by
J.W.B. King
ARC Animal Breeding Research Organisation
The Kings Buildings, West Mains
Road
Edinburgh EH9 35N, United Kingdom
Summary
A review of existing evidence in single-purpose breeds does not suggest a lack of within-breed variability or any failure to respond to selection, except perhaps in egg-laying poultry. Nevertheless it would be prudent to take various maeasures including: (i) continuation of some populations with modest selection intensities and large population sizes, (ii) avoidance of selection for one or very few characteristics to the exclusion of other traite of coramercial interest, (iii) introduction of control populations to see when and if selection responsos decline and (iv) characterization of alternative populations so that their advantages and disadvantages are shown.
17.1 Introduction
Declining variability in our breeds of livestock is a matter of concern because of its long term implications. One would like to think that current improvement plans are fashioned roughly according to the amount of genetic variation currently available in populations under selection and the real worry is to know whether such plans form a serious prejudice to progrese in future generations. Inevitably the selection process will use up genetic variability and the need is not to know whether there will be any decline, but rather to attempt to find out whether this present action prejudices the opportunity for the future changes unnecessarily.
In speaking about genetic variability I have assumed that our interests in this arises because we wish to carry out selective improvement schemes and that these schemes will take the form of selection within pure breeds to be used as such or to be followed by cross-breeding for commercial utilization. This assumption widens the remit somewhat in that the requirement is not only to have genetic variability in the population but also to be able to use this in a way which can be anticipated to give a selective responso.
In discussing the expected responso to selection and its changes over time it will be convenient to deal first with theoretical expectations, to look briefly at the conclusion from experimente with laboratory species, and then finally to examine the situation in domestic livestock and consider whether the present selection methods appear appropriate from a longer term viewpoint. Additional measures that then seem appropriate will be mentioned although fuller discussions will undoubtedly ensue in subsequent papers.
17.2 Theoretical Expectations
A loss of genetic variation in a population is expected because of the incidence of iribreeding. The amount of loss to be expected in populations of given size, with and without devioes to avoid current inbreeding, have been extensively explored and are reviewed by Yamada (this conference). With Belection the inbreeding effect is further accentuated by the fact that outstanding individuals in the parental population are more likely to produce outstanding individuals which are chosen for breeding in the next generation (Robertson, 1961). This is part of the process by which selection alters gene frequencies and will produce fixation at some loci.
The important theoretical development on selection limits springs from the work of Robertson (1960) who was able to extend oalculations on the chance of fixation of individual loci under selection in populations of different size to a more general case in which there wae selection for a quantitative character influenced by many genes. This study showed that, for additive genes, the expected limit of individual selection in any population was a function of Ni where N was the effective population size and i the selection differential in standard deviation of the units. Because the total davance possible depends on the summed effects of the size of individual gene effects and their frequency there is no way of predicting ultimate changes merely by measuring the genetic variation in the initial population. If there are a small number of genes with large effects then the expectation is that there will be a quicker responso but a amaller total one than if the same genetic variation was due to a larger number of genes with smaller effects. For recessives at low frequencies responso times would be expected to be extended but would be correspondingly reduced by dominance and epistasis. This formulation enabled shapes to be given to the time scale of selective changes. Por example, the half life of a selection programme (when progrese has been raade half way to the ultimate limit) is expected to be about 1.4N generations if the genetic variation is raainly additive. It is also possible to predict that the optimum intensity of individual selection for the greatest long term response oocurs when half the population is selected in each generation. This is important for the conservation of genetic variability but when the population is large the curve of selection limits against proportion selected can be rather flat topped allowing more intense selection without great prejudice to the ultimate selection limit.
The treatment by Robertson also allowed discussion of the effects of bottle necks in population size on the ultimate limit of selection. This emphasized the importancs of gene frequencies in the founding population in determining the ultimate limits for selection, and the idea has been extended by James (1970).
Although the general form of the selective process is clear, the changes over generations have proved difficult to tackle theoretically. One model leading to important results is that introduced by Bulmer (1971). This considere a population with a very large number of small additive contributions of equal effect from different loci. With this model the effect of selection was to reduce variability, markedly in the first generation and then by smaller amounts in subsequent generations of selection until a limiting value was reached. This effect complicates the estimation of heritabilitios and genetic correlations in populations which are subject to a selection as pointed out by Robertson (1977). The situation is further complicated where the selection methods and intensity may vary between the two sexes, as in dairy cattle, but Fimland (1979) has provided appropriate methods based on Bulmer's model. The reduotion in genetic variability under this model is due to the generation of linkage disequilibrium and if selection is discontinued, the original variability can be expected to be regenerated. With a more realistic model which involved a lesser number of loci, complete regeneration of variability would not be expected to happen.
17.3 Resulte with Laboratory Animals
It would not be appropriate to attempt any comprehensivo review of the subject but only to draw some oonclusions, drawing on past reviews of the subject and a few recent papers. The reviews of Hammond (1974) and Roberts (1974) on the use of Drosophila and mice respectively are particularly useful as is the review by Al-Murrani (1974).
