The effects of breeding and selection on lactation in dairy cattle

Abstract

Genetic selection to increase milk yield has been very successful. Genetic selection can be used to alter the shape of the lactation curve, manufacturing properties of milk, and composition associated with human health. Mid-infrared spectral data has potential for enabling low-cost, high-throughput phenotyping. There remains a disconnect between farmers’ investments in improved milk properties and payment for that milk. Lactation is a dynamic process of milk production to provide nutrition and immune benefits to the offspring while ensuring maintenance requirements of the mother are met (Darmon, 2009; Strucken et al., 2015). In dairy farming, milk is also the main product marketed for human consumption (Darmon, 2009) which is why farmers have worked to increase milk production over time. Milk production of domesticated dairy cows can reach 30–50 L per day or 6,000 to 12,000 Lover a lactation compared to 8–10 L per day and <1,000 L over lactation in feral cows (Webster, 1995). Milk production in cows follow a dynamic curve starting with an initial rapid increase in milk yield in early lactation (approx. 0–45 days-in-milk, [DIM]), followed by a peak (approx. 46–55 DIM), and then a slow decline in milk yield during late lactation (approx. 56–340 DIM) (Strucken et al., 2015). Factors such as peak milk production, persistency and lactation length determine how the total amount of milk produced is distributed over lactation (Muir, 2004). Since 1960, the amount of milk produced per cow has almost tripled (Figure 1) which is due in part to genetics (reviewed by Strucken et al., 2015; Miglior et al., 2017; Brito et al., 2021). Selection for production traits in Canada initially focussed on milk yield and milk fat yield (reviewed by Miglior et al., 2017). However, as technology developed also selection started to focus on milk protein yield and the shape of the lactation curve. In particular, the lactation curve, persistency (e.g., persistency of production or rate of decline), week of peak yield and peak yield were considered (reviewed by Miglior et al., 2017). Different milk production traits are now considered in selection indices worldwide to provide a balanced breeding goal where milk production is no longer the sole objective but also includes, for example, health and fertility traits (Miglior et al., 2017; Cole and VanRaden, 2018). The size of the US (left) and Canadian (right) national dairy herds in millions of cows (left y-axes) and the average milk production per cow in kg/y (right y-axes) between 1970 and 2020 for US and Canadian herds (data from United States Department of Agriculture, Economic Research Service & Agriculture and Agrifood Canada, respectively). Canadian data reflects only animals on herd recording. For the purpose of management decisions and dairy cow genetic evaluations, standardized lactation lengths of are used to estimate lactation yield. The International Committee for Animal Recording (ICAR) guidelines for genetic evaluation in dairy cows has standardized lactation length to 305-d (ICAR, 2021), and has adopted guidelines and approved methods for computing 24-h yield and accumulated lactation yield (ICAR, 2022). For example, in Canada the production traits that are used in national indices include 305-d fat yield, 305-d fat percentage, 305-d milk yield, 305-d protein yield, 305-d protein percentage (Oliveira Junior et al., 2021). With most breeding values estimated using a 305-day lactation, breeders are making selection decisions based on static values (Strucken et al., 2015). Genetic evaluation of lactation yield is based on the use of models that incorporate relationships from pedigree information to predict the genetic merits of animals using the best linear unbiased prediction (BLUP). Nowadays, these evaluations use both pedigree and genomic information (i.e., genomic selection) that better accounts for the realized relationships. Lactation yield evaluations were first obtained by analyzing 305-d yields (Voelker, 1981; Pander and Hill, 1993), which were accumulated from the linear interpolation of few weekly or monthly test-day (TD) records (Olori and Galesloot, 1999). Test-day records provide information on daily lactation yield (including milk, fat and protein production) collected from lactating cows by an official milk recording officer. These TD records form the official records of the cows and are usually obtained at monthly intervals through the lactation period; hence, a cow ends up with an average of ten records per lactation. Predicted total lactation yield (305-d) from TD records assumes that the records are representative average of each cows’ production in the sampled period, which may not always be the actual case (Swalve, 1995). Therefore, the true yield of a cow might be different from the accumulated 305-d yield predicted from TD records (Anderson et al., 1989). Despite the differences between the true yield and 305-d yield, several statistical methods have been used to model 305-d yield predicted from TD records (Wiggans and Van Vleck, 1979). These statistical models typically treat 305-d yield as a single trait, resulting in relatively simple models with modest computational requirements. The principal drawback of this statistical model is that it only accommodates a common environmental effect associated with all the records despite the fact that different lactation stages (i.e., trimester of lactation) could be subjected to specific environmental effects and it assumes that the lactation curve is constant for all cows (Shanks et al., 1981). However, during the different stages of lactation, a