Selection for sexual male characters and their effects on other fitness related traits in white leghorn chickens

Abstract

Introduction The cockerel comb can be regarded as a secondary sexual character or a sexual ornament, the size of which increases greatly in connection with sexual maturity ( Parker et al. 1942 ; Aire 1973; Sharp et al. 1977 ). There are two kinds of benefit from exaggerated secondary sexual characters ( Darwin 1874) : as weapons, used for fighting among individuals of the same sex and as ornaments to attract the opposite sex in the mate selection. It has been shown in red jungle fowl that male comb size is positively correlated with winning in fights between pairs of males ( Ligon et al. 1990 ) suggesting a positive relationship between comb size and testosterone level. For this reason, male comb size is probably of importance in the intrasexual selection. Because sexual ornaments are assumed to be costly to express, it has been suggested that the expression of an ornament is favourably correlated with the bearer’s condition ( Zahavi 1975; Hamilton & Zuk 1982). According to this ‘handicap theory’, sexual selection through female choice would tend to increase male comb size, but because of the cost related to the trait, there would also be a limit for too large and costly combs under natural conditions. Hyaluronic acid is a polysaccharide used in medical treatments and extracted from cockerel combs. In a breeding project together with a pharmaceutical company, the main goal were to increase the yield and quality of hyaluronic acid from cockerel combs through selection ( Tufvesson et al. 1993 , 1994; Tufvesson et al. 1998 ). in addition to the direct selection effects, this project has also made it possible to study traits, important in the intrasexual selection and sexual selection under wild conditions, and how selection for comb traits influence reproduction, physiology and fitness characters, etc. In this study genetic and phenotypic parameters are presented and direct and correlated responses for comb size, comb weight, comb shape (the way the cockerel bears the comb), body weight and mortality are quantified in two selection lines (lines S and H) and one control line (line C). Line S was selected for comb size and comb shape and line H for comb hyaluronic acid concentration (HA). The aim of this investigation was to determine the consequences of directional selection on a male sexual character, the extent to which the male traits were genetically correlated and whether selection for a secondary sexual character affected viability. Furthermore, the effect of inbreeding on the traits was studied. Materials and methods The base population, measurements and selection procedures The breeding stock originated from a number of White Leghorn lines in a selection and crossbreeding experiment with laying hens ( Liljedahl et al. 1979 ). At the start of the project, four lines were created from these selection lines, in which male comb size and comb HA concentration had been measured. A detailed description for the origin of the lines is found elsewhere ( Tufvesson et al. 1998 ). The base population was selected, and the average inbreeding coefficients in the original lines and their proportions in the base generation for the two lines (line S and H) in this study are presented in Table 1. Table 1. . Inbreeding coefficients of the original lines and their proportional contributions to the base population of lines S and H Line S Line H Originalline Inbreedingcoefficient Proportion 1 ofsires in gen. 0(%) Proportion 1 ofdams in gen. 0(%) Proportion 1sires in gen.of 0(%) Proportion 1 of dams in gen. 0 (%) 1 14 5.4 (2) 100 (228) 0 0 2 11 48.7 (18) 0 18.9 (7) 18.1 (33) 3 12 0 0 10.8 (4) 10.4 (19) 4 12 5.4 (2) 0 10.8 (4) 10.4 (19) 5 11 0 0 8.1 (3) 9.3 (17) 6 3 21.6 (8) 0 2.7 (1) 3.9 (7) 7 17 0 0 8.1 (3) 7.7 (14) 8 15 0 0 8.1 (3) 8.3 (15) 9 12 0 0 2.7 (1) 3.3 (6) 10 12 16.2 (6) 0 0 0 11 3 2.7 (1) 0 29.8 (11) 28.6 (52) 1 number of animals within parentheses Male comb size (CS) and body weight (BW) were measured at 29 weeks of age. Comb height and length were measured with a ruler, and the product of the measurements was used as an approximation of comb area. Comb shape (SH), was defined to describe the way the cockerel bears his comb. SH was assessed from generation five onwards, at 32 weeks of age and assigned to one of five categories (5, excellent shape = upright comb; 1, poor shape = pendant comb along one side of the head). At 47 weeks of age, after slaughter, the combs were weighed (CW). Mortality (M) was recorded between 16 and 47 weeks of age. Hyaluronic acid concentration in the comb (HA) was measured for all