J. Dairy Sci. 100:10251–10271
`https://doi.org/10.3168/jds.2017-12968
`© American Dairy Science Association®, 2017.
`A 100-Year Review: Identification and genetic selection
`of economically important traits in dairy cattle1
`Filippo Miglior,*†2 Allison Fleming,* Francesca Malchiodi,* Luiz F. Brito,* Pauline Martin,*
`and Christine F. Baes*
`*Centre for Genetic Improvement of Livestock, Department of Animal Biosciences, University of Guelph, Guelph, Ontario, N1G 2W1, Canada
`†Canadian Dairy Network, Guelph, Ontario, N1K 1E5, Canada
`
`ABSTRACT
`
`INTRODUCTION
`
`Over the past 100 yr, the range of traits considered
`for genetic selection in dairy cattle populations has
`progressed to meet the demands of both industry and
`society. At the turn of the 20th century, dairy farmers
`were interested in increasing milk production; however,
`a systematic strategy for selection was not available.
`Organized milk performance recording took shape, fol-
`lowed quickly by conformation scoring. Methodological
`advances in both genetic theory and statistics around
`the middle of the century, together with technological
`innovations in computing, paved the way for powerful
`multitrait analyses. As more sophisticated analytical
`techniques for traits were developed and incorporated
`into selection programs, production began to increase
`rapidly, and the wheels of genetic progress began to
`turn. By the end of the century, the focus of selection
`had moved away from being purely production oriented
`toward a more balanced breeding goal. This shift oc-
`curred partly due to increasing health and fertility
`issues and partly due to societal pressure and welfare
`concerns. Traits encompassing longevity, fertility, calv-
`ing, health, and workability have now been integrated
`into selection indices. Current research focuses on fit-
`ness, health, welfare, milk quality, and environmental
`sustainability, underlying the concentrated emphasis on
`a more comprehensive breeding goal. In the future, on-
`farm sensors, data loggers, precision measurement tech-
`niques, and other technological aids will provide even
`more data for use in selection, and the difficulty will
`lie not in measuring phenotypes but rather in choosing
`which traits to select for.
`Key words: selection goal, production trait, functional
`trait, novel trait
`
`Received March 31, 2017.
`Accepted July 9, 2017.
`1 This review is part of a special issue of the Journal of Dairy Science
`commissioned to celebrate 100 years of publishing (1917–2017).
`2 Corresponding author: Miglior@cdn.ca
`
`
`
`Genetic selection for important traits has helped
`transform and advance the dairy cattle industry. Spe-
`cific traits considered for selection in dairy cattle popu-
`lations have evolved with time as a response to changes
`to the needs of producers, consumers, and society with
`the aid of advances in technology and trait recording
`programs.
`As outlined by Shook (1989), a potential trait must
`meet several criteria before it can be considered for
`selection in dairy cattle populations. First, either it
`should have an economic value as a marketable com-
`modity or its improvement should reduce production
`costs. Second, the trait must have sufficiently large ge-
`netic variation and heritability. Third, the trait should
`be clearly defined, measurable at a low cost, and con-
`sistently recorded. Finally, an indicator trait may be
`favored if it has a high genetic correlation with the
`economically important trait, reduces recording costs,
`has a higher heritability, or can be measured earlier in
`life.
`The economic value of traits has historically been
`the driver for genetic selection. From the 1930s to the
`1970s, the focus of selection was solely on increasing
`milk production. Despite some early concern over
`selecting exclusively for yield, which was expected to
`cause a corollary decline in overall fitness, the industry
`strove to achieve maximum genetic change in the most
`financially lucrative area, which was production. The
`need to identify and select for additional traits emerged
`mainly from the recognition of the correlated genetic
`decline in other important traits. Many countries have
`shifted toward more balanced selection objectives
`by including more weight on previously undervalued
`nonyield traits (Miglior et al., 2005).
`The second criterion concerns genetic variation and
`heritability of a trait, which are central to the rate of
`genetic progress possible within a selection program.
