Monday, March 23, 2015

How would you design a practical breeding program?

From the July 2011 Dairy Newsletter

Most AI systems and Breed Associations sell linear-based or index-based “mating programs”.    These are designed to match a couple sire choices to each cow.    They make the following assumptions:

(1)  All genetic gain comes from sires selected.     (Your cows are assumed to be “zero” for all traits.)
(2)   The focus in all matings is to raise udder, feet and legs, stature, and angularity scores.
(3)   Avoiding performance depression depends only on avoiding pedigree inbreeding.

The basic design of such mating programs goes back to the 1960s.    Dr Walton of ABS pioneered the “Genetic Mating Service” [calculating an index mating composite] while Ron Long of Select perfected the “Linear Mating Service” [basing mating on two worst linear faults].      Little has changed since.

Weaknesses of “genetic” [index] computerized mating schemes

In fact, within the computer, the calculated “mating formula” would always breed every cow to the bull with the highest “index” (traits considered according to the index weighting).     Thus in such programs you are always asked to “rank” your cows (usually their DHI milk deviations) and to provide their sire.
The computer will then assign the “high” bull to your “high” cows, to the limit allowed one bull; then assign the “second high” bull to your “next high” cows, to the limit allowed one bull…   Only kicking out the high bull to your high cow if they share close pedigree relationship.    

You expected the mating program to provide trait correction – but all it does is calculate matings that will generate the highest “pedigree index”.     All the trait data the evaluator collected just goes into the AI stud’s data bank to anticipate which sires (their own and competing) might go up or down on future type proofs, so they can anticipate changes in market demand for individual bulls.

Weaknesses of “linear based” [trait corrective] mating schemes

The advantage here is that the evaluators go beyond collecting trait data for a computer to make sire choices; they usually make the choices based on their visual of the cow and memory of sire patterns.
Their computer is only used to print out results they put in (and store collected data for market uses).

The problem is with the limited number of traits (insufficient to fully describe the cow’s physique as it relates to all physical functions) and the necessity to focus on a small number of traits (two or three) to expect any heritability of the results (the more traits included, the lower the composite heritability).

Thus—typically in linear mating, you gain in stature, angularity, rear udder height, and get straighter legs with steeper foot angles—but you lose width, capacity, and strength over time (due to a constant  trait selection bias in favor of angularity).      You trade off old faults for new faults each generation.

Ultimately the weakness in both computer and hand mate systems is that the level of heritability % in individual traits, while high enough to lead us to whom the better bulls appear to be, is insufficient to insure trait correction.   

AI studs and Breed associations continue to offer “mating programs” because many dairymen want them and they have aided AI studs in gaining 100% of individual dairy’s semen purchases.    However, there has never been a University genetic research that ever proved these systems work—which is why typical geneticists advise to just random breed to high rank sires and avoid inbreeding when possible.   

Define your goals first, before selecting programs to reach them

I asked an experienced, successful dairyman what he would want as goals of a “breeding program”.
His responses were:
** Match sires to cows so as to avoid replicating faults possessed by the cows in their offspring.
** Cover the total cow physique in the mating, so that new faults (not possessed by the cows) are avoided in the offspring produced.
** Produce a physique capable of harvesting all of the genetic value gain implied in sire selections.
** Produce a physique that remains healthy, reproductive and feed efficient within the capacity gain generated by genetic selection for higher production.
** Produce cows that can sustain competitive production over long enough lifetimes that the herd is able to multiply, allowing for a second income stream from surplus cow sales, and create an opportunity for   culling as genetic improvement (rather than a constant turnover of cows from involuntary losses).

Adding the “qualitative” to the “quantitative” makes the difference

Six decades of “aAa” research reinforces the observation that “causality” in trait faults is based in the “qualitative” (analysis of the physique as a whole) rather than in the “quantitative” (measurement of traits as discretely separate from the physique).

Thus to cover the development of physiques capable of thriving under increasing production, the focus of mating should be on the physical expression—not the genetic ranking.

Selection of optimal mates is the function of genetic evaluation.    But ranking all sires on a common list without regard to qualitative differences in physique, will always lead mating in “likes to likes” direction and this “single trait selection” has been known from research to be the true cause of dairy performance depression (loss of heterosis “vigor”) (misnaming linebreeding as “inbreeding effects”).

Putting it together in a practical way

(1)    Analyse your breeding cows on the “aAa” method.
(2)    Sort the cow results according to their heterozygotic physical mating need.
(3)    Sort among ranked sires to identify the optimal choices for each group of cows.   (This is a simple process, as all AI sires are “coded” to indicate their physical mating benefit.)
(4)    Stock semen as needed in each group to breed all cows.

Knowing your needs in sires in advance can allow you to pursue acquiring semen at prices advantageous to your breeding budget.    Thus the temptation to buy “specials” (and find afterwards the sires chosen do not provide improved offspring in your herd, keeping culling costs high) can be avoided.

Will “inbreeding” be avoided?    The “aAa” groupings will prevent you from mating “likes to likes”.   The phenotype differences (between cow and sire) represent broad genotype differences, thus each mating is “heterozygous” at the level where it really counts—the genotype produced in conception.

Different phenotypes also usually result from variation in pedigrees.   So the inbreeding coefficients in “aAa” analyzed herds generally go down over time, even if some sire relationships appear.           

No comments:

Post a Comment