Response to family selection and genetic parameters in Japanese quail selected for four week breast weight

An experiment was conducted to investigate the effect of short-term selection for 4 week breast weight (4wk BRW), and to estimate genetic parameters of body weight, and carcass traits. A selection (S) line and control (C) line was randomly selected from a base population. Data were collected over two consecutive hatches for four generations. A total of 1 135 records from 156 sires and 218 dams were used to estimate the genetic parameters. The genetic improvement of 4wk BRW was 3.5, 2.7 and 0.6 g in generation 2, 3 and 4, respectively. The estimated heritability by using pedigree information was 0.35±0.06. There were a significant difference for BW, and carcass weights but not for carcass percent components between lines (P<0.01). The heritabilities and correlated responses for body weight (BW), carcass and leg weights were 0.46, 0.41 and 0.47, and 13.2, 16.2, 4.4 %, respectively. The genetic correlations of BRW with BW, carcass, leg, and back weights were 0.85, 0.88 and 0.72, respectively. Selection for 4 wk BRW improved feed conversion ratio (FCR) about 0.19 units over the selection period. Inbreeding caused an insignificant decline of the mean of some traits. Results from this experiment suggest that BW as a genetically correlated trait can be used to improve BRW.


Introduction
Family selection refers to a selection method in which family groups are ranked according to the mean performance of each family and whole families are selected or discarded.Often family average contains two different kinds of information.The first is the average breeding value of the family and the second is the environmental condition common to the whole family (Lush 1947).A major advantage with family selection is that, based on phenotypic observations from full or half-sibs, breeding values can be estimated for traits that cannot be measured on the individuals that are to be used as parents (Gjedrem 2005).Anatomical responses to selection under varying diets (Ricklefs & Marks 1985) the relationships between egg weight, hatch weight, and growth rates (Marks 1975) and survival rates (Aggrey 2002) of different lines have all been documented.Some studies show high positive correlations among live body weight and carcass traits.Redish (2004) reported that selection for pectoralis major muscle weight in maternal Japanese quail lines resulted in slight increases in the absolute weight of the pectoralis major muscle.Vali et al. (2005) found high correlations between live body weight and carcass weight components but very low correlations with carcass yield components.Breast and leg weights approximately compose 40 and 25 percent of carcass weight in quail, respectively and the remainder is the back which is less favored by consumers.So it seems direct selection for increased breast weight is more useful than indirect selection via increasing live body weight.There is limited data in the literature relative to family selection for carcass traits in quail, and particularly the most effective selection criterion to increase breast weight.Accordingly, three topics were investigated here using a selection experiment including a random-bred control.Firstly, in the present study response to selection for 4 wk breast weight (BRW) and correlated responses in carcass traits and body weight was calculated.Secondly, genetic parameters for these traits was estimated.Thirdly, the effect of inbreeding depression on the traits was assessed.

Animals
The experimental Japanese quail population (coturnix coturnix) originated from a commercial farming center in Yazd city, Iran.Around 1 000 birds were transported to the animal research station of Tehran University.Before the start of the experiment, the population was not selected for any traits.To establish a selection line (S-line) and a control line (C-line), a total of 210 birds were randomly selected from the population, then distributed equally and randomly into the two lines, and allowed to reproduce.The number of parents and progeny at 4 wk of age are presented in Table 1 by line, sex, hatch and generation.Birds in the S-line were individually leg-tagged, then two females were individually caged (25×25×30 cm) and mated to a single male every second day, while pairs of C-line females were caged together (25×25×30 cm) and mated to a single male so sex ratio was 1:2 (male:female) in each line.Birds were kept under circumstances that closely resemble commercial practice, i.e. a standard commercial feed containing 20 % CP and 2 650 Kcal ME/ kg, artificially lighted housing for 16 h per day.Food and water were available ad libitum.Eggs were collected daily and labeled by dam number to constitute pedigree.Eggs were stored up to 7 days at a temperature of 15 °C and humidity of 70 %.Eggs were set in setter for 14 days, and then, the eggs of each dam transferred to separate cells (S-line) in Hatcher trays for 3 days.At the time of hatching, the quails from the S-line were leg-tagged with a numbered plastic plate and quails from each line placed into separate pens.Quails were raised in group housing with 60 birds per square meter.quails had access to artificially lighted housing for 24 h per day, and a standard commercial feed containing 26 % CP and 2 900 Kcal ME/kg.Food and water were available ad libitum.Selection was done for three consecutive generations (a total of four generations including the base), and there were two hatches per generation in both lines.To select parents for the S-line, birds of hatch 1 were slaughtered, and then birds from the 50 % of full-sib families with the highest family breast weight (BRW) in hatch 1 were used as parents.Some females didn't lay until 69 days of age, increasing over generations.As, birds from the 50 % superior families, are selected 105, 95 and 80 birds were selected for replacement in generation 1, 2 and 3, respectively but the number of actually reproducing were fewer (Table 1).In the Cline pedigree was not recorded.A total of 105 birds of control line in hatch 2 were randomly selected.

