- Email: [email protected]
Estimates of genetic parameters for monthly egg production in a commercial female broiler line using random regression models
Livestock Science 153 (2013) 33–38
Contents lists available at SciVerse ScienceDirect
Livestock Science journal homepage: www.elsevier.com/locate/livsci
Estimates of genetic parameters for monthly egg production in a commercial female broiler line using random regression models N. Farzin a,n, R. Vaez Torshizi b, A. Gerami c, A. Seraj a a b c
Department of Animal Science, Azadshahr Branch, Islamic Azad University, Azadshahr, Iran Department of Animal Science, Tarbiat Modares University, Tehran, Iran School of Mathematics, Statistics and Computer Science, University of Tehran, Tehran, Iran
a r t i c l e i n f o
abstract
Article history: Received 18 November 2010 Received in revised form 24 January 2013 Accepted 29 January 2013
In the present study, genetic parameters for monthly egg production, collected from weeks 24–55 on 16,200 hens during 9 generations from a pedigreed commercial female broiler line, were estimated using random regression models. With and without age at sexual maturity (ASM) as a covariate, two analyses were fitted to the data for the different models. To identify the best orders of Legendre polynomials, nine different models were compared. On the basis of Bayesian information and Akaike information criteria, a random regression model with second order Legendre polynomial for fixed effect and third order for additive genetic and permanent environmental effects, was chosen as optimal model. Without considering ASM as a covariate, the heritability estimates of monthly egg productions ranged from 0.099 to 0.229. By including ASM in the model, heritabilities for the first and second monthly records decreased by 38.86% (from 0.229 to 0.140) and 51.33% (from 0.150 to 0.073), respectively. Genetic and phenotypic correlations among monthly egg records were high between adjacent periods and decreased as the time interval increased. Genetic correlations between the first monthly records and other periods changed from positive to negative (varying from 0.095 to 0.619) with considering ASM in the model. The effect of age at sexual maturity on the estimates of heritability and genetic correlations for the first month records suggests it is necessary to include ASM in the analysis of egg production to avoid biased estimates. The heritability of the fourth month egg production and its relatively high genetic correlations with all other later ages show that it could be the most appropriate period for selection. & 2013 Elsevier B.V. All rights reserved.
Keywords: Age at sexual maturity Female broiler line Genetic parameter Monthly egg production Random regression model
1. Introduction Egg number is a longitudinal trait that depends on months of production, increasing to a peak and then decreasing over time. The description of this trait is various in different studies. To analyze the egg number,
n Corresponding author. Tel.: þ 98 911 143 6115; fax: þ98 174 673 4601. E-mail addresses: [email protected], [email protected] (N. Farzin), [email protected] (R. Vaez Torshizi), [email protected] (A. Gerami), [email protected] (A. Seraj).
1871-1413/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.livsci.2013.01.015
monthly (Anang et al., 2001, 2002; Farzin et al., 2011; Nurgiartiningsih et al., 2004; Wolc et al., 2007a, 2007b; Wolc and Szwaczkowski, 2009) or cumulative (Farzin et al., 2010; Luo et al., 2007; Nurgiartiningsih et al., 2004) egg records can be used. Selection for improving egg production was usually done based on early part records, generally up to 40 week of age (Luo et al., 2007). So, to design a suitable selection program, we need to estimate the heritabilities and genetic correlations among different part records. The pattern of genetic and phenotypic variances of egg production have been analyzed in literature using single or multiple trait (Anang et al., 2000; Farzin et al., 2011; Nurgiartiningsih et al., 2004),
34
N. Farzin et al. / Livestock Science 153 (2013) 33–38
repeated records (Farzin et al., 2011; Kranis et al., 2007; Wolc et al., 2007a), fixed regression (Anang et al., 2001; Wolc et al., 2007b) and random regression models (Anang et al., 2002; Luo et al., 2007; Wolc and Szwaczkowski, 2009). Using of random regression models has been increased to describe longitudinal traits (Schaeffer and Dekkers, 1994), because of their flexibility and capability to account for time-dependent effects (Swalve, 2000). Anang et al. (2002) suggested that the random regression was the feasible model to estimate genetic parameters of egg numbers on laying chickens, when compared against the multiple-trait model. Also, similar results were indicated in turkeys by Kranis et al. (2007). In spite of several reports on laying hens, few estimates of genetic parameters have been reported for egg production in broiler lines. Moreover, the estimates of heritability for egg numbers varied strongly in various studies and depended on population, time, trait description and models used, ranging from 0.01 to 0.61, in laying hens (Anang et al., 2001; 2002; Nurgiartiningsih et al., 2002, 2004; Szwaczkowski, 2003; Unver et al., 2004; Wolc et al., 2007a) and 0.06 to 0.43, in broiler dam lines (Farzin et al., 2011; Luo et al., 2007). Also, some studies reported a high estimate of heritability for the first month of laying period (Anang et al., 2000, 2002; Kranis et al., 2007; Nurgiartiningsih et al., 2004; Wolc et al., 2007a), which may be resulted from the variation in age at sexual maturity (Anang et al., 2000). Farzin et al. (2011) estimated the genetic parameters of monthly egg production using single-trait, multiple-trait and repeated records animal models and reported that the exclusion of age at sexual maturity from the analysis of monthly egg production, especially for the first records of laying period, resulted to overestimation of (co)variance components and their corresponding parameters. Therefore, the objective of the present study was to estimate genetic parameters for monthly egg production in a female broiler line, using random regression model, and investigate the effects of including the age at sexual maturity in the model. 2. Materials and methods The data were collected from a commercial female broiler line, selected during nine generations (between 2000 and 2008), moderately for body weight and residual feed intake at seven weeks of age and mainly for the total number of chicks produced during 28 to 39 weeks of age, average egg weight at weeks 31 and 33 and age at sexual maturity (ASM) at 20 weeks of age. Body weight was recorded individually for each bird, whereas the residual feed intake was measured from the performance records of their half sibs and full sibs. Also, selection of the birds at 20 weeks of age was carried out based on information of their relatives collected from all previous generations. Chicks were raised on full feed for the first six weeks and then fed restricted amounts daily until 20 weeks. At 20 weeks of age, all selected birds (approximately 134 roosters and 1206–1474 hens per generation) were randomly assigned to floor pens, 9 to 11 hens with 1 rooster per pen. The progeny of each generation were produced from up to four different hatches (weeks 43–50 for the
first six generations and weeks 45–48 for the last three generations) through collecting hatching eggs over two consecutive weeks. For this line, weekly egg productions of 16,200 hens from 1198 sires and 7564 dams were recorded individually in trap-nests. The period of data collection was from 24 to 55 weeks of age (32 weeks). Monthly egg records were generated by summing each 4 continuous weekly eggs (e.g., M1¼sum of the eggs produced from 24 to 27 weeks of age to M8 ¼sum of the eggs produced from 52 to 55 weeks of age). For each weekly record, it was assumed that all missing values were known and equal to zero. Monthly records were described based on the following random regression model using Legendre polynomials as covariates: yikl ¼ GHi þ
q1 X m¼0
bm zklm þ
q2 X m¼0
akm zklm þ
q3 X
pkm zklm þ eikl
m¼0
