Veterinary · Pre-clinical and foundation · also MSVE AGB-211
Selection and mating systems that improve herds and flocks, sire evaluation, genomic selection and the conservation of animal genetic resources.
In the exams: ICAR AICE-JRF core group 05; ICAR AIEEA-PG 14 (AGB); EGEL Plus reproduction and genetics 15
Livestock and poultry breeding is the applied half of animal genetics. The companion subject, principles of animal genetics, explains how genes are inherited, how gene frequencies change and how heritability and breeding values are defined. This subject uses those ideas to make decisions: which animals become parents, how they are mated, how a national herd is organised so that genetic gain made in a few elite animals reaches millions of farms, and how breeds that are not commercially fashionable are kept from disappearing. Genetic change made this way is cumulative and permanent. A gain in feed efficiency or disease resistance, once fixed in a population, is paid for once and returns every generation after that, which is why breeding sits at the base of every livestock development plan.
The chapters cover selection methods and response, selection indices, selection experiments, inbreeding in closed herds, crossbreeding and heterosis, sire evaluation and progeny testing, breeding strategies and national policy, genomic selection, conservation of animal genetic resources and laboratory animal breeding. Examples come from cattle, buffalo, sheep, goats, pigs and poultry in South Asia, Africa, the Americas, Europe and Oceania, because a smallholder dairy programme and an integrated broiler company apply the same equations very differently.
The subject is examined as Animal Genetics and Breeding (AGB): in ICAR AIEEA-PG (Animal Sciences), ICAR AICE-JRF/SRF core group 05, ASRB NET and state veterinary officer papers, with recurring numericals on response, generation interval, heterosis, inbreeding and progeny-test accuracy; and in licensing and graduation tests such as Mexico's EGEL, Nepal's NLEV and the Philippine licensure test, which ask applied basics such as choosing a crossbreeding system or reading a breeding value.
Selection is the process of deciding which animals become parents of the next generation and how many offspring each is allowed to leave. Natural selection acts through differences in survival and fertility; artificial selection is imposed by the breeder. The breeder never sees an animal's breeding value (the sum of the average effects of its genes, twice the expected deviation of its progeny from the population mean). The breeder sees phenotypes, and estimates the breeding value from them. An estimated breeding value (EBV) is a weighted deviation of the available records from the mean, EBV = b(P − μ), where the weight b depends on heritability, the relationship between the recorded animal and the candidate, and the number of records.
The value of any selection method is measured by its accuracy, the correlation between the EBV and the true breeding value. Accuracy squared is called reliability. Five sources of information can be used, alone or together. Individual (mass) selection uses the candidate's own phenotype. Pedigree selection uses the records of ancestors. Family or sib selection uses the average of full or half sibs. Progeny testing uses the average of the candidate's offspring. Combined selection weights all of these together in an index or in a best linear unbiased prediction (BLUP) model. The choice depends on heritability, whether the trait is expressed in the candidate at all, the age at which records become available and the cost of recording.
With individual selection on one record, the regression of breeding value on phenotype is h² and the accuracy is h. It is simple and fast, and works well for moderately to highly heritable traits recorded early in both sexes: growth rate, weight at a fixed age, backfat depth, fleece weight, egg weight. Repeated records raise accuracy by averaging out temporary environmental effects.
Pedigree selection uses ancestors. It is available at birth, but its accuracy is capped: one parent's record gives h/2 and the mid-parent average h/√2. It is best used to choose which young animals enter a test, such as bull calves for progeny testing.
Sib selection uses the mean of full or half sibs. It is essential for sex-limited traits (milk and eggs are not expressed in males), for carcass traits that require slaughter, and for low heritability. As the number of sibs grows, the accuracy of a half-sib mean approaches 0.5 and of a full-sib mean about 0.71. Cockerels are chosen on their sisters' egg records, and full sibs of pig candidates may be slaughtered at a test station for carcass data. Family selection keeps or rejects whole families on their mean; within-family selection keeps the best in each family, which removes common-environment effects (litter, hatch) and holds inbreeding down.
Progeny testing estimates a parent's breeding value from the performance of its offspring. Each offspring receives a random half of the parent's genes, so the mean of many offspring converges on half the parent's breeding value and accuracy can approach 1. For n progeny, accuracy is √[n/(n + k)], where k = (4 − h²)/h². With heritability 0.25, k = 15, so 15 daughters give an accuracy of about 0.71 and 60 daughters about 0.89. Because k is large when heritability is low, progeny testing is most valuable for low-heritability, sex-limited traits such as milk yield and female fertility, and it is used mostly for males, which can have many offspring through artificial insemination.
Its cost is time and money. A dairy bull is five to six years old before his daughters complete a first lactation and his proof is published, so the generation interval is long and the waiting bulls must be paid for. This accuracy-versus-interval trade-off is the central problem of dairy breeding, and genomic selection was adopted largely to escape it.
Combined selection uses every relative at once, classically through a selection index and now through BLUP under an animal model, which weights own, ancestor, sib and progeny records by the information each carries.
Most breeding goals involve several traits. Three methods are classical. Tandem selection improves one trait at a time until it is satisfactory, then moves to the next; it is simple but slow, and progress in one trait can be undone later through unfavourable genetic correlations. Independent culling levels set a minimum standard for each trait; an animal failing any one is culled whatever its other merits. The method suits traits measured at different ages, because culling can be done in stages (for example, at weaning and again at yearling age), saving rearing costs. The selection index scores each animal on a weighted sum of all traits, so that excellence in one trait can compensate for a weakness in another.
