MULTIPLE SHIFTS FOR TESTING COCKERELS 271. Multiple Shifts for Testing Cockerels

Transcription

1 MULTIPLE SHIFTS FOR TESTING COCKERELS 271 vitamin E potency of natural and of synthetic alpha-tocopherol. J. Biol. Chem. 179: Hickman, K. C. D., and P. L. Harris, Tocopherol interrelationships. Advances in Enzymol. 6: Quaife, M. L., and P. L. Harris, Chemical assay of foods for vitamin E content. Anal. Chem. 20: Quaife, M. L., N. S. Scrimshaw and 0. H. Lowry, A micromethod for assay of total tocopherols in blood serum. J. Biol. Chem. 175: Rose, C. S., and P. Gyorgy, The relationship of dietary factors to the toxicity of alloxan. J. Nutrition, 39: Rose, C. S., and P. Gyorgy, Tocopherol re- DURING the first 8 years of selection at this laboratory for genetic resistance to leucosis combined with other traits of economic value, it was our practice to leave each male in his breeding pen throughout the entire hatching season of 8 to 10 weeks. By 1942 it was evident that, while marked improvement had been made in the first three years, progress thereafter had been almost nil. It was equally clear why improvement had stopped. We had tried each year to find enough good proven sires to head at least 25 percent of the breeding pens. During the first 5 years, the average number of pens used for the two resistant lines together was only 13.2, but, among the 10 cockerels or so tested annually, it was seldom possible to find 3 or 4 proven sires that could be used again without considerable misgivings. In 1940 to 1942, the average numquirements of rats by means of the hemolysis test. Proc. Soc. Expt. Biol. Med. 74: 411^15. Singsen, E. P., L. D. Matterson, A. Kozeff, R. H. Bunnell and E. L. Jungherr, Studies on encephalomalacia in the chick. 1. The influence of a vitamin E deficiency on the performance of breeding hens and their chicks. Poultry Sci. 33: Swick, R. W., and C. A. Baumann, Chemical assay for tocopherols in animal materials. Anal. Chem. 24: Zacharias, L., P. Goldhaber, and V. E. Kinsey, Vitamin E deficiency in chicks. 1. The effects of dietary supplements on plasma tocopherol levels and vitamin E deficiency symptoms. J. Nutrition, 42: Multiple Shifts for Testing Cockerels F. B. HUTT AND R. K. COLE Department of Poultry Husbandry, Cornell University Agricultural Experiment Station, Ithaca, New York (Received for publication June 21, 1954) ber of pens was increased to 19.3, and about 15 cockerels were tested annually in these three years. In spite of this increase in the number of cockerels tested, there were still never enough proven sires that could be considered satisfactory, and few over which to rejoice. The shortage resulted in part from the fact that few of the males tested were found to transmit everything that we desired, but also because not all of those tested in one breeding season were alive and in good condition for the next one. Since a few good sires were occasionally found, it was clear that the only way to get enough of them was to test more cockerels. As no additional breeding pens were available, the only way to test more cockerels was to use multiple shifts. This system was introduced in 1943 and has since been used continuously. It has proven so satisfactory that we recom-