The general, but by no means universal, finding from selection experimente is that with the course of time genetic variation declines. Ultimately it may appear that the line has reached fixation and the absence of additive variation has in some instances been demonstrated by the lack of response to forward and backward selection. In these circum-stances non-additive variation may remain and further changes in performance can be demonstrated by inbreeding and crossing. A rauch more general finding, however, has been that while the additive genetic variance may be reduced during the generations of selection, it is still demonstrably present and yet a selection limit is reached. Analysis of some experimenta in which this has happened shows a variety of factors which may be operative. One is overdominance where the heterozygotes are selected for their effect on the character under selection. A not infrequent occurrence has been the discovery of visible mutants with effects on performance. Another factor is opposing natural selection which may or may not result in heterozygote advantage opposing the artificial selection. Then there may be antagonistic genetic correlations between component charaoters of the trait under selection and also the complication of genetype-environment interactions. In characters that are particularly susceptible to inbreeding, the inbreeding depression in a small population may offset gains due to selection.
For the animal breeder these resulte suggest that the mere demonstration of significant amount of additive genetic variation is not sufficient since his primary interest will be in selection responso. Therefore any indications of complications such as heterozygote advantage, opposing natural selection and so on are very relvant to longer term predictiona of progrese. In those instances where an actual estimate of ohange due to selection is available it is therefore relevant to enquire into deterioration in other characteristics which may indicate impending complications.
The ideas about ultimate selection limits and population size introduced by Robertson (1960) have been subject to a variety of experimenta in both Drosophila and Tribolium and mica. It seems that the predicted increase in long term response with increasing size of the product Ni has been found in the majority of such studies. Roberts (1974) did, however, find that in many mouse experimente response half life was closer to O.5Ñ than the expected 1.4N. The resulte of reducing population size after a number of generations and putting the population through a bottle neck have not been so clear. The theoretical studies of James (1970) appear to be borne out in so far as the bottle neck reduces the short term response to selection but the long term consequences of the initial restriction are more unpredictable (Frankham, 1979).
17.4 Origins of Breeds of Livestock
The species of livestock in common use as domestic animals appear to have had histories with many similarities. Undoubtedly there would have been many bottle necks during the domestication procese but how restrictivo these might have been is a matter of speculation. Exchanges of genes between domesticated and wild populations will have occurred. Once we enter the historical period initial records are mostly anecdotal but they do show a great moveraent of livestock around the world. Occasionally the effective bottle neck may have been the surviving animals from a particular ship, but as transport improved in the eighteenth and nineteenth centuries there were many introductions of livestock from exotic parts which would certainly have extended the gene pool. Then followed the pedigree era which undoubtedly restricted the size of the base population. The recording of pedigrees has, however, enabled quantification of what has happened subsequent to that time. Studies have been made of the rate of inbreeding in many breeds and commonly it has been found that this is cióse to 1/2% per generation, the effective population size of the breed being much less than the census numbers due to a hierarchical system in which the stock from a few breeders tends to domínate the breed. Although a few breeds allow a grading up process the entry of new genes by this process is very much reduced by the generations of back-crossing that are required. In recent times commercial interests have tended to reverse this process, particularly in poultry and pigs, with abandonment of the purebreeding systems and at the same time breed societies have tended to develop a more liberal attitude to the introduction of animals into their herd books.
Although individual breeders will undoubtedly have practised intensivo selection during the developmont of breeds, there was probably little consensus of opinion about selection objectives between breeders and it was not until recording sobemos were introduced that selection began to be imposed on whole breeds in a coherent manner. As recording schemes have become national in scope and as breeders strive to common goals, then breeds will have begun to change in much more radical ways than in the past. Combining these pressures with, either a high natural rate of reproduction, as in poultry and pigs, or with a new high rate obtainable through the introduction of technical advances as A.I., leads to a situation where much more rapid change is possible. This in turn results in a situation in which the conservation of genetic variation within a breed can become a much more serious issue. The situation in each species will now be reviewed.