Hazel and Lush (1942) compared the three and showed that the index is never less efficient than independent culling levels, which in turn are never less efficient than tandem selection. For n uncorrelated traits of equal economic importance and heritability, an index is about √n times as efficient as tandem selection. In practice, breeding organisations often use a combination: independent culling for traits that are thresholds (structural soundness, genetic defects, semen quality) and an index for the traits that are ranked.
The response to selection (R) is the difference between the mean of the offspring of selected parents and the mean of the generation they came from. For selection on the individual's own phenotype, R = h²S, where S, the selection differential, is the mean phenotype of the selected parents minus the population mean. Because S depends on the scale of measurement and on how many animals are kept, it is standardised: the selection intensity i is S divided by the phenotypic standard deviation σP. Then R = i h² σP = i h σA, where σA is the additive genetic standard deviation. In the general form, R = i r σA, where r is the accuracy of whichever selection method is used.
Under truncation selection of a normally distributed trait, intensity depends only on the proportion selected. Standard values are 0.80 when 50% are kept, 1.27 at 25%, 1.40 at 20%, 1.76 at 10%, 2.06 at 5% and 2.67 at 1%. Males are usually selected far more intensely than females, because one male can mate many females, particularly with artificial insemination; in dairy herds almost all heifers may be needed as replacements, so female intensity is low. When male and female selection differ, the response is the average of the two sexes' contributions: R = (iM + iF)/2 × h σA for mass selection.
Observed change must be compared with an unselected control, or corrected for environmental trend, before it can be called genetic.
Breeders care about progress per year, not per generation. The generation interval (L) is the average age of the parents when their offspring that will themselves become parents are born. It is calculated separately for sires and dams and averaged. Typical ranges are about one year in poultry, one to two years in pigs, two to four years in sheep and goats and four to seven years in cattle and buffalo, depending on age at first calving, herd life and whether sires are progeny tested. Response per year is ΔG = i r σA / L.
In dairy cattle four selection pathways operate at once, as Rendel and Robertson (1950) described: sires to breed sons (SS), sires to breed daughters (SD), dams to breed sons (DS) and dams to breed daughters (DD). Annual genetic gain is the sum of the four genetic superiorities divided by the sum of the four intervals: ΔG = (iSS rSS + iSD rSD + iDS rDS + iDD rDD) σA / (LSS + LSD + LDS + LDD). The male pathways, with their high intensity and accuracy, supply most of the gain, and the dams-to-daughters pathway supplies little because almost all heifers are kept.
Every design decision trades one term against another: progeny testing raises r but lengthens L; young bulls shorten L but lower r; keeping fewer animals raises i but increases inbreeding. AI, MOET, sexed semen and in vitro embryo production shift these balances by giving each selected parent more offspring.
Selecting for one trait changes others that are genetically correlated with it. The correlated response in trait Y when selection is on trait X is CRY = i hX hY rA σPY, where rA is the additive genetic correlation. Indirect selection on X can beat direct selection on Y when iX hX rA is larger than iY hY, for example when Y is expressed late, only in one sex, or is expensive to record. Scrotal circumference in young bulls is used this way, because it is genetically correlated with the age at puberty of their half-sib heifers. Correlated responses can also be harmful: long-term selection for milk yield has been accompanied by a decline in dairy cow fertility, and rapid growth in broilers by leg and cardiovascular problems. This is why modern breeding goals include fertility, health and welfare traits.
The same trait measured in two environments can be treated as two genetically correlated traits (Falconer, 1952). When that correlation is well below 1, there is genotype-by-environment interaction: the ranking of sires in a temperate, high-input system may not hold in a hot, low-input system. This is one of the main arguments for selecting breeding stock in, or for, the environment in which their offspring will live.
Response is also limited by the loss of additive variance that selection causes through gametic phase disequilibrium (the Bulmer effect), by drift and inbreeding in small populations, by natural selection opposing artificial selection, and by recording and parentage errors, which directly reduce accuracy.
A selection index combines several measurements into one score that predicts overall genetic merit as accurately as possible (Smith, 1936, for plants; Hazel, 1943, for animals). The target is the aggregate genotype, H = a1g1 + a2g2 + … , where g is a breeding value and a its economic weight. The index is I = b1x1 + b2x2 + … , where each x is a measurement as a deviation from its mean; measurements may be records on relatives, early records or indicator traits.
The weights b are chosen to maximise the correlation between I and H, which is the same as minimising the error of prediction. They are found by solving the normal equations P b = G a, where P is the matrix of phenotypic variances and covariances among the measurements and G is the matrix of genetic covariances between the measurements and the objective traits. When the measured traits are the objective traits, are recorded on the candidate itself and are uncorrelated both genetically and phenotypically, the solution reduces to bi = ai hi²: each trait is weighted by its economic value times its heritability.
The accuracy of the index is rIH = σI/σH, and the expected response in the aggregate genotype is RH = i σI. The expected change in each individual trait j is i × Cov(I, gj)/σI. Responses in individual traits can be unexpected, including a fall in a trait with positive economic weight if it is unfavourably correlated with a more heavily weighted trait. Checking these predicted responses is part of designing any index.
The economic weight of a trait is the change in profit per unit change in that trait, other traits held constant. It comes from a profit function or bio-economic model and depends on prices, costs and payment systems, so weights differ between countries and over time: milk volume is worth less where milk is paid on solids.
Variants of the basic index serve special aims. A restricted index (Kempthorne and Nordskog, 1959) holds the genetic change in one trait at zero while improving the others, for example improving growth without increasing mature size. A desired-gains index (Pesek and Baker, 1969) finds weights that produce specified relative gains without needing economic values.
