Estimation of variance components and associated breeding values of economically important traits in dairy cows are an important strategic step for implementing efficient breeding programs. The present study investigated genetic variability of Holstein Friesian dairy cattle using single and multi-trait models combining information of milk production and fertility traits. Heritability estimates for LMY from single and multi-traits models were moderate which corresponds with most of the literature estimates for the same breed in different production environments [5, 16, 24, 25]. The moderate heritability obtained for LMY elucidates availability of reasonable genetic variability for genetic selection to improve milk production efficiency in this population. However, it was lower than values of 0.43 ± 0.24 obtained in a Thai multi-breed dairy population [26] and 0.30 for Karan Fries dairy cattle population [27]. Although the heritability of LL was low, there was exploitable genetic variance in this trait as compared with the estimated values for Ethiopian Holstein and multi-breed dairy cattle population [16, 28]. Differences between the estimates of heritability obtained in this study and estimates from other countries are most likely caused by breed, management and climate differences affecting genetic and environmental variances. Genetic parameters are specific to a population, i.e., estimates from one population in a breed may not be appropriate when applied to other herd of the same breed or a different breeds [29]. Furthermore, different estimates of the same character on the same breed show wide range of variation, some of which may reflect real differences between populations and/or the conditions under which they are studied and difference in estimation methods [30] and may change over time due to selection and management decisions [31, 32].
The estimated repeatability for LL and LMY were small to moderate estimates, respectively (Table 2). The results presented generally within the range of previous estimates [16, 27, 33]. The moderate repeatability estimates for LMY imply that information from the first parity could be used for early prediction of EBV and cow selection. This would also improve overall performance of the herd since only good producers would be retained. Low repeatability estimates of LL, however, suggested that it is more influenced by variation in management and feeding in the given environment of a particular lactation rather than of factors associated genetic makeup of the cow [33].
Accurate stages of AFC have a significant role for low replacement costs, longer fertility lifespans, and eventually boost life time dairy production efficiency [34]. The estimated heritability for the AFC was generally high from three different models (ranges 0.53 to 0.55). These relatively high heritability estimates of AFC confirmed the presence of genetic variability in the small herd thereby proper feeding and breeding management on the farm. These results correspond with the results presented by many other authors [25, 35] who reported that AFC has a relatively high heritability value than other fertility traits. Thus, higher efficiency of genetic improvement can be achieved by genetic selection. However, the estimate from the current study was higher than that of 0.031 by Brzáková et al. [36] for Czech Holstein, 0.28 by Montaldo et al. [37] for Mexican Holstein and 0.38 for Kenyan Holstein cattle [33]. The same trait is measured in a slightly different way in some other group of animals; the estimate of heritability can be expected to be different [38]. Accordingly, Pirlo et al. [39] have pointed out that estimates of heritability of AFC are in a wide range (0.05–0.75) and that inconsistencies in estimates may be caused by confounding of sire and management effects.
Heritability and repeatability estimates of CI and DO found to be low however, the heritability of CI revealed slight improvement from univariate to multi-traits models (0.12 to 0.14, Table 2). The heritability estimates for CI mostly in the range of 0.034 to 0.17 for Friesian cattle in different production environments [6, 16, 33, 36, 40], although some have reported estimates as large as 0.35 for Karan-Fries [27]. The heritability estimate obtained in this study for DO was greater than has previously been reported for Holstein Friesians in both temperate and tropical environment [16, 24, 34, 36]. VanRaden et al. [41], in a review of estimates used for genetic evaluations of fertility traits worldwide, concluded that fertility traits in dairy cattle populations have heritability of 0.04 or less. The relatively low heritability estimates for CI and DO could be explained by large environmental variance and herd management policies. However, in this study, multi-trait analysis with milk production trait indicates slight improvement in the estimates of heritability of CI. Using such kind of analysis has a great deal of importance to provide reliable and unbiased estimates of genetic parameters as a result of the model’s ability to use extra information from correlated traits [8, 9]. Thus, evaluation of fertility with milk production traits helps to increase prediction accuracy, statistical power, and reduce selection bias made by single trait analysis [10–13]. The repeatability estimates of single versus multi-trait analysis of CI and DO were small. This is as expected because genetic and permanent environmental variances of fertility traits themselves are not expected to be different, regardless of adding milk yield in a joint analysis [6]. This implies selection of animals with early lactation record would tend to be wrong decisions.