where yikl is the monthly egg record of kth hen in lth month, GHi is the fixed effect of ith generation-hatch, bm is the mth fixed regression coefficient, akm is the mth random regression coefficient for additive genetic effect, pkm is the mth random regression coefficient for permanent environmental effect, zklm is the covariate of the Legendre polynomial, eikl is the random residual effect, and q1, q2, and q3 are the orders of the Legendre polynomials for the fixed, additive genetic, and permanent environmental effects, respectively. In the repeated measurements, the residual variance can be assumed homogeneous or heterogeneous over time. To improve the accuracy of estimations in the present study, the residual variances were assumed heterogeneous throughout the laying period and divided into eight classes. Although, it was reported that inclusion of heterogeneous residual variances had little influence on the values of genetic parameter estimates (Wolc and Szwaczkowski, 2009). To investigate the effect of ASM on (co)variance components and genetic parameters, the data were reanalyzed by fitting this effect as a covariate in the model. Estimates of variance and covariance components and their respective parameters were carried out by the restricted maximum likelihood method (REML) using WOMBAT program (Meyer, 2007). Based on the phenotypic means pattern of egg productions over months, the Legendre polynomial of order 2 was used for modeling the fixed effect of month on egg production. Various orders of Legendre polynomial were characterized for additive genetic and permanent environmental effects. These orders are often chosen to be the same for simplicity of computing (Schaeffer, 2004), but in the present study, to determine the adequate model, both equal and unequal orders were tested. Nine different random regression models were compared to identify the best orders of fit for additive genetic and permanent environmental effects. Orders higher than 3 were not used, because no change in the patters of genetic variances with increasing polynomial orders was observed. Bayesian information criterion, BIC (Schwarz, 1978) and Akaike information criterion, AIC (Akaike, 1974) were used to
N. Farzin et al. / Livestock Science 153 (2013) 33–38
choose the appropriate random regression model as follows: BIC k ¼ 2logðMLk Þ þ pk logðnÞ where BICk is the Bayesian information criterion of model k, log(MLk) is the log of maximum likelihood value of model k, pk is the number of free parameters in model k; and n is the number of observations that contribute to the likelihood, and AIC k ¼ 2logðMLk Þ þ2pk where AICk is the Akaike information criterion of model k, and the other symbols are denoted as in the BIC equation. The model with smallest values of BIC and AIC was regarded as the optimal model. 3. Results and discussion The average of monthly egg production was increased until a peak in the second month and subsequently decreased till the end of the production period (Fig. 1). Also, the average ASM was 178.8 710.8 days and ranged from 162 days to 226 days. The values of BIC and AIC for several models with different orders of fit without including ASM are shown in Table 1. Since the variation patterns of these criteria were the same for the models with ASM, they were not presented. Log likelihood was increased, by changing the orders of fit from 1 (linear) to 3 (cubic), for genetic and permanent environmental effects. The similar results to the present study were also reported by Luo et al. (2007) on broiler dam line for cumulative egg numbers and by Wolc and Szwaczkowski (2009) on broiler chickens for fertility, who reported a substantial increasing in log likelihood with the increasing the degree of polynomials up to cubic, fitted to the permanent environmental effects. Also, Akbas et al. (2004) found a significant increase in the log likelihood from the linear to the quintic model for body weights in quail. The same patterns of changing in log likelihood have been also suggested for test-day milk yield in dairy cattle (Costa et al., 2008; El Faro et al., 2008; Guo and Schaeffer, 2002; Jamrozik and Schaeffer, 2002; LopezRomero and Carabano, 2003). For the present study, the model 9, with Legendre polynomial of order 3 for the additive genetic and permanent environmental effects
Average egg production
30 25 20 15 10 5 0
0
1
2
3
4 5 Month
6
7
8
9
Fig. 1. Average egg production and its standard deviation for monthly records.
35
Table 1 Order of fit for additive genetic (q2) and permanent environmental (q3) effects, number of parameters (pk), log of maximum likelihood values (log(ML)), Bayesian information criterion (BIC), and Akaike information criterion (AIC). Model
1 2 3 4 5 6 7 8 9
Order of fit q2
q3
1 1 1 2 2 2 3 3 3
1 2 3 1 2 3 1 2 3
pk
log(ML)
BIC
AIC
14 17 21 17 20 24 21 24 28
278552.88 274359.67 273580.88 274145.48 273785.21 273099.88 273236.02 272932.82 272767.63
557270.57 548919.46 547408.98 548491.09 547805.85 546482.28 546719.24 546148.16 545864.88
557133.76 548753.33 547203.77 548324.96 547610.41 546247.76 546514.04 545913.64 545591.27
Table 2 Estimates of genetic ðs2a Þ, permanent environmental ðs2pe Þ, residual ðs2e Þ, and phenotypic ðs2p Þ variances, heritability (h2) and ratio of permanent environment to phenotypic variance (pe2) for monthly egg production without ASM. Traits
s2a
s2pe
s2e
s2p
h2 7SE
pe2 7 SE
M1 M2 M3 M4 M5 M6 M7 M8
13.30 4.95 3.96 4.69 5.70 6.17 5.74 7.98
5.63 16.75 24.90 28.83 30.42 30.80 30.02 30.28
39.03 11.31 11.31 11.31 11.31 11.31 11.31 11.31
57.96 33.02 40.17 44.83 47.43 48.27 47.06 49.56
0.229 7 0.014 0.1507 0.013 0.0997 0.011 0.1057 0.012 0.1207 0.013 0.128 7 0.013 0.122 7 0.013 0.161 7 0.015
0.097 70.013 0.507 70.012 0.620 70.011 0.643 70.011 0.641 70.012 0.638 70.012 0.638 70.012 0.611 70.014
(1) M ¼monthly egg production; (2) SE¼ standard error; (3) ASM¼ age at sexual maturity.
indicated the lowest value on both AIC and BIC criteria, and therefore, was chosen as an appropriate model. Estimates of variance components, direct heritabilities and the ratio of permanent environmental variance to the phenotypic variance, without and with considering ASM as a covariate in the model, are presented in Tables 2 and 3, respectively. Without considering ASM in the model, the estimates of heritability were high at the beginning of the laying period, decreased in the second month, remained relatively stable for the third to seventh periods, and increased again in eighth month of production. By including ASM in the model, estimates of heritabilities for the first and second monthly records decreased by about 38.86% and 51.33%, respectively. The estimate of heritability for the first monthly egg numbers were reported to be high, ranging from 0.30 to 0.61, in the previous studies (Anang et al., 2000, 2002; Nurgiartiningsih et al., 2004; Wolc et al., 2007a; Wolc and Szwaczkowski, 2009). The high heritability estimate for the first egg production records has been attributed to the variation in the age at sexual maturity (Anang et al., 2000) and the rate of lay before peak (Nurgiartiningsih et al., 2004), that is in accordance with the present study, where fitting ASM in the model substantially decreased the magnitude of direct additive genetic, permanent environmental, residual and
36
N. Farzin et al. / Livestock Science 153 (2013) 33–38
phenotypic variances. Also, the pattern of reduction in the heritability of the first month records in our study can be confirmed by the reported results of Luo et al. (2007), who fitted a single-trait random regression model with fitting age at first egg and obtained a very low estimate of heritability (0.06) for the first month’s egg production in a broiler dam line. A similar results have been found by Farzin et al. (2011), who fitted single-trait and multipletrait animal models for monthly egg production in a broiler dam line and reported a reduction of about 55.14% (from 0.428 to 0.192) in the heritability of the first monthly egg records by including ASM in the model. Because of changing in heritability estimates over time, it is possible that different genes influence egg production through laying period. For early stages, the influence of age at sexual maturity is more important, whereas for the last periods, genes related to persistency may have more influence (Wolc and Szwaczkowski, 2009). In a research about random regression for milk production in dairy cows reported that the Legendre polynomials had typical border effects at the beginning and the end of the lactation (increased bias) and showed waves in the middle of the lactation (Druet et al., 2003). The ratios in variance of permanent environment effects to phenotypic variance substantially increased from M1 (0.097) to M3 (0.620), stayed relatively in a constant phase for M5 to M8 (Table 2). The very high estimate of the ratio in variance of permanent environment effects to the phenotypic variance for almost all monthly production records indicated that this effect is an important source of phenotypic variation. Table 3 Estimates of genetic ðs2a Þ, permanent environmental ðs2pe Þ, residual ðs2e Þ, and phenotypic ðs2p Þ variances, heritability (h2) and ratio of permanent environment to phenotypic variance (pe2) for monthly egg production with ASM. Traits
s2a
s2pe
s2e
s2p
h2 7 SE
pe2 7 SE
M1 M2 M3 M4 M5 M6 M7 M8
3.56 1.94 5.46 7.63 8.26 7.97 7.52 11.83
2.18 12.05 25.70 33.26 35.41 35.19 35.13 39.96
19.63 12.60 12.60 12.60 12.61 12.60 12.60 12.60
25.38 26.60 43.76 53.49 56.28 55.76 55.25 64.39
0.140 70.016 0.073 70.011 0.125 70.011 0.143 70.013 0.147 70.015 0.143 70.015 0.136 70.014 0.184 70.014
0.086 7 0.018 0.453 7 0.011 0.587 7 0.010 0.622 7 0.012 0.629 7 0.014 0.631 7 0.014 0.636 7 0.014 0.621 7 0.013
(1) M ¼ monthly egg production; (2) SE ¼standard error; (3) ASM ¼age at sexual maturity.
