! -56FALCONER: Yes, I think if I understood you right, that is a good point. The graph that you saw for litter size of the growth rate selected mice were the litter sizes of those animals actually selected, of the parent animals that were selected for their high growth rate, and I do not want you to take that experiment too seriously for that reason, because the basis of that c_mparison between the lines selected for litter size itself and the lines selected for growth rate is a litte_ different, because in the one case you havethe progeny of the selected individuals and in the other case you have the selected individuals themselves, and I think that is one of the reasons why the gene l_vel is different. : You see, the level was higher all through for the correlated responses than for the direct responses. I think that is the main reason for it. I think one can only look at the general slope of these lines. YO_: Thank you, Dr. Falconer. I think we will take a short recess now. (Recess) PART II YOHE: SECTION 2 Our next speaker is also from Edinburgh, Dr. Alan Robertson. Dr. Robertson did work at Cambridge, he is now with the Institute of Animal Genetics in Edinburgh. He tells me that hehad considerable training in chemistry, but thathls major field at present is population genetics, and its application to livestock breeding, particularly dairy catt la. He has also done work with Drosophila. Like Dr. Falconer, he is in this country to appear at the Symposlam to be held shortly on quantitative genetics. I think it is very interesting that Dr. Robertson serves very much in the same capacity as Dr. Lush does to this organization, except Dr. Robertson serves for a group of cattle breeders, the Dairy Cattle Club in Cambridge. I think it is also interesting that when Dr. Lush visited -57Britain, that he appeared before that group and today we are very happy that Dr. Robertson returned and can appear before th2s group. Dr. Alan Robertson. _ ROBERTSON: Mr. Chairman, it is indeed a great honor and a pleasure to be able to talk to such an influential group. I am rather afraid though that from your point of view I shall get more out of you than you will get out of me, but after all, that is what I came for, so why should I worry. Now, first of all, I am not responsible for my title. I think the man who wrote my title must have thought I was running for office. He says, "A theoretical and practical consideration of ceilings and plateaus." Well, it is very much a.point of view whether it is a ceiling or whether it is a plateau. However, I suppose it is Mike Lerner who brought this word plat_e_u into the subject when people were talking about cei_lings. _at I am going to talk about or use as sort oftext is a selection program with Drosophila and tie things up with that selection _rogram. is described on this sheet which is being handed out. It The actual wrlte-up was wrlt_en while I was listening to the election results last Thursday night, so it is not perhaps as coherent as it might have been. We took a large population of Drosophila which we had had in the lab for a while, a large random breeding population of about five thousand flies and selected for a character which is of very little intrinsic importance to the animal. abdomen. It is the number of bristles on particular parts of the A fly cares very little whether it has the bristles or whether it has not, at least that is my guess, the fly may think differently. little relation to reproductive fitness. It has We then took this basic population to pieces genetically. We got herltability estimates by all the means we could and then we settled down and selected, different sorts of selection, mass selection and family selections just to see whether the basic ideas worked. -58- (The following are mimeographed sheets handed out by Dr. Alan Robertson.) The discussion of selection limits will be centered on a selection experiment using Drosphila. The character s_lected was a count of abdominal bristles which is probably of little importance to the animal and not affected by natural selection. Different methods of measuring heritability agreed that it was in the neighborhood of 50%. Different programs of selection were used for up to seven generations. These included three different intensities of mass selection and also half-sib and full-sib family selection. are given over leaf. The results The agreement with prediction is reasonably good except that the response to full-sib family selection is less than expected. However, within each experiment there was definite divergences between replicates. The iO lines at thehighest intensity of mass selection were carried on for 30 generations. variation. Each line developed its own behavior, both as to mean and Some lines continued to respond to selection throughout while some responded very little in the later generations. A notable feature of these latter lines was that in many of them, a great deal of genetic variation re_ mained. Detailed analysis of lethal genes was possible and many of the lines proved to be carrying lethals at high frequencies. This suggests that the heterozygotes were being selected. Four cases were dealt with in more detail and this proved to be true in all. However, these were by no means all the story. Heritability analyses proved _ow to be valueless in predicting gain due to selection. _ Two lines are presented on the sheet. The character has altered from one of no importance to the animal to an important part of fitness. pondingly. The genetic situation has also been altered corres- The relevance of these results to the problem of selection limits will be discussed. -59Initial Population. standard deviation 3.4 Mean 35.4 Heritability estimates. Offspring -dam regression Half-sib correlation Full-sib correlation 0.51_0.07 0.46_0.11 0.52_0.07 Response to selection. Heritability of 0.50 assumed. Individual, Upwards After _ _eneratlons. 