`Traits vary in the amount of phenotypic and genetic
`variation observed, and they may be more or less heri-
`table. Traits may also be contingent on one another,
`
`10251
`
`Exhibit 1012
`Select Sires, et al. v. ABS Global
`
`

`

`10252
`
`MIGLIOR ET AL.
`
`with correlations either positive or negative and genetic
`and phenotypic correlations either strong or weak. Such
`correlations may be exploited by the use of indicator
`traits, which may be favored if they are more readily
`available than a trait of interest.
`A major boon for the progression of genetic selection
`has been the recording and access to clearly defined,
`accurate, and cost-effective phenotypes. The continu-
`ous increase in data collected throughout the produc-
`tion chain has brought forth many opportunities and
`a large number of traits with genetic evaluations for
`consideration. However, this has also resulted in a large
`number of potential traits to be considered for inclusion
`in selection programs and ultimately balanced appro-
`priately. Careless selection, changing selection goals, or
`having many different objectives can reduce selection
`pressure (Meadows, 1968) and can have an undesirable
`permanent effect on the population.
`The conception, development, and application of
`multitrait index selection has played a pivotal role in
`successful and progressive selection in many countries.
`By weighting each trait according to its independent
`effect on net profit and using genetic and phenotypic
`parameters to weight traits measured in individuals and
`relatives, the correlation of index with genetic variation
`in net profit can be maximized (Hazel, 1943; Hazel et
`al., 1994). Traits considered in selection vary between
`countries because of differences in milk and component
`prices, costs of inputs and services, production environ-
`ments, and availability of phenotypes. These factors
`can frequently change, and modifications need to be
`considered and researched continually. The identifica-
`tion of traits that are presently important for genetic
`selection and those that will be essential in the future is
`a vital aspect of selection research.
`Milk recording began in North America in 1905 and
`thus provided the foundation for selection on milk pro-
`duction. Cattle shows at county fairs made conforma-
`tion traits very popular as well. Technological advances,
`in particular the advent of AI in the late 1930s, created
`a division between producers who wanted cows to pro-
`duce milk and those who wanted good-looking cows
`that produced milk. Artificial insemination organiza-
`tions aimed their bull selection programs toward both
`types of producers. Gradually, producers recognized
`that fat and protein yields and longevity were also
`important to keeping the costs of production within
`reason. Behavior and health traits were incorporated
`soon afterward, demonstrating an increased awareness
`of the economic importance of these traits but also
`representing increasing societal concerns with intense
`production systems.
`The rapid developments in genomic information,
`automated data recording technologies, and modern
`
`Journal of Dairy Science Vol. 100 No. 12, 2017
`
`analytical techniques over the past decade are setting
`the stage for a new era in dairy cattle breeding. Here
`we review the development of phenotypes used in dairy
`cattle selection over the past century (see Appendix
`Table A1).
`
`PRODUCTION
`
`A century ago, selection in dairy cattle focused on
`high milk and fat production. In the late 1800s and into
`the 1900s, many breed associations, in addition to their
`standard herd register, worked to promote better dairy
`cows by recording cow merit through milk recording for
`the inclusion of cattle in advanced registries (Becker
`and McGilliard, 1929). These registries went by vari-
`ous names, including Advanced Registry for Ayrshire,
`Register of Production for Brown Swiss, Advanced
`Register for Guernsey and Holstein, and Register of
`Merit for Jersey. The Babcock test, invented and made
`public in 1890 by S. M. Babcock, provided an accu-
`rate and easy method for the determination of milk fat
`content in milk testing. The testing of a large number
`of purebred cows through milk testing programs and
`the organization of records into published volumes of
`the Advanced Register and Register of Merit provided
`the foundation for locating high-producing blood lines
`and the study of the inheritance of milk and fat pro-
`duction (Fohrman, 1926). Meade (1921) evaluated the
`performance of Guernsey sires and their transmitting
`ability and concluded that the best method for selec-
`tion may be to consider the percentage production of
`all advanced registry daughters based on standardized
`requirements according to age. The male line received
`the greatest attention in selection because the sire’s
`heredity was most accurately indicated by his daugh-
`ters’ production, more so than the dam based on her
`own production record (Graves, 1925). A recognized
`fault of production records in an advanced registry
`for selection purposes was that records included only
`daughters that were put on test and met advanced reg-
`istry standards. Later, some breed organizations initi-
`ated a further herd test or herd-improvement registry
`where dairy producers were required to test and report
`production of all cows in their herd. Using the descen-
`dants of 2 cows disparate for their fat tests, Burrington
`and White (1925) demonstrated that the difference in
`test could be maintained over generations. Copeland
`(1927) expressed that dairy cattle breeders had thus far
`concerned themselves chiefly with increasing milk yield
`traits and concluded that more improvement in total
`fat production could be accomplished by selecting for
`high fat percentage along with milk yield.
