Genetic analysis of growth traits in Iranian Makuie sheep breed

Abstract

The Makuie sheep is a fat-tailed sheep breed which can be found in the Azerbaijan province of Iran. In 1986, a Makuie sheep breeding station was established in the city of Maku in order to breed, protect and purify this breed. The genetic parameters for birth weight, weaning weight (3 months), 6-month, 9-month and yearling weight, and average daily gain from birth to weaning traits were estimated based on 25 years of data using DFREML software. Six different models were applied and a likelihood ratio test (LRT) was used to select the appropriate model. Bivariate analysis was used to define the genetic correlation between studied traits. Based on the LRT, model II was selected as an appropriate model for all studied traits. Direct heritability estimates of birth, weaning, 6-month, 9-month and yearling weights and average daily gain from birth to weaning were 0.36, 0.41, 0.48, 0.42, 0.36 and 0.37, respectively. Estimates of direct genetic correlation between birth and weaning weights, birth and 6-month weights, birth and 9-month weights, as well as between birth and yearling weights were 0.57, 0.49, 0.46 and 0.32, respectively. The results suggest there is a substantial additive genetic variability for studied traits in the Makuie sheep breed population, and the direct additive effect and maternal permanent environment variance are the main source of phenotypic variance.

Key Words:

• Derivate-free algorithm
• Direct heritability
• Genetic correlation
• Maternal heritability
• Restricted maximum likelihood.

Introduction

The Makuie sheep is a native breed of Iran and can also be found in Turkey (called as Ak Karaman). Its total population is estimated at approximately 2.7 millions (Abbasi and Ghafouri, 2011). It has been adapted to cold and highland environments (Safari, 1986). They are fat-tailed sheep with a medium-sized body, white in color with black rings around the eyes, nose and feet (Saadatnoori and Siahmansor, 1986). They are kept in the Eastern and Western provinces Azerbaijan and their main products are meat, wool and milk (Saadatnoori and Siahmansoor, 1986). The rearing system is mostly extensive-migratory from April to September (on natural pastures in spring and summer), and semi-intensive from October to March (on stations and fed in barns during autumn and winter). Alfalfa, barley, corn silage, concentrates and grass are the main feedstuffs used in the semi-intensive rearing period.

Breeding with a superior breeding value in growth rate, egg, meat, milk, or wool production, has revolutionized agricultural livestock production throughout the world. The scientific theory of animal breeding is based on population and quantitative genetics. Population-based genetic parameters, especially heritabilities and correlation between traits, make up the most important information required to improve selection objectives, and determine the selection gain of current selection schema.

In 1986, the Makuie Sheep Breeding Station (MSBS) was established in the city of Maku, Western Azerbaijan, Iran by the Jihad Sazandegi Ministry (Safari, 1986) in order to protect, purify and improve this sheep breed (Safari, 1986). The breeding season starts late in the summer (in September) and early in the autumn (in October). It ends by the middle of autumn (early in November). Estrus synchronization is carried out in the flock with a progesterone-releasing intravaginal (CIDR). Ewes are then bred either by artificially insemination (in the first cycle of estrus) or with controlled rams. Flushing and equine chorion gonadotrophin (ECG) injection at CIDR removal are applied to increase the litter size. Ewes are kept in the flock for a maximum of 7 parities and rams for 5 breeding seasons. Lambing occurs once a year and it starts early in the second month of winter (late January). Lambs receive creep feeding during the first two weeks of age.

Although the population has been under selection for at last 25 years, to our knowledge there has been no report that has estimated genetic parameters in this sheep breed. The objectives of the present study were to estimate genetic parameters of growth traits as well as to reveal any association between traits by using genetic correlation analysis.

Materials and methods

In this study, data of the growth traits 5,212 lambs from 130 sires and 1,320 dams recorded during the period 1989 to 2010 at the Makuie Sheep Breeding Station (MSBS). Traits were birth weight (BW), weaning weight (WW), 6-month weight (6MW), 9-month weight (9MW), yearling weight (YW) and average daily gain from birth to weaning (ADG).

