SIL Indexes, Dual Purpose Overall Index
Coopworth Technical Bulletin 11
This technical bulletin describes the basis of SIL selection indexes with reference to the SIL Dual Purpose Overall (DPO) Index used for within- and across-flock assessments by the Coopworth Sheep Society. The index is based on performance in five specific goal trait groups; growth, meat, reproduction, survival and wool. Performance on this index is the main criterion for selecting single-entered rams in the Coopworth Sheep Society.
Other indexes can be produced by the SIL genetic engine and may form the topic of a later technical bulletin. Such alternate indices are constructed in exactly the same way but differ in which traits are involved and the relative weights put on each of these.
Indexes are a proven tool in modern animal breeding. They allow selection to occur on a number of different traits, all at the same time. The relative value of these different traits is assessed in terms of their relative economic worth.
Goal traits versus predictor traits
Goal traits are what we want to improve because they clearly affect farm profitability. Sometimes they are traits we can measure (e.g. fleece weight, weaning weight), sometimes they are traits we cannot measure before selection occurs (e.g. carcass weight, adult ewe weight) or just cannot measure (e.g. number of lambs born for a ram).
Predictor traits are those we have measurements on and which are used to indicate merit in related goal traits. So we may use ultrasound fat depth, ultrasound muscle depth and body weight to predict carcass weight, a goal trait. We also use performance of relatives to predict an animal’s genetic merit e.g. under the SIL system a ram’s genetic merit for number of lambs born is a function of all lambing records for all his known female relatives.
We use predictor traits to generate breeding values (BV), our measure of genetic merit, for goal traits. The mathematics is quite simple in principle, but there are such a lot of equations to solve that a computer must be used to do this quickly and accurately.
Goal Trait Groups are combinations of breeding values that affect a particular component of animal productivity. How the breeding values are combined depends on their relative value, which we measure in cents.
The Coopworth Sheep Society combines five goal traits groups into the Dual Purpose Overall (DPO) Index used for across-flock analyses. Single-entered rams are judged on this index. The five goal trait groups in this index are Growth, Meat, Reproduction, Survival and Wool. A sixth group, Disease, can be used as part of an SIL dual-purpose index but the standard DPO Index adopted by the Coopworth Sheep Society for across-flock evaluations does not have this.
How is an index calculated?
Measurements made on farm are entered into the SIL database by the bureaus. Then the type of genetic evaluation is specified and the necessary data extracted for all the animals concerned, their relatives and for all the predictor traits. The SIL Genetic Engine then performs a series of calculations to produce breeding values for the five goal trait groups. Each of these is then weighted (multiplied) by its relative economic value, and the answers summed to give an overall measure of genetic merit, the DPO Index. At the end of this bulletin is a description of the SIL DPO Index used as the standard by the Coopworth Sheep Society together with a listing of the breeding values that make it up.
How are the economic values determined?
The economic values used have been developed to take into account:
They are relative economic values and they take account of fluctuations over time. While absolute values in the marketplace fluctuate, the DPO index is designed to produce balanced improvements in all five Goal Trait Groups.
What a jargonfest!
Yes, sorry about that. In this bulletin we consider only the standard SIL DPO index used by the Coopworth Sheep Society. However SIL offers other indexes and Goal Trait Groups with similar names. Some of these indexes are also Dual Purpose indexes (but contain different combinations of Goal Trait Groups). We could generate different names for every possible index but there are too many possibilities for this to be realistic. So to verify what is happening you should always check how a dual-purpose index was calculated.
To pick up the terminology it is helpful to think of the chain of events in the processing and use of information. Prediction traits are ones we measure. Goal traits are those that contribute to an animal attribute that we want to improve. Goal traits groups are combinations of these e.g. we want too improve growth and use calculated breeding values for carcass weight, weaning weight and ewe adult weight to do this. The DPO Index is a combination of the different goal trait groups. Combining all the trait BV’s into an index is the best way to compare animals that have different strengths or weaknesses across the traits we think are most important.