20/100 20/75 49.9 45.2 46.3 46.4 47.0 44.0 20/50 41.9 40.5 41.7 Downwards 20"100 44.0 Av___t Pred. 48.6 47?2 45.2 41.4 26.4 26.9 47.3 45.6 43.5 27.3 25.1 27.6 26.7 25.5 20/75 28.1 27.2 28.6 28.o 2502 20/50 29.2 30.3 30.4 50.0 27.5 40.4 30.5 43.0 31.0 43.8 30.0 42.4 30.5 41,7 29,1 42.8 29.1 44.2 29.7 43.5 28.5 43.5 29.1 46.5 24.5 Half-sib 20/family 2/10 Upwards Downwards Full-sib 6/family 4/20 Upwards Downwards Lines after many generations of selection _ 70.7 S.D. _ 9.8 84.6 Av. 77-7 i1.9 Heritability. Half-sib correlation Full-sib correlation 0.23 0.22 Two-way selection. Upwards Downwards Parents 0ffsoring S.D. 89.2 64.2 78.5 65.0 lO.1 4,9 The upward response had been 5 in the last 12 generations. L/5/30 _ S.D. 17.1 2.5 _, 3.7 5.1 Heritability Half-sib correlation 0.80 in _ Full-sib correlation 0'.71in _ Daughter-dam regression 0.90 in (_ 0.63 in 0.ii The downward response had been i.O in (#and 2.5 in T in the last 12 generations -60- On the whole, on this population which had not been previously selected, I they did work. You will see the basic results on:the reverse side Of this sheet, which first of all Just describeathe Standard Deviation of 3.4 and a Mean of 35.4. original population with a Theusual ways of measuring heritability are to look at the correlation between daughter and dam, to look at the similarity between offspring from the same father and also to look atthe and father. agreement. similarity between full sibs which have both the same mother These three heritability estimates were all almost too much in Daughter-dam was .51. The half-sibcorrelation of 0_46 and the full-sib correlation gave us an answer of 0.52, so everything was fine. Then, we selected; first, mass selection or individual selection with three different intensities of selection, both upwards and downwards. To take that first line on the sheet,,individual selection upwards, _20 out of i00 means that the extreme 20 flies out of i00 were chosen as parents ofjthe next generation and there were five lines. the score after five generations, 49.9, 46.3 and so on. this in detail at the moment. I have given I am going into I will not bother with the others. age over the five lines was 47.2 and downwardswe The aver_ got a considerab_l_ response with an average of 26.7. Now, it was noticeable in that early experiment that all the lines did not do the same thing. Very soon they established for themselves a definite position among the group of lines. There was one llne which was the best high llne, another line which was the worst high line, and the final differences after the five gener__ ations you will see; there was one line which got to 49.9,_which is &n improvement of 14, and there was another line Which only got to 4_, which is an improvement of 9. So, the different lines did_not all do the same thing, but on the average they did more or less the right.thing. The last two columns show -61- the average of all the lines and the predictedresponse calculated from the proportion selected, the amount of variation wehad present and the heritability, so the five up lines, 20 out of lO0, averaged47.2, dicted, 47.3. Not too bad. Well, that is somewhatof pre- an understatement shall we say. It shows how useful it is to be able to average five results when they deviate from 50 to 44 and then get Just the right average. The down lines went down to 26.7 and We predicted they would go down to 23.5, so they did not go down as far as we predicted. Now, again, we did individual selection with two other intensities, 20 out of 75 and 20 out of 50. tween You will see there again a difference be- the individual lines, but reasonably good agreement betweenthe average of all the lines and prediction in the upward direction. the downward direction we did not get as far as we expected. 20 out of 75 instead of 25.2 expected. Again in 28.0 in the 30.0 in the less intense against 27.3 expected, so it looked as though the herltability fell off in the downward direction pretty rapidly within five generations. We got some corroborative evidence as to that from the other means of s_lection. Let us look at those. The half-sib selection, we have ten families of twenty individuals. We count the twenty individuals and we pick out the best two families out of the ten or the worse two families out of the ten and from those we choose males and females which had not themselves been scored for the next generation and so on. We went upwards and downwards again, and you see that with half-sib selection we got reasonably good agreement between prediction and expectation. It did not go as fast as the individual selec- tion, but with a character of such high heritability in which the performance tells you so much about breeding value, you would expect individual selection to be faster than family selection, eapecially as in this -62- character, you can look at both sexes, which makes a lot of dl_ference. Nowever, again we dld not go down as quickly as we thought we should. In this experiment we can _ollow in detail what is going on as we go down, because within each generation we have a structure, we have ten families in each of thethree lines _nd from the variation between families we can see what is happening to the genetic variation. We did find that genetic variation fell off very rapidly in the low lines. The difference between the families over the same slre was definitely less in three generations than it had been at the start. The full-sib selection, four out of twenty, that is to say twenty familie% with the same sire and dam, six animals measured in each family and the best four families out of the twenty chosen to breed the next generation. This was deliberately chosen to keep the in-breeding the same as in the