardized.

Table 2.  Model selection determining the best subset of CHCs to explain mating success.
Model rankingNumber of CHCs in modelAdjusted r2Variables in model
  1. CHCs, cuticular hydrocarbons.

 130.09772-Me-C26, Z,Z-5,9-C27:2, 2-Me-C28
 240.08922-Me-C26, Z,Z-5,9-C27:2, 2-Me-C28,Z, Z-5,9-C29:2
 340.0877Z,Z-5,9-C25:2, 2-Me-C26, Z,Z-5,9-C27:2, 2-Me-C28
 440.08632-Me-C26, Z,Z-5,9-C27:2, 2-Me-C28, 2-Me-C30
 550.0798Z,Z-5,9-C25:2, 2-Me-C26, Z,Z-5,9-C27:2, 2-Me-C28, Z,Z-5,9-C29:2
1360.0679Z,Z-5,9-C25:2, 2-Me-C26, Z,Z-5,9-C27:2, 2-Me-C28, Z,Z-5,9-C29:2, 2-Me-C30
Table 3.  Model selection determining the best subset of CHCs to explain productivity.
Model rankingNumber of CHCs in modelAdjusted r2Variables in model
  1. CHCs, cuticular hydrocarbons.

 120.0330Z,Z-5,9-C27:2, Z,Z-5,9-C29:2
 210.0292Z,Z-5,9-C29:2
 320.0241Z,Z-5,9-C25:2, Z,Z-5,9-C29:2
 430.02302-Me-C26, Z,Z-5,9-C27:2, Z,Z-5,9-C29:2
 530.0214Z,Z-5,9-C27:2, 2-Me-C28, Z,Z-5,9-C29:2
1540.01092-Me-C26, Z,Z-5,9-C27:2, 2-Me-C28, Z,Z-5,9-C29:2
376−0.0146Z,Z-5,9-C25:2, 2-Me-C26, Z,Z-5,9-C27:2, 2-Me-C28, Z,Z-5,9-C29:2, 2-Me-C30

To test for overall difference in the direction of linear natural and sexual selection acting on the best subset of CHCs, a sequential model building approach for response-surface designs (Draper & John, 1988) was carried out in SAS using Proc GLM with use of a partial F-test (Bowerman & O'Connell, 1990) as employed by Chenoweth & Blows (2005). This method was employed as it enabled the interaction between the quantitative measure of the CHC breeding values and the qualitative measure of selection type to be investigated.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Genetic analysis

At the phenotypic level, productivity of females ranged from offspring numbers of 1–117 and masses of 0.001–0.101 g, whilst the mean number of F2 produced during the 24 h mating period by all F1 males within a family group ranged from 27 to 72. At the genetic level, the BLUPs for productivity varied between −0.03 and 0.35 phenotypic standard deviations from mean productivity, signifying that the genes of individual sires varied in their effect on mean productivity in the next generation by up to 40% of a phenotypic standard phenotypic deviation. Thirty four per cent of males (129 of 377) produced offspring with only one of two females in productivity tests. The productivity measure is indicative of male productivity over a 24-hour period when given access to two females. It is unknown whether both females were mated or number of times mating occurred within the 24-hour period. It is unlikely that female failure to reproduce was a result of female mating preferences (e.g. refusals), as all pairwise mating are successful and males are capable of inseminating greater than two females in 24 h. Heritability of offspring number was found to be 0.31 whereas mating success was found to have a heritability of 0.61. The significance of heritability values (h0 : h2 = 0) was tested using log-likelihood's. Briefly, a second analysis of variance was run within which the sire variance component was constrained at zero and the difference in likelihood's between the two models tested using chi-square. Genetic variances were found to be marginally significant for both offspring number (inline image = 3, P < 0.05) and mating success (inline image = 2.8, P < 0.05). Mating success varied between −0.50 and 0.38 phenotypic deviations from mean mating success. Transformation of mating success BLUPs to a probability of mating scale using P = exp[BLUP]/[1 + exp (BLUP)] indicated that the genes of individual sires varied in probabilities of gaining success in mating tests between 0.37 and 0.60. The heritabilities for each of the CHCs (Table 1) are similar to the range of previous estimates of CHC heritability (Blows et al., 2004).

