Livestock Research for Rural Development 26 (5) 2014 Guide for preparation of papers LRRD Newsletter

Citation of this paper

Genetic study of egg production trait of Azerbaijan's native chickens by multi-trait and random regression models

M Aghazadeh Bokat, S Alijani, H Janmohammadi, S A Raft, A Alizadeh* and P Bostanchi*

Department of Animal Science, University of Tabriz, Tabriz, Iran
sad-ali@tabrizu.ac.ir
* Breeding center of Azerbaijan's native chickens, Uremia, Iran

Abstract

In this research egg production records of 2-week periods of Azerbaijan's native chickens belonging to three generations (12, 14 and 15) were used. These records included a total of 22,955 records from 5,158 hens. Two different analyses as multi-trait model (MTM) and random regression model (RRM) were applied. The variance components were estimated using Gibbs sampling procedure. According to the Deviation Information Criterion (DIC), a RRM with Legendre polynomial of one order for fixed effect of periods, two and four orders for additive genetic and permanent environmental effects, respectively, was chosen as the optimal model.

The heritability estimates by MTM ranged from 0.0344 to 0.233 and from 0.0481 to 0.379 by RRM. Genetic and phenotypic correlations were high between adjacent periods and decreased as the time interval increased. Genetic correlation between the first two periods and other periods changed from positive to negative. Methodology based on random regression animal models could be recommended for genetic evaluation of egg production trait of Azerbaijan's native chickens.

Key words: egg production, genetic correlation, Legendre polynomial, native chickens, random regression model


Introduction

Egg production is a trait that is expressed over a long trajectory of time and as such undergoes both genetic and environmental effects. In recent years genetic evaluation of laying hens on egg production was usually based on single measurement-cumulative production. Such an approach was recommended due to simplicity and relatively small computing demands (Walk et al 2007a). In other approaches, multi-trait analysis (each record treated as a separate trait, correlated to other records), fixed regression (with average production curve included as a fixed effect) and random regression models (RRM) were suggested for analysis of longitudinal data (Anang et al 2001, 2002; Kranis et al 2007). Using of RRM has been increased to describe longitudinal traits (Schaefer and Dekkers 1994), because regression models enable a higher accuracy of selection, as they make use of information on the course of traits, and make it possible to change the course of a trait through selection (Huisman et al 2002). Anang et al (2002) suggested that the RRM was the feasible model to estimate genetic parameters of egg numbers on laying chickens, when compared against the Multi Trait Model (MTM). Also, similar results were indicated in turkeys by Kranis et al (2007).

Iranian indigenous chickens are meat-egg type or dual purpose. Growth rate and egg production under conventional rearing system in villages are very low. During the past several decades importation of exotic breeds have increased risk of extension of these chickens (Ghazikhani-shad et al 2007). The initial endeavor for breeding and extension of Iranian native fowls, in frame of national project, has been relatively successful. This project started back in 1984 in several places. All in all, six breeding stations in different regions of Iran (Mazandaran, Fars, Esfahan, West Azerbaijan, Yazd, and Khorasan) have been established. Improvement in the productivity of indigenous breeds requires attention to nutritional, breeding, health, and management aspects. From the breeding point of view, genetic improvement through selection within local chicken seems to be an attractive option and estimation of genetic parameters is the first and important step in any program.

Therefore, the objective of this study was to estimate heritability and genetic correlations of 2-week periods of egg production trait in Azerbaijan's native chickens by random regression with different order of Legendre polynomial and multi-trait models.


Material and methods

Data

Data were collected from 5,158 birds comprising 15 consecutive fully-pedigreed generations of Azerbaijan's native chickens from Breeding and Propagation Station of west Azerbaijan. The pedigree information is shown in table 1. The egg production was evaluated individually by monitoring the number of eggs obtained from 20 to 31 weeks of age for each hen. In all, 6 periods (P1 to P6) were defined, with each period consisting of 2- weeks.