For the first monthly record, the very low estimate may be attributed to the differences of the hens in starting production. These estimates declined slightly with considering ASM in the model, except for the eighth month records. The estimates of the ratios in variance of permanent environment effects to the phenotypic variance are variable in different studies. For example, Anang et al. (2002) obtained a range of 0.1–0.21 from first to eighth monthly periods for this parameter in laying hens, which is very lower than the present study. These estimates were relatively high in the study of Luo et al. (2007) and Kranis et al. (2007), who reported a range of 0.44–0.77 for the cumulative egg production in broiler dam line and of 0.45–0.70 for the monthly egg production in Turkeys. Genetic and phenotypic correlations among monthly records of egg production, without and with ASM in the model, are presented in Tables 4 and 5, respectively. The estimates between adjacent months were high and decreased as the time interval increased. Without considering ASM in the model, genetic correlations between all records were positive. The first month had a low genetic correlation with middle (M4 and M5) and latter (from M6 to M8) periods, ranging from 0.102 to 0.188. Genetic correlations between other periods were moderate to high, varied from 0.415 (between M2 and M8) to 0.980 (between M5 and M6). A similar pattern of genetic correlations to the present study were also reported by Anang et al. (2000), and Kranis et al. (2007). The corresponding estimates of phenotypic correlations were lower than the genetic correlations, with some exceptions, ranging from 0.101 (between M1 and M8) to 0.741 (between M5 and M6). With fitting ASM as a covariate in the model, genetic correlations between the first monthly records and other periods changed from positive to negative, ranging from 0.095 to 0.619. The standard errors of these estimates were mostly high (above 0.084), indicated that the results should be interpreted with caution when comparing the correlations obtained through not inclusion of ASM from the model. The genetic correlations between second and last months were moderate, ranging from 0.454 to 0.494, whereas these estimations between middle and last periods were generally moderate to high (from 0.656 to 0.979). Wolc et al. (2007a) obtained low genetic correlations between the first two periods of monthly production records with the other months and even negative between the first and the more distant periods in laying hens, using multivariate animal model. Also, negative to positive genetic correlation coefficients, ranged from 0.78 to 0.91 in three different
Table 4 Estimates of genetic (upper triangle) and phenotypic (lower triangle) correlations ( 7standard errors) between monthly egg records without ASM. M1 M1 M2 M3 M4 M5 M6 M7 M8
0.362 7 0.006 0.267 7 0.007 0.2017 0.007 0.153 7 0.007 0.121 7 0.007 0.1057 0.007 0.1017 0.008
M2
M3
M4
M5
M6
M7
M8
0.8387 0.021
0.424 7 0.055 0.832 7 0.020
0.181 70.061 0.601 70.043 0.926 70.011
0.128 7 0.060 0.469 7 0.052 0.878 7 0.018 0.964 7 0.006
0.157 70.059 0.429 70.056 0.732 70.039 0.904 70.017 0.980 70.004
0.188 7 0.059 0.458 7 0.057 0.612 7 0.052 0.833 7 0.006 0.821 7 0.037 0.947 7 0.008
0.102 7 0.057 0.415 7 0.057 0.573 7 0.056 0.546 7 0.055 0.507 7 0.053 0.558 7 0.048 0.781 7 0.027
0.6477 0.004 0.5847 0.004 0.4947 0.005 0.3987 0.006 0.3207 0.006 0.2677 0.007
0.707 7 0.003 0.637 7 0.004 0.552 7 0.005 0.433 7 0.006 0.351 7 0.007
0.730 70.003 0.667 70.004 0.573 70.005 0.428 70.006
(1) M ¼ monthly egg production; (2) ASM¼ age at sexual maturity.
0.741 7 0.003 0.669 7 0.004 0.5057 0.006
0.733 70.003 0.589 70.005
0.691 7 0.004
N. Farzin et al. / Livestock Science 153 (2013) 33–38
37
Table 5 Estimates of genetic (upper triangle) and phenotypic (lower triangle) correlations ( 7 standard errors) between monthly egg records with ASM. M1 M1 M2 M3 M4 M5 M6 M7 M8
0.052 7 0.010 0.048 7 0.009 0.089 7 0.009 0.101 7 0.009 0.088 7 0.009 0.051 7 0.009 0.0067 0.010
M2
M3
M4
M5
M6
M7
M8
0.0957 0.128
0.511 70.100 0.879 70.021
0.572 70.090 0.755 7 0.039 0.966 70.004
0.550 70.086 0.616 70.055 0.853 70.012 0.971 70.003
0.525 70.085 0.494 70.068 0.783 70.024 0.909 70.011 0.979 70.003
0.573 70.089 0.454 70.075 0.729 70.028 0.844 70.024 0.789 70.012 0.955 70.007
0.619 70.084 0.465 70.072 0.656 70.046 0.656 70.042 0.632 70.040 0.663 70.034 0.840 70.016
0.5827 0.004 0.5507 0.005 0.4827 0.006 0.3997 0.006 0.3337 0.007 0.3097 0.007
0.719 70.003 0.612 70.004 0.583 70.004 0.503 70.005 0.412 70.006
0.749 70.003 0.687 70.004 0.593 70.005 0.464 70.006
0.752 70.003 0.643 70.004 0.529 70.006
0.741 70.003 0.610 70.005
0.717 70.003
(1) M ¼monthly egg production; (2) ASM¼ age at sexual maturity.
Table 6 Estimates of permanent environmental correlations ( 7 standard errors) between monthly egg records without (upper triangle) and with (lower triangle) ASM. M1 M1 M2 M3 M4 M5 M6 M7 M8
0.313 7 0.037 0.0867 0.041 0.0357 0.044 0.0937 0.046 0.0577 0.046 0.1207 0.047 0.454 7 0.062
M2
M3
M4
M5
M6
M7
M8
0.9327 0.036
0.830 70.052 0.972 70.003
0.692 70.051 0.890 70.006 0.971 70.001
0.528 7 0.043 0.755 7 0.009 0.842 7 0.003 0.968 7 0.001
0.376 7 0.037 0.596 7 0.011 0.748 7 0.007 0.879 7 0.004 0.969 7 0.001
0.2947 0.034 0.4547 0.012 0.5767 0.009 0.7497 0.006 0.7387 0.004 0.9647 0.001
0.334 70.041 0.364 70.014 0.454 70.012 0.570 70.011 0.695 70.010 0.816 70.008 0.931 70.004
0.9657 0.003 0.8927 0.006 0.7837 0.009 0.6517 0.011 0.5367 0.013 0.4817 0.016
0.977 70.011 0.821 70.004 0.785 70.005 0.711 70.006 0.501 70.012
0.973 70.001 0.890 70.003 0.755 70.006 0.576 70.011
0.969 7 0.001 0.765 7 0.006 0.681 7 0.010
0.959 7 0.001 0.8047 0.007
0.9307 0.003
(1) M ¼ monthly egg production; (2) ASM¼ age at sexual maturity.