Statistical analysis

In model I, the direct additive genetic was considered as random effect. In model II, the direct additive genetic and maternal permanent environment was included as random effects. In the present model, genetic parameters were estimated according to genetics of the animals themselves and their permanent environment as uterus environment. It is clear that, in this model, the estimated h² was lower than those estimated by model I, because heritability has been divided into two parts: additive genetic and permanent environment. Model III included the direct additive genetic and maternal genetic. In model IV, the correlation between direct and maternal genetic effect was studied. Model V, the direct additive genetic, maternal genetic and maternal permanent environments were considered as random effects. In model VI, the random effects in model V plus correlation between maternal and additive genetic were studied.

Variance and covariance components were estimated according to an animal model with a restricted maximum likelihood (REML) approach using a derivate-free (DF) algorithm (Meyer, 1989). Maternal genetic or permanent environmental effects were taken into account by including them in appropriate models, as described by Meyer (1992). The effects of year with 21 levels, sex with 2 levels, birth type with 3 levels, and age of dam with 6 levels were considered as fixed effects. Six different univariate models were examined for each trait (Model I to Model VI). They had the same fixed effects and differed from each other in random effects and correlations (Table 1). The linear form of the 6 models was:

Table 1 The six models and their fixed and random effects.

	Fixed effects				Random effects
Model	YR	SX	BT	AD	AN	M	PE	E	rAM
I	*	*	*	*	*	–	–	*	–
II	*	*	*	*	*	–	*	*	–
III	*	*	*	*	*	*	–	*	–
IV	*	*	*	*	*	*	–	*	≠0
V	*	*	*	*	*	*	*	*	=0
VI	*	*	*	*	*	*	*	*	≠0

YR, year; SX, sex; BT, birth type; AD, age of dam; AN, animal direct additive genetic; M, maternal genetic effect; PE, maternal permanent; E: error or residual effect; r_AM, genetic correlation between direct additive genetic and maternal genetic.

where µ, overall mean of population;

Y_ijk…, each observation;

YR_i, fixed effect of year i;

SX_j, fixed effect of sex j;

BT_k, fixed effect of birth type k;

AD_l, fixed effect of age of dam l;

AN_m, additive genetic effect of animal m;

PE₀, random effect of maternal permanent environment in n level;

M_n, maternal genetic effect;

E_ijk…, residual effect of an observation ijk

The estimated parameters according to the models were: phenotypic variance (σ²p), direct additive genetic variance (σ²_a), maternal genetic variance (σ²_m), the maternal permanent environmental variance (σ²_pe), residual variance (σ²_e), direct heritability (h²_a) ; maternal heritability (h²_m) ; genetic covarance between direct additive and maternal effects (r_am) and maternal permanent environmental variance to total phenotypic variance

Model VI is a full model and the phenotypic variance (σ²_p) is partitioned to additive genetic variance (σ²_a) maternal additive genetic variance (σ²_m), δ_am is covariance between the animal additive effects and maternal additive genetic effects (σ_am). Maternal additive genetic variance (σ²_m), is a part of total phenotypic variance that is defined by maternal additive genetic. Maternal environmental variance (σ²_pe), is a part of total phenotypic variance that is defined by maternal permanent environment such as uterus environment, amount of milk produced by mother, etc. The maternal permanent environment can influence the growth indexes of lamb, as it can influence the genetic parameters estimated by different models. Generally, a big uterus produces bigger and heavier lambs. Consequently, it could have an enormous effect on the estimation of genetic parameters. The amount of milk production is another example of the maternal permanent environment. The lambs that receive more milk have a higher growth rate.

