pages/KeepEPDsReal.html

---
layout: default
title: Keep EPDs Real
imageUrl: ../img/cannulated-cows.jpg
blurImageUrl: ../img/cannulated-cows_blur.jpg
motto: Keep EPDs Real
---

<p>
    In the <a href="https://podcasts.apple.com/us/podcast/selection-for-milk-in-the-cowherd-how-much-is-too-much/id964198047?i=1000431062519">March 4, 2019 episode of BeefWatch Podcast</a>, Dr. Travis Mulliniks of the University of Nebraska-Lincoln presented the research he conducted and published in <a href="https://doi.org/10.2527/tas2016.0006">"High milk production decreases cow-calf productivity within a highly available feed resource environment."</a>
    In the paper, he discovered that milk production in beef cattle has no correlation with calf weaning weights, but has a significant negative correlation with breedback rates.
    Dr. Mulliniks' paper is in agreement with several others on this topic, and was by far the largest and most conclusive study to find these results.
</p>
<p>At the conclusion of the podcast episode, Dr. Mulliniks stated</p>
<blockquote>As producers looking at bulls: we highly suggest, in our environment in Nebraska, from an Angus standpoint, an EPD of 22 or lower. ... Above that we start seeing an increase in that milking potential. But it's something that we really need to pay attention to, 'cause as our costs continue to go up, and we're not necessarily seeing an increase in the output. ... We really need to manage and moderate milk in a lot of our cow-calf setups.</blockquote>
<p>Mathematically, he is saying:</p>
<a href="https://www.codecogs.com/eqnedit.php?latex=\inline&space;\small&space;High\&space;Fertility=Low\&space;Milk\&space;Production=Low\&space;MILK\&space;EPD\neq&space;Low\&space;Weaning\&space;Weights" target="_blank"><img class="img img-responsive math" src="https://latex.codecogs.com/svg.latex?\inline&space;\small&space;High\&space;Fertility=Low\&space;Milk\&space;Production=Low\&space;MILK\&space;EPD\neq&space;Low\&space;Weaning\&space;Weights" title="\small High\ Fertility=Low\ Milk\ Production=Low\ MILK\ EPD\neq Low\ Weaning\ Weights" /></a>
<p><em><strong>WHAT!?!</strong></em></p>
<p>
    The MILK EPD is calculated from calf weaning weights. The <a href="https://www.angus.org/Nce/Definitions.aspx">Angus Association sort of admits it</a>, the <a href="http://nbcec.org/producers/sire_selection/manual.pdf">National Beef Cattle Evaluation Consortium knows it</a>, and professors in both <a href="https://www.worldcat.org/search?q=isbn%3A9781845939816">Scotland</a> and <a href="https://www.worldcat.org/search?q=isbn%3A0646387138">Australia</a> teach students how to calculate it.
    Mathematically, that means:
</p>
<a href="https://www.codecogs.com/eqnedit.php?latex=\inline&space;\small&space;Low\&space;MILK\&space;EPD=&space;Low\&space;Weaning\&space;Weights" target="_blank"><img class="img img-responsive math" src="https://latex.codecogs.com/svg.latex?\inline&space;\small&space;Low\&space;MILK\&space;EPD=&space;Low\&space;Weaning\&space;Weights" title="\small Low\ MILK\ EPD= Low\ Weaning\ Weights" /></a>
<p>
    Of course, Mulliniks' research said very clearly:
</p>
<a href="https://www.codecogs.com/eqnedit.php?latex=\inline&space;\small&space;Low\&space;Milk\&space;Production\neq&space;Low\&space;Weaning\&space;Weights" target="_blank"><img class="img img-responsive math" src="https://latex.codecogs.com/svg.latex?\inline&space;\small&space;Low\&space;Milk\&space;Production\neq&space;Low\&space;Weaning\&space;Weights" title="\small Low\ Milk\ Production\neq Low\ Weaning\ Weights" /></a>
<p>
    So both of these pieces of information lead to something more like:
</p>
<a href="https://www.codecogs.com/eqnedit.php?latex=\inline&space;\small&space;High\&space;Fertility=Low\&space;Milk\&space;Production\neq&space;Low\&space;MILK\&space;EPD=Low\&space;Weaning\&space;Weights" target="_blank"><img class="img img-responsive math" src="https://latex.codecogs.com/svg.latex?\inline&space;\small&space;High\&space;Fertility=Low\&space;Milk\&space;Production\neq&space;Low\&space;MILK\&space;EPD=Low\&space;Weaning\&space;Weights" title="\small High\ Fertility=Low\ Milk\ Production\neq Low\ MILK\ EPD=Low\ Weaning\ Weights" /></a>
<p>I contacted Mulliniks and asked him if there was something I missed. He replied</p>
<blockquote>
    The milk EPD is actually associated with milk production.