Most dairy countries now publish a national total-merit index that combines production, longevity, fertility, udder health, conformation and calving traits in money units. Examples include Net Merit (NM$) in the United States, Breeding Worth (BW) in New Zealand, the Balanced Performance Index (BPI) in Australia, the Profitable Lifetime Index (PLI) in the United Kingdom and the Nordic Total Merit (NTM) in Denmark, Finland and Sweden. Over the last decades the share of weight on milk volume has fallen and the share on fitness traits has risen in most of these indices. Beef, sheep and pig programmes publish comparable terminal-sire and maternal indices.
The classical selection index assumes that means, variances and covariances are known and that all candidates are compared in the same environment. In real herds, animals are recorded in different herds, years and seasons, parents have been selected and matings are not random. Henderson's best linear unbiased prediction (BLUP), developed from the 1950s and adopted widely from the 1970s, solves these problems. BLUP estimates fixed effects such as herd-year-season by generalised least squares and predicts random genetic effects simultaneously through Henderson's mixed model equations.
In effect BLUP is a selection index with fixed effects estimated simultaneously and relatives weighted through the numerator relationship matrix (A). An animal model predicts every animal in the pedigree and accounts for the merit of mates; it became the standard for national dairy evaluations from the late 1980s and allows genetic trend to be estimated directly.
For practical purposes, the pedigree index of a young animal is PI = ½ EBV(sire) + ½ EBV(dam), and its accuracy is limited because it contains no information on the Mendelian sampling term, the random half of each parent's genes that the animal inherited. Only the animal's own records, its progeny or its genotype can capture that term.
A selection experiment applies known selection to a population for several generations and measures what happens. Its purposes are to estimate realised heritability and genetic correlations, to test whether theory predicts response, to find out how long response continues and to see which traits change alongside the selected one. Because environmental conditions change from generation to generation, the response in a selected line must be measured against a control. The two common designs are a selected line plus an unselected randomly bred control line, and divergent (two-way) selection with an upward and a downward line, where the difference between the lines doubles the signal and cancels the shared environmental trend. Replicate lines are needed to measure the variation caused by random genetic drift, which in small lines can be as large as the response itself.
Realised heritability is the regression of cumulative response on cumulative selection differential; in the simplest case, cumulative response divided by cumulative selection differential. In divergent selection it is the divergence between lines divided by the sum of their selection differentials. Randombred control populations have also been used to measure the genetic trend of commercial poultry stocks.
The oldest continuing experiment is the Illinois selection for oil and protein in maize, begun in 1896, which has responded for more than a century. Long-term selection in mice, Drosophila and chickens shows similar patterns; chicken lines selected divergently for high or low eight-week body weight at Virginia, USA, since 1957 now differ many-fold, with appetite and reproductive problems in the low line and obesity and reduced fertility in the high line.
Response is usually close to prediction for the first few generations, and often asymmetric between divergent lines because of inbreeding depression, unequal gene frequencies, scale and maternal effects or natural selection. Lines eventually slow and may plateau, when additive variance is exhausted or natural selection balances artificial selection; a mean that falls back when selection is relaxed shows fitness was opposing it. Robertson (1960) showed theoretically that, for genes of small additive effect, the total response cannot exceed 2Ne times the response in the first generation, and the half-life of response is at most about 1.4Ne generations, where Ne is the effective population size. Larger populations therefore give longer and larger response, and new mutations continually add variance, which is why broiler growth and feed conversion have improved for decades without a plateau. The practical lessons are to include fitness traits in breeding goals, manage effective population size, and use a control line or BLUP genetic trend to separate genetic from environmental change.
Inbreeding is the mating of animals more closely related than the average of the population. The inbreeding coefficient F (Wright, 1922) is the probability that the two alleles at a locus in an individual are identical by descent, that is, copies of one allele carried by a common ancestor. It is also the correlation between uniting gametes and the proportional reduction in heterozygosity relative to the base population. The base population, whose animals are assumed non-inbred and unrelated, must always be stated, because F is relative to it.
Wright's path formula gives FX = Σ (½)^(n1 + n2 + 1) (1 + FA), summed over every path through a common ancestor A, where n1 and n2 are the generations from sire and dam back to A and FA is the ancestor's own inbreeding. For large pedigrees the tabular method builds the relationship matrix by computer. An individual's inbreeding is half the additive relationship between its parents.
Standard values for offspring of non-inbred common ancestors are: parent × offspring and full sibs 0.25; half sibs, grandparent × grandchild and uncle × niece 0.125; first cousins 0.0625. Repeated full-sib mating raises F to 0.25, 0.375, 0.5 and 0.594 in the first four generations and to about 0.986 after twenty.
In a closed population of finite size, inbreeding rises every generation even with random mating, because all animals eventually descend from a limited number of ancestors. The rate is ΔF = 1/(2Ne), where Ne, the effective population size, is the size of an idealised population that would lose heterozygosity at the same rate. With Nm breeding males and Nf breeding females, Ne = 4NmNf/(Nm + Nf), so ΔF ≈ 1/(8Nm) + 1/(8Nf). The sex with fewer breeding animals dominates: a herd of 200 cows served by 2 bulls has Ne of about 8, not 202. After t generations, Ft = 1 − (1 − ΔF)^t.
Ne is reduced further by unequal family sizes, fluctuating numbers (Ne over several generations is the harmonic mean of the generation values, which is dominated by the smallest) and selection, which concentrates offspring in the best families. It is increased by equalising contributions: if each sire is replaced by one son and each dam by one daughter, ΔF ≈ 3/(32Nm) + 1/(32Nf) (Gowe, Robertson and Latter, 1959), roughly half the random-mating rate. Many widely used commercial breeds have large census sizes but small effective sizes, estimated in the tens to low hundreds, because AI concentrates the population on a few sires. Pedigree and genomic studies in several countries have shown rising inbreeding in major dairy breeds.