Results and discussion

Descriptive statistics

Descriptive statistics of measured traits based on age, sex and birth type (single or twin) are presented in Table 2. Triple lambing is a rare phenomena (1–2%) in Makuei sheep. As expected, the number of lambs recorded decreased with age and this decline is due to the harvest/culling and losing of lambs at different ages. The birth weight (BW) had the lowest CV. The lower CV for this trait compared to others has also been reported by other researchers (Fogarty, 1995; Yazdi et al., 1999). The amount of CV was lower than those reported by other authors (Neser et al., 2001; Mandal et al., 2006; Miraei- Ashtiani et al., 2007). This could be due to less variation and the effect of ambience, as ewes are kept in barns and fed cured roughage during late pregnancy.

Table 2 Descriptive statistics of the growth traits in Iranian Makuie sheep.

Trait	Sex	Birth type	No. of records	Mean, kg	SD	CV, %
BW	Male	Single	1992	4.52	0.54	11.91
		Twin	358	3.88	0.61	15.65
	Female	Single	2015	4.27	0.50	11.69
		Twin	373	3.59	0.51	14.08
Overall			4738	4.30	0.59	13.78
WW	Male	Single	1821	21.26	3.36	15.80
		Twin	346	17.10	5.36	31.34
	Female	Single	1955	19.70	2.94	14.95
		Twin	361	16.10	5.18	32.16
Overall			4483	19.84	3.91	19.68
6MW	Male	Single	1358	29.22	5.10	17.44
		Twin	228	25.48	4.55	17.86
	Female	Single	1673	26.46	4.03	15.22
		Twin	272	23.65	3.55	15.00
Overall			3531	27.24	4.79	17.59
9MW	Male	Single	1002	30.75	4.72	15.34
		Twin	143	27.49	5.04	18.34
	Female	Single	1411	27.40	3.96	14.46
		Twin	204	25.65	3.49	13.60
Overall			2760	28.49	4.63	16.24
YW	Male	Single	382	40.00	5.68	14.21
		Twin	45	37.62	5.88	15.62
	Female	Single	1327	31.30	4.16	13.30
		Twin	191	30.04	3.73	12.40
Overall			1945	33.03	5.76	17.45
ADG	Male	Single	1820	0.19	0.04	19.23
		Twin	346	0.14	0.04	24.93
	Female	Single	1955	0.17	0.03	18.20
		Twin	361	0.14	0.03	24.18
Overall			4482	0.17	0.04	21.47

BW, birth weight; WW, weaning weight; 6MW, 6-month weight; 9MW, 9-month weight; YW, yearling weight; ADG, average daily gain from birth to weaning; CV, coefficient of variation.

Fixed effects

The fixed effects along with their significance levels on studied traits are presented in Table 3. The effect of birth year on birth weight (BW), 6-month weight (6MW), 9-month weight (9MW), yearling weight (YW) (P<0.001), average daily gain from birth to weaning (ADG) (P<0.01) and weaning weight (WW) (P<0.05) was significant. These results were consistent with the report on Sabi sheep (Matika et al., 2003). It could be due to differences in management, food availability, disease, and climatic condition (rate of rainfall, humidity and temperature) that affect the quality and quantity of pasture forage and raising systems in different years. The age of the dam was significant on birth weight (BW) (P<0.001), weaning weight (WW) (P<0.01), 6-month weight (6MW) and yearling weight (YW) (P<0.05) traits. Young ewes tend to produce smaller lambs. Primiparous ewes are not at their mature weight and complement their growth in addition to fetal growth. This could affect the lam weight. It is well known that mothering ability, such as milk yield, increases with parity, as older ewes are usually larger and produce more milk (Dass and Acharya, 1970). However, the age of the dam did not have significant effect on 9MW trait. The same results were reported by El Fadili et al. (2000) on the Moroccan Timahdit sheep.

Table 3 Analysis of variance for birth weight, weaning weight, 6-month weight, 9-month weight, yearling weight and average daily gain from birth to weaning traits in Iranian Makuie sheep.

