Data and flock management

Data were collected from an experimental breeding flock of Boer x Central Highland goats, maintained at Sirinka Agricultural Research Center small ruminant breeding, evaluation, and dissemination site, which is located at an altitude of 1850 m.a.s.l and 11°45’ 00" N and 39°36’ 36" E. The mean annual rainfall is 950 mm. The area is a moderately warm temperature zone with mean daily temperature ranges of 16–21°C. Data were collected from 2009 to 2018. A detailed description of the data is presented in Table 1. The animals were reared under semi-intensive management i.e., animals were allowed to graze for six hours per day and sheltered in a semi-open concert barn during the night based on their physiology, sex, and age. When returned from gazing/browsing, flocks were supplemented with a concentrate mix (100–400 g day^-1 based on their physiology and age) comprised of Wheat bran, Noug cake, and salt as a mineral source. Goats were vaccinated for common diseases in the area and treated regularly. All experimental techniques and animal care were in line with FASS [15], and animal nutrition and health researchers of Sirinka Agricultural Research Center, Amhara Region Agricultural Research Institute, confirmed this during annual review forum.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Description of data structure for efficiency-related traits in Boer crossbred goats.

https://doi.org/10.1371/journal.pone.0305749.t001

The mating method was a natural controlled mating method. One buck was assigned to 20–30 breeding females, and the bucks were kept with does for forty-five days. During mating, herdsmen were assigned to each mating group to collect the mating data and pedigree information. The pure Central Highland does were crossed with pure Boer bucks to produce the F1 crossbreds with 50% Boer blood level. Male and female crossbreds were inter se mated to produce F2 crossbreds. About 25% of female crossbreds with 50% blood level were mated with 75% blood level and pure Boer buck to increase their exotic blood level to 62.5% and 75%, respectively. As a limitation, the sample size and data structure may influence the interpretation of the findings from this study to some extent. Indeed, the data was collected by researchers under on-station management of animals to improve the data quality, and this may reduce the impact of data size on genetic parameter estimates.

Evaluated traits

The investigated traits include growth efficiency from birth to weaning (GE1) = ((weaning weight–birth weight)/birth weight) x 100, from weaning to six months (GE2) = ((six-month weight–weaning weight)/weaning weight) x 100 and growth efficiency from six-month to yearling age (GE3) = ((yearling weight–six-month weight)/six-month weight) x 100. Besides, the relative growth rate from birth to weaning (RGR1) = ((ln (weaning weight)–ln (birth weight)) / 90 days) x 100, from weaning to six months (RGR2) = ((ln (six-month weight)–ln (weaning weight)) / 90 days) x 100 and from six months to yearling age (RGR3) = ((ln (yearling weight)–ln (six-month weight)) / 180 days) x 100.

Statistical analysis

The general linear model procedure of SAS [16] was used to identify the important fixed effects, which have a significant effect on the efficiency-related traits. Post-hoc test was conducted using the Tukey-Kramer test. The statistical model was as follow:

Where Y_ijklmnop is the response variables, μ is the overall mean, BT_i is the effect of i^th birth type (single and multiple births), SB_j is the effect of j^th season of birth (main rain, short rain and dry), SX_k is the effect of k^th sex of kid (male and female), DP_l is the effect of l^th parity of doe (1, 2, 3, 4, and ≥5), TB_m is the effect of m^th year of birth (2009–2018), LK_n is the effect of n^th Boer blood level (25, 50, 62.5, and 75%), GK_o is the effect of o^th filial generation (F1, F2, and F3) and e_ijklmnop is random error term associated with each observation.