half-sib selection. Again reasonably good agreement between what we got and what we predicted, but in both cases at this tlme we did not get as far as we expected. Going upwards, we expected to go up to 46.3, we got up to 43.5. Going downwards we expected to get to 24.5, and we got to 29.1. So, for some reason the full-sib family selection did not give us quite the response that we expected, but from a practical point of view, I think if one had been trying to increase the milkyield of cows or increasing the egg production of hens, one would not have been disappointed, from a practical point of view I think one is highly delighted to get even half of what one expects. think in all cases we got at least three quarters of what we expected. So, the taking to piece_ of the initial population gave us results reasonably in expectation with theory. We got our estimates of heritability, we used them to predict what -63shouldhappen and it did happen close to what we predicted, so that was reasonably satisfactory. The complications were that the different lines did not all do the same thing. This may be due to the fact that we are dealing with Dros- ophila, which only has effectively three chromosomes and has no crossing over in one sex, so that the genetic material cannot get disentangled quite as well as it can in an animal with a larger number of chromosomes. Your trouble may be that you are dealing with perhaps a smaller number of units. • I might Just put in here that in many of these lines we relaxed selectionafter the five generations, Just to see what would happen, and we Just let them go for another five generations without doing anything to them at all. This bears on the problem o_ homeostasis, which I think will be discussed later in the meeting. It was rather surprising to find that very little happened. We had selected them pretty hard and pulled the mean quite a long wa_ from that of the original population and it Just sat there and did not return to any extent back to the original mean. The population apparently did not mind very much having its bristles changed. Well, perhaps that is all right, at least there wasn't this rapid return one might expect if the situation was to any extent homeostatic. However, we can discuss that more later perhaps. Then we stopped most of these populations, we stopped all the family selections and we stopped all the intensities of individual selection,_ except the more intense, and we went on with the 20 out of i00 for thirty generations and then interesting things began to happen. Each llne rapidly got its own individual characteristics, in such Characteristics as the mean as I mentioned beforehand, definite omder became established between the different lines selected in one direction. particular line would also be rather more variable than others. One In some -6_• lines .... on the ratio of coun_males got its own individual or so generations selection. to females characteristics most was ratherpeculiar. and eventualiy of the lines had more Some had not, Each after about or less ceased some even at thlrty-five line thirty to resp6nd generations to were still going on quite happily. From general the later responses predict reason is going to happen at a high speed time that I wouid bring and then suddenly in a bottle H-1 got there, whether difficult what we expected to count was striking it, then There was another Just keeping when we measured would no a fly which which stopped had generation lines, almost the second deal and refused depended to respond the response lines shall we keep, which stopped 0fthe lines it was not going it was Still going generalization, to decide on it. gotten there much further. and it tapered off. high lines, it was after twenty- about shall we throw very fast, but at the on and it had beatenanyof as to what was going gave you the wrong and a hundred operation and we had to make a decision so any prediction invariably a great it, this line bristles and it was a major was the lowest so we kept this line because thlrty'fifth I_expected was that this llne having line which on gradually generations Now, to not a bet, at least a promise, the fly had a hundred It was going fast and then suddenly race groupj of whiskey it had or it had not, because previous some iines stop for apparently about _two months before I must admit are very However, away, future, difficult bristles. Now, this happene4 bristles ......... _ Whatsoever. a hundred it is exceedingly in the immediate There was a bet in my particular five I think, _everal " that at any particular What on merely before one can make, points. Firstly, go to selection to happen the in this answer. that almost without exception in these "65ten lines, five up and five down, when we did reach some sort of a plateau or ceiling there was a considerable amount of genetic variation, a very considerable amount, so that the population in some cases was a good deal more variable than it had been when we started .... "" We had started at about _0, we had gotten up to a level of 60 to 70, in fact in some cases over 80. The lines had stopped responding, but they \ were fantastically variable. produce uniformity. There was no doubt that selection did not We got our high lines, but lines in which some indi- viduals are high and some are only mediocre. In some cases we could take this situation to pieces in rather more detail. That is the advantage of'workingwith Drosophila, if something peculiar happens, you can utilize all the trick stocks and take out chromosomes•and make them homozygous and see if they carry lethals and so on. You can take it to Pieces and see what is going wrong. Now, out of these ten.lines about eight of them turned eut to be carrying lethal genes at high Trequencies. When I say high frequencies, I mean that per- haps thirty per cent of the chromosomes we took out would contain•a lethal and when we went Into it into more detail we would find that all these would be carrying,the same lethal . • " .C_. _ 0 / • That would suggest that we were picking out the heterozygotes the _ _'IV _ .12! whole time together.were and mating them Of c°urse' the lethals were dying_!