17.5 PoultryFor egg laying poultry there are many national statistics to show a steady increase in the number of eggs produced per hen per annum in recent decades. It is of course impossible to disentangle the many factors responsible for these increases but Clayton (1972a) has discussed some of the nutritional, husbandry and disease factors which have been operative. He concludes that conscious selection, aimed at increasing the number of eggs produced, has played relatively little part in raising the levels of production which modern poultry are capable of sustaining. Looking at more recent periods of change which are marked by the introduction of control strains, bred in such a way as to minimize genetic change, Clayton has also suggested (1968) that present selection methods are not increasing egg production. This conclusion was arrived at because the commercial layers in random sample tests maintained their absolute levels of performance whereas the control birds from the Cornell Control Strain declined over the period 1958-73 by about 12% in egg mass per hen housed. These changes were interpreted by Clayton to indicate a decline in performance of the random bred control stock rather than to a change in the environmental conditions of the random sample test. Either interpretation seems possible but Dickerson and Mather (1976) have produced evidence to show that some of the changes known to have taken place in the management system of the random sample test during the period under consideration did in fact reduce the level of performance. A control strain is intended to remain genetically constant and the only plausible explanation for a decline in performance might be if there were an egg born infection such as that described by Gavora et al. (1980). Evidence from other selection experiments such as those of Gowe (1977) and Liljedahl and Weyde (1980) and commercial programmes such as those described by Flock (1980) and Bennet et al. (1980) do, however, suggest that genetic progress is still being made in egg production. There are, however, certainly implications, probably of two kinds. The heritabilities of egg production traits are now low (see for example the results of Emsley et al, (1977) and this fact combined with serious antagonisms between traits, such as egg weight and egg quality, slows down progress. In addition it does seem from the experiments of Dickerson (1962), and Bennet et al. (1980) and the observations of Horn and Bohren (1979) that relaxation of selection does result in declines in egg production with time although it is not always clear whether this is due to a loss of heterosis or to natural selection. These complications mean that the further improvement of egg production is a challenging task.
Turning now to meat production, both broilers and turkeys can demonstrate great increases in growth rate in recent years. These appear to have been produced by mass selection methods using very large populations and, although heritabilities are probably lower than at the beginning, there has been no suggestion that genetic variation for growth rate has been exhausted. What have been observed are correlated declines in reproductive performance associated with selection for weight and body conformation. In turkeys this has reached the length that males have to be used by artificial insemination, but this proves to be an acceptable technology. What is more disturbing is a tendency for increased fatness in more rapidly growing animals which is undesirable both from the consumers point of view and also because of untoward effects on reproductive performance. Although these are complications there is no real suggestion that more sophisticated breeding programmes will not be able to cope with these particular problems.
17.6 PigsPigs have also been subject to long term improvement programmes, the oldest of which started in Denmark in 1907 and was aimed at improving the quantity and quality of export bacon. The scheme was, however, small scale and interrupted by the first world war, and did not really get into ite stride until twenty years later with the erection of additional teating stations. The ohanges that were then oreated by progeny testing have become a classioal animal improvement story and the diagrama indicating the changes in carease length and backfat thiokness are very familiar. Despite the degree of control exercised at testing stations one cannot know whether or not there were environmental changes over this time and hence how much of the observed trends were genetic. There is one analysis of these records by Smith (1963), covering the period 1952-60, suggesting that the genetic changes in backfat thickness were only one fifth of those observed. Whatever the true rate of genetic ohange over these years, the heritability of growth and carease characteristics of concern are all still in the moderate to high category. Selection experimenta directed at changing either backfat or backfat in combination with growth rate have shown that changes can indeed be readily produced by selection. Results from the field are also available in Britain where two control herds, one Large White and one Landrace, have been maintained since 1970 by the Meat and Livestock Commission and used to supply control pigs to their central testing i stations. These have shown that in recent years the rate of improvement of lean tissue food conversion has been 2.4% per annum. Big changes in performance are being made and there is, as yet, no suggestion of diminution in the rate of response.
The complioations in pig improvement which have been revealed are two fold. Firstly in some populations the selection process has been effective in increasing to intermediate frequencies a major gene with multiple effects. The halothane test described by Eikelenboom and Minkema (1974) has enabled many workers to show that the positive reaction to this test is determined by a single recessive gene. According to the selection criteria used by breeders in particular populations, the gene can reach high, but usually intermediate, frequencies. The reason for this is that desirable effeets on carease leanness and shape are counterbalanoed by harmful effeets on reproductivo performance, viability of pigs and a liabjlity to develop pale soft exudative meat in the carcass. Since not all these effeets are recessive it is possible to arrive at an intermediate frequeney of the gene which represents an intermediate optimum and provides a classical example of overdominance for eoonomic merit. This discovery indicates one way in which the selective process may lead into ohanges which are not wholely desirable. With knowledge of the genetic situation, however, this complication should be no more than a temporary hiccough and decisions should be possible on whether to eliminate this gene or to utilize it and continue the selective process.
The second item of concern in selection programmes is reproductivo performance and partioularly litter size which is an important component of commercial performance. Because of its low heritability litter size has received little attention in many national improvement programmes but nevertheless it seems desirable to monitor any adverse changos which selection for other traits might produce since it is presumably a major component of fitness of the species and therefore, from experience in many experimental situations, something which might well decline with intense selection for other characters. From analyses of the field records and from selection experimente there seems little sign of this happening but it is not something which can be overlooked. Commercial methods of pig breeding allow the maintenanoe of relatively large populations so that inbreeding depression should not be a major problem and is in any case alleviated by crossbreeding for commercial production.