Inbreeding increases homozygosity. Its first practical effect is the appearance of recessive defects, which carriers pass on silently until two carriers are mated; well known examples in cattle include bovine leukocyte adhesion deficiency, complex vertebral malformation and various forms of dwarfism and hairlessness in particular breeds. Its second effect is inbreeding depression, the fall in the mean of a trait as F rises. Depression is greatest in fitness traits (fertility, survival, disease resistance, litter size, hatchability), which have low heritability and much dominance variance, and smallest in highly heritable traits such as carcass composition. It is usually expressed as the change in the trait per 1% increase in F; published estimates for milk yield are of the order of tens of kilograms per 1% F, varying by breed and country.
Inbreeding has uses: line breeding keeps a high relationship to an outstanding ancestor, inbred lines (in poultry and pigs in the past, in laboratory animals today) provide uniform parents for crossing, and test matings expose carriers of recessive defects.
In closed herds and nuclei inbreeding is managed by using enough sires, limiting offspring per sire, mating to minimise coancestry, bringing in unrelated sires and, in organised programmes, optimal contribution selection (Meuwissen, 1997), which maximises gain for a set rate of inbreeding. Genomic inbreeding is measured from runs of homozygosity (ROH), whose length indicates how far back the common ancestor lived.
Outbreeding is the mating of animals less closely related than the population average. Outcrossing is the mating of unrelated animals within a breed, used to bring new genes into a line or reduce accumulated inbreeding. Crossbreeding is the mating of animals of different breeds. Grading up is repeated crossing of non-descript or local females, and their female descendants, with males of one improved breed; the proportion of improved-breed genes rises to 50%, 75%, 87.5%, 93.75% and 96.875% in successive generations, and in many herdbooks animals of 15/16 or 31/32 grade may be registered as purebred. Species hybridisation crosses different species: the mule (mare × jack donkey) and hinny (stallion × jenny) are almost always sterile; cattle × yak hybrids are used in the Himalayan region, with fertile females and sterile males; cattle × bison crosses have been made in North America.
Crossbreeding has three purposes: heterosis; complementarity, combining breed strengths in one product (a small fertile dam with a fast-growing terminal sire); and rapid change of breed composition, far faster than within-breed selection.
In a two-breed terminal cross all offspring are marketed; they have full individual heterosis, but the purebred dams need a separate replacement source. A three-breed terminal cross mates a terminal sire to F1 dams, adding full maternal heterosis; it dominates commercial pig production (F1 Landrace × Large White sows with a Duroc, Hampshire or Pietrain boar) and lamb production. A backcross mates F1 females back to a parental breed.
Rotational crosses breed their own replacements. A two-breed rotation (criss-crossing) retains about two-thirds of F1 heterosis at equilibrium and a three-breed rotation about six-sevenths, but breed composition fluctuates. A rotaterminal system adds a terminal sire on part of the herd.
A composite (synthetic) breed is formed by crossing several breeds and then breeding the crossbreds among themselves. It keeps a proportion of heterosis without the need for continued crossing. Expected retained heterosis is 1 − Σpi², where pi is the proportion of each breed. A composite of four breeds in equal parts keeps 75% of F1 heterosis; a 5/8 : 3/8 composite keeps about 47%. Examples include Santa Gertrudis (Shorthorn × Brahman) and Brangus (Angus × Brahman) beef cattle, the Australian Milking Zebu, the Girolando dairy breed of Brazil (standard 5/8 Holstein, 3/8 Gir) and, in India, dairy composites such as Karan Fries (Tharparkar × Holstein-Friesian), Karan Swiss (Sahiwal × Brown Swiss) and Frieswal (Holstein-Friesian × Sahiwal), and the Bharat Merino sheep.
Crossing Bos taurus dairy breeds with Bos indicus (zebu) cattle has been the main tool for raising milk yield in the tropics. A widely cited FAO review of crossbreeding for milk in the tropics (Cunningham and Syrstad, 1987) found that F1 crosses performed well, often best, in most environments, while later generations produced by mating crossbreds among themselves (F2 and beyond) declined, partly because of loss of heterosis and partly because higher exotic inheritance suits the tropics less well. The practical conclusion adopted by many national policies, including India's, was to hold exotic inheritance at about 50%, with some states or production systems allowing more under better management.
The difficulty is sustaining the F1, which needs a continuing supply of purebred zebu females; crossbred × crossbred matings drift in composition and lose heterosis. Options are rotation of exotic and zebu sires, new composite breeds and better management. Because indiscriminate crossbreeding dilutes adapted local breeds, many policies restrict it to non-descript populations and select within recognised indigenous breeds.
Heterosis, or hybrid vigour, is the superiority of the crossbred offspring over the average of the two parental breeds or lines: H = F1 mean − mid-parent mean, and heterosis percentage = (F1 − MP)/MP × 100, where MP = (PA + PB)/2. Heterosis is measured against the mid-parent, not against the better parent; a cross can show positive heterosis and still be poorer than the better breed. Reciprocal crosses (A sire × B dam and B sire × A dam) are used to separate maternal effects from heterosis.
Three kinds are recognised. Individual (direct) heterosis is the advantage of a crossbred individual in its own performance, such as growth or survival. Maternal heterosis is the advantage of a crossbred dam as a mother: higher fertility, larger litters, more milk for the young, better mothering. Paternal heterosis, the advantage of a crossbred sire, is usually small and concerns libido, semen quality and conception rate.
