SELECTION FOR INCREASED PRODUCTION IN MULTI-PURPOSE SHEEP AND GOATS

E. S. E. Galal

INTRODUCTION

Although very few breeds of sheep may be classified as specialised in producing carpet-wool, e.g. Scottish Black-face and Drydsdale, the wool produced from the majority of non-fine wool breeds can be utilized in the wool-carpet industry. Properties required by the industry can be met by blending different types of wools produced by different sheep. This paper deals with selection for increased production in multi-purpose sheep excluding fine-wool, hairy and fur breeds.

The multi-product nature of these sheep and the wide range of environments under which they produce do not make possible generalisations with regard to genetic improvement. Based on data presented by Mason (1969) and World Bank (1984), when sheep used for carpet-wool are raised for mutton and wool approximately 60% of the breeds have wool as their primary purpose; when they are raised for mutton, milk and wool the three products have an average rank of 2.1,1.9and 2.1respectively, as the primary purpose for raising these breeds. Thus, from these data it may be concluded that on the average wool is more or equally often a primary purpose for raising sheep breeds compared with mutton and both are more frequently the primary reason than milk. Very few breeds are raised primarily for pelts or fur.

Most carpet-wool sheep are raised under extensive systems, nomadic, semi-nomadic and transhumance. However, some are raised under intensive and semi-intensive systems where wool is an important by-product. Thus selection objectives and mode of disseminating genetic gain would necessarily vary according to, among other things, system of production.

TRAITS OF IMPORTANCE IN CARPET-WOOL SHEEP

Beside adaptation and fitness traits, the following are the main ones that are of economic importance:

Meat Production

This is a complex function depending on the number of sheep slaughtered and their average weight. Heritability of slaughter weight and liveweight ranges from medium to high (0.2 – 0.5) (Mavrogenis, Louca and Robinson, 1980; Guirgis, Afifi and Galal, 1982 and Rae, 1982) which implies that marketing weight is a trait that could respond to mass selection. However, slaughter weight is highly and positively genetically correlated with mature body weight and any increase in slaughter weight could be associated with increase in mature body weight which entails an increase in feed costs in the flock. This is particularly important in sheep where the cost of the breeding female constitutes a relatively larger proportion of the total costs than in other farm animals. An ideal situation would be to increase slaughter weight with a ceiling on mature body weight, i.e. selection for changing the growth curve. Other techniques to minimize the increases in mature body weight are discussed later.

Animal Production Department, Faculty of Agriculture, Ain Shams University, Shoura Al-Khaima, Cairo, Egypt.

Although slaughter weight is the most important single measurable indicator for meat production (Bradford,1974), dressing percentage and meat quality become important in some situations. Most of the carcass traits have medium to high (0.25 – 0.50) heritability.

Reproduction Rate

A convenient operational definition of reproduction rate is the number of lambs weaned per ewe joined with the ram per unit of time, LWEJ. It includes fertility, fecundity, lamb survival and maternal effects. High LWEJ is desirable when other inputs like feed and managerial skills are available. LWEJ has a low to medium repeatability (0.10–0.25) which means that taking more records on the individual substantially increases the accuracy of measurement. Although selection has been successful in increasing fecundity in Merino (Turner, 1978) and in Romney (Clarke, 1972) most of the current work emphasizes the utilization of interbreed rather than intra-breed additive genetic variance. Among all reproductive traits fecundity seems to be the one most likely to respond effectively to selection. Turner (1979) discussed conditions necessary for the improvement through crossing to exceed that through intra-breed selection.

Studies are underway to investigate the possibility of improving reproductive traits indirectly through selection for other potentially correlated traits, e.g. testis growth (Smith, 1985).

Wool Production

Fleece weight is often the most important factor, among all fleece aspects, in determining the financial returns from wool. There exist no universal standard specifications for carpet-wools. Turner (1979) cited the definition of “desirable characteristics for Indian carpet breeds” as: average fibre diameter 30–40μ, percent fibres without medullation ≥40, percent heterotype fibres (interrupted medullation) ≥20, percent hair (medulla 60% of the diameter) ≤10, percent kemp ≤ and length 7.6 – 8.9 cm. Fleece weight, fibre diameter and medullation are all highly heritable traits with heritability ranging from 0.30 to 0.60 (Rae, 1982) while heritability of kemp score was estimated as 0.43 by Guirgis et al. (1982).

