To know more…the tools of genetic improvement

1 - IBOVAL : what is it ?

• The value of indexation

A breeder is interested in what a breeding animal is going to transmit to its progeny. That is the part of progeny performance which is genetic. Assessment is not simple since the performance of an animal or a herd is subject to influences which are not genetic, such as environment, feed, temperature, etc which have not been transmitted from the parents.

The evaluation of genetic merit is via indexation for the traits which have a technical and economic interest. The indices are established from the performances of the animal but taking the breeding links ie parentage links into account. Indexation is thus a prediction made from the information available at a given moment and can thus change with time. An index is given in reference to a reference base : if this base changes the index can also evolve.

Indices are thus a tool for selection, that is to classify the animals so as to only retain the best one for a given trait.

Proposed in 1994, IBOVAL (BLUP) is a technique to evaluate beef breeds, using information from farms recording schemes from birth to weaning. Indexation thus depends upon a coherent and national management of data. It involves all the herds in the performance recording schemes in France and Luxembourg.

• Methods

To ensure that an index corresponds to a reliable assessment of the genetic value of an animal, it is necessary to take all others factors which affect performances into account (genetic and non-genetic).
Clearly, the genetic potential is expressed differently under different husbandry conditions. To calculate the true index of genetic value, the performances must be adjusted  to take environmental conditions into account.
To do that, the BLUP animal model (Best linear Unbiased predictor) was put in place.

It allows the animals own genetic performance to be taken into account as well as its parents, progeny and related animals (siblings)
For IBOVAL (BLUP), farms performance records are as follows :

Measures the ability of a cow to rear its calf from birth to weaning, ie good milk potential and good maternal behaviour. This is the maternal genetic effect on weight at weaning.

wMERPsev the maternal component giving the weaning weight

Combines the growth at weaning (CRsevr) and the milk capacity (ALait) to assess the impact of the maternal effect (MERPSEV = 0,5 CRsevr + 1 Alait). It is thus a combination of direct effect and maternal effect.

The overall multicomponent indices are :

wISEVR à Index Synthétique au sevrage

Assesses the growth potential and the morphology of an animal at weaning and only assesses the direct effects after appropriate weightings.

This index involves all types of animal, cows, calf and bull. It highlights the breed orientations as defined by the UPRA.

wIVMAT à Index Valeur Maternelle au sevrage

Assesses the ability of a (future) breeding animal to produce good quality calves at weaning in combining the direct effects of the weaning effects (ISEVR) and the milking ability. It applies solely to cows, bulls, fathers of future breeding female animals.

Genetic indices are normalised and standardised by the following principles :

w The indices are expressed in relation to a reference value which is fixed at 100. And index inferior to 100 indicates a “disimproving” effect whereas an index over 100 shows an “improving” effect.

w The IBOVAL / BLUP reference used is based on the five last years results. It is thus a shifting base which evolves. If the herd is “connected” the indices are expressed as distance from the breed reference base. For unconnected herds they are expressed as distance from the genetic level of calves born in that herd over the last breeding year.

wA bull only possesses an Alait, MERPsevr and IVMAT if he has at least 15 daughters in production and they have at least 25 recorded products. Publication of genetic values occurs only when a bull has more than 25 performance recorded products (as with any statistical test the accuracy of genetic values increases with the sample size ie number of progeny.

w Each of these indices possesses a CD (Correlation coefficient) which expresses the confidence level in the figures. The closer the figure is to 1, the more the genetic values of the animal can be regarded with confidence as different. The confidence level depends upon the data available and the heritability of the traits

• Usage

The breeding programme uses the indices to class and select the animals to breed from both male and female :

• matings of the best males and females to produce future bulls
• mating to assure the renewal of females and diffusion by “natural service” bulls ie on the farm.

The IBOVAL / BLUP indices are given on the FIVA (a cows’individual datasheet), the FIT (bull’s individual data sheet) and the BGTA (overall data base of suckler cows). The indices help farmers to choose the best breeding animals and to orientate the genetic profile of their herds.

2 - Connection

• Relevance of the results obtained

The two main advantage of the technique are :
- The herds are ranked from the most comparable to the least comparable
The CACO are published in the Overall Data Base for Suckler Cows (BGTA) and the Cow Individual Data Sheet (FIVA).