The size of heterosis differs by trait. It is largest for reproduction, survival and other fitness traits, which have low heritability; moderate for growth and milk yield; and small for highly heritable carcass and product-quality traits. There is broadly an inverse relationship between a trait's heritability and the heterosis it shows. Heterosis is also larger when the parental breeds are more genetically distant, which is why Bos taurus × Bos indicus crosses show more heterosis than crosses between two European breeds.
Three explanations are given. The dominance hypothesis holds that each parental breed carries some unfavourable recessive alleles, fixed or at high frequency, and that the cross covers them with dominant favourable alleles from the other breed. The overdominance hypothesis holds that the heterozygote at a locus is itself superior to both homozygotes. Epistasis, interaction between loci, can add to or subtract from heterosis; favourable combinations built up within a breed can be broken by recombination in later generations, which is called recombination loss.
For a single locus with dominance deviation d, heterosis in the F1 is proportional to d × y², where y is the difference in allele frequency between the two parental populations. Heterosis therefore requires both dominance and a difference in gene frequency, which is why crosses between distant breeds show more. Under the dominance model, heterosis in any generation is proportional to the expected heterozygosity relative to the F1: the F2 keeps half, a backcross keeps half, and composites keep 1 − Σpi². Crossbred performance can then be predicted by summing the breed additive effects (in proportion to breed composition), individual heterosis in proportion to the offspring's heterozygosity, and maternal heterosis in proportion to the dam's heterozygosity, as in Dickerson's crossbreeding models. Inbreeding depression and heterosis are the two sides of the same dominance: inbreeding exposes it and crossing restores it.
In pig and poultry breeding, commercial animals are crosses of specialised lines that are selected separately. Combining ability describes how well a line crosses. General combining ability (GCA) is the average performance of a line in crosses with many others, reflecting additive effects; specific combining ability (SCA) is the deviation of a particular cross from what the two lines' GCAs predict, reflecting dominance and epistasis (Sprague and Tatum, 1942, in maize). Diallel crosses, in which every line is crossed with every other, estimate both.
Reciprocal recurrent selection (Comstock, Robinson and Harvey, 1949) selects two lines on the performance of their crosses with each other, so it improves both GCA and SCA. It has been used in layer breeding, where commercial hybrids are often crosses of two or more specialised White Leghorn lines for white-egg layers and of lines from heavier breeds for brown-egg layers, and in pigs. Most commercial broilers and layers are three- or four-way crosses from a pure line, grandparent, parent and commercial tier, so that each tier's birds carry the heterosis that suits their role: the parent female is crossbred for high egg and chick output, and the broiler itself is crossbred for growth and survival.
Where the crossbred female is the product, as with F1 sows, crossbred ewes and crossbred beef cows, maternal heterosis is often worth more than individual heterosis.
A sire used in artificial insemination may have thousands of daughters, so a wrong decision about one bull affects a population. Sire evaluation estimates a sire's breeding value, or his transmitting ability, which is half his breeding value (the expected superiority of his offspring). In beef cattle it is published as the expected progeny difference (EPD); in dairy cattle as the predicted transmitting ability (PTA) or, in some countries, as a breeding value. Both are expressed relative to a defined genetic base.
Early methods used the daughter average, the daughter-dam comparison and the equiparent index, I = 2D − M (twice the daughters' average minus the dams' average, since each daughter is the mean of sire and dam). They ignore herd and year differences, the number of daughters and non-random mating.
The contemporary comparison (Robertson and Rendel, 1954, UK) compared each daughter with heifers by other sires calving in the same herd and season; the herdmate comparison, adopted in the USA in the 1960s, did the same with herdmates. The deviation was regressed by n/(n + k), k = (4 − h²)/h², so bulls with few daughters were not over-rated. These methods still assumed sires were randomly spread across herds and contemporaries were of equal merit.
BLUP replaced these methods. The sire model estimated sires' transmitting abilities while treating herd-year-season as fixed; the sire–maternal grandsire model added the merit of the dam through her sire; the animal model, now standard, estimates breeding values for all animals and fully accounts for the merit of mates and for genetic trend. Test-day models, which analyse individual milk recordings instead of whole-lactation totals, allow for the environment of each test day and give more accurate proofs for animals with incomplete lactations. National evaluations are published several times a year, and each proof carries a reliability.
Because bulls are traded internationally, their proofs must be comparable across countries. Interbull, a permanent sub-committee of the International Committee for Animal Recording (ICAR, a body distinct from India's ICAR), runs multiple-trait across-country evaluation (MACE), which treats the same trait in different countries as genetically correlated traits and publishes each bull's merit on every participating country's scale. This is why an importing country should compare foreign bulls on its own scale rather than on the exporting country's.
Beef and sheep programmes use central testing and sire referencing (reference sires used across flocks to create genetic links); pig programmes test candidates and sibs on station and on farm; poultry breeders evaluate sires on daughters and sisters within pedigreed lines.
A conventional dairy progeny test has four steps: choose young bulls from elite sires and bull mothers on pedigree index; use their semen at random across many recorded herds; evaluate them when the daughters complete a first lactation; keep the best few as proven bulls and cull the rest.
The number of daughters per bull is a compromise. With heritability 0.25, about 50 daughters give accuracy near 0.88; doubling the daughters adds little accuracy but halves the number of bulls tested with the same inseminations, lowering selection intensity. Testing as many young bulls as the recording population allows, with a moderate number of daughters each spread over many herds and with verified parentage, usually maximises gain.