Selection analysis

Using best subsets variable selection, the best subset of CHCs explaining variation in mating success was identified as 2-Me-C26, Z,Z-5,9C27:2 and 2-Me-C28 (Adjusted r2 = 0.0977) (Table 2), whilst Z,Z-5,9-C27:2 and Z,Z-5,9-C29:2 were identified as the best CHCs explaining variation in productivity (Adjusted r2 = 0.033) (Table 3). An overall best subset, on which natural and sexual selection were measured, was assembled from all the CHCs that were identified as important to either mating success or productivity, consisting of 2-Me-C26, Z,Z-5,9C27:2, 2-Me-C28 and Z,Z-5,9-C29:2. This model is necessarily a compromise between the two groups, and was the second best subset explaining mating success (Adjusted r2 = 0.089) (Table 2) and the second best subset of four CHCs explaining productivity (Adjusted r2 = 0.0109) (Table 3).

Removal of two CHCs via variable selection had little effect on selection analysis, as regression coefficients and their significance were comparable prior to and following variable selection (Tables 1 and 4). Significant positive sexual selection was identified on 2-Me-C26 (t = 2.484, d.f. = 1 and P < 0.05 initially and t = 2.634, d.f. = 1, P < 0.01 after variable selection) and significant linear natural selection on Z,Z-5,9-C29:2 (t = −2.114, d.f. = 1, P < 0.05 initially and t = −2.178, d.f. = 1, P < 0.05 after variable selection). Linear sexual selection on 2-Me-C26 was previously identified as being negative in this population when measured in phenotypic values (Blows et al., 2004). As the experimental design used here does not allow for the estimation of selection on phenotypic values (mating success and CHCs were not measured on the same individuals), we are unable to determine if the difference in selection gradients estimated here and by Blows et al. (2004) is a consequence of the removal of environmental covariance.

Table 4.  Standardized multivariate natural (βn) and sexual selection gradients (βs) on the best subset of CHCs of male D. serrata.
CHCh2Natural selectionSexual selection
βnPt (d.f. = 1)βsPt (d.f. = 1)
  1. CHC, cuticular hydrocarbon.

2-Me-C260.46−0.0110.701−0.3850.0890.0102.634
Z,Z-5,9-C27:20.310.0260.2911.063−0.0510.076−1.798
2-Me-C280.230.0050.8760.157−0.0600.100−1.662
Z,Z-5,9-C29:20.53−0.0460.032−2.1780.0120.6180.500

Natural and sexual selection were acting in antagonistic directions on the four best subset CHCs (Table 4). 2-Me-C26 and Z,Z-5,9-C29:2 were found to be under positive sexual selection and negative natural selection, while Z,Z-5,9C27:2 and 2-Me-C28 were under negative sexual selection and positive natural selection. Using the sequential model building approach (Chenoweth & Blows, 2005), it was found that the direction of linear natural and sexual selection on the best subset CHCs significantly differed (F4,162 = 2.44, P < 0.05). Although natural and sexual selection were opposite in sign on each of the best subset CHCs, only 2-Me-C26 and Z,Z-5,9C27:2 displayed significant univariate interaction of CHC by selection type terms (Table 5).

Table 5.  Sequential model building approach testing for an effect of selection type (natural or sexual selection) on the relationship between the best subset CHCs and fitness.
Sources of variationd.f.SSMSF ratioP
2-Me-C2610.0030.0030.130.723
Z,Z-5,9-C27:210.0260.0260.960.329
2-Me-C2810.0010.0010.020.885
Z,Z-5,9-C29:210.1100.1104.030.046
Selection type10.0000.0000.000.999
2-Me-C26*type10.1380.1385.090.025
Z,Z-5,9-C27:2*type10.1150.1154.230.041
2-Me-C28*type10.0510.0511.860.175
Z,Z-5,9-C29:2*type10.0880.0883.240.074

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The association between male attractiveness, mating success, and life-history traits in natural populations remains controversial. It is often suggested that if males in a population vary in genetic quality, it may be advantageous for females to mate with those which confer fitness benefits such as enhanced growth, survival (Petrie, 1994), attractiveness (Kokko, 2001) or overall viability to offspring (Kirkpatrick, 1987; Andersson, 1994). Two important meta-analyses suggest that increased survival or longevity may generally be positively phenotypically correlated with sexual traits in natural populations (Møller & Alatalo, 1999; Jennions et al., 2001). In contrast, laboratory quantitative genetic studies suggest that attractive male traits may often be costly and display negative genetic correlations between attractiveness of males and the fitness of their offspring (Brooks, 2000; Hine et al., 2002), resulting in the opposition of natural and sexual selection on male traits (Fisher, 1930). Although these two observations appear in conflict, such a pattern may be a consequence of environmental condition-dependence giving rise to positive phenotypic correlations between male display traits and fitness components (Kruuk et al., 2002), and at the same time, the evolution of an antagonistic genetic relationship between the two traits that halts the further exaggeration of male display traits as suggested by Fisher (1930).