Traits
Although some of the quail breeders started to lay eggs at 45 d of age, egg collection started at 56 d of age to obtain more eggs and chickens.The body weight (BW) was measured at 4 weeks of age.All quail from the first hatch in the S-line and approximately 80 birds of the C line were slaughtered, plucked, eviscerated and carcasses were kept for 4 h at 4 °C, then each carcass without feet was weighed (empty carcass weight).Carcass percent was calculated as the ratio of empty body weight relative to 4 wk BW.Breast and leg were separated and residual calculated as back.

Statistical Analysis
Comparison of means was done by SAS software 9.2 (SAS, 2000) using a generalized linear model: where Y ijklm was the observed trait, µ was the overall mean, L i was the fixed effect of i-th line (i=1,2), H j was the fixed effect of j-th hatch (j=1,2 for BW data), G k was the fixed effect of k-th generation (k=0,1,…,4), (LG) ik was the interaction between line and generation, ƒSex l was continuous covariate of average family sex ratio and e ijklm was the random error.Genetic analyses were carried out with the records of 1 135 (from 156 males and 218 females) fully pedigreed quail from the selection line.For all traits the initial models included the additive direct genetic effect, a maternal permanent environmental effect, an additive maternal genetic effect, and a covariance between direct and maternal genetic effects.The significance of components was determined using a likelihood ratio test (P=0.05)comparing models with and without the component.The models with maternal and permanent environmental effects were non-significant (the only exception was the permanent environmental variance for BW).The variance components, genetic parameters and inbreeding depression were estimated by ASREML software (Gilmour et al. 2000).The model used in bivariate analysis was: depression were estimated by ASREML software (GILMOUR et al., 2000 model used in bivariate analysis was: Where y 1 and y 2 represent different traits, b 1 and b 2 are the vectors o effects (including hatch, sex and generation), for trait 1 and 2, respe Vectors a 1 and a 2 are random additive genetic effects, pe 1 is maternal perm environmental effect when BW is one of the traits, and e 1 and e 2 are the r effects for trait 1 and trait 2, respectively.The incidence matrices X 1 associate elements of b 1 and b 2 with the records in y 1 and y 2 .The inc matrices Z 1 and Z 2 associate elements of a 1 and a 2 with the records in y 1 Incidence matrix W 1 associates element of pe 1 with records in y 1 (i.e.BW expectation of y 1 is X 1 b 1 , and the expectation of y 2 is X 2 b 2 .the va covariance structure of random effects of the bivariate animal model w follows: Where y 1 and y 2 represent different traits, b 1 and b 2 are the vecto effects (including hatch, sex and generation), for trait 1 and 2, re Vectors a 1 and a 2 are random additive genetic effects, pe 1 is maternal environmental effect when BW is one of the traits, and e 1 and e 2 are t effects for trait 1 and trait 2, respectively.The incidence matrices associate elements of b 1 and b 2 with the records in y 1 and y 2 .The matrices Z 1 and Z 2 associate elements of a 1 and a 2 with the records in Incidence matrix W 1 associates element of pe 1 with records in y 1 (i.e expectation of y 1 is X 1 b 1 , and the expectation of y 2 is X 2 b 2 .the covariance structure of random effects of the bivariate animal mod follows: depression were estimated by ASREML software (GILMOUR e model used in bivariate analysis was: Where y 1 and y 2 represent different traits, b 1 and b 2 are the effects (including hatch, sex and generation), for trait 1 and Vectors a 1 and a 2 are random additive genetic effects, pe 1 is ma environmental effect when BW is one of the traits, and e 1 and e 2 effects for trait 1 and trait 2, respectively.The incidence mat associate elements of b 1 and b 2 with the records in y 1 and y matrices Z 1 and Z 2 associate elements of a 1 and a 2 with the reco Incidence matrix W 1 associates element of pe 1 with records in expectation of y 1 is X 1 b 1 , and the expectation of y 2 is X 2 b covariance structure of random effects of the bivariate anima follows: depression were estimated by ASREML