An overall strategy

A general framework for a breed conservation scheme is given in Figure 15 (after FAO 1998). This design is applicable across the world and can be fine-tuned for UK conditions.

The overarching conservation strategy needs to promote the maintenance of genetic diversity, which is measured in two ways; (1) the continued existence of breeds and (2) the maintenance of the genetic diversity within them. The breeds and the data relating to them are in private ownership while the international responsibility for them resides with the UK Government.

The strategy has to be able to encompass mainstream, numerically strong breeds such as cattle as well as poorly documented and rare breeds, notably many poultry.

The general principles of conservation strategies based on live animal populations have been succinctly stated by Simm (1998, p.103):

(i) start with as variable a population as possible;

(ii) start with as large a population as possible;

(iii) turn over generations as slowly as possible;

(iv) use enough parents (especially males) to keep inbreeding at acceptable levels;

(v) minimize the variation in family sizes to reduce inbreeding; and

(vi) subdivide the breeding population to reduce inbreeding and genetic drift.

That should be at least 50, in commercial breeds as well as in breeds where genetic conservation is the priority, is a firmly embedded and enduring rule of thumb (Kristensen and Sørensen 2005). In the guidelines from FAO (1998), how was to be audited at breed level was not examined.

98 http://www.defra.gov.uk/fangr/files/BreedSocietyInventoryLetter.pdf

Figure 15: Overall strategy for conservation of FAnGR (after FAO 1998).

Recently, genetic contribution and optimization approaches; Woolliams 2007; Fernández et al. 2011) have been applied to the theory of breed genetic management, sometimes with reference to

specific breeds (Engelsma et al. 2011; Fabuel et al. 2004; Mucha and Windig 2009; Windig et al.

2007). Application to the process of choosing animals for representation in genebanks seems relatively straightforward but there do not seem to be any indications in the formal literature of these new methodologies being put into practice in the ongoing management of breeds. In principle, modifications to the software already widely used for the utilization of EBV data, would be able to perform relevant analysis (Woolliams 2007).

The theory of has led, by admittedly complicated processes, to some simple rules about genetic management and a convenient table has been produced by Woolliams (2007, p.163), redrawn as Table 17 which gives the minimum number of sires needed so that a population of particular

demographic properties will have of at least 50. The heavily outlined box represents the breeding system probably most likely in practice namely mass selection, where individuals are selected for breeding purely on the basis of their own merits for some sufficiently heritable character. Thus for example when females produce 4 breeding offspring in a lifetime and typically one male serves 5 or

more females, the breed would require 21 males and at least (21 x 5) females to achieve of 50 or more. Wright’s equation would yield of 70 for that population and the difference is because the calculations that led to Table 17 have taken account of the selection system (Woolliams 2007).

So far as number of breeding offspring produced during a female’s lifetime is concerned, the ideal is for a female to be replaced in the breed by a single daughter and if she produces more, then more males will be needed otherwise overall inbreeding will increase. The chart presented as Table 18, which is reproduced from FAO (1998), is in fact an earlier version of Table 17 and probably less applicable because the values for a given number of sires and dams are calculated on the basis of sires and dams per generation.

Species obviously differ in generation interval and litter size, but because the tables relate to numbers of animals retained for breeding rather than to total production, the differences between species do not have a very great effect on the calculations.

Table 17: Numbers of sires to be used to achieve of at least 50, under different demographic conditions and selection regimes.

Ratio of F to M

Under mass selection

Mean number of breeding offspring anticipated during lifetime of a female

Random selection

Within-family selection

4 8 12 16 20 36

5 or more 21 23 25 27 28 30 15 10

4-5 21 25 27 28 29 32 16 11

3-4 23 26 28 30 31 35 17 11

2-3 25 29 32 34 36 40 19 11

1-2 31 38 43 46 48 55 25 13

Within-family selection is the process when a breed is divided into lineages and in selection all families are represented, the best individuals from each being used for breeding. The net effect is that a sire is replaced by one of his sons and a dam by one of her daughters, certain principles of inbreeding avoidance being observed. This can greatly reduce the number of sires needed to restrict increase in inbreeding.

Table 18: Numbers of sires and dams needed per generation to achieve of at least 50.

Random selection Mass selection Within-family selection

Sires Dams Sires Dams Sires Dams

25 25 35 35 13 13

20 34 30 45 12 14

16 56 25 65 10 50

14 116 20 300 9 1000

Smaller numbers of sires: not possible

With tables like this, breeders have the basic information needed to run a selection programme which still conserves genetic variation.

Formal breeding schemes can be organized and have attracted particular attention in France but these require strong central control; they might be appropriate for breeds with perhaps no more than five flocks or herds (Danchin-Burge et al. 2010). In the UK context the scientifically robust scheme that would be most straightforward to operate would be the minimization of average coancestry (MAC; Caballero and Toro 2000). Though intuitively attractive, programmes based on maximum avoidance of inbreeding and on equalizing founder contributions are not supported by

theory. Caballero and Toro (2000) point out that though MAI will reduce or delay inbreeding in the generations immediately following establishment of such a programme, as the programme proceeds and the variance of family size declines, the degree of genetic drift will increase. Equalizing founder contributions would minimize the variance of contributions from founders, but not those

contributions from individuals of the generations intervening between the founders and their present-day descendants, so a population might show a good balance of contributions from the original founders but animals from intermediate generations may be over- or under-represented.

Guidance for organizations proposing to operate MAC programmes requires specialist input and would be expensive, particularly as follow-up analysis and auditing is needed. First steps would be:

(i) to select as mates animals whose kinships with themselves and the rest of the breed do not greatly exceed the average mean kinship, and

(ii) to pair mates with similar levels of mean kinship (Ballou and Lacy 1995).

What appears to be the only documented example of such a programme in the UK, the Cleveland Bay programme⁹⁹, is detailed by Dell (2010). While a range of software is available¹⁰⁰ specialist advice and input is necessary. A system widely used in the UK is Geneped, an add-on to the Breed Society Record package operated by Grassroots Systems Ltd. Geneped was developed in

collaboration with RBST (Townsend 2003b) and has been used to compute inbreeding coefficients and to trace founder representation in many breeds, though accessible reports are rare¹⁰¹ (other examples: Roberts 2008; Walters 2012; Wilkinson 2012). The RBST analyses Geneped outputs on behalf of breed societies, because specialist input is needed for interpretation. Recently founder representation has been emphasized less and the advice is being oriented more towards kinship analysis¹⁰².