cockerels in line H and analysed from biopsies taken at 28 weeks of age. To get a representative sample of the tissue, two biopsies from each comb were taken after anaesthetizing the cockerels. The analyses were performed by Pharmacia & Upjohn AB (Uppsala, Sweden) ( Bitter & Muir 1962). In this study, selection was carried out for 10 generations. The generation interval was 42 weeks for the first seven generations and 52 weeks for the last three generations. From generations one to five, the selection criterion in line S was male comb size at 29 weeks of age. When selecting strictly for comb area, comb size increased substantially and some of the cockerels had problems keeping their comb straight, and often it flopped down along one side of the head ( Tufvesson et al. 1998 ) . Therefore, from generation six onwards, comb shape was added as a selection criterion in this line. In generations one and two, selection was performed using a selection index. From generations three to five a single-trait animal model was used, and from generation six onwards, a multiple-trait model including both selection traits was used ( Johansson 1992). The models included the fixed effects of generation (10 classes) and rank (three classes). To avoid selecting animals whose combs were large because of their high social ranking within the male cages, an attempt was made to measure rank order. The body weight difference between the two cockerels in a cage at 16 weeks of age was used to assign them to one of three dominance-rank classes. Line H was selected for high HA in the comb at 28 weeks of age. A selection index was used in the two first generations and a single-trait animal model from generations three to 10. The fixed effect was generation (10 classes). The randomly mated line C, used as an environmental control, was housed externally for the first three generations and was available in the test house from generation four onwards. On average, there were 560 animals of each sex and selection line, with 20 sires and 120 dams used as parents per generation. Line C comprised an average of 150 animals of each sex. Housing and management At 13 weeks of age, the cockerels were randomly distributed in the test house. They were caged in pairs in a three-tier battery system for laying hens, modified for cockerels. The birds were fed ad libitum with a layers all-mash diet containing about 130 g/kg crude protein and 11 MJ/kg of energy. The cockerels were given 8 h of light per day during the first 15 weeks, whereupon the light period was successively increased by 1 h per week until 17 h of light per day was reached. The temperature was kept at 15–20°C. Statistical analyses Phenotypic line means were calculated for all traits and the variances used in the statistical analysis for testing the significance level were taken from the analysis of variance on the same data. Variances for line means were calculated ( Sorensen & Kennedy 1983) using actual mean additive relationships derived from the data. Variance components were estimated using REML (Restricted Maximum Likelihood) with an animal model. For this purpose we used the DMU package with an EM-algorithm ( Jensen et al. 1996 ). The estimation of variance components assumes unrelated, unselected and non-inbred base animals, which was not the case in the present experiment. Once the base population has been selected, bias in the analyses can be reduced by including additional relationships ( Van Der Werf & De Boer 1990). The individual relationships were, however, not known in the base population. Therefore, for each original line, average relationships were calculated and added to the base-population part of the relationship matrix. The full matrix was computed, whereupon the inverse was obtained using partitioned matrix inversion techniques. The analyses were performed in all three lines separately, and the multiple-trait mixed model used for the analyses was: y = Xb + Za + e, where y is a vector of observations for CS, BW, SH, CW and M and b contains the vector of fixed effects for each trait. For CS, b = the combined effect of generation and measuring person. For BW, CW and M, b = generation. For SH, b = generation and judging person. a is a vector of additive genetic effects; e is a vector of residuals; X and Z are known incidence matrices for fixed and random effects, respectively. To check for any correlated responses between the selected trait and the traits of interest for this study, line H was also analysed with respect to genetic correlations between HA and all other traits. First, the model described above was used, but with HA substituted for CS. Second, HA together with CS were analysed in a