The genetic parameters were estimated by the Average Information Restricted Maximum Likelihood (AI-REML) method using WOMBAT program fitting animal model [17]. The significance of random effects was tested using the log-likelihood ratio test [18], and only the selected models are presented. An effect was considered to have a significant influence when its inclusion caused a significant increase in log-likelihood, compared with a model in which it was ignored. Phenotypic and genetic correlations among investigated traits were estimated in a bivariate analysis using starting values from univariate analyses. Genetic trends were estimated by regression of the average breeding values estimated in the particular trait on the birth year. The following six-univariate animal models were fitted: (1) (2) (3) (4) (5) (6)

Where y is the vector of the records of investigated traits; b, a, m, c, and ε are vectors of fixed effects, additive direct genetic, maternal additive genetic, permanent environmental effects of the dam and residual effects, respectively; X, Z₁, Z_2, and Z₃ are incidence matrices that relate these effects to the records. It was assumed that a, m, c, and ε are normally distributed with the mean zero and variance Aσ²_a, Aσ²_m, I_pc², and I_nε², respectively. Where A is the numerator relationship matrix between animals; σ_am is the covariance between additive direct and maternal genetic effects; I_p and I_n are identity matrices with orders equal to the number of does and kids, respectively. σ²_a, σ²_m, c², and ε² are the direct additive genetic variance, maternal additive genetic variance, maternal permanent environmental variance, and residual variance, respectively.

The heritability of direct genetic effects (h²_a) = σ²_a / σ²_p; heritability of maternal genetic effects (h²m) = σ²_m / σ²_p; direct-maternal genetic correlation (r_am) was computed as the ratio of the σ_am to the product of the square roots of estimates of σ²_a and σ²_m; maternal permanent environmental variance as a proportion of phenotypic variance (c²) = σ²_c / σ²_p; residual variance as a proportion of phenotypic variance (e²) = σ²_e / σ²_p. Where σ²_a is the direct genetic variance, σ²m is the maternal genetic variance, σ_am is the direct-maternal genetic covariance and σ²_c is the maternal permanent environmental variance, σ²_e is the residual variance and σ²_p is the phenotypic variance. Total heritability (h²_t) was estimated according to Willham [19] as follows:

Where σ²_a is the additive genetic variance, σ²_m is the maternal genetic variance, σ²_p is the phenotypic variance, and σ_am is the covariance between additive direct and maternal genetic effects.

The additive coefficient of variation (CV_A) was computed as follows:

Where σ²_a is the additive genetic variance and ñ is the sample mean.

Non-genetic and genetic effects

The efficiency-related traits of Boer x Central Highland goats are presented in Table 2. The least-squares means and standard errors for GE1, GE2, GE3, RGR1, RGR2, and RGR3 were 294.0 ± 5.06, 36.6 ± 1.20, 44.9 ± 1.81, 1.46 ± 0.01, 0.32 ± 0.01, and 0.19 ± 0.01, respectively. Singletons’ pre-weaning growth efficiency and relative growth rate were higher than those of multiple-born kids. However, the post-weaning efficiency for multiple-born kids was greater than for singletons. The sex of kids considerably influenced GE1, GE3, and RGR3, and male kids had a higher value for these traits than females.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Efficiency-related traits (LSM ± SE) of Boer x Central Highland goats.

https://doi.org/10.1371/journal.pone.0305749.t002

Kids with a 25% Boer level had higher growth efficiency and relative growth rate. Likewise, they had better growth efficiency during weaning to six-month age than kids with a 75% Boer level (Table 2). The growth efficiency and relative growth rate for F1, F2, and F3 kids at all growth phases were found to be statistically similar (P>0.05). The pre-weaning growth efficiency and relative growth rate of kids born from the 2^nd parity were higher than 3^rd parity. For post-weaning efficiency-related traits, however, parity did not have a sizable effect (P>0.05). Season of kidding exerted a significant influence on pre-weaning efficiency-related traits (Table 2). Better growth efficiency and relative growth rate observed for kids born during the short rainy season than in other seasons. The year of kidding also significantly influenced all considered efficiency-related traits of the goats.

