Large Dairy Herd Management
`Third edition
`
`Section 5: Genetic Selection Programs
`and Breeding Strategies
`
`Edited by David K. Beede
`
`Exhibit 1021
`Select Sires, et al. v. ABS Global
`
`

`

`The following abbreviations may be used without definition in the book.
`
`Abbreviations
`
`
`AA
`ACTH
`ADF
`
`ADG
`
`ADL
`
`ADIN
`AI
`
`BCS
`
`BHB
`
`BLUP
`BSA
`
`bST
`
`BTA
`
`BUN
`
`BW
`
`CI
`
`CLA
`
`CN
`
`CNS
`
`CoA
`
`CP
`
`CV
`
`DCAD
`DHI(A)
`DIM
`
`DM
`
`DMI
`
`DNA
`
`EAA
`
`EBV
`
`ECM
`
`ELISA
`ETA
`
`FAME
`FCM
`
`FSH
`
`GnRH
`h2
`
`HTST
`IFN
`
`Ig
`
`IGF
`
`IL
`
`IMI
`
`LA
`
`LG
`
`LH
`
`LPS
`
`LSD
`
`LSM
`
`mAb
`
`
`amino acid
`adrenocorticotropin
`acid detergent fiber
`average daily gain
`acid detergent lignin
`acid detergent insoluble nitrogen
`artificial insemination
`body condition score
`β-hydroxybutyrate
`best linear unbiased predictor
`bovine serum albumin
`bovine somatotropin
`Bos taurus autosome
`blood urea nitrogen
`body weight
`confidence interval
`conjugated linoleic acid
`casein
`coagulase-negative staphylococci
`coenzyme A
`crude protein
`coefficient(s) of variation
`dietary cation-anion difference
`Dairy Herd Improvement (Association)
`days in milk
`dry matter
`dry matter intake
`deoxyribonucleic acid
`essential amino acid
`estimated breeding value
`energy-corrected milk
`enzyme-linked immunosorbent assay
`estimated transmitting ability
`fatty acid methyl esters
`fat-corrected milk
`follicle-stimulating hormone
`gonadotropin-releasing hormone
`heritability
`high temperature, short time
`interferon
`immunoglobulin
`insulin-like growth factor
`interleukin
`intramammary infection
`α-lactalbumin
`β-lactoglobulin
`luteinizing hormone
`lipopolysaccharide
`least significant difference
`least squares means
`monoclonal antibody
`
`
`ME
`
`MIC
`
`MP
`mRNA
`MUFA
`MUN
`
`NAN
`
`NDF
`
`NDIN
`NEAA
`
`NEG
`NEL
`
`NEM
`
`NFC
`
`NPN
`
`NRC
`
`NSC
`
`OM
`
`PCR
`
`PGF2α
`PMNL
`PTA
`
`PUFA
`QTL
`
`r
`
`R2
`
`RDP
`
`REML
`RIA
`
`RNA
`
`RUP
`
`SARA
`SCC
`
`SCS
`
`SD
`
`SDS
`
`SE
`
`SEM
`
`SFA
`
`SNP
`
`SPC
`
`TDN
`
`TMR
`
`TS
`
`UF
`
`UFA
`
`UHT
`
`USDA
`UV
`
`VFA
`
`
`metabolizable energy
`minimum inhibitory concentration
`metabolizable protein
`messenger ribonucleic acid
`monounsaturated fatty acids
`milk urea nitrogen
`nonammonia nitrogen
`neutral detergent fiber
`neutral detergent insoluble N
`nonessential amino acid
`net energy for gain
`net energy for lactation
`net energy for maintenance
`nonfiber carbohydrates
`nonprotein nitrogen
`National Research Council
`nonstructural carbohydrates
`organic matter
`polymerase chain reaction
`prostaglandin F2α
`polymorphonuclear leukocyte
`predicted transmitting ability
`polyunsaturated fatty acids
`quantitative trait loci
`correlation coefficient
`coefficient of determination
`rumen-degradable protein
`restricted maximum likelihood
`radioimmunoassay
`ribonucleic acid
`rumen-undegradable protein
`subacute ruminal acidosis
`somatic cell count
`somatic cell score
`standard deviation
`sodium dodecyl sulfate
`standard error
`standard error of the mean
`saturated fatty acids
`single nucleotide polymorphism
`standard plate count
`total digestible nutrients
`total mixed ration
`total solids
`ultrafiltration, ultrafiltered
`unsaturated fatty acids
`ultra-high temperature
`United States Department of Agriculture
`ultraviolet
`volatile fatty acids
`
`This document is registered to JOHN COLE
`
`Exhibit 1021
`Select Sires, et al. v. ABS Global
`
`