two-trait mixed model. For HA, b = generation. All models included the regression of the individual inbreeding coefficient. On the basis of the results of the multivariate analyses, which included the selection trait, the genetic level in each generation and line was calculated as the mean of the breeding values. The heritability for the categorical trait mortality, with only two values, was corrected according to Dempster & Lerner (1950) by using the following formula: h2 (corrected) = (h2p(l − p))/z2, where p = frequency of the mortality and z = the ordinate of the normal curve at the point P. Results Phenotypic and genetic trends Phenotypic and genetic trends for CS within line and generation are shown in Fig. 1. Only line S had a positive phenotypic slope over the generations, with an increase of 78% from generations one to 10. Line H showed a small, positive phenotypic trend over the first five generations. Line C did not change over time. In all lines the genetic trend for CS behaved similar to the phenotypic trend and line S showed a large amount of genetic progress. Figure 1Open in figure viewerPowerPoint Genetic (G) and phenotypic (P) trends in comb size (CS) in line S (selected for CS), line H (selected for high concentration of hyaluronic acid in the comb) and line C (a control line) Figure 2 illustrates the trends for BW. Phenotypic trends were similar for all lines, with a decrease in BW over the first eight generations followed by a rapid increase in generations 9 and 10. Line S showed a positive and line H a negative genetic trend. In line S the genetic trend showed a larger increase over generations compared with the phenotypic trend. Figure 2Open in figure viewerPowerPoint Genetic (G) and phenotypic (P) trends in body weight (BW) in line S (selected for CS), line H (selected for high concentration of hyaluronic acid in the comb) and line C (a control line) CW increased in all lines, but was most pronounced in line S ( Fig. 3). The phenotypic change between generations 1 and 10 was 84% in line S and 29% in line H. In line C the increase from generations 4–10 was 15%. For line S the genetic and phenotypic trends acted similarly. In lines H and C the genetic trends showed only small changes, whereas their phenotypic trends indicated an increase. Figure 3Open in figure viewerPowerPoint Genetic (G) and phenotypic (P) trends in comb weight (CW) in line S (selected for CS), line H (selected for high concentration of hyaluronic acid in the comb) and line C (a control line) The phenotypic score for SH between generations 5 and 10 declined in all lines ( Fig. 4), being −1.3, −1.5 and −1.0 in line S, H and C, respectively. The genetic trend for SH in line S decreased until generation seven (selection for better SH started in generation six) and increased thereafter. Line H acted similarly to line S, but here the breakpoint between decreasing and increasing slope occurred at generation five. The genetic trend was weaker in line H than in line S. Figure 4Open in figure viewerPowerPoint Genetic (G) and phenotypic (P) trends in shape (SH) in line S (selected for CS), line H (selected for high concentration of hyaluronic acid in the comb) and line C (a control line) In the two selection lines, M increased rapidly over the first five generations ( Fig. 5). In the last five generations there were large fluctuations but no sign of any further increase. In the C-line, mortality remained low in all generations except for some smaller fluctuations. In lines S and H the genetic trends for mortality decreased slightly which was opposite to the pattern of phenotypic change. Figure 5Open in figure viewerPowerPoint Genetic (G) and phenotypic (P) trends in mortality (M) in line S (selected for CS), line H (selected for high concentration of hyaluronic acid in the comb) and line C (a control line) The phenotypic trends for all traits except for BW, in line S, were significant (p < 0.001). In line H the trends were non-significant for all traits except for SH (p < 0.05). Inbreeding Figure 6 illustrates the change in inbreeding. Figures in generation zero show the average inbreeding coefficient in each of the two base populations which was dependent on the relations of the origin lines within the base populations. From generation two onwards line S showed a more or less linear increase of the inbreeding coefficient, while in line H it remained low even in the second generation and then showed a rapid increase. In line C there was a marginal increase in the inbreeding coefficient over the six generations. The regressions of inbreeding on each analysed trait were all low except for the regression between inbreeding and M in line H. This regression