(Co) variance components and heritability

Variance, covariance, and heritability estimates for efficiency traits in Boer x Central Highland goats are shown in Table 3. As per the log-likelihood ratio test, the selected models for GE1, GE2, GE3, RGR1, RGR2, and RGR3 were model 2, 6 2, 6, 6, and 1, respectively. About 18%, 3%, 23%, 20%, and 12% of the phenotypic variation in GE2, GE3, RGR1, RGR2, and RGR3, respectively were explained by the direct additive genetic effect. However, the contributions of kids’ own genes for variation of GE1 were low. The standard error for estimates in this study seems to be high. Except for RGR3, all investigated efficiency-related traits were under the influence of maternal genetic effect, and maternal heritability (h²_m) ranged from 0.09 to 0.17. Besides, the inclusion of maternal permanent environmental effects had a significant effect on heritability estimates of GE2, RGR1, and RGR2. The direct-maternal additive genetic correlation (r_am) for GE2, RGR1, and RGR2 were -0.93, -0.83, and -0.86, respectively. In this study, all investigated traits except for GE1 had moderate to high additive genetic coefficient of variation (CV_A = 10.5–29.6%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Estimates of (co) variance components and heritability for efficiency-related traits of Boer x Central Highland goats.

https://doi.org/10.1371/journal.pone.0305749.t003

Genetic and phenotypic correlation estimates

The phenotypic and genetic correlations among investigated efficiency-related traits are presented in Table 4. The phenotypic correlations (r_P) ranged from -0.43 to 0.97, and the genetic correlations (r_G) varied between -0.64 and 0.97. Growth efficiency was genetically correlated (r_G = 0.71 to 0.97) with the relative growth rate of goats in the same growth phase. However, the genetic correlations among pre-weaning and post-weaning efficiency-related traits were found to be antagonistic. The phenotypic correlations showed the same trend as the genetic correlations.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Genetic correlation (below diagonal) and phenotypic correlation (above diagonal).

https://doi.org/10.1371/journal.pone.0305749.t004

Genetic trend for efficiency-related traits

The genetic trend for growth efficiency and relative growth rate of Boer x Central Highland goats is illustrated in Figs 1 and 2, respectively. Except for GE1, the post-weaning growth efficiency showed positive genetic progress. The estimate of genetic changes for GE1, GE2, and GE3 were -0.6065, 1.2986, and 0.1398% year^-1, respectively. The relative growth rate at different growth phases had favorable genetic progress across the years. The annual genetic changes for RGR1, RGR2, and RGR3 were 0.0007, 0.0107, and 0.1819%, respectively. In general, the estimated breeding value for investigated efficiency-related traits showed an undulating trend across the years.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Genetic trends for growth efficiency traits at a specific age.

https://doi.org/10.1371/journal.pone.0305749.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Genetic trend for relative growth rate at a specific age.

https://doi.org/10.1371/journal.pone.0305749.g002

Systematic factors

Genetic, systematic, and non-systematic factors determine the performance of an animal. Thus, knowledge of the influence of systematic factors on efficiency-related traits is quite important for designing management and genetic improvement programs. The GE1 and GE2 in this study are lower than the report of Mokhtari et al. [13] for the Raeini Cashmere goat. However, Ghafouri-Kesbi and Gholizadeh [12] noted a relatively lower GE2, RGR2, and RGR3 in sheep than in the current study. This variability might be due to the variation in the management level and genetic potential of the breed/species. The effect of birth type on efficiency-related traits in this study is consistent with other studies [11, 12], which noted that the GE and RGR of triplet lambs were superior to single and twin lambs during post-weaning age. Besides, Mohammadi et al. [20] reported that single-born kids grew faster than multiple-born kids during the pre-weaning while they did not during the post-weaning period. However, pre-weaning and post-weaning growth efficiency for triple-born kids were higher than for single and twins, according to Mokhtari et al. [13]. Multiple-born kids are computing for resources during the pre-natal and post-natal periods, and thus, they try to adapt/survive forcefully. This forceful adaptive mechanism of kids could be the reason for their superiority during the post-weaning period.

