The Central Tendency of Twinning

In Katahdins, there is generally a preference for breeding ewes which have twinning genetics. We know that ewes ovulating more than one egg per cycle is heritable; and that when we select for it, we shift the bell curve toward the right: toward multiple births. So, some consider this a by-product of twinning selection, that we often get triplets or quads. This phenomenon, in itself, is a bell curve; with most ewes in a given season offering twins, and a smaller percentage having singles or triplets (and rarely, four or more). A 200% crop is fairly standard in our breed.

Because of this focus, it is considered a best practice when selecting breeding stock, to take into consideration whether the animal has twinning genetics, or not. For most people, the tendency is to ask about a particular animal “is she a twin?” This is a good start, but I’d like to illustrate why this “is a” descriptor is not as good as knowing the entire history of the dam’s twinning record. Or, better yet, her family’s entire history of twinning.

Statistics are in order, but not the kind that make some people break out in hives. These are simple statistics, which we all inherently understand, if we think about it for a minute. First off, we know that even a good ewe that’s generally a twinner, she may not have twins every single year during her whole life. She may likely single her first year. And she may single again sometime, or even have some triplets. So, a ewe’s lambing record may look like this over an eleven-year life span:

This ewe had a single her first year, six sets of twins, two sets of triplets, a single at age seven, and singled again at age eleven. She had 21 lambs in her lifetime. If we put that into a bell curve, it looks like this:

So we can see that she mostly twins: the mean is 1.9 lambs per year, and this is her “central tendency.” But she has some variation away from the mean: some years she had singles, other years, triplets.

There are two types of variation. Random variation is just that: random. It’s just the normal roll-of-the-dice consequences of something which does not have consistent performance. We typically can’t do much about random variation.

The second type of variation is special cause variation. This is variation which pulls performance away from the mean in a non-random way, from some particular root cause. In this example, year seven might have been a year of poor parasite management, which caused her body to scale back on ovulation, in response. And the years where she had triplets, those might have been really good hay years. We can do something about special cause variation, if we can determine the root cause. For desirable (positive) variation, we can try to increase that cause (like feeding better in future years). For bad (undesirable) variation, we can try to remove that cause (like being careful to de-worm when needed so ewes are in good weight heading into ovulation).

Twinning is considered “lowly heritable”, only about 20% heritable. So most of the cause of twinning (or not) is due to management, and only a small amount is hereditary. We must focus primarily on good feeding to get a good twinning rate. I know some people who “fail out” a ewe if she singles one time, and I think this may be a mistake. Occasional singles can be from random or special cause variation, and shouldn’t be a cause for alarm. We have to remember that because this trait is lowly heritable, it’s also slow to “move the needle” in either direction genetically: it’s not possible to rapidly shift to a whole population of singlers or a whole population that mostly has triplets. Genetic progress towards a new goal is slow, happening over many generations. But, we should still use this selection criteria whenever we can, because of the cumulative effect on productivity and herd genetics over time.

So, what if we are buying a lamb from this ewe in year three? We might say “yay, that lamb is a triplet!” and want to buy it and pay extra for it so we can get lots of twins and triplets! And what about in year seven? We might say “booo, that lamb is a single,we don’t want singling genetics or we’ll get singles all the time!” and we’d be unwilling to buy that ewe, or unwilling to pay good money for it.

But when we look at the bell curve of the mother, we realize, this is a logical error. The triplet from year three and the single from year seven are of the exact same  twinning genetics (assuming they have the same sire). They are both out of a ewe which has a central tendency of 1.9 lambs per year, which is pretty good! These are the genetics they are inheriting: her central tendency for number of ovulations per year. Not their single instance of birth type, which can be influenced by random, or special cause, variation.

We can see it again if we consider a lamb  from this second ewe’s record:

This ewe has spent most of her life having singles, she’s only had fourteen lambs in her lifetime (as compared to 21 lambs from the same-aged ewe above). But if we happen to buy in a year where she’s had a twin, we might say “yay, this lamb is a twin!” But this lamb is from a ewe whose central tendency is only 1.3 lambs per year! So I’d definitely prefer to by the single lamb from the first ewe than a twin from the second ewe!

See? Statistics doesn’t have to be painful. It makes perfect sense, right?

This goes to show, when buying ewes or choosing replacement ewes, and also purchasing rams which will be bred to produce replacement ewes, we should try to ask about the dam’s  lifetime twinning record, not just the year the purchased lamb was born. (And, also ask about the maternal record on the sire side, when that’s available.)

NSIP takes this a step further by calculating for us the genetic history of the lamb we are considering purchasing, combined with the records of her mother, grandmothers, sisters, aunts, female cousins etc. on both sides of her pedigree. This is way more math than we can do in our heads or in a spreadsheet! Here is a distribution of all the sheep on my farm now, and their “NLB” or “number of lambs born” NSIP scores:

I have a spread where my top ewe is likely to have 25% more lambs than the zero reference point, and my bottom ewe is likely to have 9% below the reference point. But, of course, most of them cluster towards the middle, because this is a classic bell-curve population trait. My mean is about 8%, so I’m doing  little better than the original reference point for the breed, which dates back a decade or so. (I believe it was the original mean, but they freeze it, so that over time, we can see the progress of the breed as we all move to the right of the reference point.)

If I just want to “go for max” number of lambs, I can keep selecting ewes on the right, and dropping ewes on the left. Some people don’t prefer multiple births, however, because of the extra labor, risk, expense, they incur. There is definitely a tradeoff there. Those people could “trim” on both ends of the bell curve, choosing to only keep ewes which are very close to the middle, and will mostly likely to stick to twinning most of the time.

Sires play into this too, but only for cases where we may be keeping their daughters as replacement ewes. The sire himself has no direct influence on twinning when he breeds: twinning is solely controlled by the number of ovulations in a cycle. As long as the sire has healthy sperm, he will fertilize all the eggs which are available. But, the sire does pass on maternal genetics for ovulation to his daughters. So it’s important to understand his mothers’, grandmothers’, sisters’, and female cousins’ twinning performance too, in that case. If the ram is being used mostly as a terminal sire to crank out butcher lambs, then his twinning genetics are irrelevant.

More than once, I’ve had a buyer here who is preoccupied with the “is a twin” trait, and unwilling to look past birth instance to consider birth trends. This is frustrating for me, because often I know from the data that the person is actually choosing an animal with lower twinning rate genetics than another I might recommend. But, some people are insistent they know what they’re doing! So I let it go.