`

`Large Dairy Herd Management
`Third Edition
`
`Editor-in-Chief
`
`David K. Beede
`
`Section Editors
`
`David K. Beede
`Steven P. Washburn
`Joseph M. Zulovich and Joseph P. Harner
`Normand R. St-Pierre
`Kent A. Weigel
`Robert E. James
`William W. Thatcher
`Richard J. Grant and Heather M. Dann
`Rupert M. Bruckmaier
`Joseph S. Hogan
`Trevor J. DeVries
`Carlos A. Risco
`Albert De Vries
`Stanley J. Moore and Phillip T. Durst
`Jeffrey M. Bewley
`
`Published by the
`American Dairy Science Association®
`1800 South Oak St., Ste. 100
`Champaign, IL 61820
`https://www.adsa.org/
`
`Edited and produced by
`FASS Inc.
`1800 South Oak St., Ste. 100
`Champaign, IL 61820
`https://www.fass.org/
`
`This document is registered to JOHN COLE
`
`Exhibit 1021
`Select Sires, et al. v. ABS Global
`
`

`

`American Dairy Science Association®, Champaign, IL 61820
`© 1978, 1992, 2017 by the American Dairy Science Association
`All rights reserved
`First edition published 1978
`First revised edition published 1992
`Third edition published 2017
`
`ISBN: 978-0-9634491-2-2
`
`Copyright © 2017 by American Dairy Science Association®
`
`All rights reserved. No part of this publication may be reproduced, distributed, or
`transmitted in any form or by any means, including photocopying, recording, or
`other electronic or mechanical methods, without the prior written permission of the
`publisher, except in the case of brief quotations embodied in critical reviews and cer-
`tain other noncommercial uses permitted by copyright law. For permission requests,
`contact the publisher at adsa@adsa.org.
`
`American Dairy Science Association®
`1800 South Oak St., Ste. 100
`Champaign, IL 61820
`https://www.adsa.org
`adsa@adsa.org
`
`Large Dairy Herd Management website: http://ldhm.adsa.org
`
`LDHM3_V1.1_082018_S5
`
`Cover images
`
`Top left: Example of immunofluorescent staining in prepubertal bovine mammary tissue. The cross section of the developing duct shows
`the expression of p63 (red), which indicates myoepithelial nuclei, estrogen receptor (green), about 50% of the epithelial cells, and Ki67
`(yellow), a marker for cell proliferation; DAPI staining (blue) is a general DNA stain that labels all cell nuclei.
`[Chapter 9-59: Mammary development in calves and heifers; Figure 4D]
`
`Top center: The daily trail to (and from) milking.
`[Chapter 10-67: Mastitis control in pasture and seasonal systems; Figure 3]
`
`Top right: Cow brushes are clearly a valued resource as they are used voluntarily by cows and are required by some voluntary assurance
`programs. Photo credit: DeLaval, Tumba, Sweden.
`[Chapter 11-71: Assuring and verifying dairy cattle welfare; Figure 2]
`
`Bottom left: The bedding material commonly recommended for controlling environmental mastitis is washed sand.
`[Chapter 10-65: Practical approaches to environmental mastitis control; Figure 3]
`
`Bottom center: Life cycle of a liver fluke.
`[Chapter 12-81: Parasite control in large dairy herds; Figure 2]
`
`Bottom right: Studies have shown that positive handling is correlated with cows having low fear responses to people and higher milk
`production. Some animal welfare standards now include a standardized test of avoidance distance to people as a way of screening for ap-
`propriate handling and good human–animal relationships on farms. Photo credit: University of British Columbia (UBC) Animal Welfare
`Program.
`[Chapter 11-71: Assuring and verifying dairy cattle welfare; Figure 3]
`
`This document is registered to JOHN COLE
`
`Exhibit 1021
`Select Sires, et al. v. ABS Global
`
`

