Estimation of heritability for milk urea and genetic correlations with milk production traits in 3 Danish dairy breeds

The aim of this study was to estimate genetic parameters for milk urea (MU) content in 3 main Danish dairy breeds. As a part of the Danish milk recording system, milk samples from cows on commercial farms were analyzed for MU concentration (mmol/L) and the percentages of fat and protein. There were 323,800 Danish Holstein, 70,634 Danish Jersey, and 27,870 Danish Red cows sampled with a total of 1,436,580, 368,251, and 133,922 test-day records per breed, respectively, included in the data set. Heritabilities for MU were low to moderate (0.22, 0.18, and 0.24 for the Holstein, Jersey, and Red breeds, respectively). The genetic correlation was close to zero between MU and milk yield in Jersey and Red, and −0.14 for Holstein. The genetic correlations between MU and fat and protein percentages, respectively, were positive for all 3 dairy breeds. Herd-test-day explained 51%, 54%, and 49% of the variation in MU in Holstein, Jersey, and Red, respectively. This indicates that MU levels in milk can be reduced by farm management. The current study shows that there are possibilities to influence MU by genetic selection as well as by farm management.


INTRODUCTION
Milk urea nitrogen or milk urea (MU; MUN = 0.46 × MU) are indicators of the amount of CP in the cow's diet. Dietary protein digestion results in the production of ammonia, which is converted to urea in the liver. Urea is also recycled in the rumen through saliva. Urea is primarily excreted from the body via urine but is also found in blood and milk (Moore and Varga, 1996).
The contribution of the dairy sector to the total N excretion in feces and urine in 2020 is considerable (Uwizeye et al., 2020). In the future, N excretion needs to be reduced due to European Union regulations and social demands. Measuring urea in urine is not feasible on a large scale. Spek et al. (2013) showed that urinary N and urinary urea N can be best predicted by a combination of MUN and the dietary CP concentration. Milk urea measured using infrared spectroscopy can be collected routinely on a large scale as part of the milk recording system. In combination with 3-dimensional camera technology to measure feed intake on commercial farms (Manzanilla-Pech et al., 2022), MU could be an interesting phenotype to reduce environmental eutrophication with N at farm level.
Milk urea heritability estimates reported in the literature show a wide range from 0.14 up to 0.59 (Wood et al., 2003;Mitchell et al., 2005;Miglior et al., 2007;Stoop et al., 2007;Rzewuska and Strabel, 2013) in Holstein cattle. For Jersey cattle and Brown Swiss cattle, heritabilities of 0.19 and 0.20 were reported, respectively (Beatson et al., 2019;Bobbo et al., 2020). Genetic correlations between MU and milk production trait are very low (Wood et al., 2003). In the literature, many papers state that breeding for reduced MU would be a good tool to reduce N output; however, it was shown that breeding values for MUN had no effect on the total N excretion of cows but rather influenced different excretion characteristics per urination event (Marshall et al., 2021).
Since January 2019, MU is measured for all animals participating in the Danish milk recording system. The genetic parameters for the 3 main dairy breeds, Danish Holstein, Danish Jersey, and Danish Red, are unknown. Therefore, the objective of this study is 2-fold. The first objective is to investigate the phenotypic and genetic variation of MU and estimate the heritability and the second objective is to estimate the phenotypic and genetic correlations to production traits in the 3 main dairy cattle breeds in Denmark.

Data
Phenotypic Data. No animal experiments were performed in this study, and therefore, approval from Estimation of heritability for milk urea and genetic correlations with milk production traits in 3 Danish dairy breeds A. J. Buitenhuis 1 * and N. A. Poulsen 2 the ethics committee was not required. The data used in this study were collected as part of the routine milk recording in Denmark (RYK/Viking Danmark, Aarhus N, Denmark) between January 2019 and March 2020. Milk samples were collected on regular test days and were analyzed for fat, protein, and urea concentrations at a certified laboratory (Eurofins, Vejen, Denmark) using infrared spectroscopy (MilkoScan FT+/FT6000, FOSS). Raw phenotypes were provided and stored in the Danish Cattle Database (SEGES, Skejby, Denmark). The phenotypes used for the study were milk yield (MY), fat percentage (