lines, were observed in laying hens, based on random regression model (Wolc and Szwaczkowski, 2009). We did not find any comparable estimates of genetic and phenotypic correlations between monthly egg records with and without fitting ASM in the literature. However, Farzin et al. (2010) reported that with considering ASM in the model, the genetic and phenotypic correlations between the first month and cumulative egg production decreased by 64.06% (from 0.651 to 0.234) and 50.61% (from 0.413 to 0.204), respectively. Also, besides the first monthly records of egg production, which indicated moderate genetic and phenotypic correlations with cumulative egg number (0.651 and 0.413, respectively), all other monthly records had high genetic and phenotypic correlations with this trait. The same results were reported by Luo et al. (2007) who found high genetic correlations between monthly and total egg production, except for the first month where sexual maturity and laying peak may play more important roles. The estimates of permanent environmental correlations among monthly records are presented in Table 6. Similar to the genetic and phenotypic correlations, these estimates decreased as the interval between months increased. Except for the first monthly records, permanent environmental correlations between different periods did not change basically with considering ASM in the model, varying from 0.294 to 0.972 and from 0.093 to 0.977, for without and with fitting ASM, respectively.
monthly egg production records in a random regression model resulted in an overestimation of heritabilities for the first and second month of egg productions. With the exception of the first and to some extend second month egg production records, which showed low genetic correlations with other periods, the estimates were moderate to high among all other monthly production. The estimate of heritability of the fourth month (36 weeks to 39 weeks of age) egg production records and its relatively high genetic correlations with all other later ages suggest that it could be the most appropriate period for selection. However, its implications on natural molting and brooding should be taken into account and investigated. This strategy of selection could result to reduction of generation interval, but since the genetic correlations still are not perfect, other strategies, i.e., combination of different part-periods of egg production records (36 weeks to 47 weeks of age), could be used to select for persistence of lay. Conflict of interest All authors hereby declare that there is no conflict of interest that might be construed to influence the results or interpretation of this manuscript.
Acknowledgments 4. Conclusion The findings of the current study indicated that the exclusion of age at sexual maturity from the analysis of
Gratitude is expressed to Arian Broiler Line Center, located in Babolkenar, Iran, for access to the data and to Dr M. Rokouei for helpful comments.
38
N. Farzin et al. / Livestock Science 153 (2013) 33–38
References Akaike, H., 1974. A new look at the statistical model identification. IEEE Trans. Autom. Control 19, 716–723. Akbas, Y., Takma, C., Yaylak, E., 2004. Genetic parameters for quail body weights using a random regression model. S. Afr. J. Anim. Sci. 34 (2), 104–109. Anang, A., Mielenz, N., Schuler, L., 2000. Genetic and phenotypic parameters for monthly egg production in White Leghorn hens. J. Anim. Breeding Genet. 117, 407–415. Anang, A., Mielenz, N., Schuler, L., 2001. Monthly model for genetic evaluation of laying hens I. Fixed regression. Br. Poult. Sci. 42, 191–196. Anang, A., Mielenz, N., Schuler, L., 2002. Monthly model for genetic evaluation of laying hens II. Random regression. Br. Poult. Sci. 43, 384–390. Costa, C.N., Melo, C.M.R., Packer, I.U., Freitas, A.F., Teixeira, N.M., Cobuci, A., 2008. Genetic parameters for test day milk yield of first lactation Holstein cows estimated by random regression using Legendre polynomials. Rev. Bras. Zootecn. 37 (4), 602–608. Druet, T., Jaffre´zic, F., Boichard, D., Ducrocq, V., 2003. Modeling lactation curves and estimation of genetic parameters for first lactation testday records of French Holstein cows. J. Dairy Sci. 86, 2480–2490. El Faro, L., Cardoso, V.L., Albuquerque, L.G., 2008. Variance component estimates applying random regression models for test-day milk yield in Caracu heifers (Bos taurus Artiodactyla, Bovidae). Genet. Mol. Biol. 31 (3), 665–673. Farzin, N., Vaez Torshizi, R., Kashan, N.E.J., Gerami, A., 2010. Estimates of genetic and phenotypic correlations between monthly and cumulative egg productions in a commercial broiler female line. Global Veterinaria 5 (3), 164–167. Farzin, N., Vaez Torshizi, R., Kashan, N.E.J., Gerami, A., 2011. Estimates of genetic parameters for monthly egg production traits in a commercial broiler female line. Ital. J. Anim. Sci. 10 (e12). Guo, Z., Schaeffer, L.R., 2002. Random regression submodel comparison. In: Proceedings of the 7th World Congress on Genetics Applied to Livestock Production, Montpellier, France. Jamrozik, J., Schaeffer, L.R., 2002. Bayesian comparison of random regression models for test-day yields in dairy cattle. In: Proceedings of the 7th World Congress on Genetics Applied to Livestock Production, Montpellier, France. Kranis, A., Su, G., Sorensen, D., Woolliams, J.A., 2007. The application of random regression models in the genetic analysis of monthly egg production in Turkeys and a comparison with alternative longitudinal models. Poult. Sci. 86, 470–475.
Lopez-Romero, P., Carabano, M.J., 2003. Comparing alternative random regression models to analyse first lactation daily milk yield data in Holstein–Friesian cattle. Livest. Prod. Sci. 82, 81–96. Luo, P.T., Yang, R.Q., Yang, N., 2007. Estimation of genetic parameters for cumulative egg numbers in a broiler dam line by using a random regression model. Poult. Sci. 86, 30–36. Meyer, K., 2007. WOMBAT—A program for mixed model analyses by restricted maximum likelihood. Animal Genetics and Breeding Unit, Armidale User Notes, pp. 55. Nurgiartiningsih, V.M., Mielenz, N., Preisinger, R., Schmutz, M., Schueler, L., 2002. Genetic parameters for egg production and egg weight of laying hens housed in single and group cages. Arch. Tierz 5, 501–508 . Nurgiartiningsih, V.M., Mielenz, N., Preisinger, R., Schmutz, M., Schueler, L., 2004. Estimation of genetic parameters based on individual and group mean records in laying hens. Br. Poult. Sci. 5, 604–610. Schaeffer, L.R., Dekkers, J.C.M., 1994. Random regressions in animal models for test-day production in dairy cattle. In: Proceedings of the 5th World Congress on Genetics Applied to Livestock Production, Guelph, Ontario, Canada. 18, pp. 443–446. Schaeffer, L.R., 2004. Application of random regression models in animal breeding. Livest. Prod. Sci. 86, 35–45. Schwarz, G., 1978. Estimating the dimension of a model. Ann. Stat. 6 (2), 461–464. Swalve, H.H., 2000. Theoretical basis and computational methods for different test-day genetic evaluation methods. J. Dairy. Sci. 83, 1115–1124. Szwaczkowski, T., 2003. Use of mixed model methodology in poultry breeding: Estimation of genetic parameters. In: Muir, W.M., Aggrey, S.E. (Eds.), Poultry Genetics, Breeding and Biotechnology, CAB Int., Wallingford, Oxon, UK, pp. 165–202. Unver, Y., Yavuz, A., Oguz, 2004. Effect of Box-Cox Transformation on genetic parameter estimation in layers. Turk. J. Anim. Sci. 28, 249–255. Wolc, A., Lisowski, M., Szwaczkowski, T., 2007a. Heritability of egg production in laying hens under cumulative, multitrait and repeated measurement animal models. Czech. J. Anim. Sci. 52, 254–259. Wolc, A., Lisowski, M., Szwaczkowski, T., 2007b. Genetic evaluation of laying hens under fixed regression animal models. Arch. Tierz., Dummerstorf. 50 (3), 279–287. Wolc, A., Szwaczkowski, T., 2009. Estimation of genetic parameters for monthly egg production in laying hens based on random regression models. J. Appl. Genet. 50, 41–46.
