Sexual-size dimorphism (SSD), i.e., size differences between conspecific males and females, is common in the animal kingdom and particularly marked in the mammalian species, both in wild and domestic populations (Milner et al., 2000; Poissant et al., 2008; Polak and Frynta, 2009, 2010; Gudex et al., 2009). The SSD is the outcome of combined effects of several factors such as specific selection pressure, sex-biased phenotypic and genetic variation, and it may also reflect an imperfect genetic correlation between sexes for body size (Badyaev, 2002). In a group of related species, SSD is often larger in larger species and decreases as the size of species decreases. This phenomenon is termed Rensch's rule (Rensch, 1950). In addition, in related species, SSD is normally more pronounced in wild, compared to domestic, species because of the reduced role of male combat in captive herds; the effect of reduced sexual segregation in captivity, where males usually graze together with females in mixed herds; and relaxed intrasexual selection as a result of the female-biased operational sex ratio (Polak and Frynta, 2009).

The Afshari sheep breed is one of the heaviest and largest mutton breeds in Iran and is widely distributed in mountainous areas in the west of the country. Today, a large percentage of the Afshari sheep population is raised in Zanjan province. The Afshari is a fat-tailed, carpet-wool sheep with brown color and is primarily used for meat production (Eskandarinasab et al., 2010). In Afshari sheep, there are distinct differences in growth pattern between sexes: male lambs are heavier at birth, grow faster in all growth phases, and tend to be leaner than females. On average, mature Afshari rams are over 25 % heavier than mature ewes (66 vs. 53 kg).

Although SSD is apparent in domestic animals such as sheep (Gudex et al., 2009), goat (Polak and Frynta, 2009), pig (Dunshea, 2001), cattle (Polak and Frynta, 2010), rabbit (Eason et al., 2000) and chicken (Maniatis et al., 2013), little effort has been made to investigate its genetic basis and explore its potential use in sheep breeding programs. Accurate estimation of breeding values and the definition and realization of optimal multiple-trait selection response for each sex are challenges when SSD is present in sheep production, especially if heritabilities for measures taken on males and females are unequal and (or) genetic correlations between the sexes are less than unity (Gudex et al., 2009). For this reason, the present study was designed to investigate phenotypic and genetic differences in body weights between male and female Afshari lambs and estimate the genetic correlation between sexes.

Least-squares means (±SE) for the traits studied according to year of lambing, type of birth and age of dam at lambing in males and females are shown in Tables 3 and 4, respectively. Effects of year of birth were significant in both sexes (P < 0.01). Most previous studies have also reported important year effects on growth-related traits. Variation among years in growth traits arises from differences in nutrition and agro-climatic conditions which affect animals' growth (Yilmaz et al., 2007; Baneh and Hafezian, 2009; Al-Bial et al., 2012). The sex–year interaction for weaning weight and preweaning growth rate (both P < 0.01) suggests greater sensitivity of male lambs to differences among years. For weaning weight of males the range in year means was 21.9 % of the overall mean, whereas for females, the range in year means was only 16.6 % of the overall mean. So, for postnatal growth, males were more sensitive to year effects than were females. In contrast, the range among years in birth weight was 4.9 % of the mean for both sexes, and the sex–year interaction was not significant.

Type of birth had significant effects on body weight in both sexes (P < 0.01). Single lambs were heavier than twins and triplets in all cases, in agreement with Yilmaz et al. (2007) and Al-Bial et al. (2012). Twins and triplets weighed less at birth compared to singles. In addition, competition between lambs for milk results in lower milk consumption and, consequently, lower weight gains and weaning weights compared to singles (Baneh and Hafezian, 2009). Multiplicative adjustment factors to correct weaning weights of twins and triplets to a single-lamb basis were 1.27 and 1.59, respectively, for males, but 1.23 and 1.42, respectively, for females. Thus the effect of multiple births, and especially of a triplet birth, was considerably larger in males. This was the source of the sex–type-of-birth interaction for weaning weight, and it again suggests greater environmental sensitivity in males. For birth weights, adjustments for type of birth for twin and triplet lambs were almost similar for the two sexes: 1.13 and 1.19, respectively, for males and 1.14 and 1.22, respectively, for females, indicating no sex–type-of-birth interaction for birth weight.

The age of the dam at lambing affected (P < 0.01) growth of male and female lambs. Lambs from 3- to 6-year-old ewes were heavier than lambs from 2- or > 6-year-old ewes. Our results confirmed the findings of Ganai and Pandey (2000), who reported that very young or old ewes generally produce less milk to feed their lambs and that, for this reason, lambs from second through fifth lambing were heavier than lambs from ewes at their first or greater-than-fifth lambing. For dam age, there was no obvious difference in adjustment factors between sexes for either birth or weaning weights, and therefore no dam age–sex interactions was identified.

