1. Introduction
Reproductive success is one of the main drivers of profitability in dairy herds. Every day a cow remains non-pregnant beyond the established voluntary waiting period results in reduced profit. Pregnancy loss has a substantial negative impact on reproductive efficiency and, therefore, on-farm profitability [
1]. The Committee on Bovine Reproductive Nomenclature [
2] defines embryonic loss as any pregnancy that ends prior to day 42 of gestation with pregnancy losses after day 42 considered to be fetal loss or abortion. Using a comprehensive breeding, replacement, and economic model, De Vries et al. [
3] estimated the value of a new pregnancy ranged between USD -14 and USD 551, depending on DIM at conception, milk yield, and lactation number. For low-producing cows that also get pregnant late in gestation, the value of the pregnancy was the lowest. In addition, the cost of pregnancy loss was also low among low-producing cows, because many of these cows would be culled, even when pregnant, due to low profitability. In contrast, cows with average and high milk yield that experienced a pregnancy loss represented an economic loss even greater than the value of the pregnancy. This is mainly because of increased days open, the opportunity cost of delaying the next lactation, and increased involuntary culling of profitable animals. The cost of pregnancy loss increased with the milk production level and month of gestation, being as high as USD 1373 for a high-yield primiparous cow that aborts at month 7 of gestation. Moreover, this study assumed that once aborted cows would behave as non-pregnant cows; thus, it can be speculated that the real cost of abortion would be even greater. The actual value of pregnancy and pregnancy loss changes with market values of culled cows, heifers, feed, milk, etc., but most research and simulation models support the concept that abortion reduces profitability [
4]. Wijma et al. [
5] reported that cows that abort have a greater risk of a new abortion if they conceive again, a reduced risk of pregnancy by 400 DIM, and a greater risk of exiting the herd than non-aborted cows. Furthermore, abortions can cause further health problems, such as retained placenta, metritis, endometritis, and pyometra, with the consequent impact on reproduction, milk production, and productive life [
6,
7,
8,
9]. Santos et al. [
1] estimated that 60% of pregnancies are lost from conception to term. Pregnancy losses in high-producing dairy cows range from 5.4 to 12.8% between the first and second month of gestation [
10,
11,
12], and from 1 to 3.3% during the third month of pregnancy [
13,
14,
15]. Although there are not many reports of abortion rates from 90 days to term, it was estimated at 3–5% by Hovingh [
16]. Diagnosis of the etiological cause of abortion represents a real challenge for producers and veterinarians. Among the possible causes of abortion are genetic abnormalities, heat stress, toxic agents, bacteria, viruses, and protozoa, among others [
16]. Close to 50% of the fetuses submitted to diagnostic labs yield a confirmed diagnosis, but that means that the other 50% usually remain undiagnosed [
1]. Management strategies, such as vaccination, biosecurity measures, use of mycotoxin sequestering agents, and heat stress mitigation, among others, help to reduce but not eliminate the incidence of abortions.
In the last few decades, advances in the understanding of reproductive physiology, the development of new reproductive management strategies, and genetic selection have allowed the dairy industry to drastically reverse a negative trend and increase pregnancy rates [
17]. Despite this, pregnancy losses continue to be a problem for which no clear solution has been developed. In this regard, there is an established and ongoing effort by research groups to quantify the genetic component of abortion in cattle. Several recessive haplotypes and mutations that cause embryo or fetal loss, as well as stillbirth, have been identified in the genome of dairy cattle [
18,
19,
20]. The discovery of these haplotypes allows for selective breeding of animals that carry these mutations to avoid homozygotic fetuses (and, thus, prevent pregnancy losses) and reduce the prevalence of these haplotypes in future generations [
20]. Precision mating based on genomic data to control these losses can have an annual impact of USD 11,000,000 on the USA dairy industry [
21]. Sigdel et al. [
22] identified the presence of genomic markers associated with pregnancy loss in dairy heifers and lactating dairy cows. Moreover, Gershoni et al. [
23] identified candidate markers to predict the risk of early abortion in Israeli Holstein cattle. This growing evidence that abortion incidence has a genetic component supports the fact that it can be controlled, at least in part, through genetic selection.