_ and then w_picking out the heterozygotes again in the next generation. ,'_ !i Ii_ 5 In four of the lines we actually put our fingers on the individual genes, produced heterozygous flies and saw how much they exceeded or were less than flies without this lethal gene and our expectations were justified. We found that the heterozygote was very definitely superior in high lines or inferior in the low l_hes to animals which did not have the lethal gene. It was really rather surprising that in so many of the lines we finished up with genetic variation. -66To digress a bit and discuss the extent to which this might happen in a livestock population, this was a surprise to us, .there was nothing obviously wrong with the lines and if you lo_ked at it fromthe point of view say of a population of cattle and wtnt into it in detail, it might be very difficult to realize that you had such a situation. You would find possibly that some of the bulls would have rather a lower conception rate than others, they might have a conception rate perhaps of 55 instead of 70, but some bulls do have a conception rate of 55% anyway. It was a situation where it was possible to take it to pieces with the trick stocks in the Drosophila, but if one did not realize it was there it might be very difficult to put one's finger on it if one was dealing with cattle or poultry, though one might see a decline in hatchability in the latter. Before I go on to those I will take the slides, if I may, Just to indicate the later behavior. You can see the lines in the early generations. the early generations _hey establish a definite order. In There is the best high line and there is the lowest high line and here again there is the lowest low line and the highest,high line. The lines established definite order very quickly, even though the amount of inbreeding we are doing is comparatively slight. We are p_cking out forty parents In each generation. These were a few inbreds to find out what would happen on inbreeding. This was the line which was concerned with the bottle of whiskey and you see what happened, the whiskey was drunk at about there. Well, that was certainly a little surprising, but there it is going on and then suddenly it stops and refuses to go on any further. This line, the low line, H-4, you can see it beginning to creep up, in fact it kept on going and finished up somewhere there and we gave up in disgust then, because to count ninety bristles on an individual is rather hard work. It is much nicer when you are dealing with lines like this in which you find yourself counting eight. - 67A- Fig. 1 The result of continued selection. - - - relaxed lines. K - inbred lines. -67B- 14 ; =a ,_ HIGH ,_ i'o LINES • • • / .............i" _? . • _ 4 ..... .-,_e._ ""_-- / -" \ 2r.<_ 3'o 3; .....'_\ ." I0. ds "_"_ / ', : ."........................ .__.... i.,o _ / -,.' - "... "v ....._..; _ -I .... ..,'_._.,.. H.I ...--H. 4 H 5 ,/ 2, O. • "_-'- 2 F "-'_0 t _P V 14 /'. I1 ' /H. • /." i."-....._,_._.., i/'x,, d H. 4 I_ V v ",, ""'" i'X._ ; ._. / _. \\ • #/ "_ O' GENERATIONS 2 ..," Voriability in the high "... / "--.- ! ,._-_ .".--." .t" 2" Fig. "%.."" -,.,, .. i/N 4 " / _P lines. I_ .-H o"e -'" " "-H. 40"d' -68- Here variable This again peculiar in some of the line L-5, for things lines happened. and the Instance, The females became mean performance I shall exceedingly go on into dropped more detail; very rapidly. here predictions as to what the lines were going to do did not come off. again This line L-2 was kept on, because it had responded very little in the past few generations. Then suddenly three generations after this, down it shot. There is some physiological peculiarity I think here. In all the lines, one after another, the female _uddenly went down very rapidly and became very variable. This is actual variability expressed as the coeff_clent of variation. That is the standard deviation divided by the mean. In the early genera- tions very little happens and there is this lowjline, I think it is actually L-5 females, suddenly becoming very variable, the variation going up fantastically, about six times as much as it was earlier. There is another low llne female becoming variable. some high lines. Here wehave This H-1 llne, finishing up more variable, even when we take out the mean. the base population. In absolute terms it was very much more variable than There was actually a llne which did finish up with less variability than the basepopulation Now, this is Just a slide to illustrate the peculiar behavior of females. This is the ratio of counts in male to that in females. You see again in this line more or less when it began to become more variable, up shot the ratio of males to females. other lines. There again are distinctive behaviors of some Line H-1 has a definitely different ratio of male to female count than llne L-5. Now, of these lines we took two to pieces in even more detail and those are given in the bottom half of the sheet. To discuss these two lines, H-1 in rather more detail, we have selected thls character for so long that it has become of great intrinsic importance - 69- o s to f_ 2o GENERATIONS Fig. 3 Variability in the low lines. a_ s_) . - 70- 01 ..IT MEANS l l iF\ i,/ /.\._ • .......... O ! / w it L.2 // \,II....... " _.-"'-.-b;=.-.-;"--..-_:./ ........ _/",,...