Table 1. Pedigree information of recorded animals
Information Number
Animals in total 48825
Inbred animals 40761
Sires in total 1313
Dams in total 6286
Animals with progeny 7599
Animals without progeny 41226
Base animals 396
Non base animals 48429

Two animal models were used to estimate genetic parameters. For MTM the following model was used:

Where: yik is a number of eggs per period within i-th generation-hatch for the k-th hen, is population mean, ghi is the fixed effect of i-th generation-hatch, ak is a random additive genetic effect of k-th animal, eik is random residual effect. The RRM using Legendre polynomials as covariates was as follow:

where yikl is a number of eggs per period within i-th generation-hatch of k-th hen in l-th period, ghi is the fixed effect of i-th generation-hatch, bm is the m-th fixed regression coefficient, akm is the m-th random regression coefficient for additive genetic effect, pkm is the m-th random regression coefficient for permanent environmental effect, zklm is a covariate- a value of Legendre polynomial on period, eikl is random residual effect, n1, n2, n3 are numbers of covariates, dependent on the order of Legendre polynomials. The range of order varied from one to four for n2, n3 and from one to three for n1. A total of 48 RRM were compared. The orders of polynomials are given in Table 2. The following (co)variance structure was assumed for random effects of model:

Where G was the n order covariance matrix of random genetic regression coefficient, assigned the same for all individuals, A was the additive genetic relationship matrix among the individuals in the pedigree, I was an identity matrix, P was n order covariance matrix of random permanent environment coefficient and σ2e was residual variance. The first five Legendre polynomial functions (Krikpatric et al 1990) were given as:

Where w is a standardized unit of period to lie and ranged between -1 and +1. Estimated (co)variance component were obtained by Gibbs2f90 (Misztal 2002) program based on Bayesian method. Therefore 100000 cycles of Gibbs sampling were used. Based on initial trials, a conservative burn-in period of 2000 rounds and thinning interval of 10 rounds were used for all estimated parameters. The additive genetic and permanent environmental (co)variance matrices for each period were calculated as ΦGiΦ' and ΦPiΦ'. Where, Φ is a 65 matrix with polynomial coefficient, Gi and Pi are (co)variance matrices for additive genetic and permanent environmental random regression coefficient in ith period, respectively. Diagonal of above (co) variances matrices were additive genetic variances (σ2a (i)) and permanent environmental variances (σ2pe (i)) for 1th to 6th period, respectively. For the RRM, heritability for ith period (h2i) was calculated as:

Genetic correlations between periods were calculated as:

Where covg(i,j) is the genetic covariance for egg production in ith and jth period and varg(i,i) and varg(j,j) are additive genetic variance for ith and jth period, respectively. The Deviation Information Criterion (DIC) was used to evaluate adequacy of the models, using following formula:

Where PD is effective number of parameters of the model anda measure of how well the model fits the data; the larger this is, the worse the fit. The mo- del with the minimum DIC is regarded as the optimal model.


Results and Discussion

Means and standard deviations

The means of egg number production in P1 to P6 are presented in Table 2. The average of egg production in P1 was low, then increased in P2 and subsequently decreased at the end of the production period.

Table 2. Average egg production, the respective standard deviation (SD), and coefficient of variation (CV) for P1 to P6
Period (P) Weeks of age Average (egg number) SD CV (%)
1 20 to 21 7.69 2.70 35.2
2 22 to 23 8.86 2.36 26.7
3 24 to 25 8.64 2.62 31.0
4 26 to 27 7.44 2.95 39.7
5 28 to 29 5.96 3.30 55.4
6 30 to 31 4.26 3.00 70.7

Anang et al (2002), observed similar variations for initial and final production periods. In the initial period, the phenotypic variation can be attributed to differences in the age of sexual maturity, where as the laying persistence variation, natural change might influence the phenotypic values of the final periods.

Optimal RRM

The values of DIC for the examined models are listed in table 3. These criterions indicated that model 8, including one order of Legendre polynomial in the fixed part, two and four orders for additive genetic and permanent environmental effects is the best model and therefore chosen as an optimal model for estimation of genetic parameters of egg production in 2-week periods of Azerbaijan's native chickens.