17.7 SheepIf we wanted an example of a single purpose sheep breed then it would be hard to better the example of the Australian Merino which has been developed for fine wool production since the early eighteen hundreds, Over the years there has been a great increase in fleece production per head and in recent years this has been well documented. Ferguson (1976) reviewed the production of greasy wool during the period 1935-1976. The annual trend over these years amounted to a change of 0.4% per annum. Although big management changes in pastoral system are not usually expected, he was able to aecount for this small increase by changos in the proportion of strongwoolled sheep that were kept and for an increase in super-phosphate fertilization of pastures. This observation, coupled with a review of some selection experimente where a responso had apparently ceased, led to the suggestion that a selection limit might have been reached. In a more recent survey of these and other experimenta, McGuirk (1980) decided that responsos were still taking place. In all five selection experimente he reviewed selection for increased fleece weight wae effective both when fleece weight was the sole selection criterion (in two experiments) and in three comraercially oriented selection flocks. The multi-trait selection flocks, in fact, did rather better for fleece weight than those in which fleece weight was the sole selection criterion. In one of the latter the realized heritability in the first generation of selection flock was in good agreement with estimates from the base population, at around 50% but has declined to 30$ in the 6th generation. In addition there was evidence that the superiority of the high fleece weight group was directly related to the level of performance of the control flock with which they were compared and with a fluctuating pastural environ-ment these genotype environment interactions for fleece weight could be quite dramatic. Erom chemical analyses of the wool from selected flocks it has also appeared that the level of sulphur-containing amino acids may be critical to the long term consequences of current breeding programmes for increased fleece weight. Without supplementation undesirable changos affecting physical and chemical properties of the wool may result, although it is possible that these could be overcome by changes in wool technology.
Breeds of sheep kept for moat production have not been subject to improvement programmes for any great period of time, although useful gains can already be demonstrated in some national schemes, such as those operating in Norway. There would appear to be adequate genetic variation for improvement purposes and, although Bowman (1966) comments on low heritability estimates for weaning weight in the Down breeds, this may merely reflect particular husbandry situations in which the maternal contribution to early lamb growth prodominates.
The complications arising from continued selection for a few traits are, for the most part, uncharted ground. Genetic interrelationships between characters are reviewed by Turner (1972). That antagonisms cannot be neglected, however, is shown by a selection experiment in Scottish Blackface sheep described by Purser (1980). Selection for long and short cannon bone length was found to have correlated effeets on twinning rate so that breeders seeking to produce shorter thick sot sheep might well anticipate a consequent reduction in lamb crop.
17.8 Dairy Gattle
The long history of milk recording has given many national statistics which show an increase in milk yield per cow over the years. In many countries recent changes have been around 2-3% per year. Until comparatively recently there has been no way of forming any idea of how much of this chango might be due to selection. The big stimulus to dairy cattle improvement was provided by A.I. and this, combined with milk recording, has provided opportunities for effective improvement programmes. Theoretical rates of improvement ranging up to 2% per annum for large populations are possible but do not appear to have been approached even in A.I. schemes with selection procedures that at first sight appear to be highly efficient. The reasons for this discrepaney are perhaps multiple. In the firet place it does seem from a selection experiment, by Hickman (1971), and from institutional herds where daughters from bulls with high and low A.I. proofs have been recorded that the expected differences in production are realized (and sometimes exceeded). On the other hand, measurement of genetic progresa in the field, even with best linear unbiased prodiction procedures, is not without its complications. Although estimates of genetic trend due to the sire contribution can be obtained the structure of the population may not make it legitimate to double these to estimate overall genetic changes in the population. However, taking these resulte at their face value, analysis of field records has indicated that changes of the order of 1/2to 1% per annum in milk yield are all that have been commonly achieved. In an analysis of resulte in the North Eastern United States Van Pleck (1977) isolates some of the reasons for this discrepancy. Selection in practica appears to place considerable emphasis on non production traits. Later records of cows are given too much attention and tuse of old cows and, to some extent, old proven sires to breed young bulle has extended generation intervals and reduced the annual progress that can be expeoted. There does not seem to be any reason to invoke a diminution of genetic variation as an explanation of the low rate of progress in the field.
The question of other antagonistic influences acting to slow down progress due to selection is an important one. Dairy farmers are very conscious of the importance of fertility and disease resistance as factors influencing the profitability of their cattle and questions about relationships between selection for milk yield and any effect of fertility and disease resistance are frequently raised. Although there are undoubtedly some phenotypic relationships between for example, high milk yield and the difficulty of rebreeding cows the genetic relationship is still far from clear. Uhderstanding is likely to come from two approaches. One, as exemplified in Norway, is to extend field recording so that fertility records and disease records are assembled nationally along with milk records. This is very ambitious and the answers coming out of the analysis of these records will be highly relevant. It may be not the method which can be applied in all countries and an alternative experimental method of generating high and low producing cows within carefully recorded experimental herds may be equally revealing. The latter approach extends the opportunity of recording a wider set of characteristics and carrying out disease diagnoses in more controlled conditions. Some of the first resulte from one such investigation by Shanks et al. (1978) shows the high yielding cows to have higher veterinary costs but nevertheless to make more raoney than the lower producing group. Although these are weighty problems for the dairy industry the long generation interval involved in dairy cattle breeding and the facility with which both semen and fertilized eggs can now be frozen makes the prospecte of irreparable damage by wrongly orientated selection programmes a rather remote risk.