Contents lists available at SciVerse ScienceDirect
Livestock Science journal homepage: www.elsevier.com/locate/livsci
Estimates of genetic parameters for monthly egg production in a commercial female broiler line using random regression models N. Farzin a,n, R. Vaez Torshizi b, A. Gerami c, A. Seraj a a b c
Department of Animal Science, Azadshahr Branch, Islamic Azad University, Azadshahr, Iran Department of Animal Science, Tarbiat Modares University, Tehran, Iran School of Mathematics, Statistics and Computer Science, University of Tehran, Tehran, Iran
a r t i c l e i n f o
abstract
Article history: Received 18 November 2010 Received in revised form 24 January 2013 Accepted 29 January 2013
In the present study, genetic parameters for monthly egg production, collected from weeks 24–55 on 16,200 hens during 9 generations from a pedigreed commercial female broiler line, were estimated using random regression models. With and without age at sexual maturity (ASM) as a covariate, two analyses were fitted to the data for the different models. To identify the best orders of Legendre polynomials, nine different models were compared. On the basis of Bayesian information and Akaike information criteria, a random regression model with second order Legendre polynomial for fixed effect and third order for additive genetic and permanent environmental effects, was chosen as optimal model. Without considering ASM as a covariate, the heritability estimates of monthly egg productions ranged from 0.099 to 0.229. By including ASM in the model, heritabilities for the first and second monthly records decreased by 38.86% (from 0.229 to 0.140) and 51.33% (from 0.150 to 0.073), respectively. Genetic and phenotypic correlations among monthly egg records were high between adjacent periods and decreased as the time interval increased. Genetic correlations between the first monthly records and other periods changed from positive to negative (varying from 0.095 to 0.619) with considering ASM in the model. The effect of age at sexual maturity on the estimates of heritability and genetic correlations for the first month records suggests it is necessary to include ASM in the analysis of egg production to avoid biased estimates. The heritability of the fourth month egg production and its relatively high genetic correlations with all other later ages show that it could be the most appropriate period for selection. & 2013 Elsevier B.V. All rights reserved.
Keywords: Age at sexual maturity Female broiler line Genetic parameter Monthly egg production Random regression model
1. Introduction Egg number is a longitudinal trait that depends on months of production, increasing to a peak and then decreasing over time. The description of this trait is various in different studies. To analyze the egg number,
n Corresponding author. Tel.: þ 98 911 143 6115; fax: þ98 174 673 4601. E-mail addresses: [email protected], [email protected] (N. Farzin), [email protected] (R. Vaez Torshizi), [email protected] (A. Gerami), [email protected] (A. Seraj).
1871-1413/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.livsci.2013.01.015
monthly (Anang et al., 2001, 2002; Farzin et al., 2011; Nurgiartiningsih et al., 2004; Wolc et al., 2007a, 2007b; Wolc and Szwaczkowski, 2009) or cumulative (Farzin et al., 2010; Luo et al., 2007; Nurgiartiningsih et al., 2004) egg records can be used. Selection for improving egg production was usually done based on early part records, generally up to 40 week of age (Luo et al., 2007). So, to design a suitable selection program, we need to estimate the heritabilities and genetic correlations among different part records. The pattern of genetic and phenotypic variances of egg production have been analyzed in literature using single or multiple trait (Anang et al., 2000; Farzin et al., 2011; Nurgiartiningsih et al., 2004),
34
N. Farzin et al. / Livestock Science 153 (2013) 33–38
repeated records (Farzin et al., 2011; Kranis et al., 2007; Wolc et al., 2007a), fixed regression (Anang et al., 2001; Wolc et al., 2007b) and random regression models (Anang et al., 2002; Luo et al., 2007; Wolc and Szwaczkowski, 2009). Using of random regression models has been increased to describe longitudinal traits (Schaeffer and Dekkers, 1994), because of their flexibility and capability to account for time-dependent effects (Swalve, 2000). Anang et al. (2002) suggested that the random regression was the feasible model to estimate genetic parameters of egg numbers on laying chickens, when compared against the multiple-trait model. Also, similar results were indicated in turkeys by Kranis et al. (2007). In spite of several reports on laying hens, few estimates of genetic parameters have been reported for egg production in broiler lines. Moreover, the estimates of heritability for egg numbers varied strongly in various studies and depended on population, time, trait description and models used, ranging from 0.01 to 0.61, in laying hens (Anang et al., 2001; 2002; Nurgiartiningsih et al., 2002, 2004; Szwaczkowski, 2003; Unver et al., 2004; Wolc et al., 2007a) and 0.06 to 0.43, in broiler dam lines (Farzin et al., 2011; Luo et al., 2007). Also, some studies reported a high estimate of heritability for the first month of laying period (Anang et al., 2000, 2002; Kranis et al., 2007; Nurgiartiningsih et al., 2004; Wolc et al., 2007a), which may be resulted from the variation in age at sexual maturity (Anang et al., 2000). Farzin et al. (2011) estimated the genetic parameters of monthly egg production using single-trait, multiple-trait and repeated records animal models and reported that the exclusion of age at sexual maturity from the analysis of monthly egg production, especially for the first records of laying period, resulted to overestimation of (co)variance components and their corresponding parameters. Therefore, the objective of the present study was to estimate genetic parameters for monthly egg production in a female broiler line, using random regression model, and investigate the effects of including the age at sexual maturity in the model. 2. Materials and methods The data were collected from a commercial female broiler line, selected during nine generations (between 2000 and 2008), moderately for body weight and residual feed intake at seven weeks of age and mainly for the total number of chicks produced during 28 to 39 weeks of age, average egg weight at weeks 31 and 33 and age at sexual maturity (ASM) at 20 weeks of age. Body weight was recorded individually for each bird, whereas the residual feed intake was measured from the performance records of their half sibs and full sibs. Also, selection of the birds at 20 weeks of age was carried out based on information of their relatives collected from all previous generations. Chicks were raised on full feed for the first six weeks and then fed restricted amounts daily until 20 weeks. At 20 weeks of age, all selected birds (approximately 134 roosters and 1206–1474 hens per generation) were randomly assigned to floor pens, 9 to 11 hens with 1 rooster per pen. The progeny of each generation were produced from up to four different hatches (weeks 43–50 for the
first six generations and weeks 45–48 for the last three generations) through collecting hatching eggs over two consecutive weeks. For this line, weekly egg productions of 16,200 hens from 1198 sires and 7564 dams were recorded individually in trap-nests. The period of data collection was from 24 to 55 weeks of age (32 weeks). Monthly egg records were generated by summing each 4 continuous weekly eggs (e.g., M1¼sum of the eggs produced from 24 to 27 weeks of age to M8 ¼sum of the eggs produced from 52 to 55 weeks of age). For each weekly record, it was assumed that all missing values were known and equal to zero. Monthly records were described based on the following random regression model using Legendre polynomials as covariates: yikl ¼ GHi þ
q1 X m¼0
bm zklm þ
q2 X m¼0
akm zklm þ
q3 X
pkm zklm þ eikl
m¼0