2 272 F. B. HUTT AND R. K. COLE mend it to other poultry breeders. Reasons for doing so are given in this report of its use during the past 11 years. PROCEDURE OF MULTIPLE SHIFTS Having tried 2, 3, and 4 shifts per season, we have now settled on three as the optimum number for our hatching season of 9 or 10 weeks. Pullets are used for the cockerel tests, and they are chosen from the best families as revealed by their partial tests up to the end of December. Full sisters are distributed as equally as possible to the different pens for their strain. Usually 18 are put in each pen early in January, but that number is reduced to 16 after the first eggs incubated are tested for fertility. (Some stop laying when transferred to breeding pens; a few may die, and others have been removed because of low fertility.) These 16 birds usually provide enough eggs in 18 or 19 days to ensure not only that there are 40 to 55 daughters per sire, but also that these come from an adequate number of dams. With fewer females a longer period of saving eggs would be necessary and there would be difficulty in getting three shifts of males tested per season. Our procedure in changing males now differs slightly from that described earlier (Hutt, 1949). It allows two more days between sires and discards one more day's eggs. The actual schedule followed in 1953 is shown in Table I. After males of Shift I are taken out of the pens, eggs are saved and credited to those males for the next 5 days. On the fourth of those days, males of Shift II are put in, but this is done late in the afternoon so that they cannot fertilize any of the next day's eggs. After that next day, the eggs are discarded for 6 days. Twelve days after removal of Shift I, and 8 days after the introduction of Shift II, saving of eggs is resumed, and all are credited to the second male. The transition from Shift II to Shift III follows the same routine. Before they go in the breeding pens, all replacing cockerels are exposed to artificial light for several weeks. During that period they run with large flocks of pullets. This should ensure that they are in good breeding condition when introduced but in some cases tests are made to be sure that they are producing semen. As the figures in the last two columns of Table 1 show, the procedure followed provided an ample number of daughters per male. These figures represent 24 cockerels tested in the two resistant lines, the numbers in Shifts I, II, and III being 6, 8, and 10 respectively. (Of the 30 cockerels originally started in 10 pens, two died in March, and four were so infertile that their families were discarded.) The averages and ranges in family size shown do not include 230 pullets eliminated (average number almost 9 per sire) (1) to reduce unduly large families from some sires, (2) to discard small families from some hens, and (3) to give preference to dams that had several daughters by each of the 3 males. All such eliminations took TABLE 1. Details of the operation of triple shifts in 1953 Shift In Male > put: Out Eggs saved Dates Days Eggs discarded Days Daughters per male at 6 weeks Average Range I II III Jan. 9 Feb. 16 Mar. 13 Feb. 12 Mar. 9 immaterial Jan. 30-Feb. 17 Feb. 24-Mar. 14 Mar. 21-Apr None

3 MULTIPLE SHIFTS FOR TESTING COCKERELS 273 out whole dams' families, and were made from the records without handling the birds. WASTAGE, MULTIPLE SHIFTS, AND PRESSURE OF SELECTION Perhaps the best way to evaluate the contribution of multiple shifts to the success of our operations is to examine the record. To simplify matters, the data given are restricted to our C-Resistant line, which was begun first and in which the greatest number of males has been tested. What happened to these before and after multiple shifts were started in 1943 is shown in Table 2. This record shows that 34 to 36 percent of the cockerels started on test were not available for re-use by the following January when the next season's breeding pens were made up. Some died before that time and others had to be discarded because of poor fertility or low hatchability. Although no similar record of the wastage in breeding cockerels is known to us, it seems likely that the losses shown above are not exceptional, but are fairly representative of what any poultry breeder might expect. It is also evident that after the introduction of multiple shifts in 1943 the intensity of selection, or "selection pressure," was markedly increased. Selection Cockerels started on test, no. Died before December 31 Discarded before December 31 Total wastage, no. Total wastage, % Available January 1 Re-used as proven sires, no. Proportion of those available re-used, % Proportion of those started re-used, % 8 years, 193S-' years, 1943-' pressure can be measured by the proportion used for breeding of the available animals of known performance. With respect to the cocks in our breeding program, the only adequate measure of their value is the performance of their daughters with respect to viability, productivity, and other characters of economic value. Prior to the adoption of multiple shifts, the best 40 percent of the tested cockerels still available in January were re-used to head proven-sire ma tings. That figure indicates only slight selection pressure. Some of the males were only a little better than average, but it was necessary to use them to lessen the risk of inbreeding. Had we raised the pressure of selection so that only the best 20 percent of the males were re-used, the proven-sire matings would have been reduced to slightly more than one pen per year, and, with such a practice, the resultant inbreeding might have eliminated the strain altogether. By re-using 40 percent of the males available (some of them for several years) the average number of proven-sire matings was maintained at 3 per year up to After that year, the proven-sire matings were increased so that their average annual number from" 1943 to 1952 was 4. However, since at the same time the average number of cockerels tested annually was increased from 8.7 before 1943 to TABLE 2. Record of cockerels tested before and after the introduction of multiple shifts in thereafter, there was seldom any difficulty in finding excellent males, even though the pressure of selection was raised so that only 23.9 percent of the available tested cockerels were re-used. EVALUATION OF SIRES Before the first pullets are housed about August 1, some of the cockerels under test will have died. Some may have been so low in fertility or hatchability that they have only 20 daughters or so (too few to