17.9 Beef Cattle
In contrast to other species of farm livestock, national reoords on beef cattle are distinctly rare. In Britain the only existing figures on the grwoth rates of beef breeds which were available before the advent of beef recording are those gleaned from weights taken at the Smithfield Show. Although a selected group of individuals grown for exhibition those do provide some comparativo figures over the years. They were first analysed by Hammond (1920) and subsequently by Mason (1961). The period covered was firstly 1893-1913 and then from 1950-1959. The general conclusion from both studies was that the small breeds had increased in size while the larger breeds remained stationary or decreased in size. Since the advent of more widespread beef recording the whole emphasis will have changed again for increased growth rates.
In the absence of prolonged unidirectional selection and with the results of many data analyses showing at least moderately high heritabilities for growth rate there will be little anxiety that genetic variation for growth rate is being used up. In any event, the range of body sizes available in different breeds may make this a much less serious problem that in other species. Despite this it is relevant to enquire for the future whether selection for increased growth rate will bring in its train consequent problems with reproduction and/or disease resistance. Relevant results are reviewed by Barlow (1978) and although several important antagonistas are suspected, more evidence is required.
In the past selection policies have thrown up problems which proved to be temporarily difficult until properly understood. Selection for compact conformation led to the selection of heterozygotes for a gene which when homozygous produced snorter dwarfs. Selection for visible attributes of rauscling increased the frequency of a double muscling gene reducing viability and increasing difficulty of calving in homozygous individuals. Currently selection for increased growth rate is probably bringing attendant difficulties in dystocia and although it seems that it may be possible to manipulate the growth curve to avoid most of this difficulty such selection programmes are not yet in general use. Although then selection prograrames in beef cattle have not been, and are unlikely to be devoid of problems in the future, they do not seem to me to present particular problema in the conservation of genetic variability.
17.10 General Conclusions
Any attempt to summarize and produce a viewpoint on the likely extent of the exhaustion of genetic variation is a very subjective one. My own view would be that there is not great problem except perhaps for egg laying poultry where the commercial secrecy of breeding companies is such that one does not know the true extent of the problem. Despite this optimistic note it is prudent to consider various measures which might be taken to ensure that we reap the maximum gain from investment in genetic improvement programmes. Among possible measures I would like to consider four in some detail.
17.11 References
Al-Murrani, W.K. (1974). Anim. Breed. Abstr., 42:587-592.
Barlow, R. (1978). Anim. Breed. Abstr., 46:469-494.
Bennet, G.L., Dickereon, G.E., Yashyap, T.S. and ítosley, A. (1980). In Selection Experiments in Laboratory and Domestic Animals. Commonwealth Agricultural Bureaux.
Bowman, J.C. (1966). Anim, Breed. Abstr., 34:293-319.
Bulmer, M.G. (1971). Amer. Nat., 105:201-211.
Clayton, G.A. (1968). World's Poultry Sci. J., 24:37-57.
Clayton, G.A. (1972). Ann. Genet. Sel. Anim., 4:561-568.
Clayton, G.A. (1972a). J. Reprod. Fert., Supp. 15, (1972), 1-21.
Dickerson, G.E. (1962). In Proc. Syra. 12th World's Poultry Congress, Sydney, Australia.
Dickerson, G.E. and Mather, P.B. (1976). Poultry Sci., 55:2327-2342.
Eikelenboom, G. and Minkema, D. (1974). Neth. J. Vet. Sci., 99:421-426.
Etasley, A., Dickerson, G.E. and Yashyap, T.S. (1977). Poultry Sci., 56:121-146.
Ferguson, K.A. (1976). In Sheep Breeding Proc. 1976 International Congress Muresk and Perth.
Fimland, B. (1979). Z. Tierzüchtig. Züchtgsbiol., 96:120-134.
Flock, D.K. (1980). In Selection Experimente in Laboratory and Domestic Animals. Common-wealth Agricultural Bureaux.
Prankham, R. (1980). In Selection Experimente in Laboratory and Domestic Animals. Commonwealth Agricultural Bureaux.
Gavora, J.S., Spencer, J.L., Gowe, R.S. and Harris, D.L. (1980). Poultry Sci. (in press).
Gowe, R.S. (1977). Proc. 26th National Poultry Breeders' Roundtable, Kansas City.
Hammond, J. (1920). J. Agric. Sci., 10:233-289.