Estimates of genetic and phenotypic correlations are considered as decisive tools for multiple traits genetic improvement programmes as different traits may be influenced by the same genes, implying that expression of one trait depends on the other. The genetic correlation between LMY and LL reported in this study implies that genes that positively influence LMY would result in longer LL; thus cows that milked for longer periods will be produce more LMY. The results are consistent with those reported in same breed in Ethiopian Holstein [16, 25] and Ethiopian multi-breed dairy cattle population [28].
The magnitude of genetic and phenotypic correlations between fertility traits ranged from low to high, where the highest positive correlations were observed between CI and DO. This high genetic correlation is explained by the fact that they represent almost the same overlapping traits which result in the genetic correlation close to 1 and this is an evidence for common genetic and physiological mechanism controlling these traits [29]. Similar findings have been reported for the Holstein-Friesian [16, 36]. Corroborating the present study, this high association revealed that the improvement of CI has a positive impact on DO in the anticipated direction. Again, CI and DO is strongly influenced by breeder’s decisions based on the milk-production level and management protocol [36]. In high-yielding cows, early pregnancy can negatively influence milk yield, so farmers in tropics postpone first service after calving to sustain their temporary livelihoods. To a certain extent these systematic effects of management contribute to high phenotypic correlations besides their genetic connections. Given this high genetic correlation between the two traits, selection based on CI alone could be practiced in production systems constrained by poor recording, as is the case in developing countries (16). Furthermore, milk production traits showed positive genetic and phenotypic correlations with CI and DO. This antagonistic relationship between production and fertility was, therefore, due to both genetic and environmental factors. Several authors have support the statement that animals with higher production have the poorest reproduction [16, 24, 33]. Interestingly, in this study, the negative correlation between AFC and LMY suggests that selection for milk production contributed to early calving of heifers. A high heritability estimate (0.54) and favorable genetic and phenotypic correlations of AFC with milk production traits indicates that there is potential for improvement of this trait in tropical environmental through selection. A similar correlation was reported for the same breed in tropics (16). On the other hand, a high and positive correlation was reported for Kenyan Holsteins [33].
In fact, the efficiency of selection procedures could be visualized by the determination of genetic trend and it gives us an indication of the amount of the genetic changes of the traits since the start of the genetic improvement program. The overall genetic trend of production traits with various ups and downs during different years depicted a positive trend. Similar to this, a positive genetic trend were reported for Kenyan Holsteins [33] and Ethiopian multi-breed dairy cattle population [28]. In Ethiopia, the breeding sires imported from different sources aiming to boost milk production (16). As a result, across the study years, selection program implemented in this population was able to identify sires with the best combination of genes for milk production (15.80 kg year-1 milk gains). However, the overall genetic trend for AFC was unfavorable in the studied herd. This might be due to poor breeding management and long term phenotypic performance based culling practice of heifers. Conversely, higher negative genetic trends of -0.34 month year-1 and − 0.21 month year-1 reported for the US Holstein [42] and Colombian multi-breed dairy cattle’s [43]. In line with this study favorable EBV for CI and DO were reported for the same breed and production system [33]. This can be interpreted as improvement in milk production was accompanied by a reduction in calving interval, showing that higher yields per animal were due to factors other than an increase in the lactation length. In short, the performance traits in the said herd would need further improvement. It is therefore, imperative to emphasise improvement in management and breeding evaluation techniques to make corrections on the possible mistakes made by wrong phenotypic appraisals.