cow’s yield can be influenced by metabolic disease, climate variability, feed changes and stage of pregnancy. Accounting for these specific effects in the model for each individual yield analysis representing the different lactation stages could allow for improved models. Several authors have suggested the use of a multiple-trait model, in which each lactation record is classified as a different trait (Pander and Hill, 1993; Misztal et al., 2000). Alternatively, lactation records could be considered as repeated record on the same traits and analyzed using a repeatability model (Ptak and Schaeffer, 1993; Swalve, 1995). One of the weaknesses of the repeatability model is that it assumes lactation records collected at different times have a genetic correlation of unity (Jensen, 2001). In addition, for the multiple-trait model where lactations are treated as different traits, the model could be over-parameterized and pose computational challenges. To address the disadvantages presented by the repeatability and multiple-trait models, a random regression model was proposed for longitudinal traits like lactation yield that are modeled by TD records over the course of lactation (Schaeffer and Dekkers, 1994). Random regression models accommodate the structured covariance pattern that exists among the repeated records, adequately adjust for the specific environment effect of the cows over the stages of lactation (using days in milk) and are less over-parameterized when compared to multiple-trait models. The main drawback of this methodology; however, include the large number of records required and the complexity of the models. Persistency, which can be measured based on ratios of partial and total yields, variation of yields during lactation or the shape of the lactation curve (Gengler, 1996), could give a more accurate representation of the dynamic nature of the lactation. Furthermore, it has been suggested that cows may benefit from extended lactation lengths to reduce the risk of challenging transition periods (e.g., van Knegsel et al., 2022); cows with a high persistency may be better suited to these longer lactations. Furthermore, with the increased use of sexed semen resulting in many more female replacements available (>50% of calves born are female in such herds), a rapid turn-around time from lactation to lactation is no longer necessary in nongrazing systems. Longer voluntary waiting periods have been reported to improve overall reproductive success without reducing the number of replacements available (e.g., Ma et al., 2022). Cows with a high persistency tend to produce less milk than expected in early lactation but more milk than expected during later lactation (Cole and Null, 2009). Selection for increased persistency could flatten and extend the lactation curve (Muir et al., 2004), reducing the need to dry off high-producing animals and ultimately reducing the amount of nonproductive time an animal spends in a herd. Consequently, persistency is of economic importance as well as it impacts fertility, health, and feed costs (Dekkers et al., 1998). Genetic variation in persistency also reduces the accuracy of genetic evaluations for 305-d lactation yield (Arendonk et al., 1995). However, while genetic evaluations have been developed for lactation persistency, providing separate estimated breeding values for persistency is not common in all countries (Jamrozik et al., 1997; Cole and VanRaden, 2006; Cole and Null, 2009; Strucken et al., 2015). According to Gengler (1996), heritabilities for various definitions of persistency range between 0.01 and 0.30. Others have reported heritabilities ranging from 0.09 to 0.30 (Jamrozik et al., 1998; Muir et al., 2004; Cole and Null, 2009). However, it is important to recognize that not all definitions of persistency have the same interpretation. Test-day models of persistency are typically estimated as the difference in yield between two arbitrary lactation days, one near the peak and one where the lactation curve flattens out in late lactation. If an adjustment for level of production is not made, this can result in high phenotypic and genetic correlations of yield with persistency (Dekkers et al., 1998; Swalve, 2000; Jakobsen et al., 2003; Muir et al., 2004). This presents a challenge due to the difficulty of altering yield without altering the shape of the lactation curve. Cole and VanRaden (2006) demonstrated that values produced by the best prediction of persistency have low genetic correlations with milk, fat, protein, and somatic cell score and are phenotypically uncorrelated with yield. Selection for milk production later in lactation (changing the shape of the lactation curve) also may provide an opportunity to increase milk production during a time in an animal’s life when energy allocation is possible. Research efforts have focused on predicting lactation curves using different statistical models and, more recently, machine learning approaches (Silvestre et al., 2006; Ehrlich, 2011; López et al., 2015; Liseune et al., 2021). Cole et al. (2009) reported that it can be challenging to compute curves with satisfactory properties when the number of available records is limited, for example, when working with numerically small breeds or when calculating curves for very specific situations (e.g., breed-parity-region). Appuhamy et al. (2009) showed that persistency is influenced by diseases, such as mastitis, in both the current and successive lactations. Energy is required for lactation and other life functions, including growth, reproduction, and health (Friggens and Newbold, 2007). These energy requirements and their partitioning changes throughout pregnancy, lactation