where yikl is the monthly egg record of kth hen in lth month, GHi is the fixed effect of ith generation-hatch, bm is the mth fixed regression coefficient, akm is the mth random regression coefficient for additive genetic effect, pkm is the mth random regression coefficient for permanent environmental effect, zklm is the covariate of the Legendre polynomial, eikl is the random residual effect, and q1, q2, and q3 are the orders of the Legendre polynomials for the fixed, additive genetic, and permanent environmental effects, respectively. In the repeated measurements, the residual variance can be assumed homogeneous or heterogeneous over time. To improve the accuracy of estimations in the present study, the residual variances were assumed heterogeneous throughout the laying period and divided into eight classes. Although, it was reported that inclusion of heterogeneous residual variances had little influence on the values of genetic parameter estimates (Wolc and Szwaczkowski, 2009). To investigate the effect of ASM on (co)variance components and genetic parameters, the data were reanalyzed by fitting this effect as a covariate in the model. Estimates of variance and covariance components and their respective parameters were carried out by the restricted maximum likelihood method (REML) using WOMBAT program (Meyer, 2007). Based on the phenotypic means pattern of egg productions over months, the Legendre polynomial of order 2 was used for modeling the fixed effect of month on egg production. Various orders of Legendre polynomial were characterized for additive genetic and permanent environmental effects. These orders are often chosen to be the same for simplicity of computing (Schaeffer, 2004), but in the present study, to determine the adequate model, both equal and unequal orders were tested. Nine different random regression models were compared to identify the best orders of fit for additive genetic and permanent environmental effects. Orders higher than 3 were not used, because no change in the patters of genetic variances with increasing polynomial orders was observed. Bayesian information criterion, BIC (Schwarz, 1978) and Akaike information criterion, AIC (Akaike, 1974) were used to
N. Farzin et al. / Livestock Science 153 (2013) 33–38
choose the appropriate random regression model as follows: BIC k ¼ 2logðMLk Þ þ pk logðnÞ where BICk is the Bayesian information criterion of model k, log(MLk) is the log of maximum likelihood value of model k, pk is the number of free parameters in model k; and n is the number of observations that contribute to the likelihood, and AIC k ¼ 2logðMLk Þ þ2pk where AICk is the Akaike information criterion of model k, and the other symbols are denoted as in the BIC equation. The model with smallest values of BIC and AIC was regarded as the optimal model. 3. Results and discussion The average of monthly egg production was increased until a peak in the second month and subsequently decreased till the end of the production period (Fig. 1). Also, the average ASM was 178.8 710.8 days and ranged from 162 days to 226 days. The values of BIC and AIC for several models with different orders of fit without including ASM are shown in Table 1. Since the variation patterns of these criteria were the same for the models with ASM, they were not presented. Log likelihood was increased, by changing the orders of fit from 1 (linear) to 3 (cubic), for genetic and permanent environmental effects. The similar results to the present study were also reported by Luo et al. (2007) on broiler dam line for cumulative egg numbers and by Wolc and Szwaczkowski (2009) on broiler chickens for fertility, who reported a substantial increasing in log likelihood with the increasing the degree of polynomials up to cubic, fitted to the permanent environmental effects. Also, Akbas et al. (2004) found a significant increase in the log likelihood from the linear to the quintic model for body weights in quail. The same patterns of changing in log likelihood have been also suggested for test-day milk yield in dairy cattle (Costa et al., 2008; El Faro et al., 2008; Guo and Schaeffer, 2002; Jamrozik and Schaeffer, 2002; LopezRomero and Carabano, 2003). For the present study, the model 9, with Legendre polynomial of order 3 for the additive genetic and permanent environmental effects
Average egg production
30 25 20 15 10 5 0
0
1
2
3
4 5 Month
6
7
8
9
Fig. 1. Average egg production and its standard deviation for monthly records.
35
Table 1 Order of fit for additive genetic (q2) and permanent environmental (q3) effects, number of parameters (pk), log of maximum likelihood values (log(ML)), Bayesian information criterion (BIC), and Akaike information criterion (AIC). Model
1 2 3 4 5 6 7 8 9
Order of fit q2
q3
1 1 1 2 2 2 3 3 3
1 2 3 1 2 3 1 2 3
pk
log(ML)
BIC
AIC
14 17 21 17 20 24 21 24 28
278552.88 274359.67 273580.88 274145.48 273785.21 273099.88 273236.02 272932.82 272767.63
557270.57 548919.46 547408.98 548491.09 547805.85 546482.28 546719.24 546148.16 545864.88
557133.76 548753.33 547203.77 548324.96 547610.41 546247.76 546514.04 545913.64 545591.27
Table 2 Estimates of genetic ðs2a Þ, permanent environmental ðs2pe Þ, residual ðs2e Þ, and phenotypic ðs2p Þ variances, heritability (h2) and ratio of permanent environment to phenotypic variance (pe2) for monthly egg production without ASM. Traits
s2a
s2pe
s2e
s2p
h2 7SE
pe2 7 SE
M1 M2 M3 M4 M5 M6 M7 M8
13.30 4.95 3.96 4.69 5.70 6.17 5.74 7.98
5.63 16.75 24.90 28.83 30.42 30.80 30.02 30.28
39.03 11.31 11.31 11.31 11.31 11.31 11.31 11.31
57.96 33.02 40.17 44.83 47.43 48.27 47.06 49.56
0.229 7 0.014 0.1507 0.013 0.0997 0.011 0.1057 0.012 0.1207 0.013 0.128 7 0.013 0.122 7 0.013 0.161 7 0.015
0.097 70.013 0.507 70.012 0.620 70.011 0.643 70.011 0.641 70.012 0.638 70.012 0.638 70.012 0.611 70.014
(1) M ¼monthly egg production; (2) SE¼ standard error; (3) ASM¼ age at sexual maturity.
indicated the lowest value on both AIC and BIC criteria, and therefore, was chosen as an appropriate model. Estimates of variance components, direct heritabilities and the ratio of permanent environmental variance to the phenotypic variance, without and with considering ASM as a covariate in the model, are presented in Tables 2 and 3, respectively. Without considering ASM in the model, the estimates of heritability were high at the beginning of the laying period, decreased in the second month, remained relatively stable for the third to seventh periods, and increased again in eighth month of production. By including ASM in the model, estimates of heritabilities for the first and second monthly records decreased by about 38.86% and 51.33%, respectively. The estimate of heritability for the first monthly egg numbers were reported to be high, ranging from 0.30 to 0.61, in the previous studies (Anang et al., 2000, 2002; Nurgiartiningsih et al., 2004; Wolc et al., 2007a; Wolc and Szwaczkowski, 2009). The high heritability estimate for the first egg production records has been attributed to the variation in the age at sexual maturity (Anang et al., 2000) and the rate of lay before peak (Nurgiartiningsih et al., 2004), that is in accordance with the present study, where fitting ASM in the model substantially decreased the magnitude of direct additive genetic, permanent environmental, residual and
36
N. Farzin et al. / Livestock Science 153 (2013) 33–38
phenotypic variances. Also, the pattern of reduction in the heritability of the first month records in our study can be confirmed by the reported results of Luo et al. (2007), who fitted a single-trait random regression model with fitting age at first egg and obtained a very low estimate of heritability (0.06) for the first month’s egg production in a broiler dam line. A similar results have been found by Farzin et al. (2011), who fitted single-trait and multipletrait animal models for monthly egg production in a broiler dam line and reported a reduction of about 55.14% (from 0.428 to 0.192) in the heritability of the first monthly egg records by including ASM in the model. Because of changing in heritability estimates over time, it is possible that different genes influence egg production through laying period. For early stages, the influence of age at sexual maturity is more important, whereas for the last periods, genes related to persistency may have more influence (Wolc and Szwaczkowski, 2009). In a research about random regression for milk production in dairy cows reported that the Legendre polynomials had typical border effects at the beginning and the end of the lactation (increased bias) and showed waves in the middle of the lactation (Druet et al., 2003). The ratios in variance of permanent environment effects to phenotypic variance substantially increased from M1 (0.097) to M3 (0.620), stayed relatively in a constant phase for M5 to M8 (Table 2). The very high estimate of the ratio in variance of permanent environment effects to the phenotypic variance for almost all monthly production records indicated that this effect is an important source of phenotypic variation. Table 3 Estimates of genetic ðs2a Þ, permanent environmental ðs2pe Þ, residual ðs2e Þ, and phenotypic ðs2p Þ variances, heritability (h2) and ratio of permanent environment to phenotypic variance (pe2) for monthly egg production with ASM. Traits
s2a
s2pe
s2e
s2p
h2 7 SE
pe2 7 SE
M1 M2 M3 M4 M5 M6 M7 M8
3.56 1.94 5.46 7.63 8.26 7.97 7.52 11.83
2.18 12.05 25.70 33.26 35.41 35.19 35.13 39.96
19.63 12.60 12.60 12.60 12.61 12.60 12.60 12.60
25.38 26.60 43.76 53.49 56.28 55.76 55.25 64.39
0.140 70.016 0.073 70.011 0.125 70.011 0.143 70.013 0.147 70.015 0.143 70.015 0.136 70.014 0.184 70.014
0.086 7 0.018 0.453 7 0.011 0.587 7 0.010 0.622 7 0.012 0.629 7 0.014 0.631 7 0.014 0.636 7 0.014 0.621 7 0.013
(1) M ¼ monthly egg production; (2) SE ¼standard error; (3) ASM ¼age at sexual maturity.