In this study, natural and sexual selection gradients were found to be acting in opposition on CHC genes important in explaining variation in fitness in male D. serrata. This indicates that genes for attractive CHC blends, preferred by D. serrata females, are not favoured by natural selection for enhanced productivity. This productivity cost is consistent with the previous finding of a negative genetic correlation between mating success and productivity in male D. serrata (Hine et al., 2002). Although costly, the genes for attractive male CHC blends may be maintained in the population via increasing mating success, and hence, overall lifetime reproductive success of males (Kokko, 2001) or via yet undiscovered direct benefits (Cameron et al., 2003).

The cost to productivity, associated with genes for attractive CHC blends, may potentially manifest in a number of ways. Offspring viability is often considered one of the more important fitness indicators in sexual selection (Kokko et al., 2002), and a reduction in viability in the larval stage, may account for the decreased number of offspring reaching maturity. Equally, reduced productivity may result from resource allocation between production of CHCs and other traits. For example, production of CHCs and sperm in male D. serrata may trade-off in a similar fashion to the trade-off between CHC and egg production in female D. melanongaster possessing the ovo mutation for reduced egg production (Wicker & Jallon, 1995). To further investigate the interplay between attractive CHC genes and productivity, separate components of productivity could be measured including offspring sex ratios, to find whether both sexes of offspring are affected similarly, or if increased mortality of sons (Brooks, 2000) accounted for decreased productivity.

If the force of natural selection, antagonistic to further trait exaggeration, balances sexual selection through female choice, evolution of male sexual traits may be at equilibrium (Fisher, 1930; Kirkpatrick, 1987; Kokko et al., 2002). The study by Wilkinson (1987) provided support for an equilibrium between sexual selection by male–male competition on wing length as a measure of body size and viability selection in D. melanogaster, finding a response to viability selection opposite in sign and similar in magnitude to the estimated response to sexual selection on body size. The existence of equilibrium between natural and sexual selection by female choice however, has not been quantitatively demonstrated (Kirkpatrick, 1987). Although the opposing directions of direct genotypic natural and sexual selection in this study suggest male CHC evolution may facilitate an equilibrium, comparison of selection gradients may not be an appropriate test for equilibrium for two reasons.

First, although mate choice tests may provide a relatively comprehensive estimate of sexual selection on CHCs through female choice, natural selection on CHCs, which may manifest in any combination of life-history traits (Kokko et al., 2002), was estimated only via productivity. As it is unlikely that decreased productivity represents the only cost (or benefit) of attractiveness, and to establish the existence of equilibrium, overall natural selection acting through all life-history traits would need to be compared with sexual selection. It would however be very difficult to gain an accurate overall natural selection estimate, as all correlated fitness components of natural selection would need to be measured (Rausher, 1992).

Second, even if overall natural selection was estimated, the problem of directly comparing natural and sexual selection gradients estimated using different fitness measures on different scales must be overcome. For example, in our case of mating success vs. productivity, a standardized sexual selection gradient of 0.05 and a −0.05 gradient of natural selection indicates that for a one standard deviation change in CHC mean, mating success would increase by 5% and productivity decrease by 5%. Nevertheless, as attractiveness and productivity are measured on different scales, it is doubtful that the 5% increase in mating success would produce an increase in number of offspring, which exactly offsets the 5% reduction in productivity per mating. Therefore, opposing natural and sexual selection gradients might not necessarily be equal in magnitude to represent equilibrium.

Although formal methods of selection analysis are able to quantify the strength of selection (Lande & Arnold, 1983), they may not be appropriate to quantitatively test for equilibrium between natural and sexual selection. Multigenerational studies however, may be able to provide more direct evidence on whether male CHCs continue to be exaggerated or are experiencing equilibrium. For example, an experiment that manipulates the evolutionary processes acting on male traits may be able to establish whether net selection is directional or stabilizing and provide a more conclusive test of whether opposing natural and sexual selection balance. Finally, other processes may contribute to halting the further exaggeration of male CHCs, and in particular, genetic variance in the direction of sexual selection may be limiting (Blows et al., 2004). How natural and sexual selection interact with the genetic basis of male display traits and other fitness components to result in the observed phenotypic and genetic attributes of male display traits such as CHCs in D. serrata remains to be determined.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

We thank M. Higgie and G. Joseph for technical assistance and the members of the quantitative genetics group at UQ. MWB was supported by a grant from the Australian Research Council.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References