Table 3. The order of Legendre polynomials used to fixed (n1), additive genetic (n2) and permanent environmental (n3) effects, number of parameters (Pk) and Deviation Information Criterion (DIC)
Model n1 n2 n3 PK DIC
1 1 1 1 7 109972
2 1 1 2 10 107054
3 1 1 3 14 105879
4 1 1 4 19 104427
5 1 2 1 10 105505
6 1 2 2 13 105435
7 1 2 3 17 104687
8 1 2 4 22 103269
9 1 3 1 14 105038
10 1 3 2 17 104866
11 1 3 3 21 104659
12 1 3 4 26 103379
13 1 4 1 19 104161
14 1 4 2 22 103973
15 1 4 3 26 103649
16 1 4 4 31 103280
17 2 1 1 7 105974
18 2 1 2 10 105304
19 2 1 3 14 104556
20 2 1 4 19 103334
21 2 2 1 10 105577
22 2 2 2 13 105291
23 2 2 3 17 104547
24 2 2 4 22 103351
25 2 3 1 14 105180
26 2 3 2 17 104750
27 2 3 3 21 104559
28 2 3 4 26 103354
29 2 4 1 19 104283
30 2 4 2 22 103975
31 2 4 3 26 103641
32 2 4 4 31 103360
33 3 1 1 7 105880
34 3 1 2 10 105249
35 3 1 3 14 104569
36 3 1 4 19 103406
37 3 2 1 10 105500
38 3 2 2 13 105246
39 3 2 3 17 104560
40 3 2 4 22 103347
41 3 3 1 14 105172
42 3 3 2 17 104745
43 3 3 3 21 104551
44 3 3 4 26 103354
45 3 4 1 19 104228
46 3 4 2 22 103934
47 3 4 3 26 103661
48 3 4 4 31 103325
Variance Components and heritabilities

Estimation of variance components and heritability of different periods of egg production using MTM and RRM were presented in Table 4 and 5, respectively. The means of posterior distribution of additive genetic variance for different periods of egg production are shown in Figure 2 (a) to (f). The estimated heritabilities ranged from 0.0344 (from 22 to 23 weeks of age) to 0.233 (from 30 to 31 weeks of age) by MTM and from 0.0481 (from 22 to 23 weeks of age) to 0.379 (from 30 to 31 weeks of age) by RRM. The estimated genetic variances and heritabilities were high at the beginning of the laying period, decreased in the second period and then increased to the end of egg production periods. An increase in heritability of egg production with age was also reported by Luo et al (2007) and Walk et al (2011). Changes of heritability over time may result from activation of different genes during the production cycle. Early stages of production are under the influence of sexual maturity. The high heritability in sixth period may be due to few records for this period. Kranis et al (2007), estimated heritability in turkeys from 0.05 to 0.14 (from first month to fourth month) by MTM and from 0.05 to 0.13 (from first month to fourth month) using RRM. Walk et al (2007b) for three lines A22, A88, K66 estimated heritability of monthly egg production by RRM higher than MTM. In the present study, estimated heritabilities by RRM were higher than MTM. These results show that RRM can estimate heritability better than MTM.

Table 4. Estimated values of genetic (σ2a), residual (σ2e), and phenotypic (σ2p) variances, heritability (h2) for 2-week periods egg production by MTM
Periods σ2a σ2e σ2p h2
1 0.260 6.35 6.61 0.0393
2 0.192 5.39 5.58 0.0344
3 0.254 6.25 6.50 0.0391
4 0.721 7.69 8.41 0.0857
5 1.40 8.53 9.93 0.141
6 2.15 7.05 9.20 0.233

The reported estimates of heritability for egg number varied from 0.11 to 0.53 (Francesh et al 1997; Nuriartiningsih et al 2002, 2004; Szwaczkowski 2003). Luo et al (2007) showed that egg production heritability estimates varied from 0.16 (from 26 to 27 week of age) to 0.54 (62 week of age). However, when the same authors grouped the studied periods ( from 26 to 65 weeks of age) in months, the heritability estimates varied from 0.03 (tenth month) to 0.21 (second production month). Venturini et al,(2012) reported egg production heritability from .04 (from 20 to 22 weeks of age) to 0.14 (from 23 to 25 weeks of age) by RRM.