Fixed effects	BW	WW	6MW	9MW	YW	ADG
Year	^**	^***	^***	^***	^*	^***
Age of dam	^***	^*	^**	ns	^***	^***
Sex	^***	^***	^***	^***	^***	^***
Birth type	^***	^***	^***	^***	^***	^***
R^2	0.34	0.30	0.24	0.16	0.17	0.40

BW, birth weight; WW, weaning weight; 6MW, 6-month weight; 9WM, 9-month weight; YW, yearling weight; ADG, average daily gain from birth to weaning.

* Significant at 0.05 probability level;

** significant at 0.01 probability level;

*** significant at 0.001 probability level; ns, not significant at 0.05 probability level.

Male lambs were heavier than females at all ages and these differences were significant (P<0.01). The effect of lamb sex on body weight traits at different ages has been reported in various sheep breeds (Mokhtari et al., 2008; Aghaali-Gamasaee et al., 2010). Physiological characteristics and endocrinal system (type and quantity of hormone secretion, especially sex hormones) can explain the significant influences of gender (Aghaali-Gamasaee et al., 2010).

The effect of birth type was significant on all studied traits. The frequency of single birth type was high compared to other types. A low number of triple birth types was seen so it was not included in the models. The significant effect of birth type on body weight can be due to limited uterine space during pregnancy, nutrition of the dam, especially during late pregnancy (regardless of twin or triple pregnant dams), and competition for milk sucking between multiple birth lambs during the birth to weaning period. Similar results have been reported in other breeds, such as the Hungarian Merino sheep (Komlosi, 2008).

According to the equation , 16–40% of traits’ phenotypic variances were explained by the established factors (Table 3) and the effects of sex and birth type were the most important of the traits studied (P<0.001).

Model selection

FThe model VI was the full model. The LRT test was used to determine significance between full and nest models. Statistical significance for models was set at P<0.05. The LRT value was calculated as -2 ln (likelihood for full model) + 2ln (likelihood for nested model). If the LRT value was greater than a critical value from χ² distribution with an appropriate degree of freedom (df), it can be concluded that the additional random effect has significant effect in the model and the nested model is not the best. When the differences were not significant, the nested model, which had fewer parameters, was chosen as appropriate model. The best model for all studied traits was model II (Table 4). Then, the direct additive genetic variance, maternal permanent environmental variance, and residual variance were the main sources of variation for phenotypic variance along with significant fixed effects for each trait.

Table 4 Variance components and genetic parameters of different traits in Iranian Makuie sheep breed.

Trait	MF	h^2a	PE^2	h^2m	σAM	rAM	Log-likelihood
BW	I	0.36					1281.3028
	II	0.27	0.10				1304.1027
	III	0.20		0.16			1281.3028
	IV	0.20		0.10	0.04	0.26	1281.3028
	V	0.20	0.10	0.07			1304.1028
	VI	0.20	0.10	0.11	-0.005	-0.16	1304.1028
WW	I	0.41					-6733.6159
	II	0.20	0.17				-6673.8374
	III	0.20		0.20			-6733.6159
	IV	0.20		0.10	0.39	0.35	-6733.8619
	V	0.20	0.17	0.00			-6673.8375
	VI	0.20	0.17	0.11	-0.53	-0.42	-6673.8375
6MW	I	0.48					-6086.6756
	II	0.42	0.08				-6072.7768
	III	0.20		0.28			-6086.6756
	IV	0.20		0.25	0.22	0.08	-6086.6756
	V	0.20	0.08	0.22			-6072.7765
	VI	0.20	0.08	0.31	-0.81	-0.24	-6072.7766
9MW	I	0.42					-4669.9102
	II	0.37	0.05				-4666.9063
	III	0.20		0.22			-4669.9102
	IV	0.20		0.31	-0.76	-0.24	-4669.9102
	V	0.20	0.05	0.17			-4666.9063
	VI	0.20	0.05	0.10	0.49	0.30	-4666.9063
YW	I	0.36					-3636.6781
	II	0.31	0.06				-3634.4796
	III	0.20		0.15			-3636.6781
	IV	0.20		0.25	-0.97	-0.25	-3636.6781
	V	0.20	0.06	0.11			-3634.4796
	VI	0.20	0.05	0.29	-2.02	-0.45	-3634.4798
ADG	I	0.37					13509.2322
	II	0.17	0.17				13567.2569
	III	0.20		0.17			13509.2322
	IV	0.20		0.27	-6×10^-4	-0.26	13509.2322
	V	0.17	0.17	0.00			13567.2569
	VI	0.20	0.15	0.10	-7×10^-4	-0.51	13567.2566