software (GILMOU model used in bivariate analysis was: Where y 1 and y 2 represent different traits, b 1 and b 2 are effects (including hatch, sex and generation), for trait 1 Vectors a 1 and a 2 are random additive genetic effects, pe 1 is environmental effect when BW is one of the traits, and e 1 an effects for trait 1 and trait 2, respectively.The incidence associate elements of b 1 and b 2 with the records in y 1 an matrices Z 1 and Z 2 associate elements of a 1 and a 2 with the Incidence matrix W 1 associates element of pe 1 with records expectation of y 1 is X 1 b 1 , and the expectation of y 2 is covariance structure of random effects of the bivariate an follows: depression were estimated by ASREML software (GIL model used in bivariate analysis was: Where y 1 and y 2 represent different traits, b 1 and b 2 effects (including hatch, sex and generation), for tra Vectors a 1 and a 2 are random additive genetic effects, environmental effect when BW is one of the traits, and effects for trait 1 and trait 2, respectively.The incid associate elements of b 1 and b 2 with the records in matrices Z 1 and Z 2 associate elements of a 1 and a 2 wit Incidence matrix W 1 associates element of pe 1 with re expectation of y 1 is X 1 b 1 , and the expectation of y covariance structure of random effects of the bivaria follows: Where y 1 and y 2 represent different traits, b 1 an effects (including hatch, sex and generation), fo Vectors a 1 and a 2 are random additive genetic effe environmental effect when BW is one of the traits, effects for trait 1 and trait 2, respectively.The i associate elements of b 1 and b 2 with the record matrices Z 1 and Z 2 associate elements of a 1 and a Incidence matrix W 1 associates element of pe 1 wi expectation of y 1 is X 1 b 1 , and the expectation covariance structure of random effects of the bi follows: Where y 1 and y 2 represent different traits, effects (including hatch, sex and generatio Vectors a 1 and a 2 are random additive genetic environmental effect when BW is one of the t effects for trait 1 and trait 2, respectively.associate elements of b 1 and b 2 with the re matrices Z 1 and Z 2 associate elements of a 1 a Incidence matrix W 1 associates element of p expectation of y 1 is X 1 b 1 , and the expect covariance structure of random effects of t follows: Where y 1 and y 2 represent different tra effects (including hatch, sex and gene Vectors a 1 and a 2 are random additive ge environmental effect when BW is one of effects for trait 1 and trait 2, respectiv associate elements of b 1 and b 2 with t matrices Z 1 and Z 2 associate elements o Incidence matrix W 1 associates element expectation of y 1 is X 1 b 1 , and the ex covariance structure of random effects follows: where y 1 and y 2 represent different traits, b 1 and b 2 are the vectors of fixed effects (including hatch, sex and generation), for trait 1 and 2, respectively.Vectors a 1 and a 2 are random additive genetic effects, pe 1 is maternal permanent environmental effect when BW is one of the traits, and e 1 and e 2 are the residual effects for trait 1 and trait 2, respectively.The incidence matrices X 1 and X 2 associate elements of b 1 and b 2 with the records in y 1 and y 2 .The incidence matrices Z 1 and Z 2 associate elements of a 1 and a 2 with the records in y 1 and y 2 .Incidence matrix W 1 associates element of pe 1 with records in y 1 (i.e.BW).The expectation of y 1 is X 1 b 1 , and the expectation of y 2 is X 2 b 2 .the variance-covariance structure of random effects of the bivariate animal model was as follows: matrices Z 1 and Z 2 associate elements of a 1 and a 2 with the records in y 1 and y 2 .
Incidence matrix W 1 associates element of pe 1 with records in y 1 (i.e.BW).The expectation of y 1 is X 1 b 1 , and the expectation of y 2 is X 2 b 2 .the variancecovariance structure of random effects of the bivariate animal model was as follows: Where σ 2 a1 and σ 2 a2 are direct additive genetic variances, pe 1 is maternal permanent environmental variance, σ 2 e1 and σ 2 e2 are the residual variances for trait 1 and 2, respectively; σ a1a2 is the direct genetic covariance between traits 1 and 2, and σ e1e2 is their residual covariance.A is an additive relationship matrix and I is an identity matrix.