4 274 F. B. HUTT AND R. K. COLE test for viability). All such males which could never qualify as desirable sires are eliminated as soon as they can be recognized. If that can be done before housing time, their daughters are also eliminated. There is no point in providing space, labor, and bookkeeping to test males that will never be used again. (a) Partial Tests. The next appraisal is made at the end of December, when breeding pens are arranged for the following season. By that time, some daughters of Shift I will have laid for about 5 months, but those of Shift III, hatched 7 to 9 weeks later, will have had a correspondingly shorter test period. These discrepancies present no difficulties. So long as comparisons are made within shifts, there is little difficulty in identifying males that are outstanding, mediocre, or poor, with respect to the performance of their daughters. Each sire is rated by comparison with the average for his contemporaries of the same shift and strain. When two or more strains are being selected for the same objectives (as are ours) contemporary males of both strains, but of the same shift, can be considered together. Comparisons between males of two different shifts is feasible, even in the partial tests available at December 31, but is less accurate. As was emphasized in a previous discussion of this point (Hutt, 1949), pullets from Shifts I, II, and III differ in performance up to that date because of their being hatched at successively later dates, and being thereby inevitably precipitated into different environments. Typical examples of such differences are shown in Table 3, which gives the averages for 21 cockerels of our two resistant lines tested in triple shifts in These figures show successively fewer eggs for Shifts II and III in comparison with Shift I. The three different averages for each strain could be corrected to a common base to permit direct comparisons of all sires within a strain. In our experience, that has not been necessary, as outstanding sires can easily be identified by comparisons with the averages for their shifts. One could also bring Shifts I, II and III to some common base by using correction figures determined from the records for early, mid-season, and latehatched daughters of the proven sires, each of which remains in his original breeding pen throughout the hatching season. Table 3 shows unusually early maturity for the C strain, Shift II. Lack of a corresponding deviation in the K strain from the successively later maturity that is normal in Shifts II and III shows that the aberrant figure was not induced by the environment, but must be attributed to inheritance transmitted by the two sires. TABLE 3. Some measures of performance up to Dec. 31 for pullets of different shifts and strains hatched in 1952 Strain and shift C Resistant C Resistant C Resistant K Resistant K Resistant K Resistant I II III I II III Sires Number Daughters at 6 weeks Number Median age at first egg Days Eggs to Dec. 31* Average* Mortality to Dec. 31 Percent * Uncorrected figures for 10 days in every 14.

5 MULTIPLE SHIFTS FOR TESTING COCKERELS 275 Distribution of the mortality in these pullets shows no consistent differences among shifts, but such differences frequently occur. They must be considered if accurate evaluations are to be made of the genetic viability transmitted by individual sires. Since the test periods for Shifts II and III are shorter than for Shift I at December 31, these later shifts should ordinarily show lower mortality rates then than the first one, and they frequently do. On the other hand, the later chicks are sometimes affected more by coccidia and other adverse conditions on the rearing range. By comparing averages for shifts within strains, and vice versa, and then comparing individual sirefamiles within shifts, we have no difficulty in distinguishing the best and worst families. Unfortunately the measurement of viability in these partial tests up to December 31 is not as accurate as the assay of productivity, but it must be considered when males are being selected for re-use. (b) Extension of Partial Tests. By these partial tests to December 31, the pullets show not only which sires are most desirable for re-use as proven sires, but also which families in the proven-sire matings are the most promising ones from which to select new cockerels for testing. Naturally, these last are chosen from the best damfamilies in the best sire-families. In all cases, prospective cockerels have been saved from sires previously proven to be superior, and in most cases they come also from proven dams. Some of them are from matings which, having yielded outstanding families in one year, have been repeated in the next specifically to get cockerels for testing. In such cases, the partial tests are unnecessary, or merely confirmatory. Usually, however, most of the available cockerels are not from these "repeat matings," and hence the partial tests of their sisters are important (within the limitations of sib-tests in general) in deciding which cockerels to test. Herein lies a special value of multiple shifts. The cockerels of Shift I have to be selected on the basis of performance of their sisters up to December 31, a test-period of about 3 to 5 months at best. This same limitation of a relatively short sister-test applies equally to any cockerels selected in January for use throughout the breeding season. However, our cockerels of Shift II do not go into the breeding pens until about the middle of February. By that time, sister-performances in January can be added to the previous appraisals, thus making the test period 4 to 6 months. Similarly, when cockerels of Shift III go in the pens in March, their sister-tests have been extended to 5 to 7 months by addition of the records for February. It occasionally happens that cockerels tentatively selected in January for Shifts II and III are rejected later because the extended tests have raised the mortality, or lowered the relative productivity, in the families to which they belong. (c) Complete Tests. The partial tests just discussed usually reveal not only the best prospects for immediate re-use as proven sires, but also a few males so hopelessly outclassed that they can be discarded. All the rest are kept until their tests are completed when their daughters are 500 days old. Final evaluations are then made, as before, by comparisons within shifts. It frequently happens, however, that by 500 days of age the differences among shifts so evident at December 31st have disappeared, leaving average performance for all three shifts so close together that comparisons can be made among all sires within a strain, regardless of shift. Thus,