Milk Production

Milk production in sheep is an important trait for rearing lambs and sometimes for human consumption. Some of the breeds producing carpet-wool are also renowned for their high milk yield, e.g. Awassi and Chios. Milk production has a heritability of 0.15–0.50.

RELATION BETWEEN IMPORTANT TRAITS

Genetic correlations between live weight and reproduction traits are generally positive and those between live weight and fleece weight are mostly positive and small to medium in size. Guirgis et al. (1982) reported positive genetic correlations of 0.15between weaning weight and staple length and 0.06 between weaning weight and kemp score. The estimates of the genetic correlation between live weight and milk yield are mostly positive but small.

Estimates of the genetic correlation between reproduction rate and fleece weight are variable. However, effecting desired changes in both traits simultaneously is possible (Rae, 1982). There is a scarcity of estimates of the genetic correlation between reproduction rate and milk yield but it is well known that bigger litter size is phenotypically associated with higher milk yield.

Greasy fleece weight has a high positive genetic correlation with clean fleece weight, moderately positive with staple length, low positive with each of average fibre diameter and medullation and non-consistent estimates with milk yield (Rae, 1982) and moderately negative (-0.27) with kemp score (Guirgis et al., 1982). The latter authors reported a genetic correlation of -0.13 between staple length and kemp score.

GENETIC IMPROVEMENT THROUGH SELECTION

Single Traits

This is the easiest and in many situations the only practical method of selection. Given the above described structure of the genetic correlations between important traits, it is possible to say that selecting for one trait will be associated with improvement in most other traits with little or no deterioration in a few traits. Guirgis et al. (1982) estimated direct and correlated responses expected per generation when selecting for different traits in a Barki flock at an arbitrary selection pressure of 5% in the rams and 50 % in the ewes. An extract from their results, expressing the genetic change as percentage of the trait mean, is given in Table 1. Among the three fleece traits, selection against kemp is expected to be accompanied by the largest response in yearling weight. Also, selection against kemp is expected to produce changes in the desired direction in both fleece weight and staple length.

The single trait could be a simple trait like body or fleece weight, for instance, or a complex one as kg weaned per ewe or fleece weight per unit body weight. The latter index was used as a selection criterion for the Australian Merino Society during 1963/70 (Shepherd, 1979). The relationships between the components of complex traits and their relative heritabilities are important in determining the net outcome of using such complex traits as criteria for selection. If the complex trait is a non-linear expression of its components the prediction of genetic change due to selection is often inaccurate and the selection intensity affects the contribution of each component trait to the net change. Gunset (1984) found that linearizing feed efficiency (measured as a complex non-linear function: kg feed consumed/kg gain) in a form of an additive selection index was more efficient in causing genetic gain in the ration than direct selection on the ratio when heritabilities of the two component traits were different. In practice, selection for single traits is often coupled with independent culling levels for other traits.

Selection Indices

Theoretically, selection indices are the most efficient method of selection. The construction of selection indices requires the measurement of genetic and phenotypic variation and co-variation of and between traits without error and the economic values of these traits. The availability of all these estimates would allow what Williams (1962) called “optimum index method”. However, in the absence of some of the estimates “a reduced index method” may be devised which is less efficient than the optimum but could be more efficient than single trait selection.