In India, progeny testing and pedigree selection have been run for Holstein-Friesian and Jersey crossbred cattle, Murrah and Mehsana buffalo and several indigenous breeds by state agencies, the National Dairy Development Board and ICAR institutes. The constraints are small recorded herds, incomplete recording, parentage errors and the long interval in buffalo. Where genomic selection has been adopted, young bulls are used on genomic values and daughter proofs mainly feed the reference population.
Organised breeding is usually a pyramid: a small, intensively recorded nucleus, a multiplier tier and a large commercial tier. Gain made in the nucleus flows down, and the commercial tier lags by about two generations' worth of gain, depending on the number of tiers.
In a closed nucleus, no animals enter the nucleus from below. In an open nucleus, the best animals found in the base population, usually females, are brought up into the nucleus. James (1977) showed that opening a nucleus can raise the rate of genetic gain by roughly 10 to 15% over a closed nucleus of the same size and markedly reduces inbreeding, provided the base animals are recorded well enough to be identified. Group breeding schemes, in which many cooperating farmers pool their best animals into a central nucleus and receive rams or bulls in return, were developed for sheep in New Zealand and Australia and later applied to cattle, sheep and goats in developing countries. Nicholas and Smith (1983) proposed MOET nucleus schemes for dairy cattle, in which multiple ovulation and embryo transfer produce many full and half sibs, so that bulls can be selected on sib information at a young age; the schemes promised gains comparable to or greater than progeny testing, at the price of higher inbreeding.
For smallholders with little recording, community-based breeding programmes (CBBP) let villages define the objective, record simple traits and share selected males, as with sheep and goats in Ethiopia.
A national breeding policy states, for each species and production system, the breeding objective, which breeds are improved by selection, where crossbreeding is used and at what exotic level, and how indigenous breeds are conserved. It needs identification, recording, AI and semen standards, herdbooks and an evaluation service. In Western Europe, North America and Oceania breeding is run mainly by cooperatives, breed associations and companies; in many developing countries the state or a national board runs semen production and AI.
India's policy for cattle and buffalo, broadly stated, has been selective breeding of recognised indigenous breeds within their breeding tracts; crossbreeding of non-descript cattle with exotic dairy breeds, chiefly Holstein-Friesian and Jersey, with exotic inheritance held at about 50% (higher in some states and systems); grading up of non-descript buffalo with improved breeds such as Murrah, and selection within recognised buffalo breeds. The Rashtriya Gokul Mission, launched in 2014, supports the development and conservation of indigenous breeds. National programmes have also set minimum standards for bull and semen production, expanded AI, and introduced sex-sorted semen. State policies vary, and the details change with each plan period, so current state documents should be consulted.
Other countries show other patterns. Brazil developed Girolando and selected dairy Gir; Pakistan emphasises Sahiwal cattle and Nili-Ravi buffalo; Italy's Mediterranean buffalo herdbook is selected for milk used in mozzarella; Kenya and other East African countries use dairy crossbreds in intensive smallholdings and Boran and other zebu breeds for beef; tropical countries use Bos indicus breeds and composites for heat and tick tolerance.
Dairy programmes combine milk recording, BLUP or genomic evaluation, a total-merit index, AI and increasingly sexed semen and embryos. Holstein-Friesian dominates high-input systems; Jersey suits payment on solids and hot climates; Sahiwal, Gir, Red Sindhi and Tharparkar are selected in South Asia and used in tropical crossbreeding. Beef programmes use EPDs, terminal crossing and composites, with Bos indicus types favoured in the tropics for heat and tick tolerance.
Buffalo breeding is concentrated in South Asia, which holds most of the world's buffalo, and in Egypt, Italy, China and South-East Asia. River buffalo (2n = 50) are dairy animals; swamp buffalo (2n = 48) are used mainly for draught and meat. Improvement is slowed by late maturity, seasonal breeding, long calving intervals, weak oestrus expression and the long generation interval, which makes progeny testing slow. Murrah, Nili-Ravi, Mehsana, Jafarabadi, Surti and Bhadawari are the best known river breeds, and Murrah is widely used for grading up. Crosses of river and swamp buffalo are fertile and have been used in South-East Asia to raise milk yield.
Sheep objectives differ with the product: fine wool (Merino types), meat (terminal sires on crossbred ewes, as in stratified hill-to-lowland systems), milk (Awassi, Lacaune, East Friesian) and carpet wool. The Booroola fecundity gene (FecB), a mutation that raises ovulation rate, originated in the Garole sheep of eastern India and Bangladesh and has been introgressed into other breeds. Goats are improved by within-breed selection in breeds such as Jamunapari, Beetal, Barbari, Sirohi and Black Bengal in South Asia, Boer for meat in Southern Africa and Saanen, Alpine and Anglo-Nubian for milk, often through community-based schemes and crossing.
Pig breeding is highly organised in a nucleus–multiplier–commercial pyramid run by breeding organisations. Maternal lines (Landrace, Large White or Yorkshire) are selected for litter size, piglet survival and mothering; terminal lines (Duroc, Pietrain, Hampshire) for growth, feed efficiency and lean meat. The commercial pig is usually a three- or four-way cross. Genomic selection is widely used, and the halothane (RYR1) mutation causing porcine stress syndrome has been eliminated from many lines by DNA testing.
Commercial poultry breeding is concentrated in a few international organisations that keep pedigreed pure lines and sell grandparent and parent stock. Broiler male lines are selected for growth and feed efficiency, female lines also for reproduction; layer lines for egg number, persistency, egg weight and shell quality using part-period and sib records. Sex-linked genes such as the slow-feathering allele and silver versus gold plumage colour allow sexing of day-old chicks. In India, public institutes have developed coloured dual-purpose varieties for backyard production, such as Vanaraja and Gramapriya.