Hammond, K. (1974). In 1st World Congress on Genetics Applied to Livestock Production, October 1974, Madrid, Spain.
Hickman, C.G. (1971). J. Dairy Sci., 54:191-198.
Horn, P. and 3ohren, B.B. (1979). Poultry Sci., 58:275-278.
James, J.W. (1970). Genet. Res., 16:249-250.
Liljedahl, L.E. and Weyde, C. (1980). In Selection Experiments in Laboratory and Domestic Animals. Commonwealth Agricultural Bureaux.
Mason, I.L. (1961). Agriculture, 68:71-77.
McGuirk, B. (1980). In Selection Experiments in Laboratory and Domestic Animals. Commonwealth Agricultural Bureaux.
Porser, A.P. (1980). A.B.R.O. Annual Report.
Roberts, R.C, (1974). In 1st World Congress on Genetics Applied to Livestock Production, October 1974, Madrid, Spain.
Robertson, A. (1960). Proc. Royal. Soc. B. (Biological Sciences), 153:234-249.
Robertson, A. (1961). Genetic Res., 2:189-194.
Robertson, A. (1977). Z. Tierzüchtg. Züchtgsbiol., 94:131-135.
Shanks, R.D., Freeman, A.E., Berger, P.J. and Kelley, D.H. (1978). J. Dairy Sci., 61:1765-1772.
Smith, G. (1963). Anim. Prod., 5:259.
Tumer, H.N. (1972). Anim. Breed. Abstr. 40:621.
Van Fleck, D.L. (1977). Proc. Int. Conf. on Quantitative Genetics 1976, Ames, USA. 543-561.
La courbe de réponse à la sélection a généralement une forme caractéristique dans laquelle une période de changement linéaire est suivie d'une baisse progressive avec diminution de la variation génétique jusqu'à stabilisation et atteinte d'une limite finale. Le traitement théorique des réponses à long terme et la prévision du seuil ultime de sélection se sont reveles difficiles et tout ce qui a été fait dans ce domaine derive des travaux de Robertson (1960). II a demontre que la limite atteinte était une fonction de Ni, dans laquelle N était la taille effective de la population et i l'intensité de sélection. La forme de la courbe de réponse a été raesurée par la demi-période et il a été établi que, lorsque la variance génétique était entièrement additive et due à plusieurs gànes ayant de faibles effets, la demi-période prévisible était de 1,4 N générations, mais qu'elle était inférieure si la dominance et l'épistasie étaient importantes.
Des expériences faites sur des animaux de laboratoire ont confirmé les principes généraux de cette théorie mais ont également mis en évidence certaines difficultés. Dans de nombreuses expériences, la demi—période semble avoir été beaucoup plus courte et fréquemment de l'ordre de 0,5N générations. Dans certains cas, il a été demontre qu'á la limite il y a eu épuisement de la variance additive avec fixation complète, mais dans d'autres oas, la réponse à la sélection a cessé avant que la ségrégation à l'intérieur de la lignée ait disparu. Cette situation a été fréquemment attribuée aux effets contraires de la sélection naturelle, bien que l'on ait également invoqué d'autres explications telles que la dégénérescence consanguine et les liaisons avec des gènes désavantageux.
Les animaux domestiques ont fait l'objet d'une longue domestication, avec une sélection probablment tres faible (excepté eut-être au début en ce qui concerne les gènes qui influent sur le comportement). Jusqu'à la formation des races et à l'enregistrement des pedigrees, les échanges d'animaux de populations locales ont probablement donné des populations de grande taille effective. L'introduction d'un systàme d'élevage de race pure caractérisé par une hiérarchie des troupeaux, aura réduit la taille des populations de chaque race. En dépit de ces contraintes, la tendance générale à éviter la consanguinité a donné des taux de consanguinité qui ne sont en general que de 0,5 pour cent par génération et qui correspondent ainsi à une taille effective de population relativement importante. Avant que le contrÔle objectif des performances soit entrepris sur une grande échelle, l'intensité de sélection pour des caractères particuliers était probablement encore faible etvariait d'un éleveur à l'autre. L'introduction de systèmes de contr3le des performances et l'accord qui s'est fait sur les objectifs de la sélection, ainsi que les progrès techniques réalisés notamment en matiére d'insémination artificielle, ont conduit à une situation dans laquelle la conservation de la variation génétique au sein d'une race peut devenir un probleme plus sérieux.
On a très peu de preuves de l'appauvrissement de la variation génétique chez les animaux domestiques. Chez la volaille, certaines observations donnent à penser que la réponse à la sélection pour la production d'oeufs a cessé, mais la majorité des intéressés estiment que l'on peut encore obtenir des réponses utiles du point de vue du nombre d'oeufs, même si les éleveurs ont décidé d'élargir la gamme des objectifs de la sélection. íki ce qui concerne les poulets de chair, chez lesquels les réponses initiales à la sélection pour le tatue de croissance ont été particulièrement spectaculaires, la réponse s'est certainement affaiblie mais, jusqu'ici, il ne semble pas qu'une limite ait été atteinte. Ce qui est préoecupant, ce sont les variations accessoires de fécondité habituellement liées à l'adiposité des reproducteurs. Chez les dindons, la sélection intense pour le taux de croissance a fait que les mSles les plus gros ne peuvent Stre utilisés que pour l'insémination artificielle.