Selection indices in the absence of economic values: Elston (1969) proposed a non-linear index , where P is the phenotype of the trait, coded by subtracting the minimum of each respective trait and n is the number of traits involved. Baker (1974) found that I = σ_p2 P1 + σ_p1 P₂ is a good linear approximation of Elston's index. Dividing the index by a σ_p1 σ_p2 it becomes I = P₁/ σ_p1 + P₂/ σ_p2, i.e. the linear approximation σ_p1σ_p2 amounts to equal weighting per phenotypic standard deviation. Elston's index does not even require estimates of genetic parameters. Pesek and Baker (1969) proposed another index which is free of economic weights but instead, the desired genetic changes are stated and taken into consideration. If the desired genetic changes were arbitrarily set equal to the genetic standard deviations, Pesek and Baker index and that of Elston would give similar results. Baker (1974) compared the two indices with selection for two individual traits-under 12 different combinations of genetic correlations and heritabilities. He found that the expected responses by using either selection index was higher than when selection was performed on individual traits when the magnitude of the genetic correlation was ≤ 0.65. However, when the genetic correlation gets high enough (viz.0.8 in his study) the high correlated response in one trait associated with the selection for the other makes the overall response in the two traits in the case of individual trait selection some times higher than selection by either index especially when the selection is performed primarily for the trait with the highest heritability.,

Restricted selected indices: When a trait is genetically correlated with other trait(s) in a selection index it is expected to change with changes in these other traits. However, the change in such a trait could be controlled (or restricted) (Kempthorne and Nordskog, 1959; Tallis, 1962; Cunningham, Moen and Gjedrem, 1970 and Brascamp, 1984). A trait in question is the ewe live weight where it is in general genetically correlated in a positive way with other traits of interest mainly lamb growth. It is desired to restrict the increase of ewe weight, while improving the others, in order to minimize the maintenance costs of the breeding ewes. To effect such restriction, Cunningham et al. (1970) used a dummy variable in the basic selection index, I = b₁ × fleece weight (FW) + b ₂ × number of lambs produced (NL) + b₃ × weaning weight (WW) + b₄ ewe weight (EW) + b₅ × dummy variate, to select for an aggregate genotype defined as T = 2FW + + 22NL + 1WW + OEW. They compared four situations, ewe weight is completely ignored in the selection index, not included in the index but restricted through the dummy variate, included in the index but not restricted and included in the index and restricted through the dummy variate. They found under sets of genetic parameters differing mainly in genetic correlations that the maximum genetic response to the index was obtained when EW was not restricted. Only very slight loss in genetic response occurred when EW was restricted but included in the index while the loss was substantial when EW was restricted and not included in the index. Table 2, from Cunningham and Gjedrem (1970), compares the correlation between the index and the aggregate genotype, as a measure of the relative response in each case, under different situations. They also found that change in EW could be brought to zero with negligible effect on the response of other traits and of the aggregate genotype and that unless EW is being restricted there is no need to include it in the index. That work showed that WW and NL were the highest in terms of their contribution to the overall genetic gain and they were the least affected traits by restricting EW. It also showed that the relationship between the degree of the restriction of EW and the change in genetic gain in other traits and in aggregate genotype was linear. Cunningham and Gjedrem (1970) also examined the stabilization of EW by assigning to it a negative economic value in the aggregate genotype but they recommend against such procedure.

While selection indices are considered the most efficient in causing genetic changes, they are also the most expensive to construct. The cost of genetic gain should be taken into consideration when deciding on an improvement scheme. For instance, in circumstances where phenotypic parameters are estimated from previous records, when great accuracy in selection is not required and when traits may be measured at different stages to enable early culling of surplus animals the method of the independent culling levels may be preferred to that of the selection index (Young and Weiler, 1959).

DISSEMINATION OF GENETIC GAIN

An efficient dissemination of genetic gain often requires a degree of organized flock structure and sheep record keeping. The stratification of the flock into two tiers, nucleus and base flocks, or into three tiers, nucleus, multiplier and commercial flocks is followed in major sheep countries. The nucleus may be closed or open to allow the movements of the genes upward in the hierarchy.

In developing countries two situations exist, one where the flock (usually extensively run) size is substantial but taking records is impractical and the other where the flock is small (often below 20) and selection within it is not possible due to the diminished selection differential and to avoid inbreeding. In these circumstances the only practical means to spread the genetic gain is through the distribution of improved rams produced in governmental or semi-governmental flocks either directly to commercial and village flocks or after they have multiplied in selected larger flocks. Artificial insemination could accelerate the rate of genetic improvement but for practical considerations its use is limited in such countries.