- The other looser means of connection such as the inter herd movement of breeding stock and/or usage of non-connector AI bulls are now taken into account.

This additional data has allowed the evaluation of new bulls in herds which were non-connected or weakly connected up to now.

• 2) Method

The objective is to obtain an “indicator” of herd connection which would allow a good correlation coefficient, ie a strong confidence assessment of the figures for an animal. Recent changes allow the calculation of inter-herd connection taking into account more factors than the simple system used up to now (usage of non-connected AI bulls, exchange of breeding animals). The previous method was thus refined : one is no longer in an all or nothing situation.

• The principle of the CD correlation coefficient (confidence limits)

It depends upon the estimation of a correlation coefficient between two herds over 4 breeding seasons. The correlation coefficient allows an estimate of the quality (0-1) of inter herd data. The closer the figures are to 1 then the more the genetic values between the two herds are the same. In general statistical terms, the lower the figure (towards) zero the higher the confidence that two sets of data differ.

All the herds are compared, two by two, to obtain an overall figure of the correlation coefficient (of comparison). On this basis, herds are classed following their level  of “connection” to the rest of the herds. A hierarchie is then established between herds, “well-connected” or “poorly connected”.

A network is established. For that a criteria of admission to the ranks of connected herds, (CACO), which varies from 0-1 is used. The CACO corresponds to the correlation coefficient of comparison which allows two herds to enter into the connection network. Each herd thus has a CACO and a rank. When the CACO is 0, the herd is completely noncomparable to other herds. In contrast, when it is 1, the herd is totally in fact identical to other herds considered connected.

• Example of an application of the method

- If a herd has a CACO (correlation coefficient of comparison) greater or equal to 0,40 then its last 4 breeding seasons are considered connected.

- If a herd has a CACO less than 0,40, only the breeding seasons containing at least 5 progeny of recorded and evaluated connector bulls.

It is the same for breeding seasons further back (before 1998). The CACO value cannot be guaranteed a priori. To ensure that his herd is connected, a breeder has to ensure enough AI calves from connector bulls, so that at least 5 progeny can be weighed, scored and form part of the indexation of that breeding season.
Solely the indices of known bulls active and connected are publishable. With respect to the minimum of 25 progeny (overall weaning index = ISEVR) and 15 daughters and 25 progeny (maternal weaning index = IVMAT), there are unchanged. In addition the connection of bulls will remain based on 10 progeny checked and recorded at weaning over a single or several breeding seasons.

If one takes the example of 2 situations. In both cases, the descendants of 4 bulls are dispersed in 4 herds.

Scenario 1

To compare sires 1 and 2 is possible since they have progeny in the same herd (as have sires 1and 4 and, 3 and 4.) Indirectly sires 2 and 4 are equally comparable since both  are linked to sire 2. All are thus connected.

• Results

Note that before 1996, Salers selection was based solely upon phenotype, ie what the animal looked like / adult linear scoring. After 1996, IBOVAL ie EBV (BLUP system) in any event a scientific assessment of an animals genetic value from its ancesters ans its progeny was used. Plus more use of AI.

Over the years, there has been a marked increase in muscularity (DM), skeletal development (DS), more or less size, and growth rate to weaning (CRsevr). This has led to a slight decrease in the FN (Facilité de naissance), literally translated birth ease but in fact an indexmainly based on birth weight (plus some morphology, ie taking shape into account). However, SO FAR, this increase in calf size has not resulted in more calving problems since the AVEL (calving aptitude) is stable, indeed it may even have improved slightly over the last 10 years. Very careful attention is paid to pelvic structure in France and Salers remain by some way the best calvers in France.

Note that milking ability (Alait) has not been impaired despite the impressive increase in muscularity (DM), again because of attention to IBOVAL evaluations and continued selection for this trait. How many French breeds imported to Ireland for example have had their milk potential bred out ?

Scenario 2

Here it is impossible to compare the genetic value of sires 2 and 3. The spread of data in the herds is not adequate. The ensemble is not connected. However the bulls and herds 1 and 2 on the one hand and 2 and 3 on the other are comparable two by two.

The quality of a connection scenario depends both on the structure (the distribution of descendants of bulls in herds) and on the sample size (the number of descendants in the herds).