What is BLUP?


BLUP is the shortened version of Best Linear Unbiased Prediction, or breeding quality number.

Blup calculations are used to predict everything from hog birth rates to volcanic activity, but has not been used in the breeding of other horse breeds than the Icelandic. By compiling fixed numerical values assigned to the traits of any given individual, or group of individuals (hogs, volcanoes, horses), it is possible to track and predict those traits objectively through mathematical calculations. A Blup rating is linear because of the constant updating of the numbers and unbiased because there is little room for subjective opinion. The resulting numbers allow breeders to make informed predictions when choosing breeding stock.

On average, a horse that comes to an evaluation, gets 7-7.5 for individual characters.

The Blup is normal-distributed with an average at 100, i.g. the ultimately average horse meeting up at evaluation has an average at 100, a nag has Blup much lower than 100 (for example 90) and a good horse has a Blup higher than 100 (for example 110). The higher the BLUP is over 100, the more likely it is that the horse and it's offspring will be good.

What BLUP is, is a prediction of the breeding potential of the individual, that is, if you breed the horse, how likely is it that the horse will improve the offspring, or make an offspring of a lesser quality.

A horse has a Blup for many characters. The blup is calculated for all the characters that are evaluated, and for the main score. If a horse gets 100 in Blup for a character A, it means that if you breed that horse, it's not likely to influence that character A in a positive or negative way, it's breeding potential is simply average for that character. If you have a horse, where the available information about the character A is over average, then that horse gets over average in Blup for the character A.

When the blup of a horse X for a characteristic is calculated, all availabel individual evaluation scores that are connected to that horse are used for the calculations, both the scores of the horse X itself, it's parents, it's offspring, it's grandparents, and lots of other relatives. Virtually all evaluation scores of all known relatives are used. Takes BIG computers to do it.

Individual horses can be evaluated many times. You can have the horse evaluated when it's 4 year old, and many times each summer all it's life if you want to. On the other hand, if a horse gets a good evaluation, let's say 8.5 for tolt, when it's young, it gets higher BLUP than if it gets the same evaluation when older. That's because we want our horses to mature quickly, and that you don't need to train them for years to become good. Taking horses to 2-3 evaluations is common for stallions, showing them as 4-5 year olds once, and again when they're fully trained. Taking them to too many evaluations doesn't change the BLUP much, but can be bad for the reputation. If it takes many tries to get one good average evaluation then people tend to think there is something wrong with the horse.

A Blup is based on the relatives of the horse, and evaluations of the horse itself. If a horse for example has 10 evaluated relatives that all have a pretty head, it has a high BLUP for head (that is, it is likely to have a pretty head). If the horse is then evaluated, and there the score for head is low (that is the head prooves to be ugly in appearance) the BLUP gets much lower.

The evaluations that have been made in Iceland in the last 40 years are now put into a large database, Fengur, with evaluation of more than 100.000 horses, and on this data have been made the calculations which are the basis for the breeding value calculations on Icelandic horses, the BLUP! The BLUP is expected to be used as supplement to other evaluations, and insight, but not to rely on, one and alone.

The exact calculation of the Blup is a very complicated statistical procedure, and will not be covered here.

EXAMPLE:

So, if I take one of the supergood stallions as an example, a stallion that has 145 in Blup for the character tolt. It's the stallion B. If I check him quickly, and his pedigree (and forgetting and not checking lots of other info on his pedigree, it's just to give you an idea), the stallion O:
Got on one evaluation the character 8.5 for tolt.
Got on one evaluation the character 8 for tolt.
Got on one evaluation the character 9.5 for tolt.
Got on one evaluation the character 9.5 for tolt.
Has got about 400 registered offspring, where of lots of them are evaluated, and most of the evaluated offspring get high marks for tolt.
Has got a father with high BLUP for tolt (116), and many evaluated offspring, and good evaluations for tolt.
His father's parents were evaluated.
Has gotten an unevaluated mother, that has given a number of evaluated offspring, all with good evaluations for tolt, so she's got good BLUP for tolt.

There is info on more distant relatives that are also calculated into this. All this gets him to be the stallion that has one of the highest BLUP numbers of all stallions in Iceland (145) and with a very high "security" (öryggi) as there is so much info on him. This means he's likely to make the tolt in his offspring better.

Take one young stallion E for example and the character "legs".
He's not been evaluated. His offspring have not been evaluated. His father was evaluated His mother was evaluated, got 7.5 (average) for legs of feet, and she has no evaluated offspring. His mother's father was evaluated and has many evaluated offspring, he's got 99 in BLUP for legs. His father's parents were evaluated with evaluated offspring, and have near 100 in blup for legs. So there is quite some info on his grandparents, but none about E or his offspring. The info on his grandparents point to average BLUP. E has a calculated 102 in BLUP for legs, based mostly on info on his grandparents, but also on info of further relatives which I won't dig into here (father's father's father, uncles, nieces etc.). But the security of this info is rather low, only 56%. That means he's likely to make the quality of feet of his offspring neither better nor worse, as the BLUP is virtually 100.

If we'd find a stallion where all the info is bad about a character, the BLUP will be bad, and he's likely to have bad influence on this character in his offspring.

If you breed together a stallion with 120 in BLUP for tolt, and a mare with 80 in BLUP for tolt, it's likely that you'd get a neither good nor bad offspring from this combination, and the offspring will get the BLUP 100 for tolt untill further information about it is collected (the offspring is evaluated, or it's offspring, or other relatives).

Also, when BLUP is calculated, it works all ways. Let's say you take the above stallion O, and breed him to a mare with high blup for tolt (let's say 120), the offspring B will have a high blup when it's unevaluated (maybe something like 132). When the offspring is evaluated, and it gets good evaluations for tolt, it will help raising the BLUP of the stallion O a bit.

On the other hand, if O is bred to a mare with low BLUP (let's say 90) and bad evaluation for tolt (let's say 7), the offspring P will get a lower BLUP (maybe 110). If the offspring is evaluated, and it gets an equally good evaluation as B, this will improve the BLUP of the stallion O MORE than the same evaluation of B. The reason is, that the stallion has shown/proved it's breeding potential to be much better, if it could raise the offspring of the bad mare up to being a good horse, than if it could raise the offspring of the good mare to be an equally good horse.

So, if both B and P go to an evaluation, and both get 9.0 for tolt, the effect of P on it's father's blup will be a lot more.

And more such stuff is calculated along with this, but these are the most important factors.

So, how can you use this information in real life? If you for example have an unevaluated mare, that has a good tolt but is very reluctant to trot, and you are rather certain that it's a fault she's born with (not training mistake), you might want to improve this in the offspring. Then it's best to search for a stallion that has good tolt, but which has gotten supergood evaluation for trot, and preferably good Blup for trot too. That way, you know that the sire has both proven himself as an individual with good trot, and that his pedigree and offspring have also had a tendency to have good trot. Another way to use this in real life is when choosing youngsters. If you are looking at several youngsters, it can help to check on their BLUP factors, to see how good they are likely to become, but remember this is a hint, not the final truth.