Tips and Scenarios

Animal Behaviour

This can be important, for example to ensure that younger bulls are not over-dominated by older bulls, resulting in few matings actually made by them, and possible trauma from fighting. Make a Group for younger bulls and constrain their selection into MMG that specify that male Group. A soft way of doing this is possible: you can make a ‘trait’ in your main datafile that reflects bull age (set values for females to zero) or other ‘social’ grouping, and emphasis that trait(s) appropriately to engineer that male groups are populated with compatible animals.

Differential use per male

It may be sensible to allow for more males in a group if these males are young and unable to cover as many females. As an example, with equal-sized female groups, set some male group sizes to 5 and some to 3. Use Preassigned Groups to constrain the groups of 5 males to be young males. If female groups vary in size, you should make adjustments accordingly. You could also use the soft approach in the example above.

Mating young bulls only to young cows or heifers (or only to older cows)

You can use Pre-assigned Groups for this, or the soft approach: Use a ‘trait’ in your main datafile that reflects animal age, allocating values for both sexes. Selecting downwards emphasis in an MMG will attract young males and females into that MMG. If you want to mate young bulls only to older cows, you could simply negate the dam ages in the main datafile.

Allocating high calving-ease bulls to heifers

Pre-assigned Grouping is the obvious way to do this. The obvious soft approach is to attract heifers into a nominated group by selection on a ‘trait’ that is young age, for females only (set values for males all to 0), and attracting High-CE bulls to the same group by a ‘trait’ that is EBVs for bulls only (set EBV for the females all to zero).

Multi-sire joining plus Single sire joining

If you want to do some single sire joining, for example for what you perceive as elite males, you can simply set some of the male group sizes to 1. However, a better idea might be:

1. First make an initial run to get a mating list just for your single-sire mating groups. This will give you full control over all component objectives for this part of your breeding program.
2. Then remove the selected males and females from candidacy in the main datafile, but add these matings to CommittedMatings.txt.
3. Lastly, make the MMG analysis as described above, but with Committed Matings enabled. The single-sire matings will be accommodated when the multi-sire matings are being evaluated.

Backup matings; Clean-up matings

You can use Mixed Mating Groups to run backup matings to cover those females that do not fall pregnant to your main-round matings.

Some operations use groups of bulls (or rams, etc) to carry out clean-up matings following a main mating round using natural and/or AI matings. This requires fewer mating paddocks, each containing multiple bulls and multiple cows, and might make for easier management. However, genotyping is probably required for identifying sires of the resulting progeny.

You might choose to do this by entering the main round matings as Committed Matings in CommittedMatings.txt and using Mixed Mating Groups (MMG) to set up the male and female groups for the backup matings. In this way, your backup mating decisions are conditioned on the main round matings you have already made. You can probably do this backup MateSel analysis immediately after you have completed the main round MateSel analysis.

This is not a perfect solution for a number of reasons – the prime one being that we do not know which females will not get pregnant to their main round matings, such that the MMG backup run has to cover all possibilities. There is not much we can do about that, except to remove any females that are known to be pregnant. Moreover, there will be many fewer progeny coming from the backup matings, compared to a normal MMG run, and any effects of this have not been tested. However, progeny that come from these backup matings should reflect the OCS policy (e.g. Target Degrees) used for the backup MMG run.

Using this MMG backup approach, you can set up backup mating groups where the bulls and cows are appropriately matched e.g. targeted at mating heifers with high calving ease EBV bulls.

Targeting numbers of males and/or females per Group WITHIN a Mixed Mating Group

For example, you might want 5 males from pre-assigned Group A plus 11 males from pre-assigned Group B in a single Mixed Mating Group.

One way to do this is as follows: Make a dummy trait that has value 2 for group A males, 0 for Group B males, and 0 for all females. Make the run and use Average Progeny Value - Set Target to make that dummy trait average 0.3125 = 5/(5+11), and that should give you 5 Group A males plus 7 Group B males in the mix. By relaxing the weighting on that manipulation, and not achieving 0.3125, you can discover any compromises made by sticking to your target. To add such control on the female side, make a second dummy trait and proceed as for the males.

Cohort Management: Selecting groups of animals for later mating decisions, including topping up of animals at AI stations and multiplier herds

As an example, consider that we need to maintain populations of boars at AI stations at three divergent locations, plus associated populations of sows. You need to enable pre-assigned groups for both sexes. Set MatingGroup in the main datafile to 1, 2 and 3 for individuals that are currently at the three locations – this will ensure that they stay at these locations. Set MatingGroup to 0 for animals that are new candidates for sending from the nucleus farm to each of these three locations (MatingGroup 0 means that MateSel is free to choose which location, if any, is best for each of these candidates).