`Traits also considered by breeders to aid in the se-
`lection of animals to improve production were body
`
`Exhibit 1012
`Select Sires, et al. v. ABS Global
`
`

`

`100-YEAR REVIEW: SELECTION GOALS IN DAIRY CATTLE
`
`10253
`
`conformation traits. The thought was based on the
`view that conformation of the cow shows her probable
`production and that the conformation of the sire will be
`transmitted to his daughters, implying their probable
`production (Gowen, 1926). Gowen (1920) attempted to
`study conformation and its relation to milk-producing
`capacity through the calculation of phenotypic correla-
`tions. He confirmed the existence of the relationship
`between body conformation and production that was
`common belief among producers, although he conclud-
`ed that conformation was a poor guide for milk produc-
`tion. Body weight and other measurements indicating
`size and shape were related to milk production (but not
`fat percentage) and were inferior to production records
`of ancestors for breeding for milk production (Gowen,
`1926). Consideration of conformation traits remained
`in the forefront for breeders because an easily measured
`indicator for production was in demand. The true value
`of conformation in breeding for production was unclear,
`though, and Copeland (1941) reported only a slight re-
`lationship. Tyler and Hyatt (1948) found an intraherd
`relationship of 0.19 between conformation and average
`fat record and stated that selecting within a herd for
`conformation would not substantially improve fat pro-
`duction. Harvey and Lush (1952) found a genetic cor-
`relation of 0.18 between conformation and fat produc-
`tion from daughter–dam pairs. O’Bleness et al. (1960)
`studied the genetic correlations between production
`traits and individual conformation traits, the strongest
`of which were fat with pin bone width (0.39–0.40), rear
`udder shape (−0.54), and fore teat length (0.42) and
`with milk, dairy character (0.95–0.98). However, with
`an index, selection on the basis of milk production alone
`would be almost as effective as including conformation
`traits as well.
`More measures of producing ability related to the
`lactation curve and persistency began to appear. The
`2 main factors in total yearly milk and fat production
`were the yield during the maximum month and the
`persistency of production or the rate of decline (Turner,
`1926). Turner (1926) suggested inheritance of fat pro-
`duction during the month of maximum production
`and endorsed consideration of persistency in select-
`ing breeding animals. Copeland (1937) studied Herd
`Improvement Registry records of the American Jersey
`Cattle Club, discovered that some cows maintained
`their production longer than other cows under similar
`conditions, and ruled that lactation persistency was an
`inherited trait.
`Gaines and Overman (1938) discussed the American
`Dairy Cattle Club’s step toward requiring estimates of
`milk protein yield on the advice of the club’s geneticist
`that protein was the most biologically valuable milk
`component. However, at the time no practical field
`
`test for protein existed. The importance of the non-
`fat component of milk, or SNF, increased due to the
`awareness of the nutritional value of SNF and its effect
`on milk processing. Data presented by Richardson and
`Folger (1950) suggested that SNF contents were inher-
`ited and that its relationship with fat was nonlinear.