`

`Large Dairy Herd Management
`Third Edition
`
`Section 5: Genetic Selection Programs and
`Breeding Strategies
`Note
`
`This section is an excerpt from Large Dairy Herd Management, third edition. As such, links within these
`chapters to other sections of the book are not functional. Other sections of the book are available for
`purchase at https://ldhm.adsa.org/. The front matter and index of the complete book are available as free
`downloads.
`
`Preface
`
` K. A. Weigel
`05-23: Improving production efficiency through genetic selection
`
` J. B. Cole and D. M. Spurlock
`05-24: Improving health, fertility, and longevity through genetic selection
`
` R. R. Cockrum, K. L. Parker Gaddis, and C. Maltecca
`05-25: Making effective sire selection and corrective mating decisions
`
` K. A. Weigel and T. J. Halbach
`05-26: Capitalizing on breed differences and heterosis
`
` C. D. Dechow and L. B. Hansen
`05-27: Genomic selection and reproductive technologies to optimize herd
` replacements
` F. Peñagaricano, A. De Vries, and D. T. Bennink
`
`05-28: Genomic selection and reproductive technologies to produce
` elite breeding stock
` H. J. Huson and J. Lamb
`
`
`
`329
`
`331
`
`341
`
`357
`
`369
`
`379
`
`389
`
`Sponsorship
`
`The ADSA® Foundation gratefully acknowledges the generous exclusive sponsorship of Large
`Dairy Herd Management, Section 5: Genetic Selection Programs and Breeding Strategies by the
`following company:
`
`This document is registered to JOHN COLE
`
`Exhibit 1021
`Select Sires, et al. v. ABS Global
`
`

`

`Large Dairy Herd Management, 3rd ed.
`https://doi.org/10.3168/ldhm.0523
`© American Dairy Science Association®, 2017.
`Improving production efficiency through genetic selection
`John B. Cole*1 and Diane M. Spurlock†
`*Animal Genomics and Improvement Laboratory, Agricultural Research Service, USDA, Beltsville, MD 20705-2350
`†Department of Animal Science, Iowa State University, Ames 50011
`
`SUMMARY
`
`Genetic selection has been a very effective tool for achieving lasting gains in animal production and efficiency.
`Prediction of the genetic merit of animals for a variety of traits occurs through the integration and analysis of
`multiple types of data, including genotypes that describe variation in DNA sequences among animals. These data
`are gathered, maintained, and analyzed through the efforts of multiple organizations working together in the
`dairy industry. The success of this genetic evaluation program is evidenced by improvements in the genetic merit
`and actual performance of cows for milk, fat, and protein yields. Although these production traits will continue
`to be important to US dairies in the future, interest in the ability to select animals for improved efficiency of
`production has increased in recent years. Estimation of genetic merit for feed intake, feed efficiency traits, or both
`will likely be added to US genetic evaluation programs in the future.
`
`INTRODUCTION
`
`GENETIC SELECTION AND GENOMIC PREDICTION
`
`The goal of dairy cattle breeding is to increase pro-
`ductivity and efficiency by means of genetic selection.
`This is possible because related animals share some of
`their DNA, and we can use statistical models to predict
`the genetic merit of animals based on the performance
`of their relatives. Historically, production goals focused
`on the amount of milk, butterfat, and protein produced.
`Although these traits remain an important part of most
`selection strategies, other traits, such as longevity and
`fertility, have increased in importance. A substantial
`amount of research now focuses on production effi-
`ciency, often calculated as a function of individual feed
`intake or greenhouse gas emissions. Interest in selection
`for production efficiency, rather than total production
`of milk, fat and protein, is increasing because of greater
`competition for feed and water, as well as growing
`demand for animal protein from the growing global
`middle class.
`The goals of this chapter are to describe how genetic
`and genomic selection programs work, and to demon-
`strate how these tools are being used to produce dairy
`cattle that are efficient producers of food for human
`diets.
`
`1 Corresponding author: john.cole@ars.usda.gov
`
`Genetic gains differ from those due to improvements
`in an animal’s environment (e.g., housing or feeding)
`because they are cumulative and can be transmitted
`from parents to offspring. Selection programs are de-
`signed to identify the animals in a population with the
`highest genetic merit. The animals with highest genetic
`merit may then be selected as the parents of the next
`generation, resulting in genetic improvement of the
`population.
`
`
`
`Genes Versus Environment
`
`An animal’s performance for a trait of interest (its
`phenotype) is influenced by the genes it inherited from
`its parents, as well as the environment in which it is
`placed. Records from all animals in a population can
`be used to estimate the amount of variation in the
`trait, and that variation can be broken down into por-
`tions due to genetics and to the environment (Falconer
`and MacKay, 1996). Traits that are influenced mostly
`by environment are said to have low heritabilities,
`and traits with large genetic components are said to
`have high heritabilities. In general, traits with higher
`heritabilities respond more rapidly to genetic selection
`because genetic merit for highly heritable traits can be
`estimated more accurately than that for lowly heritable
`traits. Table 1 shows heritabilities for traits included in
`the lifetime net merit economic index (NM$), which
`represents a broad array of phenotypes. The traits in
`
`331
`
`This document is registered to JOHN COLE
`
`Exhibit 1021
`Select Sires, et al. v. ABS Global
`
`