6 276 F. B. HUTT AND R. K. COLE whereas the average (uncorrected 1 ) figures for egg production in the C-Resistant line's 3 shifts at December 31, 1952, were 57, 56, and 40 (Table 3), the corresponding (corrected 1 ) figures at 500 days, when all groups were comparable with respect to age, were 192, 212, and 208, and there was little difficulty in ranking the 9 sires in the 3 shifts. EVALUATION OF DAMS Although designed primarily to test additional sires, multiple shifts also permit a better evaluation of dams than is possible in single-male matings. In the latter, when some hen's daughters excel the average for those of the sire, one cannot be sure if that should be credited to inheritance from the dam or to "nicking" of the particular sire and dam concerned. If the hen be tested with two males, and produces superior daughters by both, the probability that she is transmitting desirable genes becomes greater, and the possibility that nicking is responsible becomes less. With the triple shifts that we use, it seems reasonably certain that the hens whose daughters by all 3 sires con- 1 For trapnesting 5 days in 7. siderably excel the daughter-averages for those same sires are the kind of birds that we want. Accordingly, such outstanding dams are mated with our very best sires. Since they were first used in the cockereltesting pens when 10 to 12 months old, they can be evaluated by partial tests of their daughters up to December 31, and used as proven dams when 2 years old or less. These dams are re-appraised as tests of their daughters are extended and final evaluations are made in ample time to decide which qualify for use again as proven dams when 3 years old. The procedure is illustrated by the data in Table 4, which show the pertinent figures for one of our pens in which 3 full brothers of the C-Resistant line were mated in turn with 17 pullets. Among these, three proved to be infertile with all 3 cockerels, two did not have daughters by all 3, and four had fewer than 9 daughters altogether. After exclusion of these dam-families, the records for the remaining 8 dams were as given in Table 4. It is clear that only 2 dams, E and F, produced consistently superior daughters by all 3 sires, that all G's daughters were very poor, and that the other hens yielded TABLE 4. Evaluation of dams according to deviations of average egg productions of their surviving daughters by 3 full brothers from the means for all surviving daughters of those same sires Dam A B C D E F G H Daughters per sire, no. Their mean production* Daughters per dam, number First brother Daughters from: Second brother Deviations* Third brother Dam's rating Average Poor Poor Average Superior Superior Bad Poor * To 500 days, uncorrected figures for 10 days in every 14; actualfiguresare greater by 40 percent.

7 MULTIPLE SHIFTS FOR TESTING COCKERELS 277 FIG. 1 Production indices, or hen-housed averages, to 500 days of age for unculled flocks of the C-Resistant strain hatched annually from 1935 to In the last 9 years, the average number of pullets housed per year was about 850, of which two-thirds were sired by cockerels, one-third by proven sires. The figure for 1952 was added after the straight line had been fitted (by least squares) to the other points. good results with one or more sires, but poor results with others. One must not assume that all plus deviations indicate significant superiority. In practice, statistical tests to determine which families of 3 to 5 daughters are significantly better than average are not worth the time required to make them. The experienced breeder can recognize by the magnitudes of the deviations from the mean, coupled with consideration of the numbers of daughters per dam in each case, just which hens are best. In this case, only 2 of 17 pullets originally put in the breeding pen qualified as proven dams. SOME RESULTS Multiple shifts were introduced in our program of selection as the only feasible way to provide more outstanding sires than we had previously been able to find. Since facilities were very limited, any controlled experiment to demonstrate to others the merits of multiple shifts over other methods of testing cockerels was out of the question. Nevertheless, some evidence of their value is indicated in Figure 1. This shows the production indices, in the C-Resistant line, for all annual flocks hatched from 1935 to While the production index, or henhoused average, is scarcely a satisfactory measure of progress in selecting either for egg-production alone, or for viability alone, it has the convenience of combining in one figure the separate influences of these variables. Since we have selected for both, we use it here to save space. It will be noted that from 1938 to 1941, inclusive, there was no improvement whatever, and that the rise in 1942 was very slight. Since the introduction of multiple shifts in 1943, results have varied considerably, but the general trend has been steadily upward. The index for the 1947 flock was greatly lowered by the