Le porc a également fait depuis longtemps l'objet de programmes d'amélioration le premier ayant été lancé au Danemark en 1907 dans le but d'améliorer qualitativement et quantitativement la production de viande. Bien qu'aucune population témoin n'ait été disponible jusqu'à une apoque récente, d'importants changements phénotypiques se sont produits, Avec l'introduction d'autres programmes nationaux et l'intensification de la sélection, la possibilité d'appauvrissement génétique devient réelle. Toutefois, le nombre de populations témoins dont on dispose est très faible de sorte qu'il est pratiquement impossible de mesurer l'affaiblissement de la réponse.
Chez le mouton, l'augmentation du rendement en laine du merinos est peut-être l'un des changements les plus longs sur lequel on possède des informations. Au cours de récentes expériences de sélection, on a émis des doutes sur la continuité de la réponse, mais ceci était probablement dû à une limitation tenant à l'environnement et il semble possible de réaliser encore des progrès. On ne peut en dire autant des races à viande et la faible héritabilité du taux de croissance observé par certains chercheurs pourrait simplement refléter les origines très limitées de nombreuses races "Down".
Le contr3le laitier est aussi pratiqué depuis longtemps chez les bovins et des rendements accrus ont été enregistrés tout au long de cette période, bien que la meilleure alimentation ait probablement eu une influence beaucoup plus grande que les changements génétiques au cours des deux dernieres décennies. Grâce à l'insémination artificielle, on peut effectuer une sélection réellement intensivo et certains pays ont choisi d'utiliser des populations de petite tailie effective. Les estimations du changement génétique obtenues par des méthodes statistiques sont maintenant tout à fait precises, mais la constitution de populations témoins permettrait de mieux mesurer les modifications défavor-ables des caractères tels que la fécondité ou la sensibilité aux maladies.
Bien que les bovins de boucherie aient eux aussi une longue histoire de sélection généalogique, les changements d'objectifs de la sélection au cours des années ont probablement empêché une perte importante de variabilité génétique. L'existence de nombreuses races différentes fait que le problème est moins grave que chez beaucoup d'autres espèces.
Pour assurer les meilleures perspectives d'une réponse génétique durable, il faut dono:
- Maintenir quelques races on les soumettant à une sélection de faible intensité et en gardant des populations de grande taille, pour maximiser la réponse telle qu'elle est mesurée par la demi-période.
- Eviter de sélectionner pour un ou quelques caractères seulement à 1'exclusión d'autres.
- Introduire plus de populations témoins pour voir si et quand les réponses s'affaiblissent, pour mesurer les changements des caractères secondaires et pour servir de population de réserve le cas échéant. Avec l'introduction des techniques de oongélation, ceci devient économiquement possible.
- Caractériser d'autres populations qui ont été préservées de facon à connaitre les avantages ou les inconvénients particuliers que chacune peut offrir dans différentes circonstances.
La curva de respuesta a la selección suele seguir una forma característica, en la que a un período de cambio lineal suceda una disminución gradual con variaciones genetipicas reducidas hasta llegar a la fijación y a ciertos limites definitivos. El tratamiento teórico de respuestas a largo plazo y la predicción de limites definitivos de selección han resultado difícles, y derivan de la obra de Robertson (1960). Robertson demostró que el limite alcanzado era una función de Ni, donde "N" era el tamaño de la población efectiva e "i" la intensidad de la selección. La forma de la curva de respuesta se medía por la mitad de h. vida y se demostraba que cuando todas las variancias genéticas eran aditivas y debidas a muchos genes con pequeños efectos, la mitad de la vida esperada era de 1.4N generaciones, pero resultaba inferior si la dominancia y la epistasis eran importantes.
Los experimentos con animales de laboratorio han oonfirmado los principios generales de esta teoría, pero también revelan algunas complicaciones. En muchos experimentos, la mitad de lavida perece haber sido mucho mis corta, y frecuentemente del orden de 0,5 N generaciones. En ciertos casos, se ha demostrado que al límite ha habido agotamiento de variancia aditiva con fijación compita, pero en otros casos, la respuesta de selección ha cesado antes de desaparecer la segregación dentro de la línea, frecuentemente, esta situación se ha atribuido a los efectos opuestos de la selección natural, si bien se han invocado también otras explicaciones como la degeneración consanguínea y el ligamiento con genes deletéreos.