How can you have breeding values for traits you don’t measure?
At first glance it looks strange to generate a breeding value for a trait you did not measure. We can do this because traits are related to each other. This may be because of something in the environment e.g. eating more will lead to an animal having a heavier body weight and a heavier fleece weight (it may also be fatter). So you can predict the BV for a trait from other measurements made on that animal. Similarly, you can use information from relatives to predict the breeding value of an animal.
Statisticians call this sort of relationship a correlation. In our prediction of breeding values we are specifically interested in genetic correlations because they allow us to predict genetic merit for one trait from measurements made of other traits and/ or measurements made on related animals. The basis for a genetic correlation is that some of the genes that make animals grow larger carcasses are the same genes that make them grow more wool, AND relatives share some genes in common. Use of predictor traits and of information from relatives to produce breeding values for traits we didn’t measure has been rigorously tested and proven in New Zealand conditions and overseas.
The Coopworth angle
Your society has carefully weighed up all the options to arrive at the SIL DPO index described above as the standard by which all Coopworth sheep are assessed for within- and across-flock analyses. There is great benefit to you as breeders from benchmarking your sheep against those of other breeders. Use of this standard SIL DPO index allows this.
The standard SIL DPO index used by the Coopworth Sheep Society is constructed as follows. Predictor traits and pedigree information are used by SIL to calculate breeding values for a number of goal traits. Each of these breeding values is combined into one of five goal trait group sub-indexes. Each breeding value is multiplied by its relative economic value (REV). These five goal trait group sub-indexes are then combined into the Dual Purpose Overall Index.
Looking at one component, the SIL Dual Purpose Meat Goal Trait Group, we see that it is a combination of two breeding values, lean weight (LEANBV) at 158 cents a kilogram and fat weight (FAVBV) at –132 cents a kilogram.
| SIL Dual Purpose Overall (DPO) = | SIL Dual Purpose Growth (with Meat) |
| + SIL Dual Purpose Meat (with Growth) | |
| + SIL Dual Purpose Reproduction | |
| + SIL Dual Purpose Survival | |
| + SIL Dual Purpose Wool |
| SIL Dual Purpose Growth (with Meat) | = (36 * WWTBV) + (106 * CWBV) - (10 * EWTBV) |
| SIL Dual Purpose Meat (with Growth) | = (158 * LEANBV) - (132 * FATBV) |
| SIL Dual Purpose Reproduction | = 1720 * NLBBV |
| SIL Dual Purpose Survival | = 3400 * SURBV |
| SIL Dual Purpose Wool | = (126 * FW12BV) + (378 * LFWBV) + (350 * EFWBV) |
BV signifies a breeding value; ‘*’ means ‘multiplied by’; WWT = weaning weight; CW = carcass weight; EWT = ewe adult weight; LEAN = carcass lean weight; FAT = carcass fat weight; NLB = number of lambs born; SUR = survival of lambs; FW12 = 12 month fleece weight; LFW = lamb fleece weight; EFW = ewe fleece weight
NB: there are negative relative economic values for ewe body weight and carcass fat weight.
Coopworth Technical Bulletin 12: SIL Goal Trait Groups
This note advocates the use of SIL goal trait groups rather than breeding values in selection. The views put forward are designed to stimulate discussion. There is no right or wrong answer, but the SIL system offers the opportunity to review our accepted approach to using genetic information.
It is now easier to summarise and simply the process of selection of sheep in our breeding flocks and to present information to ram buyers that is more easily understood. Given the increasing quantity of information we have available, we must work to incorporate relevant information, but not at the expense of the ease with which we can use it.