As shown in Table 2, male lambs were 6.6 % heavier at birth, had 14.4 % higher growth rates, and consequently were 16.0 % heavier at weaning compared to female lambs. The M / F ratios, which were used to express the magnitude of SSD (Lovich and Gibbons, 1992) were, respectively, 1.07, 1.14 and 1.15 for BW, WW, and GR. These estimates showed low levels of SSD for these traits. In addition, our findings regarding levels of SSD were closer to those reported by Polak and Frynta (2009) in domestic (1.41), compared to wild (1.67), species of sheep.

Differences between male and female lambs reflect difference in the endocrine environment, and associated differences in nutrient requirements, between the sexes. Regulatory mechanisms controlling growth hormone (GH) secretion are sexually dimorphic (Gatford et al., 1996; Jaffe et al., 1998). Gatford et al. (1996) reported that sex significantly affected patterns of changes in circulating growth hormone, IGF-I (insulin-like growth factor), and IGFPB-3 (insulin-like growth factor binding protein) concentrations in growing prepubertal lambs. Mean plasma GH concentrations, GH pulse amplitude, and integrated plasma GH concentrations were greater in rams than in ewes at a variety of different ages. They concluded that differences in circulating patterns and concentrations of GH, IGF-I, and IGFBP-3 may arise from the action of steroids during sexual differentiation and before puberty. Steroids such as testosterone and estrogen affect the growth of males and females differentially. In males, testosterone is produced in large quantities, whereas in females estrogen and progesterone predominate. Testosterone stimulates muscle growth by affecting the rate of protein synthesis, protein breakdown, and the net gain or loss of muscle protein (Mateescu and Thonney, 2002). In addition, exposure to high estrogen levels limits growth of long bones and consequently affects body size in females.

Results of univariate animal models including data from lambs of both sexes are shown in Table 5. Heritability estimates for BW, WW, and GR were 0.32, 0.17, and 0.12, respectively. Maternal permanent environmental effects (c2) were largest for BW (0.18) and lower (0.08) for WW and GR. Heritability estimates for WW and GR were quite close to literature averages reported for these traits in models that fitted maternal effects, but the heritability estimate for BW was somewhat larger than the literature average (Safari et al., 2005). Maternal effects in our analyses were anticipated to approximate the sum of the additive and permanent environmental maternal effects but were generally smaller than averages reported for this sum in the literature (Safari et al., 2005).

Estimates of variance components and genetic parameters from the bivariate analysis are presented in Table 6. In most cases, estimates of additive genetic variances were higher in females, while estimates of maternal permanent environmental and residual variances were higher in males. As expected, results from the univariate model in Table 5 were generally intermediate to, but not simple averages of, the sex-specific estimates in Table 6, indicating that differences between the sexes are reflecting variation in overall (co)variance structures. Our results regarding additive genetic variance were in agreement with Gudex et al. (2009) – who estimated greater additive genetic variance in females than males for weaning and post-weaning body weights in Australian Coopworth, Poll Dorset and White Suffolk sheep breeds – but contradicted Milner et al. (2000), who found higher additive genetic variance in male lambs for body weight, hind leg length and incisor breadth in a free-living population of Soay sheep. High values for VC and VE in males for WW and GR were consistent with the hypothesis of greater environmental sensitivity in males. This observation was consistent with the generally lower prenatal survival rates observed for male lambs (e.g., Dalton et al., 1980; Riggio et al., 2008), even after accounting for the potential confounding influence of sex differences in birth weights (Hatcher et al., 2009), meaning that, even though males are heavier and may seem more robust than females, they also seem to be more sensitive to environmental stresses. Accordingly, abortion cases and mortality rate should be higher in males compared to females, a phenomenon which has also been documented in human studies (Cunningham et al., 2010). The ratio of VA/VE in both the sexes was maximum for BW. However, in males, the ratio was close to unity. If this ratio were low, phenotypic response to selection would be slower because environmental variation could offset phenotypic response to selection even when there is high additive variation in the trait of interest.

Estimates of h2 for BW, WW and GR were 0.39, 0.15 and 0.11, respectively, in males and 0.29, 0.22 and 0.16, respectively, in females. Heritability estimates for growth-related traits in sheep vary considerably among breeds and populations and depend on the method of estimation. However, current estimates were in the range reported in different breeds of sheep by Safari et al. (2005). The higher heritability estimates for WW and GR in females were consistent with results of Gudex et al. (2009), who reported higher heritability estimates for post-natal body weights in female lambs of the Australian Coopworth, Poll Dorset and White Suffolk breeds, although in most cases the differences were not significant. By contrast, estimated values of h2 and CVA for BW in males were, respectively, 34 % (0.39 vs. 0.29) and 14 % (7.3 vs. 6.4 %) greater than that in females. Houle (1992) proposed that the “coefficient of additive genetic variance” is often more informative than the heritability. The CVA scales the component of additive genetic variance by the trait mean instead of the total variance and so is not confounded with the magnitude of other variance components (Kruuk et al., 2000). The “VA/x¯” ratio measures “variability” or capability to change, while the ratio “VA/VP” measures “heritability” of the trait. Our estimates of CVA show more additive genetic variability, i.e., more potential for response to selection, in BW of male compared to female lambs. In both sexes, estimates of c2 and tM were approximately the same for male and female lambs at birth, but for weaning weight tM and c2 were higher for males. However, approximate standard errors and resulting 95 % confidence intervals for both direct and maternal variance components did not indicate significant differences between sexes. A previous study by Gudex et al. (2009) on sexual dimorphism in sheep generally agrees that differences between male and female lambs in estimates of direct and maternal variances are not significant. These results may reflect a lack of sexual selection in domestic animals which has resulted in decreased SSD (Polak and Frynta, 2009).