`Around the 1950s, changes to the milk pricing structure
`were proposed to better reflect the importance of, and
`compensate producers for, both fat and SNF contents
`in milk. With an economic value projected for SNF,
`dairy cattle breeders would seek to improve contents
`in their cows. The heritability presented by Johnson
`(1957) for SNF was 0.34 for both Holstein and Jersey
`breeds. This study also reported very strong genetic
`correlations between different milk components, deduc-
`ing that selection for one would equate to selecting for
`the others but with less pressure. Nonfat components
`were expensive and difficult to obtain for large numbers
`of cows, thus limiting the number of observations for
`this trait. Legates (1960) and Laben (1963) reviewed
`the many factors affecting SNF. Selection at the time
`emphasized total milk yield per cow, which returned the
`greatest value for milk, but farmers were encouraged to
`test for SNF to help build knowledge of its variation
`and possible future economic gains (Laben, 1963). The
`cost of testing milk for protein was an added expense
`that could potentially be avoided, as protein still in-
`creased by selecting for correlated fat and milk yields
`(Van Vleck, 1978).
`Infrared methods for the analysis of milk samples for
`fat, protein, and lactose content delivered more rapid
`and less expensive measures than those previously
`available (Biggs, 1967). Protein testing became much
`more commonplace in the 1970s using this technology
`and became standard in milk testing. Milk processors
`began paying premiums for protein, and the number
`of cows with a higher genetic propensity to produce
`protein increased rapidly as protein was added into
`national selection indices (Shook, 2006). A shift toward
`increased emphasis on protein yield in selection indices
`occurred in many countries.
`Although total protein content in milk was the pri-
`mary consideration for selection, genetic variability and
`milk protein variants garnered additional attention. As-
`chaffenburg and Drewry (1955) first reported a genetic
`polymorphism of β-LG producing 2 different forms.
`Genetic variation was next reported in α-LA (Blum-
`berg and Tombs, 1958) and in the CN (Aschaffenburg,
`1961). The effects of the many discovered milk pro-
`tein variants were considered important to achieving
`specific requirements for the dairy industry, including
`cheese manufacturing. Furthermore, several of the dif-
`ferent protein variants have been associated with milk
`production and with fat and protein yield and percent-
`
`Journal of Dairy Science Vol. 100 No. 12, 2017
`
`Exhibit 1012
`Select Sires, et al. v. ABS Global
`
`

`

`10254
`
`MIGLIOR ET AL.
`
`ages (Ng-Kwai-Hang et al., 1984, 1986; Aleandri et
`al., 1990; Bovenhuis et al., 1992; Tsiaras et al., 2005).
`However, many contradictions in their effects exist in
`the literature, and it is not clear whether the protein
`genes or linked genes produce the effect (Bovenhuis et
`al., 1992).
`The shape of the lactation curve became a trait of
`interest for selection once again because of issues of cow
`health from the stress of high peaks in production and
`for use in an index to improve total yield. Aspects of
`the lactation curve considered by Shanks et al. (1981)
`included the lactation curve, persistency, week of peak
`yield, and peak yield. Generally low heritabilities were
`found, except for peak yield, which had heritabilities
`ranging from 0.16 to 0.23. Ferris et al. (1985) reported
`low heritabilities with large standard errors for lacta-
`tion shape measures, and indices formulated to flat-
`ten the lactation curve did so at the expense of milk
`yield. Jamrozik et al. (1997) applied random regression
`models to test-day yields to generate EBV for partial-
`lactation yields and persistency for animals with even
`single test-day records. Genetic evaluations are now
`performed in several countries for lactation persistency.
`Much success has been achieved in the improvement
`of production traits as a result of genetic selection.
`The dominant role production traits held in selection
`programs for many decades has been diminishing as se-
`lection goals become broader. Selection for production
`traits needs to be examined in tandem with relevant
`nonyield traits.
`
`CONFORMATION
`
`Conformation, or type, of an animal has been of
`interest to dairy producers since the beginning of the
`selection process. The archetypes for conformation and
`beauty in dairy cattle have been passed down through
`time and conveyed from past breeders (Copeland, 1941).
`Producers strongly considered conformation traits for
`breed standards of perfection for registration and in the
`show ring as well as to garner top prices in public sales.
`The aesthetic aspect of the animal was the main reason
`for selection, but conformation traits were increasingly
`used to select dairy cows for other characteristics, such
`as higher production and longevity.