h_2a, direct heritability; h^2m, maternal heritability; σ_AM, co-variance between direct additive and maternal additive genetic effects; r_AM, correlation between direct and maternal additive effects; PE², heritability due to permanent environment; MF, fitted model; BW, birth weight; WW, weaning weight; 6MW, 6-month weight; 9WM, 9-month weight; YW, yearling weight; ADG, average daily gain from birth to weaning.

Heritability and (co)variance components for the traits

Birth weight

Direct heritability was estimated for birth weight (BW) using model I. It was 0.36 and was higher than that estimated for other breeds such as Dorper (Neser et al., 2001). However, using model II and introducing maternal permanent environment to the model, the direct heritability decreased to 0.27. The maternal permanent environment effect was 0.1. This indicates that the 10% of BW variation was explained by maternal permanent environment which was lower than that reported by other researchers (Neser et al., 2001). This suggested we should not use direct heritability as the only criteria for breeding programs.

Weaning weight

The estimates of h²_a and PE² by using model II for weaning weight (WW) were 0.20 and 0.17, respectively; this was in agreement with other researchers such as Mandal et al. (2006). The high value of PE² for weaning weight suggests the importance of the permanent maternal environment. Including direct maternal genetic correlation in model VI yielded a large negative estimate for this parameter in all studied traits while h²_a and h²_m values were dramatically increased. This has also been reported in Australian Charolais Cattle (Meyer, 1992).

Negative correlations between direct and maternal effect for early growth traits are common but are not biologically possible (Mantiatis and Pollot, 2003). According to previous researchers (Meyer 1992), the possible reasons for the negative correlation may be environmental stresses (such as udder problems, and non-sufficient nutrition), different methods of animal management, different methods of data collection, size of data set, and data structure (from field condition or experimental condition). The experimental flocks due to standardized management are much less subject to having negative correlation between direct and maternal effects. Another reason for this negative correlation may arise from the number of offspring recorded per mother as a low number of offspring produce high negative correlation (Mantiatis and Pollott, 2003). It has been proved that the relative values of h²_a and h²_m were greatly influenced by the model used in the analysis of data(Koyuncu and Duru, 2009). From a developmental perspective, some negative correlations provide checks and balances between direct and maternal effects for growth traits. Whereas this condition could be involved in preventing the species from becoming much larger or producing more and more milk (Cundiff, 1972).

Six-month weight

The highest h²_a for this trait were achieved by using models I and II, 0.48 and 0.42, respectively. For 6-month weight (6MW), 9-month weight (9MW) and yearling weight (YW) permanent maternal environment had less effect while the animal effect and maternal genetic played an essential role. In a previous study on the Sangsari breed, another Iranian fat-tailed sheep breed, the estimates of h²_a and PE² for 6-month weight (6MW) were higher than our estimates (Miraei-Ashtiani et al., 2007). According to the results obtained by Mantiatis and Pollott (2003), parameter estimates in different populations were strongly influenced by the number of progenies per dam and the proportion of mothers with recorded performance.

Nine-month weight

Our estimates based on models II were 0.37 and 0.05 for h²_a and PE², respectively, which were between the ranges of values estimated for the Kermani breed, another Iranian fat-tailed sheep breed (Mokhtaria et al., 2008). The maternal permanent environment, like 6-month weight (6MW) declined, whereas maternal genetic proved to be more dominant than direct genetic and permanent maternal environment.