Considering the economic impact of pregnancy losses discussed in the previous paragraphs, the development of a multi-trait selection index that includes abortion risk would help select and breed more profitable cows. Selection indices allow dairy breeders and producers to select animals based on a combination of different traits, each of which receives a relative weight. Selection indices provide a way to combine information about many traits into a single number that can rank animals and inform breeding decisions. Initially, selection indices were built accounting only for production traits, hence indirectly selecting against fertility and health traits [
24,
25] due to the negative correlation between the traits. In the last few decades, the dairy industry has become more conscious that boosting the fertility and health of dairy cows is paramount to maximizing profitability, reducing antibiotic use, assuring animal welfare, and complying with sustainability standards. As a result, modern selection indices include not only production traits but also fitness traits. Attending to these needs of dairy producers and the dairy industry, Zoetis developed the Dairy Wellness Profit
TM (DWP
$) index in 2016. DWP
$ is an economic multi-trait selection index that was formulated to estimate the potential lifetime profitability an animal would generate under dairy economic conditions in the United States and includes cow and calf wellness, production, fertility, functional type, longevity, livability, calving ability, and milk quality traits [
26,
27]. This index has been validated in commercial dairy farms to select more profitable and healthier cows [
27]. In 2020, DWP
$ was updated to include additional traits shown to impact the lifetime profitability of a dairy animal; one of those traits was cow abortion.
The objectives of this study were to (1) describe the development of the genomic predictions for cow abortions in Holstein dairy cattle based on producer-recorded data and ssGBLUP methodology and (2) evaluate the efficacy of genomic predictions for cow abortions in commercial herds of US Holstein cows using data from herds that do not contribute phenotypic information to the evaluation. In the current study, we hypothesized that cows with the highest genetic risk for abortion would have a higher phenotypic incidence of abortion.
4. Discussion
Although genetic and genomic associations with abortion in cattle have been previously described in the literature [
22,
23,
35,
36], to the best of our knowledge this is the first report to describe abortion risk in a genetic evaluation and demonstrate efficacy to predict actual abortion events using real commercial herd data, and subsequently include it in a lifetime merit selection index.
The additive genomic variance observed in this study was greater than previously reported by others for abortion and/or embryo loss [
23,
35,
36], which explained more than 12% of the variability in abortion incidence. The heritability estimates for abortion were similar to those reported in previous studies [
22,
23,
35,
37], although they were lower than the estimates reported by Bamber et al. [
38] and Sigdel et al. [
22] (multiparous cows only). Even though these studies, including this one, have used different approaches to evaluate genomic markers of abortion risk, the fact that heritability and additive genetic variance are greater than 0 supports the hypothesis that it is feasible to select toward reduced abortion risk based on genomic evaluations.
Incidences of abortion (42 to 260 days after AI) in the genetic evaluation (11.7%) and the validation population (12.8%) were similar to previous reports [
1]. Abortion incidence distribution across gestation periods for the validation data set was also in accordance with what has been previously described by others, with greater incidence in the first trimester (i.e., 58%). The incidence of abortion increased with parity within the validation population, similar to what was reported by Keshavarzi et al. [
39], but different from what was observed by others [
40]. Regardless, results presented in
Figure 3 clearly indicate that for lactations 1 to 4, the predicted probability of abortion increases with lower Z_Abort STA values. It can be appreciated that the magnitude of this effect is greater for more mature cows. This may have an important economic significance because mature cows that get pregnant early in lactation (and maintain that pregnancy to term) are more profitable [
41], and abortions of those tend to represent the greatest economic cost [
3].
The influences of the HH defects in this study were minimal, with only four abortions involving mates of carriers. The low frequency of HH carriers among the genomically confirmed AI service sires, the lack of difference in the proportion of carriers among cows that did or did not abort, and the similarity in the frequency of carrier animals through Z_Abort quartiles for most HH allow us to assume that the abortion incidence was not influenced by the matings of HH carriers. Despite this, HH2 and HH4 were (or tended to be) less frequent in the best quartile for Z_Abort STA. This is not surprising because although HH need to be in a homozygous state to be lethal, the possibility that these markers are also related to some genes with an additive effect cannot be ruled out. Furthermore, we cannot disregard the possible existence of other unknown HH that could be associated with Z_Abort predictions.
Genomic STA for Z_Abort showed a positive correlation with other wellness traits related to reproductive efficiency and animal health. The positive genetic correlation between reduced abortion risk and reduced risk of twin gestation has been previously described by our group [
28], and a phenotypic correlation between these two traits has been described in several reports [
42,
43,
44]. Moreover, we observed a positive genetic correlation between Z_Abort and reduced risk of retained placenta and metritis, which is in accordance with phenotypic observations where aborted cows had a greater probability of suffering retained placenta [
6,
8] and metritis [
7,
9]. Furthermore, in the present study, we observed that aborted cows had an increased incidence of retained placenta (3.0 vs. 1.2% for aborted and not aborted cows, respectively) and metritis (7.6 vs. 4.6% for aborted and not aborted cows, respectively). These observations, in addition to a positive genetic correlation with a genomic prediction for cow livability, suggest that selecting cows with reduced abortion risks would also contribute to a reduction in the incidence of other health events that directly impact reproduction, cow livability, and profitability.