`Several breed associations started classification pro-
`grams to appraise conformation of all animals in the
`registry based on a scorecard or scale of points. In 1929,
`Holstein cattle in the United States were classified for
`4 major categories: general appearance, dairy charac-
`ter, body capacity, and mammary system. Soon after,
`in 1932, the American Jersey Cattle Club established
`a similar classification program. The data collected
`through the classification programs confirmed an exist-
`
`Journal of Dairy Science Vol. 100 No. 12, 2017
`
`ing relationship between conformation and producing
`ability, highlighting the fact that both conformation
`and production should be considered in selection pro-
`grams (Copeland, 1938).
`The stability of conformation classification and the
`repeatability of the measurement were questioned in
`early studies (Johnson and Lush, 1942; Hyatt and
`Tyler, 1948). For example, Johnson and Lush (1942)
`reported high variation between evaluators and low
`repeatability for conformation traits ranging from 0.34
`to 0.55. The heritability estimates reported for confor-
`mation traits were moderate to low. Tyler and Hyatt
`(1948) reported a heritability estimate for conformation
`of approximately 0.30. O’Bleness et al. (1960) reported
`heritability estimates for 27 different conformation
`traits ranging from 0.00 to 0.33.
`In an effort to establish a more objective way to mea-
`sure conformation traits, studies were conducted using
`data recorded from a conformation appraisal program
`initiated in 1953 in New York. The studies showed ef-
`fects of appraiser (Van Vleck and Albrectsen, 1965),
`age, and lactation stage (Norman and Van Vleck, 1972)
`on conformation trait measures. Using data from the
`same appraisal program, Van Vleck (1964b) estimated
`heritabilities for conformation traits higher than those
`estimated previously and much closer to those reported
`in more recent studies (e.g., Rupp and Boichard, 1999).
`In 1967, the Holstein-Friesian Association of America
`introduced a descriptive classification program that
`included an assigned code value for 12 conformation
`traits in addition to the 4 scorecard traits already re-
`corded since 1929. These measures were also recorded
`on unregistered cows. The implementation of this sys-
`tem provided a large amount of data, allowing a more
`precise evaluation of conformation traits. Because of
`their negative correlation with milk production, confor-
`mation traits should be included in selection objectives
`to maintain cow appearance (Grantham et al., 1974).
`Due to increased attention on linear conformation
`appraisal in the 1980s, a substantial change occurred
`in the methods used for estimating genetic values for
`conformation traits. The aim was to score conformation
`traits using a wider range of numerical scores (i.e., a
`50-point basis). This measurement method presented
`several advantages, the major one being that it allowed
`analyses on a continuous scale and with mixed-model
`evaluation, as described by Thompson et al. (1983).
`The predictive ability of conformation traits for ad-
`ditional traits of interest, other than production or
`longevity, was considered in several studies. Udder con-
`formation traits showed varying but usually positive
`correlations with milking ability (Blake and McDaniel,
`1979) and favorable correlations with udder health
`(Monardes et al., 1990; Rogers et al., 1991). As well,
`
`Exhibit 1012
`Select Sires, et al. v. ABS Global
`
`

`

`100-YEAR REVIEW: SELECTION GOALS IN DAIRY CATTLE
`
`10255
`
`mostly favorable correlations have been found between
`conformation and fertility (Dadati et al., 1985, 1986).
`The relationship between conformation and calving
`ease was negative when considering the conformation
`of the calf and positive when considering the conforma-
`tion of the dam (Cue et al., 1990).
`
`LONGEVITY
`
`Longevity in dairy cattle has many different defini-
`tions and encompasses traits referring to the length
`of time a cow remains in the productive herd or its
`ability to remain in the herd. Measures of longevity
`have included age at disposal or last calving, number
`of lactations, and survival to a fixed age or lactation
`number. Cow longevity is a fundamental component
`of profitability in dairy production and, apart from
`production traits, has the greatest economic value (Al-
`laire and Gibson, 1992). Longevity reduces the costs
`of replacements and maximizes the profitable period
`following the recovery of initial breeding and rearing
`costs. In addition, improving longevity could aid in
`breed development and genetic improvement because
`it would allow for more voluntary culling and greater
`selection intensity if fewer replacements were required.