-_..J_ .............. _,,., Z..... "_'-=.............. ..,r....... _ . "-" _...... _.s ""_"-:--"-:---_ ....... I0 IS liO iS GENERATIONS Fig. 4 Ratio of d I counts to _ counts. "" IO i_ -71- i , to the animal. . f The animal previously did not mind whether it had _orty bristles or thirty'five. Now, it is of greatest importance to _!ttwhether it has forty or thlrty-flve, because we are going to:let itproduce off- spring in one case and not in the other. We have made this character, previously unimportant to it, of the highest importance. We have almost made,it a part of reproductive fitness, and we will now expect it to behave•in the genetically complicated manner that we would expect components of fitness to behave, and this I suppose is an•explanation of why we get peculiarities. I think I myself would expect characters which are not of importance to the animal to be pretty simply lnherited or thevariation to be simply additive, the heterozygote intermediate between the two hom0zygotes and that is what seems to happen in the base population. _, I would expect characters closely connected with reproductive_fi_en_2 _ . .pm the whole to be inherited in a fairly complicated way. _ If they were inherited simply, naturally selection would have picked out those genes and fixed them. You expect, to my mind, characters llke shall we say butter•fat content in cattle to have a high heritability and to respond to selection, because there has been no natural selection for them previously. I would expect characters like egg laying in poultry to havea lowerS"\ heritability, because a hen which lays more eggs has always been liable tol / produce more offspring than a hen which laid less. Now, we have made this bristle characteristic into something like the number of eggs and we get a more complicated situation. H-l-30, that is the first high line at the thirtieth generation, with a male average of 71 and a female average of 85, an average of both sexes of 78, and there you ' have the standard deviation. Now, this line had responded very little in the last twelve generations. It responded flve bristles in twelve genera- tions, in spite of the fact its previous level of response had been in the -72" order of four a generation. That is to say, it had effectively s_opped, it behaved as though there was no genetic variation left, but on analysis there turned out to be a lethal gene present which in the heterozygote increased the number of bristles by about fourteen. You may then ask where does this come from if in the initial population your variation was in the order of threep but that is a matter for discussion. We could not get any further, so we measured our heritability by halfsib correlation_and by full-sib correlation. We got a heritability of Well, now, that is a reasonable sort of heritability. ' It is the he_ftability of milk production in cattle and one hopes that one can improve milk production in cattle by selection. Now, we take the population and we take the top 20% of animals and we mate them together, which we do normally under selection, that is the usual selection procedure in this experiment, and we take the bottom 20% and we mate them together. The results of that are given half way down. The top 20_, the parents average 89.2, and the offspring went down to 78.5, which is very little difference from the generation in which they started, which was half a bristle lower, so we tried to go upwards, we did not get anywhere. We tried to go downwards, the parents were 695.2, and the offspring were 65.0. So in a delightful situation in which you try to go upwards you pick out your high parents and you go back to where you started. You try to go downwards, you pick out your low parents and you go down to where they ar_ a heritability of 100% going downwards and a heritability of nought going u_wards. _: Actually the explanations are quite simple. Just the segregating heterozygotes. These high animals are You try to go downwards, you pick o. out animals without the lethal and you mate it to itself, and of course you get down. If you try to go down any further your response will be much lower, but in that generation you are picking out animals _hich do not -73contain the lethal, so down you go with a bump. So, there was a situation in which having taken it to pieces by the usual heritability method one might have expected to resPOnd, we found that it would in one direction and it would not in another, because it had this lethal in,it. Lastly, the low line, L-5. I do not understhad it even yet. Now, this really was a stinker, because count was 3.7. all. The male count was 17.1, the female There were a large number of females with no bristles at It must have been very cold I should think. The females were frightfully variable. The distribution of females looked something like this, going from zero to 20, there were a certain number of zeros and there were some flies going up to 22 with a mean of about 4.0, but there had been very little change in the female mean in the last twelve generations. The female mean had gone down from six to four in the last twelve generations, so here we are again in a situation which we cannot get any further by selection, but the population is fantastically variable. We do some heritability estimates by our half-sib correlation and we get an answer of 90_. We do it by the full-sib correlations and we get an answer for that of 60%. We do it by daughter-dam regression, that is to say we pick out the high females and low females and we get a very low answer. That is reasonable. That is a replicate in a sense of what we are doing in actual selection, and the offspring of the low females is very little different from the offspring of the high females. Going into this in detail, it turned out to be an exceedingly complex situation which we never got down to analyze properly because the the Ph.D. student who was doing the work decided he wanted his Ph.D. and not to do any more work and I found myself shot