Interestingly, although previous studies have shown an increase in pregnancy loss in cows suffering from mastitis [
39,
45] and we have observed a similar trend (15.1 vs. 11.2% for cows with and without a mastitis event, respectively), the observed correlation between Z_Abort and Z_Mastitis is weak. An explanation for this can be that most of the studies reviewed by Dahl et al. [
45] were focused on the pre-fetal period (i.e., the first 45 days of pregnancy) whereas we only evaluated those pregnancy losses that occurred beyond day 42 of gestation. The positive correlation with Z_Respiratory might be explained by increased resistance to respiratory viruses, which are usually associated with infectious abortions. This holds for Z_Calf Respiratory, although in a smaller magnitude. Interestingly, there was a positive genetic correlation between Z_Abort and Z_Calf Livability. Previous studies have suggested that markers for abortion are related to genes governing the development of the placenta and fetal immune system, among other functions [
22,
37]. Therefore, it is possible that cows with high risks of abortions will give birth to calves that have compromised in utero development and with a greater risk of early death or disease. This may have an important implication for the selection of recipients for high-value embryos, because, besides increasing the probability of the gestation to term, it would increase the probability of survival for the calf.
The positive genetic correlation of Z_Abort with DPR and CCR gPTA was relatively low, but it should be noted that the correlations are not adjusted for the reliability of the predictions (42% for Z_Abort). The magnitude of the correlation between DPR gPTA and Z_Abort STA and the observation that the maximum overlap between quartiles is 8% (
Table 10) suggest that these two traits are predicting the risk of abortion through different mechanisms. This confirms that Z_Abort adds important and valuable information to abortion risk prediction and selection for more fertile, healthier, and profitable cows. The negligible magnitude of the correlation with HCR is within expectations because only lactating cows were included in the current evaluation.
There is strong evidence supporting that high-yield dairy cows have lesser plasma concentrations of progesterone [
46], which is associated with decreased embryo quality and impaired embryonic and fetal development. Thus, the negative correlation between Z_Abort and important production traits (milk, fat, and protein) is not surprising. On the other hand, neither we nor other authors who analyzed the effect of milk production on abortion incidence observed a significant effect [
39,
40]. Considering that most evidence regarding the increase of pregnancy loss with high milk yield points towards a large effect on early embryo mortality and the negative correlation of Z_Abort with production traits, it could be hypothesized that there might be an association between Z_Abort and early embryo loss. It is important to note that, although in the current study the days of pregnancy at the time of abortions were registered as after day 42, in most of the cases it was not possible to determine the exact day of abortion because during the first trimester most abortions are diagnosed at pregnancy reconfirmation (usually after day 50) or because cows are observed in estrus. Therefore, some of these pregnancy losses diagnosed as abortions could actually be embryonic losses. Considering that close to 60% of abortions occurred during the first 90 days of gestation, it can be hypothesized that many of the abortions predicted by Z_Abort were actually embryonic losses. Although this hypothesis requires further research to be confirmed, it is of great scientific and economic significance because the prediction and control of early pregnancy losses are some of the most important challenges in the dairy industry today.
The negative correlations with production traits might be a drawback for including Z_Abort as a selection trait in a genetic improvement program. Including Z_Abort (5% weight) in the DWP
$ index, along with other fertility and health traits, resulted in positive selection towards cows with fewer abortions. Estimated response to selection for DWP
$ (updated April 2022) shows that an increase of 1 SD would result in an increase of 219 pounds for Milk gPTA, 20 pounds for Fat gPTA, and 11 pounds of Protein gPTA coupled with an increase of 0.19 in Z_Abort STA (
Table A1). Conversely, the negative correlation of Z_Abort STA and NM
$ index suggests that the use of this index for selection will continue to increase the incidence of abortions.
The magnitude of the improvement gained in Z_Abort through DWP$ might seem marginal if only the female selection is considered. Having the abortion trait predictions for sire selection will likely lead to the removal of some outliers with poor genetic merit and the impact could be of greater magnitude.
Over the last decades, advancements in reproductive management and genetic selection have allowed the dairy industry to reverse a negative trend in reproductive performance and increase pregnancy rates. Despite this, controlling pregnancy loss and abortion are still a pendent assignment. The evaluation presented and validated in the present study provides the industry with a new tool that could greatly impact reproductive efficiency and cow profitability for future generations of dairy cows. To assess the impact of this trait, we performed a simulation analysis using the model described by DeVries [
47] in which we compared two 1000 milking cows herds with different abortion incidences for Lact 1 to 4. One herd was assigned an abortion incidence for the worst Z_Abort quartile and the other for the best quartile. The model estimated a USD 17,000 difference in annual profit in favor of the herd with abortion incidence that corresponded to the best 25% for Z_Abort STA. In addition, the model does not account for the increased incidences of metritis, retained placenta, and other health events related to abortion. Thus, it can be assumed that the real difference in profitability would be even greater, providing evidence of the economic impact that reducing abortion risk through genetic selection can have.