El ganado ha tenido una larga historia de domesticación, probablemente con una selección extremadamente suave (excepto quizás al principio, para genes que afectaban al comportamiento). Hasta la formación de razas y el registro de ascendencias, los intercambios de animales de poblaciones locales resultaron probablemente en poblaciones efectivas muy numerosas. El establecimiento de un sistema de pura raza, con una jerarquía característica de manados o rebaños dentro de ese sistema, reducirá el tamaño de la población de cada raza. A pesar de estas constricciones, la tendencia general a evitar la endogamia ha dado lugar a mas tasas de consanguinidad que por lo general no pasan del V2 por ciento por generación y que corresponden, por lo tanto, a una población efectiva relativamente grande. Hasta que no se dià comienzo al registro objetivo en gran escala, las intensidades de selección para caracteres específicos seguían siendo probablemente más o menos bajos, según el interés que cada criador dedicara a esa actividad. La introducción de planes de registro y el logro de un consenso de opinión sobre objetivos de selección, junto con adelantos técnicos como la inseminación artificial han llevado a una situación en la que la cuestión de la conservación de la variación genetipica dentro de una raza puede adquirir mayor gravedad.
Las indicaciones relativas al agotamiento de la variancia genética en ganado doméstico son extremadamente escasas» En lo que se refiere a las aves de corral se ha indicado que ha cesado la respuesta a la selección para producción de huevos, pero según la opinión general todavía pueden conseguirse respuestas útiles en el número de huevos si bien los orladores han optado por una mayor variedad de objetivos de selección. En los pollos asaderos, donde las respuestas iniciales a selección por índice de crecimiento han sido especialmente espectaculares, la tasa de respuesta ha disminuido, sin duda alguna, pero -por el momento - nada indica que se haya llegado a un limite. Lo que preocupa son los cambios secundarios en la fertilidad, normalmente asociados al engorde excesivo de las animales reproductores. Eh los pavos, la selección intensiva para acelerar la tasa de crecimiento ha llevado a una situación, en la cual los machos de mayor tamaño sólo pueden emplearse mediante la inseminación artificial.
El cerdo ha sido también objeto de programas de mejora a largo plazo, el más antiguo de los cuales empezó en Dinamarca, en 1907, y tenia por objeto mejorar la cantidad y la calidad de la carne. Aunque no se ha dispuesto de poblaciones testigo hasta hace poco tiempo se han operado grandes cambios fenotípicos. Al introducirse otros programas nacionales y acrecentarse la intensidad de selección, se hace real la posibilidad de agotamiento genético. Sin embargo, el número actual de poblaciones testigo es muy pequeño, de modo que apenas cabe la posibilidad de detectar disminuciones en la tasa de respuesta.
En los ovines, el ausento de rendimiento de lana en los merinos escpizá uno de los cambios a más largo plazo sobre los que existe información factible» En recientes experimentos de selección han existido dudas sobre la continuidad de la respuesta, pero ello s e ha debido probablemente a limitaciones ambientales y aún parecen posibles ulteriores progresos. No se dispone de pruebas similares en lo que respecta a las rasas de carne y es posible que la baja heredabilidad para la tasa de crecimiento, observada por algunos investigadores, indique simplemente los orígenes muy limitados de numerosas rasas Down.
El ganado vaouno lechero también tiene una larga historia de registro de la producción de leche y de aumento de rendimientos en este período, si bien hasta los dos últimas decenios las mejoras en la nutrición probablemente hayan constituido una aportación mayor que los cambios genéticos. Ia Inseminación artificial permite obtener una selección realmente intensiva y ciertos países han decidido emplear tamaños pequeños de población efectiva. Actualmente, las estimaciones de los cambios genéticos obtenidos por métodos estadísticos son bastante precisas, pero el establecimiento de poblaciones testigo ofrecería mis campo de acción para medir los cambios desfavorables de características tales como la fertilidad y la incidencia de enfermedades, las cuales constituyen motivo de preocupación.
El ganado ganado de carne tiene también una historia igualmente larga de crianza genealógica, pero probablemente los fluotuantes objetivos de selección a lo largo de los años han impedido que se produjerán pérdidas graves de varianoia genética. La disponibilidad de muchas rasas diferentes hace que la cuestión sea menos grave que en muchas otras especies.
Fura asegurar las mejores perspectivas de respuesta genética a largo plazo, es necesario:
(1) Seguir manteniendo ciertas poblaoiones de cría con intensidades de selección modestas y grandes tamaños de población, para maximizar la respuesta esperada de semivida.
(2) Evitar la selección de una o varias características con exclusión de otros caracteres.
(3) Introducir mis poblaciones testigo, para ver cuándo y si las respuestas disminuyen, para medir los cambios en características secundarias, y para que sirvan de población de reserva en caso de necesidad. Con la introducción de técnicas de congelación este sistema resulta económicamente factible.
(4) Caracterizar poblaciones alternativas que se han conservado, a fin de conocer las ventajas e inconvenientes específicos que cada una de ellas pueda presentar en circunstancias diferentes.