was 0.50 (% M/% inbreeding). Figure 6Open in figure viewerPowerPoint Inbreeding coefficients (I) in line S (selected for CS), line H (selected for high concentration of hyaluronic acid in the comb) and line C (a control line) Heritabilities and correlations The heritabilities were moderate to high for all traits except mortality ( Table 2). There were small differences in heritability between the lines for all traits. Genetic and phenotypic correlations in line S, H and C are given in Tables 3, 4 and 5, respectively. The genetic correlations were similar between lines. There was a high positive genetic correlation between CS and CW. Both CS and CW showed a negative correlation with SH and a positive one with BW. The genetic correlation between BW and SH was low in all lines. In line H the genetic correlations of HA with CS, BW, CW, SH and M were 0.15, 0.03, 0.11, 0.05 and 0.05, respectively. Table 2. . Heritabilities (h2) with standard errors, within parenthesis, and phenotypic standard deviation (σp) in cm 2 , g, score units, kg and percentage resp. for comb size (CS), comb weight (CW), comb shape (SH), body weight (BW) and mortality (M) in the three lines Line S Line H Line C Trait h 2 σp h 2 σp h 2 σp CS 0.56 (0.03) 18.6 0.54 (0.03) 14.7 0.53 (0.05) 14.0 CW 0.54 (0.03) 21.6 0.63 (0.03) 17.8 0.64 (0.05) 22.9 SH 0.43 (0.03) 86.4 0.48 (0.03) 103.0 0.50 (0.05) 95.1 BW 0.59 (0.03) 20.2 0.68 (0.02) 17.5 0.51 (0.05) 19.9 M 0.09 (0.05) 32.1 0.07 (0.01) 27.1 0.09 (0.03) 17.1 Table 3. . Genetic (above) and environmental correlations (below) with standard error, within parenthesis, between comb size (CS), comb weight (CW), comb shape (SH), body weight (BW) and mortality (M) in line S Trait CS CW SH BW M CS 0.89 (0.02) −0.49 (0.05) 0.29 (0.04) −0.06 (0.17) CW 0.27 (0.03) −0.39 (0.06) 0.36 (0.04) 0.07 (0.17) SH −0.11 (0.03) 0.02 (0.04) 0.12 (0.06) −0.08 (0.19) BW 0.18 (0.03) 0.24 (0.03) 0.15 (0.04) −0.08 (0.16) M −0.04 (0.02) −0.19 (0.02) −0.03 (0.02) −0.18 (0.02) Table 4. . Genetic (above) and environmental correlations (below) with standard error, within parenthesis, between comb size (CS), comb weight (CW), comb shape (SH), body weight (BW) and mortality (M) in line H Trait CS CW SH BW M CS 0.86 (0.02) −0.58 (0.05) 0.36 (0.04) 0.29 (0.11) CW 0.39 (0.03) −0.38 (0.05) 0.44 (0.03) 0.47 (0.11) SH −0.11 (0.03) −0.05 (0.04) 0.02 (0.06) −0.21 (0.13) BW 0.21 (0.03) 0.13 (0.04) −0.04 (0.04) 0.28 (0.11) M 0.06 (0.02) 0.14 (0.02) 0.04 (0.02) −0.17 (0.02) Table 5. . Genetic (above) and environmental correlations (below) with standard error, within parenthesis, between comb size (CS), comb weight (CW), comb shape (SH), body weight (BW) and mortality (M) in line C Trait CS CW SH BW M CS 0.86 (0.03) −0.73 (0.05) 0.17 (0.08) −0.19 (0.26) CW 0.48 (0.05) −0.55 (0.06) 0.26 (0.07) −0.38 (0.32) SH −0.22 (0.06) −0.09 (0.07) −0.14 (0.08) 0.37 (0.32) BW 0.39 (0.05) 0.42 (0.06) −0.09 (0.06) −0.32 (0.28) M −0.10 (0.05) 0.38 (0.06) 0.02 (0.05) −0.09 (0.05) Discussion Consequences of directional selection on comb size All models of sexual selection assume that traits are costly to produce and maintain ( Møller 1996). The phenotypic and genetic trends for CS ( Fig. 1) in combination with its high heritability illustrate the potential for selection for this trait. The genetic progress in line S, with a nearly twofold increase in CS, indicates that if there is sexual selection for the trait one would expect a continuous increase in CS if other fitness related traits are not negatively related to CS. von S chantzet al. (1995) used cockerels from generation nine to show that the males with largest combs were preferred as mating partners by the females even when the average comb size was nearly doubled, due to the artificial selection. The theories, according to which secondary sexual characters under sexual selection pressure will confer a fitness cost, would explain why natural and sexual selection under wild conditions do not result in an increase of CS over time. According to this line of argument, environmental conditions should remain unchanged over time. If the environmental conditions were to change one might expect a new optimization process to result in the most favourable combination of fitness-related traits. If the quality of the environment declines, CS will probably tend to get smaller because under these new conditions the cost of the character will increase. On the other hand, CS probably will tend to get larger under more favourable environmental conditions which should partly explain the large genetic progress in line S. The small positive genetic and phenotypic trends in line H might due to the positive correlation between HA and CS (0.15) in