and return to estrus cycling (Friggens and Newbold, 2007). During early and peak lactation cows typically experience an energy deficiency as they cannot meet the high energy requirements for milk production because of physiological restraints on feed intake and mobilization of body energy reserves (Strucken et al., 2015). This negative energy balance can affect large numbers of cows with between 52% and 75% of cows in excessive negative energy balance reported (Macrae et al., 2019). While there have been improvements in dairy cattle nutrition management (van Knegsel et al., 2007; Esposito et al., 2014), strong genetic selection on mainly one aspect, lactation, has changed how energy is allocated for other functions (Friggens and Newbold, 2007). This has led to challenges with metabolic disorders, fertility, immunity, and heat tolerance. These antagonistic relationships between lactation and functional traits have been explained in part by unfavorable genetic correlations, which means that including them in a selection index allows for continued, although slower, genetic progress in milk production, while simultaneously maintaining the resilience of animals (Pryce and Harris, 2006). Energy mobilization can be observed in dairy cattle through the recording of an indicator trait, body condition score. In Canada, this score is categorical ranging from 1 to 5, where animals with the lowest score are thin with no fat reserves and animals with the highest score are over-conditioned (Mongeon et al., 2020). The drop in body condition score is indicative of the mobilization of energy reserves, which can be seen, for example, at the beginning of a lactation period. Several studies have observed that heifers and cows that produced more milk at the beginning of a lactation period, often had larger decreases in body condition scores during this same period (Pryce and Harris, 2006; Manríquez et al., 2021; Tribout et al., 2022). This rapid mobilization of energy reserves can result in debilitating metabolic disorders in the short-term and have long-term effects on fertility and immunity. Researchers have investigated the relationship between body condition score and female fertility, finding that genetic selection on higher milk production in early lactation likely exacerbates the already existing negative energy balance, seen in the larger decrease in body condition score, and having a negative effect on fertility (Van Arendonk et al., 1991; Dechow et al., 2002; Berry et al., 2003). With larger decreases in body condition score and body weight shortly after calving being associated with metabolic disorders, including ketosis, displaced abomasum and metritis, milk fever, ketosis, and fatty liver (Frigo et al., 2010; Koeck et al., 2012; Stevenson et al., 2020). In addition, higher milk yield was associated with hyperktonemia, and fat and protein production were associated with ketosis and displaced abomasum (Koeck et al., 2013). Greater body weight and a smaller change in body weight were favorably genetically correlated with infectious diseases and other diseases (Frigo et al., 2010) and milk yield was unfavorably correlated with ovarian cysts and clinical mastitis (Gernand et al., 2012; Koeck et al., 2014), likely because of the negative energy balance too. Even more far-reaching effects were observed in a study of the effect on females born to mothers lactating while pregnant, where the greater the milk yield of mothers, the larger the negative effects of prenatal programming on milk production, metabolic efficiency and lifespan (González-Recio et al., 2012). Muir et al. (2004) estimated genetic correlations between persistency and reproductive traits that ranged from −0.17 to 0.43. In addition to increasing the overall amount of milk, fat, and protein, selection for desirable milk composition and quality is also possible. Some aspects of this selection are subject to fundamental laws of biochemistry—the osmotic properties required for milk secretion in the alveoli of the mammary gland are fixed—but are For example, many desirable of milk are (e.g., and are associated with both human health and desirable manufacturing properties for is also to the fatty of milk to more it with human et al., 2009; et al., 2022). milk quality traits, such as somatic cell provide an indicator of the quality of the milk, but also the health of the and have been in most genetic selection However, for traits is often by to and milk recording are based on relatively simple of milk fat true protein, other not fat, and somatic cell Some milk also have the to record spectral fatty composition typically is using which is and has low information milk protein composition is usually obtained through which has While has been proposed as an that is and it often has To these a small number of could be with a larger number of using models et al., The use of these in breeding allow breeders to produce more milk and with a more desirable of specific over time. Furthermore, it allow production of milk that has more desirable properties for both and making selection for associated traits like (e.g., et al., 2022). of milk to predict milk of milk to predict milk Genetic selection for improved and cow fertility remains a challenge because the traits used focus on days and rate (e.g., VanRaden et al., 2014), both of which have low While these are to on a they are likely to be for the actual of traits, such as reproductive size and score et al., and and fertility may produce of genetic than fertility traits because they have higher fertility remains a many aspects of production a genetic et al., but fertility, measured as has a near and However, correlations of with