Analysis
Variance components and genetic parameters were estimated within breed for the milk production traits using a REML repeatability model in DMU (Madsen and Jensen, 2007): where Y ijklmn = dependent variable corresponding to the nth test-day observation of cow m at DIM k on herdtest-day l during parity j in year-season i; µ = general mean; YS i = fixed effect of year-season (i = 1, …, 16); parity j = fixed effect of parity (j = 1, …, 4+); dim = DIM modeled as a Wilmink curve (Wilmink, 1987); htd l = random effect of the herd-test-day; A m = random additive genetic effect of animal m; pe n = random permanent environment effect of animal m; and e ijklmn = random residual effect.
The variance components for MU were estimated using model 2: which is the same as model 1, but the dim k = fixed effect with 20 levels representing intervals of 15 d. Heritabilities and repeatabilities were estimated using univariate analysis. Correlations were estimated using bivariate analysis of equations 1 and 2.
Heritability (h 2 ) was calculated as where σ A 2 = additive genetic variation, σ pe 2 = permanent environmental variation, and σ e 2 = residual variation. Repeatability (r) was calculated as The proportion of variance due to the herd-test-day (h htd ) was calculated as where σ htd 2 = herd-test-day variation. The figures were made in R (version 4.2.0) using the libraries ggplots2 (version 3.3.6) (https: / / cran .r -project .org/ web/ packages/ ggplot2/ ggplot2 .pdf) and ggpubr (version 0.4.0; https: / / rpkgs .datanovia .com/ ggpubr/ ). Table 1 shows the mean and coefficient of variation for the different parities for each breed across the lactation. The highest mean MU level of 4.82 was found for Danish Jersey cows in parity 1, and the lowest mean MU level of 4.37 was found for Danish Holstein cows in parity 3. In general, Danish Holstein had the lowest MU compared with Danish Jersey and Danish Red. For Danish Jersey and Danish Red, the highest mean MU was found for parity 1 and subsequently lowered for parities 2 to 4+. Danish Holstein cows showed the same pattern, except for parity 4+, where increases in mean MU level were observed (Table 1). Figure 1 shows the mean MU content over the lactation for Danish Holstein, Danish Jersey, and Danish Red. One can see that the beginning of the lactation is Buitenhuis and Poulsen: GENETIC PARAMETERS FOR MILK UREA lower compared with the rest of the lactation, but after the onset of the lactation, the urea content has a flat curve. For Danish Holstein cattle, the lowest mean MU was found during the first 30 DIM (parity 1: 4.12 ± 1.03; parity 2: 4.07 ± 1.12; parity 3: 3.88 ± 1.11; parity 4+: 3.74 ± 1.12). This is the same for Jersey (parity 1: 4.47 ± 1.14; parity 2: 4.58 ± 1.26; parity 3: 4.39 ± 1.27; parity 4+: 4.22 ± 1.29) and Red (parity 1: 4.36 ± 1.07; parity 2: 4.36 ± 1.20; parity 3: 4.14 ± 1.21; parity 4+: 3.38 ± 1.16). Based on the least squares means, we detected a significant difference between all parities for all 3 breeds ( Figure 2). The smallest difference between parities was found for Danish Red, whereas the largest difference between parities was found for Jersey. The largest difference in least squares means is between parity 1 and parity 4 for all 3 breeds.

Variation of Milk Urea
We detected a significant year-month effect on the MU content in the milk of all 3 breeds (Figure 3). The highest least squares mean MU was observed in April for the Holstein cattle (4.78 ± 0.03). For both Jersey (5.11 ± 0.06) and Red (4.92 ± 0.05), the highest least squares mean for MU was observed in June.

Variation of Production Traits
In Table 1, the mean and the coefficient of variation is given for the production traits for each parity. The MY is highest for the Danish Holstein compared with Danish Jersey and Danish Red breeds, whereas FP and PP are highest for Danish Jersey and lowest for Danish Holstein. For all 3 breeds, the MY is increasing with each parity. Table 2 shows the heritability, repeatability, and herd-test-day effect for the Danish Holstein, Danish Jersey, and Danish Red breeds. Heritability for MU in Danish Holstein (0.22), Danish Jersey (0.18), and Danish Red (0.24) are higher than the heritability for MY (0.16, 0.14, and 0.22, respectively) but lower than the heritability for FP and PP.

Heritability, Repeatability, and Effect of Herd-Test-Day
The repeatability for MU is relatively low for all 3 breeds (Danish Holstein 0.38, Danish Jersey 0.34, and Danish Red 0.39). The highest repeatability was found for PP in the Danish Holstein breed (0.60).
For MU, the herd-test-day explained 51%, 54%, and 49% of the total variation for Danish Holstein, Danish Jersey, and Danish Red, respectively. The highest variation explained by the herd-test-day was 72% for MY for Danish Holstein, whereas for Danish Jersey and Danish Red herd-test-day explains 58% and 59% of the total variation for MY. The lowest variation explained by herd-test-day was 8% for PP in the Danish Red breed. The ratio of genetic variance to herd-test-day variance showed that the genetic effects were much larger for FP and PP in all 3 breeds, whereas the opposite was shown for MU and MY. Table 3 shows the phenotypic and genetic correlations among MU and the production traits for all 3 breeds. The phenotypic correlations between MU and Buitenhuis and Poulsen: GENETIC PARAMETERS FOR MILK UREA