this line. Correlated responses in body weight CS showed a positive correlation with BW in all lines ( Tables 3, 4, 5) which explains the positive phenotypic and genetic trends in line S for BW ( Fig. 2). In the perspective of sexual selection, if both CS and BW are of importance in the male–male competition and/or in the female mate choice the genetic response for each trait would be favoured by the positive genetic correlation between CS and BW. In the broiler industry ( Siegel & Dunnington 1985; Robinson et al. 1993 ) it is well known that large animals very often have reproductive difficulties. In fitness terms, this correlated response could be a handicap when the BW increases to the point that it exceeds its optimum for the specific environment. Line H, on the other hand, showed a negative genetic trend in BW over the first four generations. The negligible correlation between BW and the selection trait HA (0.03) suggests that a correlated response is not the cause of the negative trend. One reason for the negative BW trend might be that the two most represented original lines, lines 2 and 11 ( Table 1), were also the two heaviest lines, and over the first generations, as line H became more and more genetically homogenous, a normalization of the BW at a lower level occurred, caused by natural selection. Correlated responses in comb weight The genetic correlation between CS and CW was high in all lines ( Tables 3, 4, 5) which also explains the positive phenotypic and genetic responses in CW in line S ( Fig. 3). Accordingly, those traits should have many genes in common, and measures of CS on the live animal should give a good prediction of comb mass. CW may also be relevant to the sexual selection theories given that high comb weight as well as large comb size confer a cost to the bearer. Correlated responses in comb shape It is likely that both CW and CS affect the shape of the comb (SH), where the latter character may further add to the impression of comb size as visualized by females during mate choice and by males during male–male competition. One reason for the negative phenotypic trends for SH in all lines, including line C ( Fig. 4), could be that the measuring persons have lowered the scale over the generations. When comparing the genetic trends ( Fig. 4), they started at generation zero but the measurements of SH were not started until generation five. The predicted breeding values were thus based on the relationship matrix together with the genetic covariances between SH and the other traits for the generations with no SH measurements. The negative genetic correlation between CS and SH ( Tables 3, 4, 5) is shown in the genetic trend for line S. During the first seven generations the only selection was for CS, and the genetic correlation caused a negative response in SH. Consequences of directional selection on both comb size and comb shape From generation eight onwards line S was selected for both CS and SH which resulted in a positive genetic SH trend. The initial drop in the genetic trend for line H is more difficult to interpret because there was no sign of a correlated response between SH and HA (rg = 0.05). The reason to the initial drop for line H might have been the lower SH level in the base population of line H which was visualized in generation five once the measurements of SH had started. In lines H and C the genetic correlation was less negative between SH and CW than between SH and CS ( Tables 4, 5), but in line S the two correlations did not differ. There is probably a positive relationship between SH and comb thickness because a thick comb is probably easier to keep straight up compared with a thin one. For that reason one would expect the correlation to be less negative between with SH and CW than between SH and CS. If, in choosing a mate, hens prefer males whose combs are both large and well shaped, then they would be more likely to choose a male with a heavy comb, compared with the situation where comb size alone is used as the selection criterion. In this case, two sexual traits should be considered when assessing the value of the comb as a sexual ornament. Correlated responses in mortality The genetic correlation between CS and M was low and with large standard error, in all lines. Von Schantz et al. (1995) used logistic regression and found no relationship between CS and M but got a positive regression of comb size at 16 weeks of age on M in line S and in two other lines, selected for increased comb size at 23 and 29 weeks of age. In line H, mortality increased over generations but the genetic trends showed a slight decrease ( Fig. 5). The high regression coefficient for inbreeding with M in