MateSel will then select males and females to be sent from the nucleus to each of the three locations. This can be fixed numbers, or in competition with animals currently at the three locations (set their MustUse/ABSminuse values to 1 to prevent culling). In all cases, selections are made to maximise the overall objective function, globally, but also including any specific traits/markers/inbreeding etc. specified separately for each location.

More information can be found here.

Mating groups selected from family tanks of fish or insects or similar

As an example, consider that full sib families of fish are raised in separate tanks without individual-animal identification. All family members have the same EBV, based on the genomic EBVs of the parents. However, at mating time, the biggest fish from each tank are selected from the tanks, with more fish selected from better families. Here are three strategies:

1. Enter just one male and one female to represent each family in the main datafile. This means two data lines per candidate family in the main datafile, and the parents in these two lines are the sire and dam that made that family. However, the pedigree above these ‘candidates’ should be the actual individuals in the pedigree. With MaxUse set sufficiently high, the normal MateSel results (without MMG) tell you how many fish of each sex (or even what volume of eggs, for females) to select from each family, and what families to mate with each other. With a cross-classified mating pattern (MOETing disabled) each family/sex can mate with multiple families of the opposite sex. However, you are left with the issue of dividing these matings into the number of mass spawning tanks you have available. Moreover, this strategy does not accommodate within-family mass selection on phenotype at selection time.
2. To accommodate the last point, enter say 10 lines per sex per family in the main datafile, and augment the EBV in each line by adding a normally distributed deviation appropriately calculated to reflect the value of the phenotypic information used for within-family mass selection at selection time. This means that you will have a range of EBV merit within each family and each sex, and that mass selection within families is being addressed. Without this, higher-EBV families would be over-rated, to the detriment of genetic diversity, as in reality they would benefit less from within-family selection. If there are 80 fish per tank, then each selection in the mating list actually reflects 80/10 = 8 fish. Of course, you could enter the actual number of fish per family/sex, if known. So, pick the best fish of each sex from each tank, as recommended by the mating list, and construct matings according to the mating list. However, you may still have a problem in dividing these matings among the number of mass-spawning tanks you have available – and you may prefer the control that the MMG feature gives you in setting up your mixed mating group sizes and attributes …
3. As 1. or preferably 2. above, but using the MMG feature as described above to give more control over your mass spawning group sizes and trait/marker attributes. In particular, your results will conform to the number and capacities of the mass spawning tanks that you have available.

Unknown sex

In some cases, the sex of some or all the animals cannot be determined. If the mating groups are large and mating ratio is reasonably close to equal, then random allocation of individuals to one or other sex may give a reasonable result for the initial main run. (If there is sufficient demand for the following …) Under MMG, the method considers that animals of unknown sex are in the ‘wrong’ sex group 50% of the time, and can be crossing with members of their own group, as well as with incorrectly assigned animals in the other-sex group. Where all animals are of unknown sex, the impact of matings are evaluated under the assumption of random mating among all members across the ‘male’ and ‘female’ groups allocated to each other. Other strategies involving MateSel’s BiSexual machinery can be contemplated.

Open pollination

[With thanks to Wallace Cowling for guidance.] Some plant species are out-crossers, and in some cases matings/crosses are made by growing male and female plants in proximity to each other. There are many different ways to carry this out, largely dependent on species. In a simple example, consider that as part of the breeding program, a number of genotypes of female plants (could an obligate out-crossing species, or bred for male-sterility by various means) are grown in proximity to a number of genotypes of male plants, possibly in a greenhouse and probably with bees or other pollinators introduced. Of course, MMG can be used to form the groups of male genotypes and female genotypes, and to dictate group allocations for the greenhouses available. Unlike cattle and fish, a male or female plant genotype is generally represented by multiple individuals. This means that each genotype can be used across greenhouses, and that the extent of use of any genotype can vary (and be optimised) across all greenhouses.

To do this in the current version, simply make multiple ‘clones’ of each candidate genotype that can be used more than once, with a slightly different ID for each in your input dataset. Without further action these will be treated as full siblings, but you can do a proper job and enter the correct self-relationships in CandGRM.txt and read these in (even if the relationships are actually from pedigree alone). With sufficient demand, we will consider a more streamlined approach for this.

Where Progeny Inbreeding is important, there will be a tendency for each genotype to be used in just one or a few greenhouses, but this depends on other non-additive affects that are targeted, and on optimal contributions.

Conservation programs

As noted here, there is some potential to apply MMG to wild isolated populations, especially where inference about animal relationships can be made from visual or electronic observations and/or genomic information. In such cases, two or more groups, each consisting of both males and females, are set up using some form of physical separation.