Friday, December 17, 2021

Alternatives to ranking indexes


CONCEPTIONS   Dairy Route Newsletter                                     Fall  2020


Many who were early adapters of Genomic selection who are now milking third or fourth generation heifers are seeking answers to why milk production is no longer improving.    The answer?    Look to the way your index ranking is calculated.  Milk may not be the goal of the ranking you have been following.

Calculation of “Lifetime Net Merit”  (LNM$)    as revised August 2020
There is a graphic explanation of the formulas for LNM$ and Holstein TPI on page 25 of the new International Protein Sires bull book.     To summarize, overall weighting is 44% Butterfat and Protein yields;  16% Type composites;  and 40% conception, calving ease and health indicators.  There are 14 individual traits included, five of which are composites of a dozen sub-traits.   Only 0.7% of the calculation derives from Milk yield volume.    LNM$ has a different goal.

“Net Merit” is an attempt to combine several traits that affect the cash flow we would get from producing milk.     The current formula is day and night difference from the original “Net Merit” of the 1970s, which was weighted 70% Milk yield and 30% Butterfat yield – no type, no fertility, no size considerations, no health calculations.   The inventors just  assumed  that all those traits would stick around as we focused on “single trait” production gains.     As we learned however, the traits we were ignoring mostly got worse.    Crossbreeding was the solution many tried, in a belief that “hybrid vigor” would fix what “indexing” had lost.    “Lifetime” Net Merit was thus an attempt from geneticists to regain our attention by providing a comprehensive index.

How heritable are indexes??
This has always been a question in my mind.    If you take a set of traits that are 25% heritable (butterfat, protein, milk yield) and combine them with a dozen other traits that are only 15% to as little as 2% heritable, what is the heritability of an unequally weighted composite index??   
No one knows, all that you are given is the “Reliability” (Rel%) of the statistical calculation.    In the case of Genomic estimates, these come out around 75% production, 60% type, 50% health.    As for the individual traits that add up to the Genomic rankings, geneticists still caution us that a progeny evaluation for any individual trait could be widely different from an original Genomic imputation for that trait (which is, 40% “Parent Average” + 60% “gene marker possession”).

What is the multi-generation effect of only using a selection index for breeding decisions?
The effect we are seeing is, in fact, “random” for production (a high milk, low bf% and pr% and a low milk, high bf% and pr% bull can be “equal” for LNM$ within the overall calculation).    The 56% of the selection index that is composed of generic type, health, fertility and calving linears in fact will become the overriding driver of the genes passed through each generation, until you reach that point where your cow genotypes are mirror images of the new bull genotypes.  
     Any chance of “heterosis” from “like to like” mating declines with each additional generation.

Genomics is just another attempt at “single trait” selection
What is the problem with focusing on a single trait over multiple generations?   It is the problem of “inbreeding depression”.    Traits we ignore get worse, while the lack of remaining heterosis means what we wanted quits responding further.    All our experience tells us that every four generations, geneticists throw out their latest formula as the population data says “progress” is not occurring at the rate predicted by their theory.   They will announce a “new and improved” Net Merit formula, but busy dairymen do not read articles with heavy math in them, and no one “gets it” that what they have been depending on may be failing them.     Why fail?   Won’t the new formula solve all the problems??    No, because all the old “parent average” pedigree values built into the indexes have screened out all the “outcross” sources prior to changing the formula.     With each new iteration, regardless of calculation, you have fewer ancestors available to produce new heterosis vigor.

In prior generations of cattle, under previous indexing systems, when “inbreeding depression” occurred the result was usually slower growth rates, slow repro, more stillbirths, harder heifer calving, more health problems, and ultimately shorter herdlife.    Visually, such cows would look “frail”, their udders would outgrow their frames and collapse at the age they should peak milk.

Today, under composite indexes dominated by “Health and Fitness”, many have initially seen a reduction in the above problems—usually a three generation success.    Then we get another kind of “inbreeding” depression – heifers no longer reach peak milk, and lack persistence.    In a broader view, we have turned “dairy” cows into “beef” cows, who only continue to milk under a very specific, expensive, high-input feeding and management regime.

How do you pick an “outcross” to restore heterosis to such a herd ??
You exchange “index “ selection (which is a “sire stack” method) to “cow family” selection.    You re-emphasize “maternal traits” looking for sires whose cow lines show annual improvement in actual yields, efficient calving intervals, and higher lifetime production totals.   “Longevity”  (lifetime measurement) is completely different from the “Productive Life” index for sires, which is a composite of traits assumed to prevent early age involuntary culling (added up before the first offspring ever calve and enter production).     Common sense tells us that if a cow has many calvings, producing over a full lifetime, staying in the herd in competition with all the new heifers that enter it, she has the good traits in her genotype that in the combination she represents actually achieve Longevity.       Where “Genomics” estimates potential, maternal line selection represents when it actually happened.      Plus, we get added bonuses: no cow lives a long time if she is a hard breeder or has a bad disposition or is prone to mastitis.    Thus the cows with proven longevity over multiple generations are just better cows.

Zoetis research herds from which their “wellness traits” data is derived, still show that of those cows who reach third and fourth calvings and are still physically functional, their production is usually 40% higher than what they did as first calf heifers.    Longevity maximizes production.

No comments:

Post a Comment