line H (0.5) indicated a sensitivity of inbreeding to M in this line which may explain the increase in M. Intrasexual selection The role of the comb at intrasexual selection depends upon whether it is just an effect of a correlated response with aggressive behaviour via testosterone level as described by Ligon et al. (1990) or if the impression of the comb also is of importance for the ranking order among the males. In an earlier study by Tufvesson et al. (1993) using lines S and C, the testes at 52 weeks of age was significantly heavier in line C (16.8 g) than in line S (11.0 g) after nine generations and the genetic correlation between testes weight and CS was negative (−0.23). On the other hand the correlation between testes weight and BW was positive (0.26). The blood level of testosterone was not measured in the present study but it will be measured in further studies. If the comb itself is of importance in the intrasexual selection one may expect a larger importance of SH than if the only effect of the comb size is aggressive behaviour via the level of testosterone. The authors’ experience indicates that SH is much more dependent on the cockerel’s condition at a certain time than CS is. SH often deteriorates just before the cockerel is recorded to be in a pathological condition (T ufvesson M., personal observation; Zuk et al. 1990 ). The reason for the low environmental correlation between SH and M in all lines ( Table 3,4,5) would, accordingly, be that all cockerels were SH-judged at 32 weeks of age but M could occur long after, namely up to the age of 47 weeks. Environment-specific optimal level of expression When placing a population into a new, richer environment such as climate-controlled facilities, ad libitum feeding, etc. one expects a rapid selection response, because more of the genetic variation will be expressed in the phenotype. The response however, soon reaches a new level at which further increase is limited by the environment ( Beilharz et al. Beilharz, R. G., ; Luxford, B. G. & ; Wilkinson, J. L. Quantitative genetics and evolution: is our understanding of genetics sufficient to explain evolution? J. Anim. Breed. Genet., 110 (3), 161–170. ) . The environmental conditions were obviously favourable for the trait CS as indicated by the high heritabilities for CS in all lines and the large selection response in line S. As CS approaches its environmental limit, the heritability and genetic progress can be expected to decline. It seems that 10 generations of selection for increased CS is not enough to reach the environmental limit at which CS is expected to stabilize at an optimum size determined by natural selection. This fact can also explain the low genetic correlation between M and CS in line S ( Table 3). The size and shape of the cockerel comb, can be regarded as secondary sexual characters or sexual ornaments. Sexual characters are assumed to be costly to express and the expression of a secondary sexual character is suggested to be favourably correlated with the bearer’s condition or fitness. The aim of this study was to clarify the effects of selection for male sexual characters, the correlated responses in other male traits and whether selection for sexual characters affects viability. Two selection lines (lines S and H) and a control line was used for 10 generations. Line S was selected for male comb size at 29 weeks of age and from generation six onwards, comb shape (the way the cockerel bear his comb) was added. Line H was selected for high concentration of hyaluronic acid (HA) in the comb at 28 weeks of age. The average number of animals of each sex and selection line was 560 and the randomly mated control line comprised an average of 150 animals of each sex per generation. Traits recorded and analysed were comb size (CS) and body weight (BW) at 29 weeks of age, comb shape (SH) at 32 weeks of age, comb weight (CW) after slaughter and mortality (M). In line S, the genetic and phenotypic trends increased for CS, CW and BW. Both CS and SH are traits involved in the impression of comb size as visualized by females during mate choice and by males during male–male competition. With artificial upward selection for the male character CS (line S), CS, CW, BW and M also increased but SH was impaired. When adding SH to the selection criteria in line S, the negative genetic trend for SH was changed to positive. As CS approaches its environmental limit, the heritability and genetic progress can be expected to decline. It seems that 10 generations of selection for increased CS is not enough to reach the environmental limit at which CS is expected to stabilize at an optimum size determined by natural selection.