`

`332
`
`GENETIC SELECTION PROGRAMS
`
`Table 1. The heritability of traits in the lifetime net merit economic
`index, which range from 0.01 (heifer conception rate) to 0.40 (body
`size composite); the emphasis placed on each trait in the index is a
`function of its heritability and economic value1
`
`Trait
`
`Heritability
`
`Milk yield
`Fat yield
`Protein yield
`Somatic cell score
`Productive life
`Udder composite
`Feet/legs composite
`Body size composite
`Daughter pregnancy rate
`Heifer conception rate
`Cow conception rate
`Sire calving ease
`Daughter calving ease
`Sire stillbirth
`Daughter stillbirth
`1Data from VanRaden and Cole (2014).
`
`0.20
`0.20
`0.20
`0.12
`0.08
`0.27
`0.15
`0.40
`0.04
`0.01
`0.02
`0.086
`0.048
`0.030
`0.065
`
`NM$ have heritabilities ranging from 1% for heifer
`conception rate to 40% for body size composite. Gener-
`ally, traits associated with physical shape and size have
`high heritabilities (≥40%), traits related to milk and
`solids yield have moderate heritabilities (15 to 30%),
`and traits associated with health and fertility have low
`heritabilities (≤10%). If there is no variation in a trait
`attributable to genetics, such as for sire conception
`rate, then population performance cannot be improved
`using genetic selection.
`
`How Does Genetic Selection Work?
`
`Genetic selection works by improving the average
`genetic merit of animals in the population each genera-
`tion. In dairy cattle, this is accomplished by selecting
`bulls that have high genetic merit for traits of interest
`as sires of the next generation of cows. The breeder’s
`equation, shown in Equation [1], describes how different
`aspects of traits under selection affect the rate of ge-
`netic gain in the population (see, e.g., Bourdon, 1999):
`
`∆G
`
`year =
`
`reliability
`
`×
`
`selection intensity
`×
`gen
`eeration interval
`
`genetic variance
`
`.
`
`
`
`[1]
`
`In this equation, ΔGyear is the annual rate of genetic
`change in the population, reliability is a measure of
`the precision with which genetic merit is estimated,
`selection intensity is a measure of how selective we
`are when choosing the parents of the next generation,
`genetic variance is the variation among animals in the
`population that is attributable to genetic differences,
`
`and generation interval is the average age of parents
`when their offspring are born. Reliability and genetic
`variance differ from trait to trait, whereas selection
`intensity and generation interval are generally proper-
`ties of a population and do not depend directly on the
`trait under selection. The use of genomic information
`allows us to increase the rate of gain by computing
`high-reliability genetic evaluations early in an animal’s
`life, affecting both the reliability and the generation
`interval.
`
`Calculation of Breeding Values
`
`For many years, dairy cattle genetic improvement
`programs have been based on the mixed model meth-
`odology developed by Henderson (1984). In a simple
`mixed model analysis, the phenotype is modeled as a
`function of fixed (e.g., sex) and random (e.g., genetic)
`effects:
`
`
`
`y = Xb + Zu + e,
`
`[2]
`
`where y is a vector of phenotypes; X and Z are matri-
`ces that link observations to fixed and random effects;
`b is a vector of values for fixed effects; u is a vector
`of random animal breeding values; and e is a vector
`of residual error effects. This equation describes the
`phenotype measured on an animal as a combination of
`genetic and environmental effects, as well as unknown
`effects that we cannot measure individually (error).
`When these equations are solved, each animal in the
`pedigree receives an estimate of genetic merit known
`as a predicted transmitting ability (PTA), which is the
`estimated animal breeding value from u divided by 2.
`The animal model used for most traits in the United
`States (VanRaden and Wiggans, 1991) is more com-
`plex than the example described above, and includes
`sophisticated contemporary groups, herd-by-sire, and
`cow permanent environmental effects.
`
`Genomic Prediction
`
`Genomic prediction occurs when information about
`an animal’s DNA is used along with its performance
`and pedigree data for the estimation of genetic merit.
`Genotypes describe DNA inherited from each parent
`at specific markers known as single nucleotide poly-
`morphisms (SNP). Each SNP represents a position in
`the DNA sequence that has only 2 possible variants.
`Genotypes at these SNP can be read in large numbers
`at low cost, and they describe which variant(s) at each
`SNP an animal has inherited. These genotypes provide
`information about an animal’s genetic composition that
`
`This document is registered to JOHN COLE
`
`Exhibit 1021
`Select Sires, et al. v. ABS Global
`
`