Al-Saef [21], Ghafouri-Kesbi and Rafie-Tari [11], and Ghafouri-Kesbi and Gholizadeh [12] have reported the superiority of males over females in terms of efficiency-related traits. According to Baneh and Hafezian [22], the estrogen hormone limits the growth of long bones in females more than in males. This might be a possible reason for the inferiority of females over male kids. A filial generation did not have a significant influence on investigated traits. Boujenane et al. [23] have made a similar observation regarding the growth traits of sheep. The influence of dam parity/age was noted in previous studies [24, 25], which could be explained by dams’ degree of maturity and milk production potential. A sizable effect of season and year on efficiency-related traits was noted elsewhere [1, 13, 20, 21]. The variability of climatic variables, forage availability, and disease infestation could be the reason for observed variation across years and seasons.

Genetic parameters

The contribution of direct additive genetic effect to phenotypic variation of pre-weaning growth efficiency (GE1) seems low relative to GE2 and GE3. This low contribution of kid’s own genes for phenotypic variation of GE1 indicates that maternal genes and genes that have a non-additive effect could control GE1. In addition, the contribution of kids’ genes is pronounced after weaning, which could be the reason for a relatively higher direct additive genetic effect after weaning age. Yogesh et al. [26] and Latifi et al. [27] have made a similar observation. The observed high standard error for estimates in this study could be ascribed to the sample size, data structure, or a small number of progeny per dam and limited information on linked offspring and dams’ performances. There are few reports of heritability for growth efficiency and relative growth rate of small ruminants in the literature. Mokhtari et al. [13] noted a relatively higher h²_a estimate for GE1 and GE3 in goats and relatively high estimate was reported by Ghafouri-Kesbi and Gholizadeh [12] in sheep than in the current study. However, the h²_a estimates for RGR1, RGR2, and RGR3 in this study are higher than the estimates reported by Ghafouri-Kesbi and Gholizadeh [12] and Ghafouri Kesbi and Rafie-Tari [11].

The maternal heritability (h²_m) of GE1 in this study is well consistent with the report (0.10) of Mokhtari et al. [13]. However, Ghafouri-Kesbi and Gholizadeh (12] noted a lower estimate (0.04) for GE1 than the current result in Baluchi sheep. The estimate of h²_m for RGR1 is higher than the estimate reported by Ghafouri-Kesbi and Gholizadeh [12]. The higher h²_m estimate for GE1 and GE3 than h²_a indicates that genes controlling maternal performance (uterine nourishment, milk production, cytoplasmic effect, and mothering ability) had a higher influence on these traits than the genes carried by the kids. Likewise, genes contributing to maternal performance had a significant contribution to RGR1 and RGR2. The influence of the maternal genetic effect could be attributed to differences in the quality and capacity of the uterine space for the growth of the fetus [28]. Thus, a lack of consideration of maternal genetic effects makes it difficult to select superior goats, as the true genetic potential of kids is masked by maternal performance. Therefore, maternal effects should be included in the model to obtain accurate estimates of genetic parameters, to have accurate estimates of breeding values, and to be useful for setting selection strategies for genetic improvement.

Higher and negative additive-maternal correlations (r_am) were noted for growth traits and Kleiber ratio by different scholars [13, 27, 29–31]. The negative r_am estimate may be due to natural selection for an intermediate optimum [32], poor data structure, true genetic antagonism [18], and negative environment correlation between the dam and offspring [33]. If it is due to true genetic antagonism, the negative estimate implies that selection for increasing these traits in kids unfavorably affects the maternal ability of dam for these traits. Thus, the selection program should consider both direct additive and maternal genetic effects to maximize genetic gain. In addition, Gutierrez et al. [34] suggested that the antagonistic relationship could be compensated by improving management practices and using supplementary feeding.