    If you reread the description for milk EPD, <strong>it is a direct correlation between milk</strong> and mothering ability <strong>on weaning weight</strong>, which would be the maternal component of weaning weight.
    So using bulls with high milk EPDs will increase milk production of your cowherd if you retain heifers out them.
    <br />
    &hellip;
    <br />
    Selecting for low milk EPD does not mean you will necessarily decrease calf weaning weight.
    You can still select for growth through WW EPD that would be independent from maternal traits.
    <br />
    <em>[Emphasis mine]</em>
</blockquote>
<h2>This was the last straw</h2>
<p>
    I was formally introduced to EPDs at the age of 13 by my local semen sales rep.
    He gave a slide presentation with lots of pretty pictures: one had lots of bell curves to illustrate accuracies, another had colored bars in four colors to illustrate SNPs, and another one had little bull silhouettes in little pens to illustrate contemporary groups.
    He even defined Expected Progeny Difference: <em>expected</em> means a statistical prediction, <em>progeny</em> means the offspring - not the animal itself, and <em>difference</em> means compared to &hellip; something.
    Not the breed averages, though.
    That's why you really need to compare your potential sires against the breed averages, which are handily published in every sire catalog.
</p>
<p>
    At the end of the presentation, he asked for questions.
    I had one.
</p>
<p>
    <em>&quot;How are they calculated?&quot;</em>
</p>
<p>
    He knew the answer to that, too. It turns out that EPDs are generated by computers that take into account <b>all</b> of an animal's relatives, from his sons and daughters to his distant cousins.
    It reminded me of the Yahtzee game that was installed on my computer.
    That answer didn't satisfy me.
    If you're reading this, I'm guessing that answer doesn't satisfy you, either.
</p>
<p>
    I knew some extension agents and professors through 4-H and started making phone calls to get a real handle on how EPDs are calculated.
    One of these calls yielded the name and phone number of a professor (both of which I have now forgotten) at Colorado State University who specialized in these things.
    I eagerly called him.
</p>
<blockquote>
    <table>
        <tr>
            <th>Professor</th>
            <td>Hello?</td>
        </tr>
        <tr>
            <th>Me</th>
            <td>Hello, my name is Thomas Christensen. I'm a 4-H member interested in learning about EPDs.</td>
        </tr>
        <tr>
            <th>
                Professor
            </th>
            <td>
                <em>(Repeat slideshow presentation from semen rep, except exclude the pretty pictures, because FaceTime wasn't a thing, yet.)</em>
            </td>
        </tr>
        <tr>
            <th>Professor</th>
            <td>
                Does that answer your questions?
            </td>
        </tr>
        <tr>
            <th>Me</th>
            <td>Well, I was really looking for information on how they're calculated.</td>
        </tr>
        <tr>
            <th>Professor <small>(indignantly)</small></th>
            <td>You can't possibly understand that. The Angus Association has all the resource you need to understand EPDs as a producer: you don't need to know how they're calculated.</td>
        </tr>
        <tr>
            <td></td>
            <td><em>Click</em></td>
        </tr>
    </table>
</blockquote>
<p>
    Maybe it's a good thing that I've forgotten his name, after all.
</p>
<p>
    From that day forward, I read everything I could get my hands on about EPDs, and asked anyone who might know something about how they were calculated.
    Every time it was the same non-answers that the semen rep had used: usually his exact language.