`The main goal in selecting for longevity is to decrease
`the premature disposal of cows or involuntary culling.
`By reducing the involuntary culling rate, dairy produc-
`ers can consequently increase the voluntary culling rate
`and keep only the most productive animals. Reasons
`and strategies for culling are vast, may differ between
`years, and vary greatly between producers because
`they depend on the situation of the herd and involve
`a great degree of subjectivity and personal preference.
`Cows may be removed for voluntary reasons such as
`herd reductions, old age, level of production, body or
`conformation, management or workability, and sale for
`beef. Involuntary culling may occur for various reasons
`including reproductive performance, general health or
`illness, injuries, and accidents. Therefore, selection for
`longevity incorporates the improvement of many differ-
`ent components.
`Automatic selection for increased longevity is pre-
`sumed because cows remaining in the herd longer would
`produce more progeny and thereby contribute more to
`the succeeding generations (Parker et al., 1960). Be-
`cause of the aforementioned importance of the trait
`to producers, further deliberate and direct selection
`for longevity was attractive and warranted. Measures
`of longevity were more easily recorded and accessible
`than records for fitness traits. Asdell (1951) examined
`DHIA herd culling records and stated that work needed
`to be done to develop longer living cows and reduce
`the loss of aging cows to sterility and udder troubles,
`
`which were on the rise. When studying the occur-
`rence of cystic ovaries in a herd, Casida and Chapman
`(1951) found that there was a significant daughter–dam
`correlation for time spent in the herd. Wilcox et al.
`(1957) estimated in a single herd a heritability of 0.37
`for longevity measured as number of parturitions. In a
`herd that had experienced no deliberate selection for
`conformation or production, Parker et al. (1960) found
`a near-zero heritability for longevity in terms of age at
`last calving. In general, the heritability of longevity in
`dairy cows is low (White and Nichols, 1965; Miller et
`al., 1967; Hargrove et al., 1969; Schaeffer and Burnside,
`1975; Ducrocq et al., 1988; VanRaden and Klaaskate,
`1993). Variation in reported heritabilities could be at-
`tributable to single or small numbers of herds used in
`early studies and differences in culling reasons between
`the populations.
`To qualify a direct record for longevity, the cow or
`daughters of a sire must have reached the end of their
`productive life, which means the cow is no longer avail-
`able and the generation interval is increased in evalu-
`ated sires. To overcome this and the low heritability,
`early measures for indirect selection for longevity were
`investigated. Parker et al. (1960) found a significant
`correlation between first-lactation fat production and
`longevity recorded as age at last calving. Gaalaas and
`Plowman (1963) found a tendency for better producing
`young cows to stay in the herd longer using an intrasire
`regression of age at last calving on production. The
`propensity for high first-lactation producers to com-
`plete more lactations was substantiated by Van Vleck
`(1964a), White and Nichols (1965), and Hargrove et al.
`(1969), indicating that selecting young sires on daugh-
`ter first-lactation production records would indirectly
`improve longevity. This conclusion was in contrast to
`the belief of many producers, who thought that many
`young high-producing cows leave the herd early and do
`not live up to this high production later in life (Van
`Vleck, 1964a).
`The physical characteristics of a dairy cow were as-
`sumed to be related to its longevity. Conformation traits
`were widely available for classified cattle, known early
`in life (usually first lactation), and heritable, making
`them attractive indicators of longevity. Specht et al.
`(1967) reported a correlation of 0.20 between overall
`first classification score and longevity of Holstein-
`Friesian cows. They found similar correlations between
`individual conformation traits and longevity. Using the
`daughters of AI sire Holsteins, Van Vleck et al. (1969)
`examined the relationship between 66 type categories
`measured in first lactation and longevity, determined
`as number of recorded lactations. The type traits with
`the strongest correlations with longevity were plumb
`rear teat position (0.38), sharp dairy character (0.35),
`
`Journal of Dairy Science Vol. 100 No. 12, 2017
`
`Exhibit 1012
`Select Sires, et al. v. ABS Global
`
`

`

`10256
`
`MIGLIOR ET AL.
`
`intermediate thurls (0.26), and typical head (0.25).