away to look after a farm for three months, so there we more or less stopped, but I think -74•this would deserve more consideration. As a complicated situation which one might expect to hold very well for characters which are of great importance to the animal, like fertility and egg laying and so on. So, in these later generations, when we are dealing with a character of considerable importance to the animal, which has been of importance to it for twenty or thirty generations, our theory does break down. Now, to generalize a bit, under what situations would one expect our theory not to give us the right answer;in response equal the expected response. terms of will the prediction of What sort of conditicns in order for us to get our right answer? must hold Well, that entails an examination of the conditions which must hold for the theory to be correct. Now, I will list reasons why the theory would go wrong. First of all, if we had major genes segregating, I think we can expect the theory to break down. This may be in the sense of giving either a larger or smaller response to selection than expected. The theory is based on the assumption of a large number of genes operating each with fairly small effects. When I say fairly small effects, in this particular population, my standard deviation was 3.5- I would expect the genes that were operating to have an effect of about i, and then I would call that a mlnor gene. If major genes come into the situation at all, we will expect to get crazy answers. / I may say that one major gene affecting this character did suddenly pop into the population. I gave the class to whom I was lecturing a popu- lation to do some family selection, inbreeding and so on, just to show theory working in practice, and to my surprise it did not work at all, because one of the inbred lines did not play the game and there turned out to be a major gene segregating, which in females instead of having a mean of 37 ha@ a mean of 84. The females were sterile. So even though they were very high, I could not do anything about it because they would not breed \ -75" and I out-crossed them and all sorts of things to try to get rid of the sterility and get a pure llne at that level, but no luck, so here again is something which would be useful If only It _._as not lethal or it was not sterile in thls particular case. The males would breed. The second condition that must hold, the second situation In which the theory will break down is if there is a great deal of natural selection, that is to say if the lethal genes are at all important in controlllng the situation. In a population of cattle or of poultry, you might have a lethal gene important and nevertheless not detected unless you were specifically looking for it. We had the lethals in our populations and we only found them because we had specific methods of analysis which are only possible in Drosophila. The major gene I referred to earlier has got a mixture of natural selection and a major gene put into It, but if natural selection is at all important, I think we can expect that our heritabilitM estimates will not be borne out in terms of response to selection. ThirdlY, do complications such as epistasis enter into the picture at all, so that perhaps they may require both Gene A and Gene B that Dr. Lush mentioned thls morning. Now, I do not myself see that such a situation would hold up selection. t On this I rather agree with Dr. Lush, one would eventually get to a situation in which you would fix both A and B and I think that in the process of fixation your theory would give:vou reasonably the right answer about what was going to happen. I cannot see myself that epistasis is going to complicate the issue very much as one i_ selecting. Fourthly, will over-domlnance complicate it? Here again -- I have possibly not given this as much thought as I might, but I feel inclined to give the answer probably not. Now, there is a fifth reason; why you do not get the improvement that -76you would think you ought to. It is not really relevant to anything that I previously said, but a situation in which the environment changes from generation to generation. Now, this could occur if disease is at all important and if the pathogens that the flock is meeting in one year are different from the pathogens that the flock are meeting in the next year. One might see considerable heritable variation in the flock meeting Pathogen A in a certainyear, one would select for it, but one would not see any improvement in it next year because Pathogen A would not be important and Pathogen B would. If that is at all important, I think that might explain why we are not getting the response that we might. Now, those five reasons to my mind pretty well include all the reasons. There may be other reasons which I have not properly thought about, but I put most of the emphasis therefore on the natural selection and some on the possible occurrance of major genes. There will be one great question mark in your mind, as there is in mine, if these things can happen with Drosophila, what sort of experiment can we do with populations of cattle or with population of poultry to make sure we are not in the same situation. Now, I am not prepared to give an answer to that. I would llke to. I think that is what we must do in the next year or so perhaps, take some of these populations which got into a peculiar situation, play all sorts of tricks on them, inbreed them, select them, get heritabilities by dlfferent methods, do what I would call diagnostic experiments, see if one can find out what is wrong without utilizing the tricks of Drosophila, because quite obviously some of these populations superficially should have responded to selection and we took them to pieces and we realized why they should not, but it might have been difficult to