Table 5. Estimated values of genetic (σ2a), permanent environmental (σ2pe), residual (σ2e), and phenotypic (σ2p) variances, heritability (h2) for 2week periods egg production by RRM
Periods σ2a σ2pe σ2e σ2p h2
1 0.499 2.80 3.60 6.90 0.0723
2 0.288 2.10 3.60 5.99 0.0481
3 0.466 2.92 3.60 6.99 0.0667
4 0.504 3.69 3.60 7.79 0.0647
5 1.10 5.79 3.60 10.5 0.105
6 4.15 2.71 3.60 10.5 0.379

The estimated heritabilities in the present study was lower than the other studies, but were similar to estimated heritabilities by Kranis et al (2007) and Venturini al., (2012). Also the observed gradual increase in heritability with age does not agree with the finding of the U-shaped pattern of heritability described in previous studies (Anang et al 2002; Walk and Szwaczkowaczk 2009). The difference may be due to different environment, breed, population size and long of periods. In this research the aim was not to develop a method for selection of chicken based on two weeks egg production, but to derive some information on egg production in 12 weeks period. For selection of chickens based on egg production, we will extend our study to longer periods and probably full periods in future.

Figure 1. Estimates of heritability for 2week laying periods by MTM and RRM
Figure 2. The posterior distribution of additive genetic variance for P1 to P6 (a to f)
Genetic correlations

The genetic correlations between 2-week periods by MTM and RRM represented in table 6 and 7, respectively. The genetic correlations ranged from -0.596 to 0.741 under MTM and from -0.618 to 0.887 under RRM. Most periods were negatively correlated with egg production of the 1th period.

Table 6. Estimated values of genetic (upper triangle) and phenotypic (lower triangle) correlation between 2-week periods egg production using MTM
Periods 1 2 3 4 5 6
1 0.470 0.183 -0.160 -0.277 -0.596
2 0.355 0.441 0.284 0.0262 -0.00653
3 0.115 0.359 0.682 0.640 0.379
4 -0.0110 0.171 0.371 0.741 0.459
5 -0.0729 0.0488 0.180 0.319 0.692
6 -0.169 0.0398 0.0986 0.238 0.285  

Correlation coefficients under RRM were lower than under the MTM and often negative. The difference may be due to different model components implemented in models: the use of permanent environment variance in random regression, which allowed the genetic term to better accommodate to the genetic changes in the variance of egg production along the trajectory.

Table 7. Estimated values of genetic (upper triangle) and phenotypic (lower triangle) correlation between 2-week periods egg production using a RRM
Periods 1 2 3 4 5 6
1 0.301 -0.256 -0.491 -0.336 -0.0951
2 0.445 0.801 0.360 -0.331 -0.618
3 0.136 0.427 0.798 0.0828 -0.372
4 -0.0650 0.200 0.499 0.655 0.235
5 -0.141 -0.00273 0.124 0.510 0.887
6 -0.147 0.0000378 0.0414 0.177 0.431  

Kranis et al (2007) estimated genetic correlation between -0.09 and 0.95 under a RRM and between 0.05 and 0.88 under MTM. Walk et al (2007) estimated genetic correlation between -0.78 to 0.94 (A22 line), -0.53 to 0.91 (A88 line), -0.54 to 0.94 (K66 line) by RRM and between -0.17 to 0.94 (A22 line), 0.14 to 0.99 (A88 line), -0.14 to 0.93 (K66 Line) under MTM. In this research, Correlations were higher for consecutive periods, whereas negative estimates were mostly obtained for the first two periods with other periods. Low correlations between the first and later periods were confirmed on phenotypic level. This may suggest a different genetic background of initial egg production. Such results were also observed for egg production in laying hens under MTM and RRM by Anang et al (2000, 2002), Szwaczkoski (2003), Luo et al (2007) and Walk et al (2007b).