Genetic (ra(m,f)) and maternal permanent environmental correlations (rc(m,f)) between sexes were 0.986 and 0.723 for BW, 0.995 and 0.983 for WW, and 0.995 and 0.966 for GR, respectively, which highlighted BW as somewhat dimorphic at the level of the maternal effects (P < 0.15; Table 7). Gudex et al. (2009) estimated ra(m,f) for weaning, post-weaning, yearling and hogget weights in Coopworth, Poll Dorset and White Suffolk breeds of sheep. Their estimates of ra(m,f) ranged from 0.59 for yearling weight in White Suffolk to 0.97 for weaning weight in Poll Dorset and White Suffolk breeds. Based on a literature survey, Poissant et al. (2010) founded that, except for fitness-related traits, genetic correlations between homologous traits expressed in males and females were predominantly large and positive. Because of genotype–environment interactions, ra(m,f) is environment-specific, and, due to differences in allele frequencies, estimates of ra(m,f) are also population-specific (Falconer, 1989; Poissant et al., 2010). Large and positive ra(m,f) show that growth-related traits in males and females are controlled by the same genes and that selection on males (i.e., in prospective sires) will improve both male and female body weights.

If the traits of interest were controlled by genes that are differently expressed in lambs of the two sexes, the ra(m,f) would be significantly less than 1.0, and, in such a situation, divergent selection could be used to diverge male and female phenotypes toward desirable outcomes. For example, the relationship between lamb birth weight and perinatal survival is curvilinear, with survival rates commonly optimized at intermediate optimum birth weights (Hatcher et al., 2009). Sex-specific selection could thus allow increases in birth weights in females, to improve perinatal survival, but without changing birth weights in males, to avoid increases in difficult births. Deviations of ra(m,f) from unity may result from activation of loci on sex chromosomes. Wittenburg et al. (2011) quantified effects of sex-linked genes on birth weights of Landrace piglets. They partitioned the additive effects into autosomal and gonosomal effects and reported that the Y chromosome contributed significantly to genetic variation in birth weight, though the effect of the X chromosome was not significant. Such Y- or X-linked loci activation has been documented in other species (Lindholm et al., 2006; Willmore et al., 2009). However, ra(m,f) in our data were very close to 1.0 for all measured variables, apparently precluding an important role for nonautosomal genes.

Departure of rc(m,f) from unity would indicate differential maternal effects of the dam on male and female offspring. In our analyses, maternal effects would include both additive maternal effects controlled by genes carried by the dam and permanent environmental effects of the dam that are presumably mediated by early-life influences on the dam and retained across lambings. Values of ra(m,f) for weaning weight and preweaning gains were very close to 1.0, whereas the correlation of maternal effects on BW across sexes was considerably lower (0.73), though still not significantly different from 1.0 (P < 0.15). The potential for sex-specific maternal effects at either an additive genetic or permanent environmental level is intriguing given the potential for lifetime, or even intergenerational, fetal programming events and epigenetic modifications in females that can affect future progeny birth weights in mammalian species (Drake and Walker, 2004). Differential effects of such modifications on birth weights of lambs of different sexes have, to our knowledge, not been documented but would be consistent with the role of IGF-2 in placental and fetal growth and the paternal and maternal imprinting of IGF-2 and its receptor, respectively, in sheep (Lawrence et al., 2012).

In conclusion, male lambs were heavier at birth, grew more rapidly to weaning, and were therefore heavier at weaning compared to female lambs. Effects of lamb sex on weaning weight and preweaning rate of gain interacted with effects of birth year and birth type. These interactions, and the generally greater residual and maternal variances observed for male lambs, suggest greater environmental sensitivity of male, compared to female, lambs. Additive genetic variances for preweaning gains and weaning weights were somewhat larger in female lambs. However, additive genetic correlations for body weight traits between male and female lambs were large, suggesting that these traits were controlled by similar genes. Opportunities for sex-specific selection to create or modify sexual-size dimorphism in sheep therefore appear limited and cannot be used to diverge phenotypes for body weights between the sexes.