Genomic selection, proposed by Meuwissen, Hayes and Goddard (2001), predicts breeding values from thousands of single-nucleotide polymorphism (SNP) markers spread across the whole genome. Because the markers are dense, most genes affecting a trait are in linkage disequilibrium with at least one marker, so the combined effect of all markers captures much of the additive genetic variance without needing to know which genes are causal.
First, a reference (training) population with genotypes and accurate phenotypes or proofs is used to estimate marker effects; then genotyped young candidates receive genomic estimated breeding values (GEBV), checked on a validation set whose records were withheld. Common methods are genomic BLUP (GBLUP), which replaces the pedigree relationship matrix with a genomic relationship matrix (G) built from the SNPs (VanRaden, 2008); Bayesian methods such as BayesA, BayesB and their successors, which allow some markers to have large effects and others none; and single-step genomic BLUP (ssGBLUP), which combines genotyped and non-genotyped animals in one evaluation by blending the pedigree and genomic relationship matrices.
Accuracy depends on the size of the reference population, the heritability of the trait, the effective number of independent chromosome segments in the population (which depends on Ne and genome length) and how closely the candidates are related to the reference animals. An approximate expression (Daetwyler and colleagues, 2008) is r = √[N h² / (N h² + Me)], where N is the number of reference animals and Me the number of independent segments. Low-heritability traits need much larger reference populations.
Schaeffer (2006) calculated that genomic selection of young bulls could roughly double the rate of genetic gain in dairy cattle while cutting the cost of proving bulls by more than 90%, mainly by shortening the generation interval on the sire-of-sons and dam-of-sons pathways. The United States published official genomic evaluations for dairy cattle from January 2009, and most major dairy countries followed within a few years. Young genomically tested bulls are now used heavily from about one year of age, and elite heifers and embryos are genotyped to choose bull mothers. Rates of genetic gain for production and for low-heritability traits such as fertility and health have risen markedly.
It helps low-heritability, sex-limited, late-expressed and expensive traits most, because a GEBV is available at birth and captures part of the Mendelian sampling term. New traits such as feed efficiency or disease resistance can be added once a reference population is recorded. Pig and poultry nuclei and several sheep and beef programmes also use it.
Limitations remain. The reference must be refreshed each generation, because accuracy decays as candidates become less related to it. Across-breed prediction is poor, a major obstacle for indigenous and small breeds. Inbreeding per year has risen in some dairy populations as intervals shortened. In smallholder systems the bottleneck is phenotype recording, not genotyping, although low-density panels with imputation cut costs.
Animal genetic resources (AnGR) are the species, breeds and populations of domestic animals and the genetic variation within them. FAO's first report on the State of the World's Animal Genetic Resources for Food and Agriculture (2007) found that a large proportion of recorded breeds were classified as at risk or of unknown status, and that breeds continued to become extinct. The main causes are the replacement of local breeds by a few high-output international breeds, indiscriminate crossbreeding within the breeding tracts of local breeds, changes in production systems and markets, loss of grazing land, disease epidemics and the culling that follows them, war and weak policy.
Local breeds carry adaptations that are hard to recreate: trypanotolerance in N'Dama and West African Shorthorn cattle, heat tolerance and tick resistance in zebu, drought survival and valued product qualities. They also have option value for future environments, markets and diseases, and cultural value. An extinct breed cannot be recovered.
Conservation begins with characterisation: surveys of population size, distribution and production systems, phenotypic description and molecular characterisation of diversity with recommended marker panels, now increasingly SNP arrays. Breeds are then classified by risk, based on the number of breeding females and males, the trend in numbers and the proportion of females bred pure. FAO's categories include extinct, critical, critical-maintained, endangered, endangered-maintained, not at risk and unknown; the numerical thresholds are set out in FAO guidelines and differ in national systems.
The international framework includes the Convention on Biological Diversity (1992), the FAO Global Plan of Action for Animal Genetic Resources adopted at Interlaken in 2007, and the Nagoya Protocol on access and benefit-sharing (adopted 2010, in force 2014). Countries report to FAO's Domestic Animal Diversity Information System (DAD-IS). In India, the ICAR National Bureau of Animal Genetic Resources (NBAGR) at Karnal characterises and registers breeds and gives each registered breed an accession number; the number of registered breeds increases as new breeds are recognised.
In situ (in vivo) conservation keeps live animals with farmers in their own production system, supported by incentives, breed societies and niche markets; it keeps adaptation and lets the breed evolve. Ex situ in vivo conservation keeps live animals on farms or centres away from their normal environment; it protects against local disasters but is costly and may lose adaptation. Ex situ in vitro conservation, or cryoconservation, stores frozen semen, embryos, oocytes, somatic cells or DNA in gene banks. It halts genetic change, needs little space and is insurance against catastrophic loss, but restoring a breed from semen alone requires backcrossing onto females of another breed for several generations, whereas embryos allow direct reconstitution. Techniques are well established for semen of cattle, buffalo, sheep, goats and pigs; freezing poultry semen gives variable fertility and bird embryos cannot be cryopreserved in the same way, so poultry conservation also relies on primordial germ cells and live flocks.
Genetic management of a conserved population aims to hold the rate of inbreeding low. A commonly used guideline is an effective population size of at least 50, equivalent to ΔF of 1% per generation, for short-term viability, with larger sizes needed to maintain adaptive potential over the long term. Practical rules are to use as many sires as possible, equalise the number of offspring per parent (for example replacing each sire by a son and each dam by a daughter), avoid mating close relatives, lengthen the generation interval and use optimal contribution methods. When funds are limited, priority among breeds may be set by combining risk status, the breed's contribution to overall genetic diversity (for example Weitzman's diversity approach) and its special traits.