Yearling weight

The estimates of h²_a and PE² for YW trait by using model II were 0.31 and 0.06, respectively. These results were in the range of other reports (Komlosi, 2008). The negative correlation between direct and maternal genetic in yearling weight (YW) trait as in Komlosi’s (2008) studies in Hungarian merino and meat sheep breeds may be due to fewer progeny recorded per dam (Mantiatis and Pollott, 2003).

Average daily gain from birth to weaning

Average daily gain from birth to weaning (ADG) is the important factor in sheep breeding systems, especially in mutton systems; therefore, it could be included in the selection index as a selection criterion. The average daily gain of the Makuie breed was 170 g in this study. Using model II, direct heritability and heritability due to maternal permanent environment were estimated to be 0.17. Estimated values agreed with the value estimated for the Arabi sheep breed (another Iranian fat-tailed sheep) using the same model (Mohammadi et al., 2010).

Correlation study

A bivariate model was used to define the correlation between studied traits (Table 5). Genetic correlations between the direct effects on all traits were positive and ranged from 0.32 to 0.98. These are in the range of reports on Sangsari sheep (0.17 to 0.99) (Ashtiani et al., 2007). Direct genetic and phenotypic correlations were higher in adjacent weights than non-adjacent ones, as reported by Boujenane and Hazzab (2008) in Draa goats. The phenotypic correlation among body weight traits was lower than their corresponding genetic correlations, except for those between weaning weight (WW) and average daily gain from birth to weaning (ADG). The phenotypic correlations ranged from 0.15 between birth weight (BW) and average daily gain from birth to weaning (ADG), to 0.96 between weaning weight (WW) and average daily gain from birth to weaning (ADG). These are close to results reported by Gizaw et al. (2007) in Menz sheep. The results of the present investigation showed that due to positive genetic and phenotypic correlations, selection for any of the studied traits could result in an increase in phenotypic magnitudes and genetic potentials for body weight traits.

Table 5 Correlation between traits in Iranian Makuie sheep breed from bivariate DFREML analysis (based on model II).

Trait 1	Trait 2	rp12	ra12	rpe12	re12
BW	WW	0.31	0.57	0.44	0.18
BW	6MW	0.25	0.49	0.41	0.08
BW	9MW	0.22	0.46	0.04	0.12
BW	YW	0.23	0.32	0.22	0.20
BW	ADG	0.15	0.33	0.31	0.02
WW	6MW	0.68	0.98	0.89	0.49
WW	9MW	0.63	0.93	0.87	0.45
WW	YW	0.50	0.84	0.57	0.30
WW	ADG	0.96	0.92	0.95	0.98
6MW	9MW	0.75	0.93	0.81	0.62
6MW	YW	0.59	0.80	0.76	0.43
9MW	YW	0.73	0.87	0.78	0.64

r_p12, phenotypic correlation between trait 1 and trait 2; r_a12, direct additive genetic correlation between trait 1 and trait 2; r_pe12, maternal permanent environmental correlation between trait 1 and trait 2; r_e12, residual correlation between trait 1 and trait 2; BW, birth weight; WW, weaning weight; 6MW, 6-month weight; 9WM, 9-month weight; YW, yearling weight; ADG, average daily gain from birth to weaning

Conclusions

Estimates of genetic parameters from the present study suggest that there is substantial additive genetic variability in the population for all studied traits. The high genetic correlation between weaning weight and yearling weight indicates that breeders’ rams could be selected at an earlier age. According to the LRT test we used, model II is the best model. This indicates that we need to include maternal permanent variance as an important proportion of phenotypic variance in breeding objectives.

Acknowledgments:

the authors are grateful to the MSBS staffs who were involved in the data collection.

The excellent schematization of MSBS in 1968 by Dr. Alex Safari is highly appreciated.