For the first monthly record, the very low estimate may be attributed to the differences of the hens in starting production. These estimates declined slightly with considering ASM in the model, except for the eighth month records. The estimates of the ratios in variance of permanent environment effects to the phenotypic variance are variable in different studies. For example, Anang et al. (2002) obtained a range of 0.1–0.21 from first to eighth monthly periods for this parameter in laying hens, which is very lower than the present study. These estimates were relatively high in the study of Luo et al. (2007) and Kranis et al. (2007), who reported a range of 0.44–0.77 for the cumulative egg production in broiler dam line and of 0.45–0.70 for the monthly egg production in Turkeys. Genetic and phenotypic correlations among monthly records of egg production, without and with ASM in the model, are presented in Tables 4 and 5, respectively. The estimates between adjacent months were high and decreased as the time interval increased. Without considering ASM in the model, genetic correlations between all records were positive. The first month had a low genetic correlation with middle (M4 and M5) and latter (from M6 to M8) periods, ranging from 0.102 to 0.188. Genetic correlations between other periods were moderate to high, varied from 0.415 (between M2 and M8) to 0.980 (between M5 and M6). A similar pattern of genetic correlations to the present study were also reported by Anang et al. (2000), and Kranis et al. (2007). The corresponding estimates of phenotypic correlations were lower than the genetic correlations, with some exceptions, ranging from 0.101 (between M1 and M8) to 0.741 (between M5 and M6). With fitting ASM as a covariate in the model, genetic correlations between the first monthly records and other periods changed from positive to negative, ranging from 0.095 to 0.619. The standard errors of these estimates were mostly high (above 0.084), indicated that the results should be interpreted with caution when comparing the correlations obtained through not inclusion of ASM from the model. The genetic correlations between second and last months were moderate, ranging from 0.454 to 0.494, whereas these estimations between middle and last periods were generally moderate to high (from 0.656 to 0.979). Wolc et al. (2007a) obtained low genetic correlations between the first two periods of monthly production records with the other months and even negative between the first and the more distant periods in laying hens, using multivariate animal model. Also, negative to positive genetic correlation coefficients, ranged from 0.78 to 0.91 in three different
Table 4 Estimates of genetic (upper triangle) and phenotypic (lower triangle) correlations ( 7standard errors) between monthly egg records without ASM. M1 M1 M2 M3 M4 M5 M6 M7 M8
0.362 7 0.006 0.267 7 0.007 0.2017 0.007 0.153 7 0.007 0.121 7 0.007 0.1057 0.007 0.1017 0.008
M2
M3
M4
M5
M6
M7
M8
0.8387 0.021
0.424 7 0.055 0.832 7 0.020
0.181 70.061 0.601 70.043 0.926 70.011
0.128 7 0.060 0.469 7 0.052 0.878 7 0.018 0.964 7 0.006
0.157 70.059 0.429 70.056 0.732 70.039 0.904 70.017 0.980 70.004
0.188 7 0.059 0.458 7 0.057 0.612 7 0.052 0.833 7 0.006 0.821 7 0.037 0.947 7 0.008
0.102 7 0.057 0.415 7 0.057 0.573 7 0.056 0.546 7 0.055 0.507 7 0.053 0.558 7 0.048 0.781 7 0.027
0.6477 0.004 0.5847 0.004 0.4947 0.005 0.3987 0.006 0.3207 0.006 0.2677 0.007
0.707 7 0.003 0.637 7 0.004 0.552 7 0.005 0.433 7 0.006 0.351 7 0.007
0.730 70.003 0.667 70.004 0.573 70.005 0.428 70.006
(1) M ¼ monthly egg production; (2) ASM¼ age at sexual maturity.
0.741 7 0.003 0.669 7 0.004 0.5057 0.006
0.733 70.003 0.589 70.005
0.691 7 0.004
N. Farzin et al. / Livestock Science 153 (2013) 33–38
37
Table 5 Estimates of genetic (upper triangle) and phenotypic (lower triangle) correlations ( 7 standard errors) between monthly egg records with ASM. M1 M1 M2 M3 M4 M5 M6 M7 M8
0.052 7 0.010 0.048 7 0.009 0.089 7 0.009 0.101 7 0.009 0.088 7 0.009 0.051 7 0.009 0.0067 0.010
M2
M3
M4
M5
M6
M7
M8
0.0957 0.128
0.511 70.100 0.879 70.021
0.572 70.090 0.755 7 0.039 0.966 70.004
0.550 70.086 0.616 70.055 0.853 70.012 0.971 70.003
0.525 70.085 0.494 70.068 0.783 70.024 0.909 70.011 0.979 70.003
0.573 70.089 0.454 70.075 0.729 70.028 0.844 70.024 0.789 70.012 0.955 70.007
0.619 70.084 0.465 70.072 0.656 70.046 0.656 70.042 0.632 70.040 0.663 70.034 0.840 70.016
0.5827 0.004 0.5507 0.005 0.4827 0.006 0.3997 0.006 0.3337 0.007 0.3097 0.007
0.719 70.003 0.612 70.004 0.583 70.004 0.503 70.005 0.412 70.006
0.749 70.003 0.687 70.004 0.593 70.005 0.464 70.006
0.752 70.003 0.643 70.004 0.529 70.006
0.741 70.003 0.610 70.005
0.717 70.003
(1) M ¼monthly egg production; (2) ASM¼ age at sexual maturity.
Table 6 Estimates of permanent environmental correlations ( 7 standard errors) between monthly egg records without (upper triangle) and with (lower triangle) ASM. M1 M1 M2 M3 M4 M5 M6 M7 M8
0.313 7 0.037 0.0867 0.041 0.0357 0.044 0.0937 0.046 0.0577 0.046 0.1207 0.047 0.454 7 0.062
M2
M3
M4
M5
M6
M7
M8
0.9327 0.036
0.830 70.052 0.972 70.003
0.692 70.051 0.890 70.006 0.971 70.001
0.528 7 0.043 0.755 7 0.009 0.842 7 0.003 0.968 7 0.001
0.376 7 0.037 0.596 7 0.011 0.748 7 0.007 0.879 7 0.004 0.969 7 0.001
0.2947 0.034 0.4547 0.012 0.5767 0.009 0.7497 0.006 0.7387 0.004 0.9647 0.001
0.334 70.041 0.364 70.014 0.454 70.012 0.570 70.011 0.695 70.010 0.816 70.008 0.931 70.004
0.9657 0.003 0.8927 0.006 0.7837 0.009 0.6517 0.011 0.5367 0.013 0.4817 0.016
0.977 70.011 0.821 70.004 0.785 70.005 0.711 70.006 0.501 70.012
0.973 70.001 0.890 70.003 0.755 70.006 0.576 70.011
0.969 7 0.001 0.765 7 0.006 0.681 7 0.010
0.959 7 0.001 0.8047 0.007
0.9307 0.003
(1) M ¼ monthly egg production; (2) ASM¼ age at sexual maturity.