`

`CHAPTER 5-23: PRODUCTION EFFICIENCY AND GENETIC SELECTION
`
`333
`
`is not influenced by the environment, thus improving
`the accuracy with which genetic merit is estimated.
`
`Genotyping Chips
`
`When the initial sequencing of the bovine genome
`was completed in 2009, an international consortium
`of government, university, and industry cooperators
`worked with Illumina Inc. (San Diego, CA) to develop
`a set of SNP to be included on a genotyping chip for
`cattle. This resulted in a set of 54,001 SNP included
`on the Illumina BovineSNP50 BeadChip, which be-
`came publicly available in December 2007. Genotypes
`were pooled from the Beltsville Agriculture Research
`Center (Maryland), University of Missouri (Columbia),
`and University of Alberta (Edmonton, AB, Canada)
`to identify SNP useful for genomic evaluation. Mark-
`ers were excluded from genomic evaluation for many
`reasons, such as low call rates (genotypes frequently
`could not be determined), low minor allele frequencies
`(variants occurred too rarely), or high correlations with
`adjacent SNP (markers added little new information).
`In addition to computing genomic breeding values,
`the SNP also are used to identify and correct pedi-
`gree errors. As of June 2016, 1,418,194 genotypes from
`21 different chips are included in the National Dairy
`Database, including low-density (6,000 to 9,000 SNP),
`medium-density (30,000 to 60,000 SNP), and high-
`density (77,000 to 777,000 SNP) chips. The majority of
`the genotyped animals are young females that do not
`
`yet have completed lactation records. This represents
`the largest database of genotyped animals in the world,
`but most of the ~9 million cows in the national herd do
`not receive genomic tests.
`
`Genomic PTA
`
`Effects of alternative SNP variants associated with
`traits of interest are estimated using phenotypes,
`pedigrees, and genotypes from a group of animals with
`high-reliability PTA, known as the predictor population.
`Once these SNP effects have been estimated, genomic
`PTA can be computed for animals that have genotypes
`but no available phenotypes. The marker effects must
`be periodically re-estimated, so phenotypes must be
`collected continuously. The final genomic predictions
`combine 3 terms using selection index procedures: (1)
`direct genomic predictions based on the SNP effects;
`(2) parent averages (PA) or PTA computed from geno-
`typed ancestors using traditional relationships; and
`(3) traditional PA or PTA computed using pedigree
`and phenotype information. Reliabilities of genomic
`predictions are approximated using the genomic rela-
`tion of each animal to the predictor population and
`the reliability of the predictor evaluation. The official
`evaluation for the bull Bacon-Hill Pety Modesty-ET
`(HO84003013654627), the top-ranked bull for lifetime
`net merit in the December 2015 evaluation, is shown in
`Figure 1 as an example of the results from the evalua-
`tion system.
`
`Figure 1. December 2015 official bull evaluation for the Holstein sire Bacon-Hill Pety Modesty-ET (HO84003013654627). Source: Council on
`Dairy Cattle Breeding (https://www.uscdcb.com/cgi-bin/general/Qpublic/proc.Q.cgi?qname=getbull&single&id=HO840003013654627).
`
`This document is registered to JOHN COLE
`
`Exhibit 1021
`Select Sires, et al. v. ABS Global
`
`