DISCUSSION
In our study, we showed that the MU [MU(mmol/L) × 60.06 = MU(mg/L)] was consistent over parities for Danish Holstein and Danish Red. This agrees with other studies of Holstein cattle where the MU content of the milk was consistent over parities (Wood et al., 2003;Atashi et al., 2021;Chen et al., 2021). The Danish Jersey breed showed the highest MU compared with the Danish Holstein and Danish Red breeds for all parities but also had the largest reduction in MU from first parity (4.82) to fourth and higher parities (4.54), indicating that there are differences between breeds for MU. So far, there are no studies on the Danish Jersey and Danish Red breeds regarding their MU content. For the studies on the Holstein breed, large differences in MUN concentration exists, for example, from >20 mg/dL (Stoop et al., 2007;Atashi et al., 2021;Chen et al., 2021) to <13 mg/dL (Wood et al., 2003;Miglior et al., 2007).
In the literature, the heritability for MUN (mg/dL) in the Holstein breed is in the range of 0.14 (Stoop et al., 2007) to >0.30 (Wood et al., 2003;  For Brown Swiss, the heritability was 0.20 across parities (Bobbo et al., 2020). van den Berg et al. (2021) showed that heritability for Australian Holstein and Jersey was 0.10 and 0.15, respectively, and heritability for New Zealand Holstein and Jersey was 0.28 and 0.24, respectively. The heritabilities for MU (mmol/L) in our study estimated across parities were 0.22, 0.24, and 0.18 for Danish Holstein, Danish Red, and Danish Jersey, respectively. The proportion of the total variance explained by herd-test-day in our study was 0.51, 0.49, and 0.54 for Holstein, Danish Red, and Jersey, respectively. This is lower than the proportion of variance explained by herd-test-day of 0.58 that was found by Stoop et al. (2007) in first parity Holstein cattle, but it indicates that there is considerable influence of management factors such as diet on MU. This influence of herd-test-day on MU was also found by, for example, Wood et al. (2003) and Bastin et al. (2009). Furthermore, Figure 3 shows the least squares mean MU value changes over the year and between years for all 3 breeds. The variation over time is in line with the study of Bastin et al. (2009).
Phenotypic correlations between MU and milk production traits in this study were low for all 3 breeds. This is in line with Miglior et al. (2007). They also found a negative phenotypic correlation between MU and MY and a slightly positive correlation between MU and fat and protein percentage, respectively. In addition, Stoop et al. (2007) and Bobbo et al. (2020) also . The LSM MU is on the y-axis. The differences between parity 1 and parity 2 (1-2), parity 1 and parity 3 (1-3), parity 1 and parity 4 (1-4), parity 2 and parity 3 (2-3), parity 2 and parity 4 (2-4), and parity 3 and parity 4 (3-4) are on the x-axis.
showed low phenotypic correlations between MU and milk production traits in Holstein and Brown Swiss, respectively. The low correlation between MU and the milk production traits could be due to management when, with a given TMR, low-producing cows receive more protein than they require, leading to an excess in N in their body and, therefore, a higher MU content.
In our study, we found a negative but low genetic correlation between MU and MY, for all 3 breeds. Miglior et al. (2007) and Chen et al. (2021) found a negative genetic correlation between MU and MY for Holstein and Bobbo et al. (2020) found a negative genetic correlation between MU and MY for Brown Swiss, respectively, whereas others found a positive genetic correlation between MU and MY (Wood et al., 2003;Stoop et al., 2007). The genetic correlation between MU and FP was positive and moderate to high, whereas the genetic correlation between MU and PP was positive but low. This is in line with the results of Miglior et al. (2007) for the Holstein breed, even though the genetic correlations by Miglior et al. (2007) were somewhat higher in comparison with our study. In this study, we use a relatively simple statistical model to analyze the MU data. Others have used a Wilmink function to model the MU (e.g., Stoop et al., 2007). However, the phenotypic data of all 3 breeds showed a lower MU content in the beginning of the lactation and, once the lactation progressed, the average MU content was constant (i.e., it did not follow a classical lactation curve; Wilmink, 1987). Therefore, it was decided to fit the DIM as a fixed factor analogous to Bobbo et al. (2020). Spek et al. (2013) showed that the variation of MU can be attributable to several factors, which are related to, for example, the ability to select the sources of protein in the feed provided, the cows' appetites, the fermentation patterns in the cows' forestomachs, the metabolic efficiency in the liver, the AA uptake by the udder, and the renal urea clearance. In this study, the MU was measured in milk samples of cows participating in the Danish milk recording system. However, no information in the aforementioned factors were Buitenhuis and Poulsen: GENETIC PARAMETERS FOR MILK UREA  available. This suggests that the heritability for MU is pointing to a general heritable component rather than a heritability affecting an N efficiency component. Marshall et al. (2021) estimated breeding values for MUN; however, when cows were fed different diets of ryegrass, they found no effect of the MUN breeding value on the total daily urinary N excretion. This is in line with the conclusion of Spek et al. (2013) that MUN is heritable; when estimating breeding values for MUN, these have no relation to the efficiency of N utilization. In Denmark currently, several farms have been equipped with 3-dimensional camera technology to measure feed intake (Manzanilla-Pech et al., 2022). In combination with a feed analysis, this equipment could provide the amount of N intake of the individual cow. With this information, MU could play a role in the research of N utilization. However, more research in this area is needed.

CONCLUSIONS
The aim of this study was to estimate genetic parameters for MU in the 3 main dairy breeds in Denmark. The heritabilities of MU in the 3 breeds were moderate. Genetic correlation between MU and MY was negative and positive for FP and PP. The results of this study show that there is a genetic component underlying MU in the Danish main dairy breeds and herd practice can influence the MU content.