Results
Descriptive statistical parameters of the traits analyzed are presented in Table 2 for both lines.Coefficient of variation was larger in the selected line due to the effect of selection.Genetic improvement was 3.5, 2.7 and 0.6 g for generations 2, 3 and 4 respectively.Selection for 4 wk BRW improved feed conversion ratio (FCR) 0.19 units over the selection period.Least squares means (family LSM for S-line) and standard errors by hatch and generation for different traits are shown in Table 3.The quails from the second hatch generally were heavier (P<0.01).There were a significant difference for all traits considered (except carcass percent components) between the two lines (P<0.001) from generation 2 and onwards.Egg weight was larger in the selected line from generation 1 and onwards.Heritability and variance component are presented in  matrices Z 1 and Z 2 associate elements of a 1 and a 2 with the records in y 1 and y .
Incidence matrix W 1 associates element of pe 1 with records in y 1 (i.e.BW).The expectation of y 1 is X 1 b 1 , and the expectation of y 2 is X 2 b 2 .the variancecovariance structure of random effects of the bivariate animal model was as follows: Where σ 2 a1 and σ 2 a2 are direct additive genetic variances, pe 1 is maternal permanent environmental variance, σ 2 e1 and σ 2 e2 are the residual variances for trait 1 and 2, respectively; σ a1a2 is the direct genetic covariance between traits and 2, and σ e1e2 is their residual covariance.A is an additive relationship matrix and I is an identity matrix.

Results
Descriptive statistical parameters of the traits analyzed are presented in Table for both lines.Coefficient of variation was larger in the selected line due to the effect of selection.Genetic improvement was 3.5, 2.7 and 0.6 g for generations 2, 3 and 4 respectively.Selection for 4 wk BRW improved feed conversion ratio (FCR) 0.19 units over the selection period.Least squares means (family LSM for S-line) and standard errors by hatch and generation for different traits are shown in Table 3.The quails from the second hatch generally were heavier (P<0.01).There were a significant difference for all traits considered (except carcass percent components) between the two lines (P<0.001) from generation 2 and onwards.Egg weight was larger in the selected line from generation 1 and onwards.Heritability and variance component are presented in  where σ 2 a1 and σ 2 a2 are direct additive genetic variances, pe 1 is the maternal permanent environmental variance, σ 2 e1 and σ 2 e2 are the residual variances for trait 1 and 2, respectively; σ a1a2 is the direct genetic covariance between traits 1 and 2, and σ e1e2 is their residual covariance.A is an additive relationship matrix and I is an identity matrix.

Results
Descriptive statistical parameters of the traits analyzed are presented in Table 2 for both lines.Coefficient of variation was larger in the selected line due to the effect of selection.Genetic improvement was 3.5, 2.7 and 0.6 g for generations 2, 3 and 4 respectively.Selection for 4 wk BRW improved feed conversion ratio (FCR) 0.19 units over the selection period.
Least squares means (family LSM for S-line) and standard errors by hatch and generation for different traits are shown in Table 3.The quails from the second hatch generally were heavier (P<0.01).There were a significant difference for all traits considered (except carcass percent components) between the two lines (P<0.001) from generation 2 and onwards.Egg weight was larger in the selected line from generation 1 and onwards.Heritability and variance component are presented in Table 4 based on bivariate models.Heritabilities ranged from 0.20 for back weight to 0.47 for leg weight.Likewise genetic and phenotypic correlations are shown in Table 5.There were high genetic and phenotypic correlations between BRW and the other traits.
The mean inbreeding for all birds and inbred birds are presented in Table 6 by generation and sex.The mean percentage of inbreeding for all birds and inbred birds was 0.64 and 11.3, respectively.Estimates of inbreeding depression are shown in Table 7. Effects of inbreeding were generally not significant.Figure 1 shows the genetic trend for 4 wk BRW in S-line based on predicted breeding values from the bivariate mixed model.The response was approximately constant over generation.