`

`334
`GENETIC SELECTION PROGRAMS
`THE US DAIRY GENETIC EVALUATION SYSTEM
`
`Input Data
`
`Selection programs make use of many types of infor-
`mation about the animals in the population, including
`phenotypes, pedigrees, and genotypes. Phenotypes are
`measurements of individual animal performance, such
`as a test-day milk yield or breeding record. Pedigrees
`describe genetic relationships among animals in the
`population. Genotypes describe the DNA inherited from
`each parent using SNP markers. The information pro-
`vided by records can vary considerably. For example,
`a record for a trait with high heritability, such as milk
`yield, provides more information about an animal’s
`genetic potential than a record from a trait with a low
`heritability, such as conception rate. Pedigrees differ in
`their quality (error rate) and completeness, with scien-
`tific studies reporting error rates of 15 to 20%. These
`errors introduce bias into genetic evaluations and can
`reduce rates of genetic gain. Genotypes can provide
`varying amounts of information based on the number
`of markers on the chip, but this can be accounted for
`using a process called imputation.
`The most familiar phenotypes are those related
`directly to cow productivity, such as milk, butterfat,
`and protein yields. In typical milk recording programs,
`those yields are recorded on a monthly basis under the
`supervision of a technician using certified meters to en-
`sure accurate observations. Milk samples taken during
`the test are sent to a laboratory for measurement of
`fat and protein contents and somatic cell score. The
`test-day milk yield and composition information are
`then sent to a dairy records processing center, where
`the observations are adjusted to a mature-equivalent
`basis. Corrections are also applied to account for dif-
`ferences in lactation length and milking frequency so
`that production records from animals in different en-
`vironments are comparable. This is similar in principle
`to comparing feed properties on a dry matter basis.
`Finally, lactation yields are estimated from test-day
`observations using best prediction (VanRaden, 1997).
`Schmidt et al. (1988) cover these topics in much greater
`detail, including many worked examples.
`
`The US Genetics Industry
`
`The Council on Dairy Cattle Breeding (CDCB;
`https://www.uscdcb.com/) is responsible for receiving
`data and computing and delivering genetic evaluations
`for US dairy cattle, a role that it assumed from the
`Animal Improvement Program (AIP; http://aipl.
`
`arsusda.gov/) of the USDA’s Agricultural Research
`Service in December 2015. The Council has 3 voting
`board members from each of the 4 major sectors of
`the dairy improvement industry. The Purebred Dairy
`Cattle Association (PDCA; http://www.purebred-
`dairycattle.com/) represents the national dairy breed
`registries, such as the American Jersey Cattle Asso-
`ciation and Holstein Association USA. The National
`Association of Animal Breeders (NAAB; http://www.
`naab-css.org/) represents the individuals and organi-
`zations that produce semen for use in AI. The dairy
`records processing centers (DRPC) receive data from
`farms and milk recording laboratories, use those data
`to provide herd management tools, and forward data
`to other organizations, such as CDCB and PDCA, on
`behalf of their customers. The national Dairy Herd
`Information Association (DHIA) represents all of the
`organizations involved in dairy production recording,
`including the DRPCs, maintains national standards
`for data recording, and certifies devices for use in milk
`recording programs. Other industry groups, such as
`genotyping laboratories, have nonvoting representation
`on the CDCB Board of Directors.
`
`Evaluation Releases
`
`Official genetic evaluations are released 3 times each
`year (April, August, and December). In addition to
`providing authoritative genomic PTA for all animals
`evaluated in the United States, these releases include
`information from the multiple-trait across-country
`evaluations computed by the International Bull Evalu-
`ation Service (Interbull; Uppsala, Sweden). Additional
`information, such as animal rankings for the lifetime
`net merit, cheese merit, fluid merit, and grazing merit
`indices (VanRaden and Cole, 2014), are provided when
`new official releases are published. Monthly genomic
`evaluations for young bulls are provided to the organi-
`zation that nominated each animal for genotyping, and
`evaluations for females are sent to the respective breed
`associations. These evaluations include genomic PTA,
`reliabilities, and genomic inbreeding values. Recently
`received genotypes are processed each week to generate
`approximate genomic evaluations for new animals, but
`those releases include only approximated reliabilities
`and genomic inbreeding because of computational limi-
`tations. Animals with genotypes that became usable
`since the previous weekly evaluation (e.g., because of
`corrected pedigrees) also receive weekly evaluations.
`The goal of this release schedule is to provide animal
`owners with accurate information for decision making
`as quickly as possible.
`
`This document is registered to JOHN COLE
`
`Exhibit 1021
`Select Sires, et al. v. ABS Global
`
`

