Other sources of variation fluence the generation of variability.

106 W .G. Hill Livestock Production Science 63 2000 99 –109 of the best of a group of small lines is expected to knowledge and technology. There are other possi- be greater than that of one large population. This bilities. is, of course, the main function of inbreeding- selection schemes. 1. Marker assisted introgression is essentially a way 3. It is important to distinguish between the total of backcrossing in a particular region QTL or amount of variation present within and among a gene, whereby performance of the recipient popu- set of populations and that utilisable by selection lation is increased whilst retaining most of its in finite time. Thus if unselected populations are previous variability. The problem is to locate the maintained alongside selected populations to con- QTL, and then to do the backcrossing with the serve variation, the total variation in the set of recipient population of sufficient effective size populations is increased, but it may not be that substantial background variation is not lost, possible to utilise the conserved variation in an except in the region very close to the introgressed effective and timely way as the unselected popu- QTL. A first and major problem is to identify the lations lag further behind. If an unselected or QTL to introgress. These are likely only to be backward population is to be used, the loss of found by intensive study of well-recorded popula- performance from bringing in inferior genes has tions or of crosses between populations. Detailed to be made up quite quickly, either by backcros- QTL mapping in large numbers of populations sing out all but specified loci, which requires would be a major undertaking. information on relevant marker genes, or because 2. Induced mutagenesis of unspecified quantitative genetic variation is much higher. Assuming the variation is an old idea that seems still to be used infinitesimal model, the increment in additive in a limited way in plant breeding, but has not variance is proportional to the genetic distance been applied successfully in animals. The reasons between the populations which is proportional to are not entirely clear, but traditionally X-rays the sum of their inbreeding coefficients sub- have been used; they cause chromosomal damage, sequent to derivation from the same source. With so mutations are ‘unfit’. Chemical mutagenesis is accurate selection, such that response is nearly being widely used to screen for mutants in mice proportional to standard deviation, the propor- to serve as models for particular diseases in man. tional increment in œV , about half that in V , is For quantitative traits, because mutants are at low A A relevant. These issues need further consideration frequency, those of small effect take a long time in design of conservation programmes. to become useful; and laboratory experiments 4. A more interesting and difficult case is where have typically been too short. Experiments using selection objectives change in the medium or long mutagenesis have in fact indicated an increase in term. If these go through a process of gradual genetic variance, but these have not been regarded evolution rather than revolution, variation present in most cases as large enough to be worthwhile. in current commercial populations can be utilised. Spectacular changes have, however, been found Populations maintained just to increase the total in some cases, for example from mobile P-ele- variation present among and within populations ment mutagenesis in Drosophila by Mackay remain unlikely to be able to contribute usefully 1985, and smaller responses using retroviral unless relevant QTL have been identified. insertions in mice Keightley et al., 1993. Per- haps the subject needs revisiting, if only to provide a better understanding of how to in-

6. Other sources of variation fluence the generation of variability.

3. Site directed mutagenesis is a technique that has In the previous discussion we have considered been limited by two restrictions: an inability to solely classical sources of variation, namely that perform the technique except in a specific line of present initially in populations and that arising by mice using embryo stem ES cells, and a lack of mutation. This is a very restricted viewpoint in light knowledge as to which genes to mutate. A of new developments in genetic and reproductive solution to the former seems likely through the W .G. Hill Livestock Production Science 63 2000 99 –109 107 use of the nuclear transfer ‘Dolly’ techniques of of such long term response or opportunity in terms of Wilmut and co-workers Schnieke et al., 1997. potential economic benefit and or political oppor- The second will take some time while information tunity from having unique material, and cultural accumulates, but already there are candidates for benefit from maintaining our heritage? The other is knockouts or partial change in function. For essentially a genetic issue and the subject of the example, null mutations of the myostatin gene previous discussion: what factors determine the lead to muscular hypertrophy in mice and double availability of useful genetic variation? muscling in cattle Grobet et al., 1997, with The long term selection experiments undertaken in several independent mutants identified in cattle; laboratory animals have spanned many tens of an obvious experiment is therefore to manipulate generations, equivalent to many centuries in cattle the gene or its promoter in these and other species terms, which is well beyond the time horizon where to modify its effects. consumer demands and resource supplies can be 4. Gene transfer through addition experiments has predicted. In order to assess the value of retained not yet been successful in livestock improvement, variation in enabling selection response, the formal, perhaps mainly because expression of the trans- hard-line, approach is to discount returns from future genes such as for growth hormone could not be years to the present in order to assess opportunity controlled. It has, however, enabled new functions costs properly. Even if discount rates as low as 3 such as the expression of human proteins in the per year are taken, which is well below the level milk to be incorporated. With the opportunity for which a business would use for investment, this specific gene modification, however, rather than translates in typical breeding programmes to a gene insertion via traditional transgenic technolo- discount rate of almost 5 per generation in pigs and gy, gene transfer is likely to use the same 15 per generation in cattle. Thus, with cattle, processes as site directed mutagenesis so as to improvements five or more generations later become make a single copy of a complete gene from rather economically unimportant when viewed from another species. the present day. Consider the simple trade-off be- tween short and long term improvements as func- These routes whereby individual genes can be tions of selection intensity and effective population constructed and incorporated have unlimited poten- size. As James 1972 showed, with high per tial, although they are obviously expensive compared generation discount rates, selecting very intensively with traditional selection. Whilst there are obviously at the expense of population size best uses fixed ethical and public acceptance issues which arise with resources. germ-line manipulation, opinions can change with Such arguments do not accord with what is time and barriers should not be put up to the considered socially acceptable, in relation to our acquisition of knowledge. As genes can be changed, descendants’ benefits and opportunities. Even so, the transferred between species, and presumably ‘in- issues of maintenance of genetic variation in lives- vented’ with defined expression patterns, it does not tock and of conservation in general cannot be seem sensible to ignore them in any discussion of rationally discussed without a consideration of the long-term maintenance and generation of genetic time horizon. We do not know how demands for variation and constrain discussion solely to retaining animal products will change, and we do not know what is currently present. what new technologies will come along. For exam- ple, genetic conservation programmes in cart-horses would have had no impact in the tractor age;

7. Time horizons technologies such as blood typing of livestock have