Genetic Improvement and Correlated Responses
The mean 4 wk BRW in S-line and C-line in the last generation were 47.1 and 40 g, respectively.This is equal to 15.1 % cumulative genetic improvement, or 5.0 % improvement per generation.Genetic improvement was 3.5, 2.7 and 0.6 g for generations 2, 3 and 4, respectively.The responses to short-term selection for body and carcass weights has been reported previously (Nestor & Bacon 1982, Toelle et al. 1991, Turkmut at al. 1999, Baylan et al. 2009).Brah et al. (2001) reported 9 generations of selection for 4 week body weight had increased the superiority of the selected strains over the control line to 44.2 and 35.9 %.Reddish (2004) reported that 6 generations of selection for pectoralis muscle weight resulted in 2.5 and 27 g improvement in pectoralis muscle weight and body weight in Japanese quail, respectively.The different responses to selection in different experiments can be due to selection intensity, accuracy of selection, genetic variance and different environmental conditions.
The results showed that selection for 4 wk BRW resulted in a correlated response especially in BW, carcass weight components and egg weight and less so in carcass percent components.Mean BW in S-line and C-line in the last generation were 195.2 and 169.5 g, respectively (Table 3).This represents 13.2 % total increase, or 4.4 % per generation.Correlated responses for carcass and leg weights were 16.2 and 4.8 % total response or 5.4 and 1.6 per generation, respectively (Table 3).These results indicate that BRW is favorably correlated with BW and carcass weight.Carcass, breast and leg weights in the S-line were higher than in the C-line (P<0.01).Generally differences between trait means in generation 3 and 4 are lower than in previous generations (Table 3).That is likely due to the substantial reduction in actual reproducing females because the progenies were available from fewer families (Table 1).Mennicken et al. (2005) reported divergent selection for ω3:ω6 had no effect on fertility and hatchability.Age at first egg, laying intensity and egg weight were also not different between the selected lines.
The mean of commercial FCR for S and C lines in the first generation was 2.55 and 2.59 and in the last generation, 2.37 and 2.60, respectively (0.19 unit improvement) or equals 8.8 % total response.Improved FCR to a certain body weight could be partially due to lower maintenance costs and lower fat deposition of birds with higher growth rate (Pym 1990).Knizetova (1996) concluded that live weight at 4 wk of age affected the relative growth rate and feed efficiency (weight gain/feed).Generally there is a favorable correlation between growth and FCR because of enhanced pulsative growth hormone release (Buyse et al. 1999, Leclercq et al. 1989).

Genetic Parameters
Redish (2004) reported 6 generations of selection for pectoralis muscle weight and obtained a realized heritability of 0.25.Vali et al. (2005) reported heritability for breast, leg and carcass weights of 0.26, 0.28 and 0.27, respectively.Falconer (1960) reported that heritability for a particular trait can take different values according to the population, the environmental condition surrounding the animal and the calculation method.Prado-Gonzalea et al. (2003) reported that differences in heritability may be due to method of estimation, population genetic structure, environmental effects and sampling error from small data set or sample size.As these changes are dependent on the number, effects and frequencies of the genes which influence the quantitative trait, long-term experiments may provide more detailed information about its underlying inheritance (Hill et al. 1992).As just pointed out correlated responses are due to high genetic correlation between BRW with BW and carcass traits (Table 5) that are in agreement with Gaya et al. (2006) and Vali et al. (2005).Shahin et al. (2005) reported genetic and phenotypic correlation between body and carcass weight were 0.83 and 0.88, respectively that are in agreement with current study.Redish (2004) reported that selection for pectoralis major muscle in maternal Japanese quail lines resulted in slight increases in absolute BW.

Inbreeding and Inbreeding Depression
Inbreeding caused a decline in the mean of some traits (Table 7).Values had a range of −0.06 to 0.06 for carcass weight and 4wk BW, respectively but its effect wasn't significant.Inbreeding depression has previously been reported for these traits (Khaldari et al. 2010).Brah et al. (2001) with the avoidance of mating between relatives reported Inbreeding levels of 0.32 to 0.43 % per generation that did not appear to be of any significance in affecting the response.Abplanalp (1967) reported that in quail populations three times more inbreeding depression for their entire reproductive cycle relative to domestic fowls can be observed, so in a closed population of quail a loss of 1 % hatchability would be sustained for every increase in the degree of inbreeding (about twice as severe as compared to chicken).
Results from this experiment do not suggest between family selection as the preferred method to increase breast weight because breast weight is highly correlated to BW, and BW is easier to record and can be recorded on selection candidates.Thus selection for BW would efficiently increase BRW, as shown by Khaldari et al. (2010).
by ASREML software (GILMOUR et al., model used in bivariate analysis was: by ASREML software model used in bivariate analysis was: by ASREML soft model used in bivariate analysis was: by ASREML model used in bivariate analysis was:

Table 1
Number of parents and progeny in each line by hatch and generation Table 4 based on bivariate models.Heritabilities ranged from 0.20 for back weight to 0.47 for leg weight.Likewise genetic and phenotypic Table 4 based on bivariate models.Heritabilities ranged from 0.20 for back weight to 0.47 for leg weight.Likewise genetic and phenotypic

Table 2
Descriptive statistics for body weight, carcass traits and egg weight of selected (S) and control (C) lines a : additive variance, σ e : environmental variance, σ pe : maternal permanent environmental variance