`

`CHAPTER 5-23: PRODUCTION EFFICIENCY AND GENETIC SELECTION
`SELECTION FOR INCREASED PRODUCTION
`
`335
`
`Response to Selection
`
`The production of milk, fat, and protein by cows in
`the US dairy herd has increased dramatically in the past
`75 years. During this time, the number of cows in the
`national herd has decreased from 23.67 million in 1940
`to 9.26 million in 2014, whereas average annual milk
`yield has increased from 2,097 to 10,092 kg over the
`same time period (Figure 2a,b). The average Holstein
`in 2014 produced almost 5 times as much milk as cows
`in 1940, and much of this improvement in productivity
`is due to genetic selection. The genetic improvement for
`milk yield averaged 65.9 kg per year between 1960 and
`2013, accounting for roughly half of the total improve-
`ment in milk yield observed over the same period. Simi-
`larly, genetic improvement for fat yield averaged 2.3 kg
`per year from 1960 to 2013, whereas that for protein
`was 2.1 kg per year between 1970 and 2013. The rate
`of genetic improvement has remained similar for cows
`and bulls. Although the emphasis placed on milk yield
`in selection indices has decreased over time [from 52%
`in the 1977 PD$ (Predicted Difference Dollars) index
`to 1% in 2014 NM$], milk volume remains important
`in some markets (Florida and Southeast marketing
`areas) and the emphasis on milk solids has remained
`largely consistent (48% in PD$ and 42% in NM$). Even
`though milk receives relatively little emphasis in NM$,
`genetic merit for volume continues to increase due to its
`correlations with the other traits in the index (Table 2).
`The CDCB publishes 4 selection indices to account
`for differences in how farmer are paid for their milk,
`fat, and protein (VanRaden and Cole, 2014). The net
`merit index (NM$) is based on the average value of
`
`Figure 2. The number of cows in the US national dairy herd
`(a) and the average milk yield per cow per year (b) between 1940
`and 2014. Source: Milk Production, Disposition, and Income Annual
`Summary
`(http://usda.mannlib.cornell.edu/MannUsda/viewDocu-
`mentInfo.do?documentID=1105) .
`
`Table 2. Correlations of predicted transmitting abilities (PTA) for individual traits in the lifetime net merit (NM$), cheese merit (CM$), fluid
`merit (FM$), and grazing merit (GM$) indices with the overall index, and the expected genetic response of those traits to selection on the index
`per year and decade1
`
`Correlation of PTA with index
`
`Expected genetic progress from NM$
`
`Trait
`
`2014
`NM$
`
`2014
`CM$
`
`2014
`FM$
`
`2014
`GM$
`
`
`
`PTA
`change/year
`
`Breeding value
`change/decade
`
`0.56
`0.64
`0.60
`0.62
`Protein yield
`0.65
`0.69
`0.69
`0.70
`Fat yield
`0.39
`0.62
`0.38
`0.46
`Milk yield
`0.70
`0.64
`0.68
`0.68
`Productive life
`−0.43
`−0.36
`−0.46
`−0.44
`Somatic cell score
`0.11
`0.08
`0.09
`0.09
`Udder composite
`0.11
`0.09
`0.11
`0.11
`Feet/legs composite
`−0.19
`−0.20
`−0.20
`−0.20
`Body size composite
`0.49
`0.29
`0.37
`0.35
`Daughter pregnancy rate
`0.23
`0.15
`0.14
`0.15
`Heifer conception rate
`0.48
`0.31
`0.35
`0.34
`Cow conception rate
`Calving ability dollars2
`0.41
`0.36
`0.36
`0.37
`1Data from VanRaden and Cole (2014; http://aipl.arsusda.gov/reference/nmcalc-2014.htm).
`2Calving ability dollars is a weighted average of sire and daughter calving ease and stillbirth.
`
`4.7
`7.2
`134
`0.64
`−0.04
`0.04
`0.05
`−0.09
`0.22
`0.10
`0.34
`2.8
`
`94
`144
`2,679
`13
`−0.75
`0.75
`1.04
`−1.80
`4.4
`2.0
`6.7
`57
`
`This document is registered to JOHN COLE
`
`Exhibit 1021
`Select Sires, et al. v. ABS Global
`
`