female fertility traits also are and there is no by which a cow’s lactation can the reproductive of to climate change and other this also be considered for selection as it has been that production selection reduces heat (e.g., and 2000; et al., while changes in milk yield, lactation or persistency may also alter et al., 2012). might be that the to heat is not per to but there are of cows in that are relatively and modest in their to and produce milk may to decreases in cattle in environmental and increased quality of life for from are and are influenced by the the and among all these et al., 2018). are correlated with milk fat production, selection for is also likely to affect milk This can be through use of uncorrelated et al., 2021), but it may not be through genetics to produce cows with low total However, the amount of by dairy cows per of milk, fat, and protein produced to While current are focused on because of the in production, there also may be to reduce the amount of in and in US genetic selection indices and the on each somatic cell score, and body weight calving calving calving cow cow health feed early first and Predicted and in US genetic selection indices and the on each somatic cell score, and body weight calving calving calving cow cow health feed early first and Predicted and Selection indices to with focus from production and fertility a on cow selection be with from all of the (Cole et al., 2021), but their indices for (e.g., et al., 2020). However, the milk recording has not been as in to meet current the of information between different in the US milk The in the current and of but that only one of and the of milk recording genetic evaluation need to be to data from more of such information is in and milk (e.g., manufacturing that not information to a (e.g., and that not in milk recording While data from is the for milk can be to records that are collected of data between different of a national milk recording including and test-day milk collected on milk composition information produced by milk standardized records by dairy records and breeding values by the genetic evaluation selection for traits has focused on milk and fat and protein yields, with to persistency of lactation. This has been very although challenges have been the of yield with however, more is possible. is being in fertility traits that are to actual than or days and there are to produce milk with better manufacturing properties and However, farmers are to for traits that have for the milk then there to be a for them to of the of that milk. This challenge exists many traits, including that have high recording and it is not that farmers all the of data while of the of the milk and Milk in dairy countries is on that not only but also In many however, farmers are only for of the specific payment many for management and breeding for investments in and milk (e.g., et al., 2019). This to the that there is a disconnect between farmers’ investments in improved milk properties and payment for that milk. in milk that are as such as the of has not led to to the a to the process that they for This it for farmers to in feed and other because there is no that they can also is important to that the environment in which the cow is and the management available in that are also genetics In this management all aspects of animal including and effects of heat include and dry fertility, and been for et al., is also that environmental have effects on animals that during (e.g., et al., 2022). is to can the effects of heat on milk production et al., and has focused on in the of (e.g., et al., and 2003; and but there are that are to be improved by breeding In particular, to the potential of genetics lactating cows be of maintaining dry and intake to lactation, and to Selection for greater can these but it cannot the of the environment on One of this is that there is likely no cow for in the genetics be more to the in which be Genetic selection to increase milk yield has been very and there is no to that selection are being While most selection has focused on yield, there are also for the of lactation manufacturing properties of milk, and composition associated with human health. Mid-infrared spectral data has potential for enabling low-cost, high-throughput and genetic evaluations based on that technology are now the However, there remains a disconnect between farmers’ investments in improved milk properties and payment for that milk that may the potential of these Cole is the for Research and the of dairy and cattle semen in the in was a Research and Research of Animal and where more than Cole on for the of Animal and the International Committee for Animal focus on genetic of fertility, health, and traits in dairy of selection for dairy pedigree and of genetic in is a at the of is the of genetic to the genetic potential of animals and how this a in the health and of these goal is to that in animals genetics to meet the of the in a and has in working with and using both pedigree and genomic is a Research at the for Genetic of at the of has been a in several countries the and where has used of genetics and computational to study genetic and in van is a at the of the for Genetic of and the for the of Animal the and over a range of animal health and of animal production and animal including and more both fundamental and aspects of animal to animal to production is a and Canada Research in at the of with to incorporate breeding for improved health, and of has in the of genetics and statistical as it to the genetic and genomic evaluation of on various and and in Canada and and has on animal and in and This was by and Research of Canada and

Topics & Concepts