    What they all amounted to was &quot;you don't need to know <b>how</b> EPDs are calculated, you just need to know that they work.&quot;
    It was almost like I was asking sacrilegious questions, like Martin Luther of his bishops.
    Every time I was rebuffed, I grew even more upset at those who perpetuated these hand-waving definitions..
</p>
<p>
    I eventually learned how EPDs were calculated after all. In a chance encounter at a library in a far-away state, I read <a href="https://www.worldcat.org/search?q=isbn%3A0646387138"><em>Animal Breeding: Use of New Technologies</em> by Brian Kinghorn, Julius van der Werf and Margaret Ryan</a>.
    (I highly recommend this book, but it's extremely hard to find a copy in this country.)
    I learned the truths that the sire rep and the BIF had been hiding (or ignorant of) the entire time.
    It wasn't that mysterious of difficult, after all.
    Most of the things that had been said were, in fact, lies, but only lies because they had been deprived of their mathematical context.
    Knowing the truth became a whole new burden, though, as I became more angry every time I heard the half-truths and misunderstandings of the High Priests of Beef.
    It was upon this sore toe and disgruntled spirit that the hammer of Mulliniks' ill-fated words fell.
    As Marty Robbins put it, &quot;I couldn't stand no more.&quot;
</p>
<h2>Something had to be done</h2>
<div class="row">
    <div class="col-md-3">
        <img class="img img-responsive img-rounded" src="../img/milk-is-not-milk-button.jpg" alt="MILK &ne; Milk" />
    </div>
    <div class="col-md-9">
        <p>
            So I founded the #KeepEPDsReal movement.
            The movement officially launched at the Albany County Fair 2019, where my colleagues and I wore and gave out buttons with the message MILK &ne; Milk.
            The premises of the movement are simple:
        </p>
        <h3>
            Measure what you actually want to measure
            <br />
            <small>MILK &ne; Milk</small>
        </h3>
        <p>
            EPDs are useful tools to select for things that they actually measure, but if you try to use <code>MILK</code> to select for preg-rates or use <code>BW</code> to select for calving ease, then it's like you're using a wrench to pound nails with: you'll eventually get the job done, but it will take longer and the results might not be pretty.
            The sire catalogs are riddled with these kinds of fake math, though.
            Worse yet are the dollar-indexes that assume perfectly uniform conditions across the entire country all year long.
            <a href="https://archive.org/details/animalbreedingpl032391mbp/page/n170">Selection indexes are important</a>,
            but using them as the sole criteria for sire selection leads to <a href="https://archive.org/details/animalbreedingpl032391mbp/page/n164">genetic stagnation in the long term</a>.
            And if you're assumptions are wrong going into a selection index system, then you're results will be flawed, too.
            <a href="https://blog.steakgenomics.org/2018/12/bif-genetic-prediction-decision-support.html">Custom selection indexes</a> will never be a substitute for using the proper measurements at the proper time, either.
        </p>
        <h3>
            Ranchers can do math, too
            <br />
            <small>Why doesn't everyone know <em>Y</em>=<em>Z&mu;</em>+<em>X&beta;</em>+<em>&epsilon;</em>?</small>
        </h3>
        <p>
            EPDs are not hard to understand, and what with matrix algebra built into most scientific calculators, there is no technical reason ranchers shouldn't be taught how to create their own EPDs.
            I'm convinced that as scientists, being open and transparent about the methods used to make EPDs will give producers more realistic expectations, wider acceptance, and lower skepticism of EPDs and genomic tools.
            And I mean <strong>real</strong> transparency. Defining EPD by <a href="https://blog.steakgenomics.org/2019/03/epds-101-use-information-to-improve.html">reading the dictionary definitions of "Expected," "Progeny," and "Difference" and showing a bunch of pretty bell curves</a> doesn't cut it.
            I'm trying my best to cover this point by publishing documents and computer programs for ranchers to calculate their own EPDs at <a href="https://github.com/millironx/beefblup">github.com/millironx/beefblup</a>, but I need help.
            Like I said, I'm not an expert, so my programs and explanations are clunky at best.
        </p>
    </div>
</div>