lines, were observed in laying hens, based on random regression model (Wolc and Szwaczkowski, 2009). We did not find any comparable estimates of genetic and phenotypic correlations between monthly egg records with and without fitting ASM in the literature. However, Farzin et al. (2010) reported that with considering ASM in the model, the genetic and phenotypic correlations between the first month and cumulative egg production decreased by 64.06% (from 0.651 to 0.234) and 50.61% (from 0.413 to 0.204), respectively. Also, besides the first monthly records of egg production, which indicated moderate genetic and phenotypic correlations with cumulative egg number (0.651 and 0.413, respectively), all other monthly records had high genetic and phenotypic correlations with this trait. The same results were reported by Luo et al. (2007) who found high genetic correlations between monthly and total egg production, except for the first month where sexual maturity and laying peak may play more important roles. The estimates of permanent environmental correlations among monthly records are presented in Table 6. Similar to the genetic and phenotypic correlations, these estimates decreased as the interval between months increased. Except for the first monthly records, permanent environmental correlations between different periods did not change basically with considering ASM in the model, varying from 0.294 to 0.972 and from 0.093 to 0.977, for without and with fitting ASM, respectively.
monthly egg production records in a random regression model resulted in an overestimation of heritabilities for the first and second month of egg productions. With the exception of the first and to some extend second month egg production records, which showed low genetic correlations with other periods, the estimates were moderate to high among all other monthly production. The estimate of heritability of the fourth month (36 weeks to 39 weeks of age) egg production records and its relatively high genetic correlations with all other later ages suggest that it could be the most appropriate period for selection. However, its implications on natural molting and brooding should be taken into account and investigated. This strategy of selection could result to reduction of generation interval, but since the genetic correlations still are not perfect, other strategies, i.e., combination of different part-periods of egg production records (36 weeks to 47 weeks of age), could be used to select for persistence of lay. Conflict of interest All authors hereby declare that there is no conflict of interest that might be construed to influence the results or interpretation of this manuscript.
Acknowledgments 4. Conclusion The findings of the current study indicated that the exclusion of age at sexual maturity from the analysis of
Gratitude is expressed to Arian Broiler Line Center, located in Babolkenar, Iran, for access to the data and to Dr M. Rokouei for helpful comments.
38
N. Farzin et al. / Livestock Science 153 (2013) 33–38
References Akaike, H., 1974. A new look at the statistical model identification. IEEE Trans. Autom. Control 19, 716–723. Akbas, Y., Takma, C., Yaylak, E., 2004. Genetic parameters for quail body weights using a random regression model. S. Afr. J. Anim. Sci. 34 (2), 104–109. Anang, A., Mielenz, N., Schuler, L., 2000. Genetic and phenotypic parameters for monthly egg production in White Leghorn hens. J. Anim. Breeding Genet. 117, 407–415. Anang, A., Mielenz, N., Schuler, L., 2001. Monthly model for genetic evaluation of laying hens I. Fixed regression. Br. Poult. Sci. 42, 191–196. Anang, A., Mielenz, N., Schuler, L., 2002. Monthly model for genetic evaluation of laying hens II. Random regression. Br. Poult. Sci. 43, 384–390. Costa, C.N., Melo, C.M.R., Packer, I.U., Freitas, A.F., Teixeira, N.M., Cobuci, A., 2008. Genetic parameters for test day milk yield of first lactation Holstein cows estimated by random regression using Legendre polynomials. Rev. Bras. Zootecn. 37 (4), 602–608. Druet, T., Jaffre´zic, F., Boichard, D., Ducrocq, V., 2003. Modeling lactation curves and estimation of genetic parameters for first lactation testday records of French Holstein cows. J. Dairy Sci. 86, 2480–2490. El Faro, L., Cardoso, V.L., Albuquerque, L.G., 2008. Variance component estimates applying random regression models for test-day milk yield in Caracu heifers (Bos taurus Artiodactyla, Bovidae). Genet. Mol. Biol. 31 (3), 665–673. Farzin, N., Vaez Torshizi, R., Kashan, N.E.J., Gerami, A., 2010. Estimates of genetic and phenotypic correlations between monthly and cumulative egg productions in a commercial broiler female line. Global Veterinaria 5 (3), 164–167. Farzin, N., Vaez Torshizi, R., Kashan, N.E.J., Gerami, A., 2011. Estimates of genetic parameters for monthly egg production traits in a commercial broiler female line. Ital. J. Anim. Sci. 10 (e12). Guo, Z., Schaeffer, L.R., 2002. Random regression submodel comparison. In: Proceedings of the 7th World Congress on Genetics Applied to Livestock Production, Montpellier, France. Jamrozik, J., Schaeffer, L.R., 2002. Bayesian comparison of random regression models for test-day yields in dairy cattle. In: Proceedings of the 7th World Congress on Genetics Applied to Livestock Production, Montpellier, France. Kranis, A., Su, G., Sorensen, D., Woolliams, J.A., 2007. The application of random regression models in the genetic analysis of monthly egg production in Turkeys and a comparison with alternative longitudinal models. Poult. Sci. 86, 470–475.
Lopez-Romero, P., Carabano, M.J., 2003. Comparing alternative random regression models to analyse first lactation daily milk yield data in Holstein–Friesian cattle. Livest. Prod. Sci. 82, 81–96. Luo, P.T., Yang, R.Q., Yang, N., 2007. Estimation of genetic parameters for cumulative egg numbers in a broiler dam line by using a random regression model. Poult. Sci. 86, 30–36. Meyer, K., 2007. WOMBAT—A program for mixed model analyses by restricted maximum likelihood. Animal Genetics and Breeding Unit, Armidale User Notes, pp. 55. Nurgiartiningsih, V.M., Mielenz, N., Preisinger, R., Schmutz, M., Schueler, L., 2002. Genetic parameters for egg production and egg weight of laying hens housed in single and group cages. Arch. Tierz 5, 501–508 . Nurgiartiningsih, V.M., Mielenz, N., Preisinger, R., Schmutz, M., Schueler, L., 2004. Estimation of genetic parameters based on individual and group mean records in laying hens. Br. Poult. Sci. 5, 604–610. Schaeffer, L.R., Dekkers, J.C.M., 1994. Random regressions in animal models for test-day production in dairy cattle. In: Proceedings of the 5th World Congress on Genetics Applied to Livestock Production, Guelph, Ontario, Canada. 18, pp. 443–446. Schaeffer, L.R., 2004. Application of random regression models in animal breeding. Livest. Prod. Sci. 86, 35–45. Schwarz, G., 1978. Estimating the dimension of a model. Ann. Stat. 6 (2), 461–464. Swalve, H.H., 2000. Theoretical basis and computational methods for different test-day genetic evaluation methods. J. Dairy. Sci. 83, 1115–1124. Szwaczkowski, T., 2003. Use of mixed model methodology in poultry breeding: Estimation of genetic parameters. In: Muir, W.M., Aggrey, S.E. (Eds.), Poultry Genetics, Breeding and Biotechnology, CAB Int., Wallingford, Oxon, UK, pp. 165–202. Unver, Y., Yavuz, A., Oguz, 2004. Effect of Box-Cox Transformation on genetic parameter estimation in layers. Turk. J. Anim. Sci. 28, 249–255. Wolc, A., Lisowski, M., Szwaczkowski, T., 2007a. Heritability of egg production in laying hens under cumulative, multitrait and repeated measurement animal models. Czech. J. Anim. Sci. 52, 254–259. Wolc, A., Lisowski, M., Szwaczkowski, T., 2007b. Genetic evaluation of laying hens under fixed regression animal models. Arch. Tierz., Dummerstorf. 50 (3), 279–287. Wolc, A., Szwaczkowski, T., 2009. Estimation of genetic parameters for monthly egg production in laying hens based on random regression models. J. Appl. Genet. 50, 41–46.