`

`1Data from Bull Lists by Breed–Sorted by Merit Index, December 2015 (https://www.uscdcb.com/dynamic/sortnew/current/index.shtml).
`BRANDVALE STOIC 266-ET
`NO-FLA BARCLAY
`MR PRE DIRECTOR 57512-ET
`S-S-I MONTROSS JEDI-ET
`TRIPLECROWN GATEDANCER-ET
`MR MEGA-DUKE 54608-ET
`MR MOGUL DELTA 1427-ET
`SEAGULL-BAY CHARISMATIC-ET
`SEAGULL-BAY SUPERSIRE-ET
`BACON-HILL PETY MODESTY-ET
`
`SEAGULL-BAY COMANCHE-ET
`
`
`
` NO-FLA ALTAFIREUP
` ABS RAIDEN-ET
` MR MEGA-DUKE 54608-ET
`
` MR OAK DELCO 57279-ET
` MR MOGUL DELTA 1427-ET
`
`
`
`
`
`Bull name
`
`
`
`FM$
`
`Bull name
`
`
`
`NM$
`
`Bull name
`
`336
`
`GENETIC SELECTION PROGRAMS
`
`Cheese merit
`
`Fluid merit
`
`Lifetime net merit
`
`December 2015 genetic evaluations; bulls have similar rankings across different indices1
`Table 3. The name and predicted transmitting abilities of the top 10 available Holstein bulls for lifetime net merit (NM$), fluid merit (FM$), and cheese merit (CM$) from the
`
`milk and components across marketing orders, but the
`fluid merit (FM$) and cheese merit (CM$) indices
`provide tools for farmers whose milk is used principally
`for bottling or cheese manufacturing, respectively. The
`grazing merit index is useful for pasture-based dairies
`who often want greater emphasis on fertility in the bulls
`they use. Purebred dairy cattle associations also have
`their own selection indices for ranking animals, such
`as Holstein Association’s Total Performance Index and
`the American Jersey Cattle Association’s Jersey Per-
`formance Index. Rankings are generally similar across
`indices, but the top animals for each index sometimes
`differ, as shown in Table 3 for NM$, FM$, and CM$
`(grazing merit is not included Table 3 because sorted
`bull lists are not currently published for that index).
`For example, 6 bulls appear in the top 10 for both NM$
`and FM$, but they are ranked differently. Most of these
`sires are young bulls with genomic evaluations but no
`daughters. Before the introduction of genomics, semen
`was not offered for sale until a bull completed progeny
`testing, with the objective of obtaining performance
`data from about 100 milking daughters.
`
`Genetic Lag
`
`Cows on commercial dairies typically have lower ge-
`netic merit than elite cows used to breed young sires
`and bull dams, and this difference is referred to as
`genetic lag. This lag is often easy to see when PTA for
`bulls and cows are plotted together, such as in Figure
`3, which shows the genetic trend for Holsteins born
`between 1957 and 2014. It seems intuitive to many
`people to conclude that routine genomic testing of all
`heifer calves would provide a one-time increase in aver-
`age genetic merit but that the trend would not change.
`However, this is not true. The routine use of genomic
`testing provides better information than traditional
`PTA about an animal’s true genetic merit because it
`tracks the chromosomes actually inherited from each
`parent, rather than assuming that hypothetical “aver-
`age” chromosomes were inherited. Genomic selection,
`as anticipated, is producing changes in the parameters
`in the breeder’s equation. The first major change was
`to the reliability term in the numerator of Equation
`[1] (VanRaden et al., 2009). García-Ruiz et al. (2016)
`also recently documented large changes in the selection
`intensity term in the numerator, particularly for lowly
`heritable traits such as fertility, and the generation in-
`terval in the denominator of Equation [1]. This means
`that the slopes of the genetic trend lines are changing,
`as well as their heights, and that PTA for bulls and
`cows both are improving faster under genomic selection
`than