`Schaeffer and Burnside (1975) looked at sire proofs for
`the survival rates of 2-yr-old daughters making a record
`at 3 and 4 yr of age and resolved that improvement in
`longevity could best be achieved using type and milk
`proofs opposed to longevity directly. With the arrival of
`linear type traits, more research into their correlation
`with longevity was completed. Several studies high-
`lighted the relevance of many udder characteristics, feet
`and legs, and dairy character in improving selection for
`longevity (Rogers et al., 1988; Foster et al., 1989; Short
`and Lawlor, 1992).
`An improvement of the longevity definition was sug-
`gested to better direct selection toward increasing the
`ability of cows to survive irrespective of production
`(Van Arendonk, 1986). Miller et al. (1967) examined
`longevity by dividing cows into opportunity groups
`to enable comparisons before all cows had died and
`further adjusted longevity for milk production. They
`found that heritabilities decreased when the effect of
`first-lactation milk was removed. Later, Ducrocq et
`al. (1988) suggested 2 measures of longevity: (1) true
`longevity not adjusted for yield, describing the ability
`of the cow to remain in the herd, and (2) functional
`longevity, linearly adjusted for the cow’s milk yield
`relative to the herd, representing the ability to delay
`involuntary culling. The correction of longevity for milk
`production should expose differences between animals
`culled for nonproduction reasons. Given culling levels
`for production, adjustment of longevity for production
`was recommended to eliminate bias from culling for
`production (Dekkers, 1993).
`Research over the past decades has shown that lon-
`gevity is heritable and that selection is possible. Thus,
`many major countries in dairy breeding have included
`longevity in routine genetic evaluations (Miglior et al.,
`2005). Multiple-trait evaluations combining indirect
`measures of longevity with direct measures are help-
`ful to improve the accuracy of longevity evaluations.
`There is currently no consensus in the trait definition
`and methodology for evaluation across countries. The
`United States considers productive life, which combines
`direct longevity defined as total months in milk through
`84 mo of age, along with SCS, udder, body size, feet
`and leg composites, and milk, fat, and protein yields
`(Cruickshank et al., 2002). In Canada, genetic evalua-
`tions for direct longevity are from a 5-trait animal mod-
`el including cow survival from first calving to 120 DIM,
`from 120 to 240 DIM, from 240 DIM to second calving,
`survival to third calving, and survival to fourth calving
`to account for differences in the genetic background of
`survival at different time points (Sewalem et al., 2007).
`Complementary indirect longevity evaluations in Can-
`ada are based on dairy strength, feet and legs, overall
`
`Journal of Dairy Science Vol. 100 No. 12, 2017
`
`mammary, rump angle, SCS, milking speed, nonreturn
`rate in cows, and interval from calving to first service
`(Sewalem et al., 2007). In the future, the incorporation
`of additional traits relating to longevity, which includes
`many health traits, may benefit evaluation and selec-
`tion for longevity.
`
`FERTILITY
`
`Because of the economic importance of reproductive
`efficiency, much attention has been given to fertility
`traits and to their relationship with production over
`the years. Moreover, genetic correlation with produc-
`tive life indicates that fertility plays a major role in
`longevity of the cow (VanRaden et al., 2004). The ad-
`vent of AI activities actualized the problem of fertility,
`and the possibilities of breeding for reproduction had to
`be investigated in a completely new light. Principally,
`the attention and research on fertility has been directed
`toward female fertility. The consequences and varia-
`tion in sire fertility are seldom regarded in the genetic
`improvement of fertility. This is despite the fact that
`different fertility measures in dairy breeding can be af-
`fected by only the cow or bull or a combination of both
`male and female fertility, such as conception rate. Early
`measures of female fertility were the number of services
`required for conception, nonreturns to first service, the
`interval from calving to first insemination, and calving
`interval. The disadvantages of interval from calving to
`first insemination were that it might be influenced by
`farmer decisions or by seasonal calving. A late insemina-
`tion could also be the result of estrus detection failure,
`so that the cow was cycling successfully but did not
`have the opportunity to conceive. On the other hand,
`success traits suc