Then & Now
By now, most breeders will have seen SIL indexes on their reports and some will have had Goal Trait Group information as well. Previous genetic evaluation systems generally reported index values together with some or all of the breeding values that went into the calculation of the index. So what is the difference? Basically, Goal Trait Groups are sub-indexes. That is, they are combinations of breeding values focussing on just one area of an animal’s productivity. In some cases there are several breeding values involved, in others there is only one (see Table 1). When there is more than one, this reflects the fact that the trait affects farm productivity in several ways eg. Wool production, where income comes from the wool of slaughtered lambs, of replacement hoggets and of the ewe flock, or that we need several measures to characterise the trait (e.g. Disease, where internal parasite resistance needs to be assessed at different times in an animals life).
Table 1. Breeding values contributing to the six SIL Goal Trait Groups for the SIL Dual Purpose Overall Index.
| Goal Trait Group | Breeding Values Used | Economic Weights (cents) |
| Growth | WWT CW EWT | +36 +106 -10 |
| Meat | LEAN FAT | +158 -132 |
| Wool | LFW FW12 EFW | +378 +126 +350 |
| Reproduction | NLB | +1720 |
| Survival | SUR | +3400 |
| Disease | FEC1 FEC2 AFEC | -180 -180 -370 |
Calculating Goal Trait Group sub-indexes
Goal Trait Groups are combinations of breeding values that affect a particular component of animal productivity. How the breeding values are combined into each sub-index depends on their relative value which we measure in cents (see Table 1). The overall index (Dual Purpose Overall , DPO) used by Coopworth breeders is simply the sum of the Goal Trait Group sub-indexes.
Goal traits are what we want to improve because they clearly affect farm profitability. Sometimes they are traits we can measure (e.g. fleece weight, weaning weight), sometimes they are traits we cannot measure before selection occurs (e.g. carcass weight, adult ewe weight) or just cannot measure (e.g. number of lambs born for a ram).
We use measurements we have made on what are called predictor traits to produce breeding values for traits that are used in the Goal Trait Group sub-indexes. These breeding values are based on an animal’s own performance plus that of it’s relatives and take into account the degree to which each trait is controlled by genes. As mentioned in a previous note, the mathematics is quite simple in principle, but there are such a lot of equations to solve that a computer must be used to do this quickly and accurately.
Goal Trait Groups vs Breeding Values
Many breeders are used to interpreting an index in light of its component breeding values. However, this is not easy when there are many breeding values. Also, breeding values have quite different units of measurement that are not readily comparable. That experienced breeders were able to interpret previous indexes on the basis of breeding values says a lot about their skill and dedication.
The SIL system was designed to summarise information for key areas of productivity in order to simplify the interpretation of genetic evaluation reports. Now that we have more sophisticated ways of analysing the data and because we can collect and process so much more information, we are in danger of information overload if we do not.
It can be argued that the component breeding values are useful when two animals have a similar Goal Trait Group sub-index. However, it must be accepted that the genetic merit is, over most situations, equivalent in economic terms. A breeder may want to know this information in order to plan matings so that particular combinations of traits are more likely to occur in the progeny. Such thinking is well entrenched in livestock breeding. However, on a flock basis, there is the risk that average genetic gains will be reduced. It needs sophisticated methods to plan matings to ensure this does not occur. When such methods are appropriately employed they can actually increase responses to selection compared to random mate allocation.
Perhaps the best argument for use of Goal Trait Group sub-indexes, is in the selling of rams. Many buyers do not have the experience or knowledge that a breeder has. So a simpler way of presenting information is highly desirable. Goal Trait Groups have an easy to understand focus (Growth, Meat, Wool, Reproduction, Survival or Disease) and a common unit of measurement (cents). All breeders have much to gain from a greater number of buyers having confidence in using reports from genetic evaluations to help buy suitable rams. Creating simple, user-friendly reports is important if we are to increase the number of buyers who have this confidence.
Conclusions
Mark Young
Senior Adviser, SIL
NB: This report was written for the Coopworth Sheep Society. It assumes that breeders are selecting for growth AND meat using the SIL DPO index.