About 60 to 80% of the cost of animal production is feed cost [35]. Moreover, nowadays, the land use pattern has changed, i.e., grazing land is shrinking at the cost of extensive crop production, and the goat grazing/browsing resources are disappearing at a fast rate [36]. In this situation, selecting efficient animals would be imperative to ensure the sustainability of goat production. In small ruminants, an increased growth rate up to market weight with relatively little increase in mature size could help to improve production efficiency by reducing production costs. This can be achieved by selection based on relative growth rate [3, 11] and growth efficiency. However, the heritability estimates for relative growth rate and growth efficiency in this study suggest that genetic progress through selection would be slow under prevailing management conditions.

When there is a lot of residual error variance in a trait, heritability is not a viable way to measure the amount of genetic variation in the variance of a trait [5]. In such a situation, Houle [37] suggested using additive coefficients of variation, which is a measure of additive genetic variation that standardized by the trait mean and thus independent of other sources of variance, unlike heritability [38]. The moderate genetic coefficient of variation for most of the investigated traits in this study shows the possibility of including efficiency-related traits in the selection index of the breeding program to improve the efficiency of goat production. Besides heritability and genetic variability, the economic importance of a trait and its genetic correlation with other traits are important to allow a trait to be included in the selection index [39]. According to Cassel [40], improvement through mass selection (i.e., selection based on own performance) may be difficult if the heritability estimate is <0.15. In this situation, selection of goats using single source of information may not be sufficient and thus, selection using breeding values estimated from different sources of information would be important.

The genetic correlations of GE1-RGR1, GE2-RGR2, and GE3-RGR3 in this study are higher than the estimate (r_G = 0.429 to 0.654) reported by Ghafouri-Kesbi and Gholizadeh [12]. This means selection for higher growth efficiency would positively influence the relative growth rate of goats. The antagonistic genetic correlations among pre-weaning and post-weaning efficiency-related traits are in agreement with several studies [5, 11–13]. Therefore, selection for GE1 and RGR1 could not positively impact post-weaning efficiency traits, i.e., selection for more efficient animals in the pre-weaning age would tend to decrease the growth efficiency in the post-weaning period. This could be explained by the compensatory growth of kids. This result suggests that not all efficiency-related traits can be genetically improved concurrently through selection, and thus selection should focus on the trait of the highest importance. Nearly six months of age was when farmers sold crossbred goats. In such a scenario, improving the growth rate till the marketing age could increase Boer crossbred goat productivity efficiency.

Genetic trend

There are no reports of genetic trends for efficiency-related traits of goats in the literature so far. Mahala et al. [31] noted positive genetic progress for other pre-weaning efficiency-related traits such as absolute weight gain (0.5477) and Kleiber ratio (0.0299). Contrary to the current study, the negative progress for absolute weight gain (-0.0451) and Kleiber ratio (-0.0065) during the post-weaning age (six months to yearling) was reported by Mahala et al. [31]. The level of selection implemented and the culling rate of animals as per their estimated breeding value explain being favorable and unfavorable genetic progress. In this study, selection and culling were not that much strong, i.e. selection was conducted based on their physical performance, and small numbers of goats were culled due to their poor performance. Thus, these issues could explain the observed undulating genetic trend for investigated trends across years. The absence of a selection of dams based on breeding value could be the possible reason for observed unfavorable genetic progress for pre-weaning growth efficiency, as the pre-weaning performance of kids is under the sizable influence of the gene controlling maternal ability.

The genetic parameter estimates are affected by animal management, breed/population, selection pressure, data size and structure. Hence, the study is limited to one flock, providing information about that specific flock. In addition, the data structure and sample size may have some influence on how the study’s conclusions are interpreted. Nevertheless, the data was collected by researchers managing the animals under on-station, which would have lessened its influence on estimates of genetic parameters. Anyways, the small dataset and data structure in this study could increase the standard error of the estimate. Thus, estimating genetic parameters using large datasets and good data structure would be important to improve estimation accuracy.