Laboratory animals are bred to known genetic definitions so that experiments are reproducible. An inbred strain results from at least 20 consecutive generations of brother × sister (or parent × offspring) mating, giving F of about 0.986, so its animals are practically identical and homozygous. Branches separated for some generations are substrains; long-separated mouse substrains differ at several loci, which can change results. F1 hybrids between two inbred strains are uniform and vigorous but do not breed true.
Outbred stocks are closed colonies kept heterozygous, usually with ΔF below about 1% per generation, using many breeders and systematic (often rotational) mating that avoids relatives and equalises contributions. Other defined types include congenic strains, made by backcrossing a gene or chromosome segment into an inbred background for at least ten generations, so that, counting the F1 as N1, the unlinked donor genome left at generation Nn is about (½)^n, roughly 0.1% at N10 (marker-assisted 'speed congenics' reaches this in fewer generations); coisogenic strains, differing from the inbred strain at a single locus by mutation arising within it; recombinant inbred strains, derived by inbreeding from an F2 of two inbred strains and used for gene mapping; consomic strains, in which one whole chromosome has been substituted; and transgenic, knockout and genome-edited lines.
Mice and rats are bred in monogamous pairs (accurate pedigrees, use of postpartum oestrus), trios or harems. Mouse gestation is about 19 to 21 days and rat gestation about 21 to 23 days; both are weaned at about three weeks. The hamster's gestation is about 16 days, the rabbit's about 31 days and the guinea pig's much longer, about 59 to 72 days, with precocial young. Timed matings in mice are confirmed by the vaginal plug. Pheromone effects affect breeding: the Whitten effect (male urine synchronises oestrus in grouped females), the Bruce effect (exposure of a newly mated female to a strange male can block implantation) and the Lee–Boot effect (suppression or prolongation of oestrous cycles in females housed together without a male).
An inbred strain is kept as a foundation colony (strict pedigreed brother × sister mating), an expansion colony and a production colony a limited number of generations from the foundation, which limits drift. Strain identity is checked with genetic markers, now usually SNP panels. Health status is defined: specific-pathogen-free animals lack a listed set of pathogens; germ-free and gnotobiotic animals carry none or only known microorganisms. Cryopreserved embryos and sperm protect lines against loss and drift and reduce live animals kept, in line with the 3Rs.
International committees set mouse and rat nomenclature: inbred strains take capital letters or letter-number codes, substrains follow a slash, and F1 hybrids are named from both parent strains, female first.
A student needs to read and use the core formulas quickly: R = i r σA/L, heterosis percentage, retained heterosis, F from a pedigree, Ne from the number of sires and dams, and progeny-test accuracy. In entrance and PSC papers these give easy marks if units and bases are handled carefully. Know the difference between breeding value, transmitting ability and EPD, the order of efficiency of tandem, independent culling and index selection, and the parent breeds of the main composites. Practise explaining why a crossbred F1 does best and why inter se crossbreds fall back. That explanation is asked in viva and interviews again and again.
A field veterinarian or livestock officer uses breeding daily without calling it that. The work includes choosing semen for a smallholder's cow by reading PTAs, reliabilities and the index used in the country of proof, and advising against the use of a herd's own son on his half-sisters. It means recognising when a buffalo herd has too few bulls and planning a crossbreeding system to fit a farm's feed and management. It also means explaining to farmers why a crossbred bull from a crossbred cow is not a good source of replacement genes, and checking parentage and identification before animals enter a progeny-testing programme. The officer must understand and apply the state's breeding policy on exotic inheritance.
A senior officer designs and runs breeding programmes. That work means setting a breeding objective with economic weights, sizing a progeny-testing or genomic programme to the recording population, and managing a nucleus or open-nucleus scheme. The officer also monitors inbreeding in bull stations and semen banks and decides which indigenous breeds to conserve and how. The senior role also carries accountability. Programmes fail more often from poor identification, recording and parentage than from poor statistics, so the senior's job is to build and protect data systems, set semen and bull standards, and resist pressure to import genetics that the local environment cannot support.
A senior researcher works on the open questions: across-breed and multi-country genomic prediction for small and indigenous populations, and genomic evaluation with few phenotypes in smallholder systems. Other questions are breeding goals that include resilience, methane, feed efficiency and welfare, and genotype-by-environment interaction under climate change. Some work covers optimal contribution selection with genomic relationships, the value of gene editing alongside selection, and the design of conservation programmes that combine cryobanks with in situ incentives. Such research must be judged by whether it changes decisions on farms and in national programmes, and by honest measurement of realised genetic trend against prediction.
Core genetics theory (Hardy-Weinberg, variance components and estimation of heritability) is left to the animal-genetics-principles manual. This manual does not give FAO's numerical risk-status thresholds, current counts of registered Indian breeds, country-by-country semen standards or current exotic-inheritance limits state by state, because these change and differ between documents; consult the current FAO guidelines, NBAGR and state policies. The formulas for national total-merit indices are described but not reproduced, because their weights are revised often. Mixed-model equations, variance component estimation (REML) and the algebra of single-step genomic evaluation are described in words, not derived. Breed-by-breed descriptions and home tracts belong to the livestock production subjects.
Original practice questions written to the syllabus and pattern of the exams named in each question's examTags. They are not past papers and do not reproduce any board's questions. Every numerical answer was computed by script and the working is shown in the explanation.
These notes are free to read here, with no account.
What the Laisha app adds:
Free for ten days, then a subscription. We would rather write that here than have you find it later.