Genetic gain realized under one environment (governmental station) is likely to be transferable to sheep raised under commercial situations. Results of research have shown that genetic-environment interactions in sheep are not very important unless the genotypes are to be used over a wide range of environments.

MAJOR GENES

Major genes can be important in introducing characters that are originally lacking in certain populations. One example is the discovery of the N multiple allelic series (N^t, N^d, N^J, n) in the Romney sheep and the formation of Tukidale, Drysdale and Carpetmaster sheep “breeds” in New Zealand (Wickham, 1978 and Wickham and Rae, 1977) for specialized carpet-wool production. The homozygous NN genotypes have the highest density of halo hair but the heterozygous Nn has a varying density of halo hair depending on the allele.

The Booroola gene in the Australian Merino (Piper and Bindon, 1982) is another example for major genes which increase ovulation rate. It offers opportunity for increasing fecundity in Merino sheep (Robertson, 1979).

REFERENCES

Baker, R. J., 1974. J. Anim. Sci. 54: 1–8

Bradford, G. E. 1974. Proc. 1st Wrld. Cong. Genet. Appl. Anim. Prod. Madrid 1: 725–738

Brascamp, E. W. 1984. Anim. Breed. Abst. 52: 645–654

Clarke, J. N. 1972. Proc. NZ. Soc. Anim. Prod. 32: 99–111.

Cunningham, E. P., Moen, R. A. and Gjedrem, T. 1970. Biometrics 26: 67–75

Cunningham, E. P. and Gjedrem, T. 1970. Acta Agric. Scand. 20: 193–204.

Elston, R. C. 1969. Biometrics 29: 85–97

Guirgis, R. A., Afifi, E. A. and Galal, E. S. E. 1982. J. Agric. Sci. Camb. 99: 277–285.

Gunsett, F. C. 1984. J. Anim. Sci. 59: 1185–1193

Kempthorne, 0. and Nordskog, A. W. 1959. Biometrics 15:10–19.

Mason, I. L. 1969. A world dictionary of livestock breeds, types and varieties. Commonwalth Agric. Bur., Farham Royal, U.K.

Mavrogenis, A. P., Louca, A. and Robinson, 0. W. 1980. Anim. Prod. 30: 271–276.

Pesek, J. and Baker, R. J. 1969. Can. J Plant Sci. 49: 803–804.

Piper, L. and Bindon, B. M. 1982. Proc. Wrld. Conf. Sheep and Beef Cattle Breeding.

Rae, A. L. 1982. In Sheep and goat production (ed. I. E. Coop) pp. 15–55 Elsevier Sci. Publ. Co., Amsterdam.

Robertson, E. E. 1979. In Sheep Breeding (ed. G. L. Tomes et al. ) pp. 351–355. Butterworths.

Shepherd, J. H., 1979. In Sheep Breeding (Ed. G. L. Tomes et al. ) pp. 235–246. Butterworths

Smith, C, 1985. In Genetics of Reproduction in Sheep (ed. R. B. Land and D. W. Robinson). Longman.

Tallis, G. M. 1962. Biometrics 18: 120–122.

Turner, H. N. 1978. Aust. J. Agric. Res. 29: 329–350

Turner, H. N. 1979. In Sheep Breeding (ed. G. L. Tomes et al. ) pp. 81–99. Butterworths.

Wickham, G. A. 1978. Wrld. Rev. Anim. Prod. XIV, No. 2.

Wickham, G. A. and Rae, A. L. 1977. Proc. NZ Soc. Anim. Prod. 37:213–217.

Williams, J. S. 1962. Biometrics 18: 3/b-393.

Winrock International, Assessment of the current status and potential role of sheep and goats in developing countries. Prepared for World Bank, Washington DC.

Young, S. S. Y. and Weiller, H. 1960. J. Genetics 57: 329–338.

TABLE 1. Effect of selection (as % of the mean) in Barki sheep

Source: Guirgis et al. (1982)

TABLE 2. Accuracy of selection index with and without restriction on ewe bodyweight (EW)

Source: Cunningham and Gjedrem (1970)