Conclusion


Acknowledgments

We thank breeding and propagation station of Azerbaijan's native chicken and agriculture organization for data collection.


References

Anang A, Mielenz N and Schuler L 2000 Genetic and phenotypic parameters for monthly egg production in white leghorn hens. Journal of Animal Breeding Genetics.117: 407 -415

Anang A, Mielenz N, Schuler L and Preisinger R 2001 The use of monthly egg production record for genetic evaluation of laying hens. Journal of Ilmu Ternak dan Veteriner 6: 230-234

Anang A, MielenzI N and Schuler L 2002 Monthly model forgenetic evaluation of laying hens II.regression Random. Journal of British Poultry Science 43: 348-390

Francesh A, Estany J, Alfons L and Iglesias M 1997 Genetic parameter for egg number, egg weight and egg shell color in three Catalan breed. Journal of British Poultry Science 76: 1627–1631

Ghazikhani-Shad A, Nejati-Javaremi A and Mehrabani-Yeghaneh H 2007 Animal models estimation of genetic parameters for most important economic traits in Iranian native fowls. Journal Pakistan Biology Science 10: 2787–2789

Huisman A, Kranis E, Van A and Arendonk J 2002 Genetic parameters for various random regression models to describe the weight data of pigs. Journal of Animal Science 80: 575-582

Kranis A, Su G, Sorensen D and Wooliams 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. Journal of British Poultry Science 86: 470–475

Krikpatric M, Lofsvold D and Bulmer M 1990 Analysis of the inheritance, selection and Evaluation of growth trajectories. Genetics 124: 979-993

Luo P T, Yang R.Q and Yang N 2007 Estimation of genetic parameters for cumulative egg numbers in a broiler dam line by random regression model. Journal of British Poultry Science 86: 30-36

Misztal I 2002 GIBBS2F90 Manual Disponsible en: http://nce. Ads.uga. edu /igancy / num pub/ blupf90/docs/gibbsf90.Pdf

Nurgiartiningsih V M, Mielenz N, Prreisnger R, Schmutz M and Schueler L 2002 Genetic parameters for egg production and egg weight of laying hens housed in single and group cages.Journal. Arch. Tierz Dummenstorf 5: 501–508

Nurgiartiningsih V M, Mielenz N, Prreisnger R,Schmutz M and Schueler L 2004 Estimation of genetic parameters based on individual and group mean records in laying hens. Journal of British Poultry Science 5: 604–610

Schaeffer L R and Dekkers J C M 1994 Random regression in animal models for test day Production in dairy cattle. Proc. 5th World Congress of Genetics Applied to Livestock Production. Guelph, Ontario, Canada XVIII: 443-446.

Szwaczkowski T 2003 Use of mixed model methodology in poultry breeding: Estimation of genetic parameters. In: Poultry Genetics, Breeding and Biotechnology (eds. W.M. Muir, S.E. Aggrey) CAB International, 165–202

Venturini G C, Grossi D A, Ramos S B, Cruz V A R,Souza C G, Ledur M C, EifaroI L, Schmidt G S and Munari D P 2012 Estimation of genetic parameters for partial egg production periods by mean of random regression models . Journal of Genetic molecular Research 11:1819-1829

Walk A, Arango J, Settar P, O’sullivan N P and Dekkers J C M 2011 Evaluation of egg production in layers using random regression models. Journal of British Poultry Science 90: 30–34

Walk A, LisowskiI M and Szwaczkowski T 2007a Genetic evaluation of laying hens under fixed regression animal models. Journal Arch. Tierz Dummenstorf 50: 279-287

Walk A, LisowskiI M and Szwaczkowski T 2007b Heritability of egg production in laying hens under cumulative, multitrait and repeat measurment animal models. Journal of Animal Science 52: 2545-259

Walk A and Szwaczkowski T 2009 Estimation of genetic parameters for monthly egg Production in laying hens based on random regression models. Journal Applied .Genetics. 50:41-46


Received 30 December 2013; Accepted 18 March 2014; Published 1 May 2014

Go to top