take them to pieces had they been populations of hens or populations of cattle. I do not know whether I should Just throw in at the end of the possibility that some of these plateaus reached after long generations of -77- selection can be br6ken interest lation by mutation. in the possibility by mutation, either by spontaneous is to the extent quantitative characters, acter, to be gleaned which would ing, the amount arising is about mutation by x,radiation population broke way, through experiments this, which up very strikingly much faster Now, there are reasons due to a crossing matter over happening of operational We have variation has been fact, irradiated that we could disappointing. the controlled we are used inbred more work along place. they took a so I think Thank However, through it and extent has been we must you. but as a by irradiation. we could get new have responded those lines. be an end. There of two genera- and on the whole lines, bnt the rate of response I think that might in which and irradiated was broken lines in of new that that was not mutation, by selection to in our populatlon, to arise of ten generations. to see what The irradiated experiments, at the moment. in a matter in a peculiar it the plateaus. in Italy for a matter lines utilize minds quickly the plateau it a figure, of production a plateau for thinking of variation for very much the rate had reached aris- in the base population, through down is char- variation our selection is in some people's There this particular To give we find in these nom-existent. small. with of increasing the plateau tions and went interfere arises or the amount is very of into a popu- of spontaneous to let us break interesting like variation, amount or by irradiation. with one could not hope by mutation are some quite is almost of the variation the possibility a certain new variation the amount by mutation not materially the population Now, that of spontaneous and to put it another mutation out of some experiments one-thousandth so that would to which spontaneously, suggest spontaneously has been that one can get new variation Our evidence a little There much our answer faster smaller do considerably than than •o78YOHE: Do we have some questions? DICKERSON: I wonder if Alan would mind explainlng the difference in a situation in which a homozygote is lethal but the heterozygote is better than the homozygote. ROBERTSON: You mean explain the theory oF what would happen? DICKERSON: Well, if you discount the importance of over-dominance, but_you say that this lethal situation is very important, I am a little ' confused. ROBERTSON: I knew this would cause an argument. Well, a lethal situation will behave like this, suppose one takes one's population here and these are parents and you mate the extremes together, you mate high to high, middle to middle and low to low. Here we have the offspring and experimentally and theoretically one gets this, this is _in which the heterozygote is greater than the homozygote. Now, to my mind in theory if one was dealing with a situation with genotype AIAI, A2A2, and AIA2 in order of preference and one got into a balance between these two by selection, because this was higher than the other two and one did the same thing, I am open to correction here and probably will be, I would expect little response to back selection. I take it that the parent-offspring regression, over-dominance would be zero. DICKERSON: I fail to see the difference in definition between the situations. ROBERTSON: In this situation we never see the animals aa at all, they die. DICKERSON: That is right. ROBERTSON: I think the difference is that in one case you get no response to back selection and in the other you do. DICKERSON: I guess I am a little dense. ROBERTSON: All right, we will fightlt Nevermind, out later. go ahead. -79- FORBES ROBERTSON: You drew an analogy between the situation in these long selected lines and the genetic phenomena turned up withwhat in characters that are more directly concerned wlthfltness; happens one of the essential features of characters that are directly concerned with fitness, as we all know, is great sensitivity to environmental conditions. As you know in the ordinary unselected stock, provided you do not starve your animals, the bristle number is comparatively unaffected by the sort of environmental variation which is very important say in butter fat production and so forth. Now, what happened in the tall end of your selection, is there any evidence of the environmental variation going up or not? ROBERTSON: Could we have the second slide? You see What happens to the mean in different generations here, it is pretty regular. When we get down here it begins to wander about. think we are finding the same phenomena there. I Certainly it is my impres- sion with the low lines that they are much more sensitive to slight changes of temperature or food or what have you. I think we have the same thing coming Inthere. ' WARREN: I have no question. I have a comment that might bear on this and I believe i am not giving away anybody's thunder, but in the laboratory at Purdue, where we have been carrying on selectiMe experiments, one of Dr. Bells' students worked with a population that had a plateau apparently for egg production and analyzed it for lethals and found little or no evidence of lethals. You are working here, of course, with a trait and whether that is Just a peculiarity, I do _ot know_ _ ROBERTSON: It might possibly happen that you would not get the same thing with egg production and it might also be a function of the initial population that we started with, which we know did have a fair number of -8o- lethals in it, as do most populations of Drosophila, but it was surprising that so many of them seemed to have an effect on the bristle character as well. GOWE: Dr. Robertson, did you ever take two lines, say two of your high lines that were approximately the same mean and plateaued and mate them together and see if you re-created your genetic variance? ROBERTSON: Yes, we did, but as far as I remember the mean changed very little, that is to say we mated H-I and H-5 and it was more or less intermediate and we then selected and we did not get a very striking response on selection. We got a little response, but I would not say that _t was particularly greater than the slight remaining response in the two individual lines. GOWE: Was that done with several lines or several crosses? ROBERTSON: It was done with the two _ighest lines, but_there are a certain amount of theoretical reasons why in Drosophila you might find it difficult to do that because you rely on crossing over to do it for you and you have only got three chromosomes and so it might not be a very useful thing, one should not draw too many general conclusions from that. In any case, you might expect more response if the two lines were quite different in origin. SHOFFNERi What about the possibility, I suppose you call it a physiological limit, in which you, after you are getting so many bristles on a given space; unless you get bristles on the bristles, would that make sort of a ceiling Situation for this? ROBERTSON: Well, a physiological limit I presume is a situation in which genetic substitutions do not make any difference, that is putting the physiologic thing in genetic terms. Now, here we have a situation in which there are very considerable genetic variations. We Just cannot fix them. Surely a physiological -81- limitwould definition bristles you a population whatever genes and you would YOHE: tainly give you put in, there expect Any further recessed Dr. S. S. Munro, Chairman THRUSDAY pretty well first describe large group of outstanding man, and the second a research Dr. Munro Wisconsin, with Hatcheries personally, He went wrong down to the States. the He joined a have in the States that he went wrong I can here from is that he was a government man, and he is now a commercial. graduated from McDonald degree gOvernment Coop Hatcheries College, then went down to Madison, and then he has a lot in common from Scotland He then went back the Canadian Western time that arehere in Scotland. Coop. I SESSION geneticists_.we got his Master's the fellows P.M .... him as a man twice gone wrong. coming I cer- this afternoon. o'clock Western EVENING by be anymore of the group Since many of you do not know Dr. Munro time, he was a Canadian, Canada, SECTION goingto discussions at four-thirty beCause to vary conslderably. on behalf for the excellent PART Ill HIGGINS: Well, variation, arenot such a population questions? thank you gentlemen ... The session with very little because and worked for about he got hls Ph. D. degree in Edinburgh. fifteen with years Then he worked and Is now with the in Washington. Dr. Munro. MUNRO: here guess Thank you. that I was brought I have the distinct from the West Coast impression to start it is true that I used to do a lot of arguing, quite a lot, physiologically mine, the first recognize me. who you were. time I have at least, because getting a controversy. hair down I but I have mellowed I met an old classmate seen him for twenty-three He sald If you kept a little since years I think of and he did not I would have known -SZ- Some twenty years ago I got a little mixed up in what they now call !' population genetics and studied inheritance of egg production and was a little bit surprised myselfto find the figures I had at least seemed to show that genetics was not very important and I published a paper or two at that time. It was hard work because I was at an institution which to say nothing of not having an academic atmosphere had practically no library facilities or no technical people to turn to and I had to dig all this stuff out myself. I was rather concerned then at that time because accord- ing to figures I could gather there was not more than - I think I said 25% heritability in egg production and I thought I was quite high because there were a lot of things I could not remove from my data, like hatch date, which made full sisters more alike than half sisters. I do not think much thought had been given to the subject, because that started quite a ruckus and a lot of dispute, most of which I never heard anything about until years after- ward. Now it is generally agreed that the figure is more likely lO%but at that time I was expecially concerned, perhaps more concerned than most with theoretical possibilities of what I called the gene to gene interaction and the gene to environment interaction, which I mentioned in my papers. For a long time I was an outcast in that regard, but I am commencing to feel quite at home now, because we have this over-dominance and we have a recognition of eplstasis and we have these buffering genes now talked about and I do not have much to argue about, so my task is going to be quite easy here tonight. I Just have to introduce these people and tell you about their pedigree. The first man on the program is Dr. Ed Godfrey. This man was born in New Hampshire, I guess he is a real New Englander, but he spent some time in the Navy up there in the Artlc regions floating around and I am sure he got a lot Of experience about buffering genes in sub-zero weather up there. He has a Bachelo2's degree from New Hampshire, a Master's and a Ph.D. -83! at Ohio, then he spent one year at the University of Tennessee,.and then he too went commercial in 1953. As far as his work is concerned, I think most of you will recognize him from the work he did on the inheritance of body welght in fowl and, also, thework sex reversal in turkeys. Dr. Godfrey. he did on physiological basis of
© Copyright 2025 Paperzz