8 278 F. B. HUTT AND R. K. COLE advent in mid-winter of Newcastle disease. With the single exception of that year, all of the production indices since 1943 have been higher than any of the preceding ones. In 1948, an outbreak of blue-comb disease, to which this C strain proved more susceptible than others (Cole, 1950) caused a set-back at housing time, which is reflected in the index for that year. In spite of such misfortunes, the index has risen steadily in the 5 years since 1947, and the rate at which it has done so suggests that the end-point is not yet in sight. For the flock of 1952, mortality from 42 to 500 days of age was 26.9 percent, average production of survivors 209 eggs, egg size 59.9 grams and body weight 2113 grams. We believe that the progress made since 1943 is attributable in large measure to the fact that multiple shifts, by providing more tested cockerels, have permitted greater selection pressure than was previously possible. As was shown earlier, without multiple shifts 40 percent of the tested cockerels still available were re-used as proven sires. With multiple shifts, only 24 percent were so used. It would be still better to use only the best 10 percent. The most remarkable thing about the results of selection shown in Figure 1 is not the continuous progress made since 1943, but the fact that any progress whatever was possible in that period. The average number of breeding pens used for the C line was only 10.5 per year. Four of these were allotted to proven sires, the remainder for testing cockerels. Selection was made not only for viability and productivity, but also for fertility, hatchability, egg-size, body weight, and freedom from undesirable defects. Had we used only one cockerel per pen per season, the number tested in 11 years could not have exceeded 72. The actual number tested was 172. The progress made with a few pens, but many objectives, must be attributed in large part to the multiple shifts that provided an additional 100 tested males among which to seek superior sires. CRITICISMS OF PUTATIVE PATERNITY Crew (1926) and Warren and Kilpatrick (1929) found that when one male replaces another, the influence of the first is lost in 7 to 10 days, and frequently earlier. We had considered, therefore, that in crediting eggs to the replacing male 7 days after his introduction, and 9 days after removal of his predecessor, the risk of having a second-shift chick sired by a first-shift male is negligible. To make assurance doubly sure, the procedure was modified two years ago, and eggs are now not credited to the replacing male until the 8th day after his introduction, which is 12 days after removal of the preceding male. Cary and Godfrey (1952) found in their trials that errors in assigning paternity occurred in 8 of 15 pens. Their procedure differed from ours in that eggs were credited to the replacing male as early as the 6th day after his introduction. It is scarcely surprising that this 6-day period proved inadequate. If the sixth-day eggs had been discarded in their first trial, 2 more pens would have been free of errors. In the 2 remaining pens of that trial which showed "errors," it seems possible that these might have been fewer if the transition of the replacing cockerels from celibacy on the cockerel range to the responsibilities of polygamy in the breeding pens had been somewhat less abrupt. Further experiments to determine the rate of change in paternity following replacement of one male by another are reported by Cole and Hutt (1955). It should be remembered that a few errors in

9 MULTIPLE SHIFTS FOR TESTING COCKERELS 279 assigning paternity can do little or no harm. In a modern breeding program, the performance of a hen's daughters is more important than her pedigree. Moreover, from our cockerel-test matings, all the male chicks are usually eliminated early, and prospective breeding cockerels are saved only from proven-sire matings, where no shifts of sires are made. Even if the errors in assumed paternity number 2 percent, one illegitimate daughter in a cockerel's family of 40 or 50 under test will neither make him nor break him. The tests reported by Cole and Hutt show that with the schedule shown in Table 1 errors in assigning paternity are considerably less than 2 percent. CRITICISM OF "SYSTEM Di" The use of multiple shifts has been criticized by Lerner and Dempster (1951). These writers had not tried multiple shifts. They gave no data from breeding trials. Merely by invoking certain symbols, I, h 2, r, and I, by assigning arbitrary values to some of these, and by making calculations therewith, they were able to conclude that "System Di" (multiple shifts) compared with "System Si" (single shift) is "relatively inefficient, except possibly in very small flocks...." There may be a need for this sort of thing, but, unfortunately, its relation to poultry breeding is none too clear. So far as can be determined, the theories of Lerner and Dempster about multiple shifts seem to be (1) that the genetic value of each cockerel is inadequately measured, and (2) that selecting 20 cockerels to test (in two shifts) automatically lowers selection pressure below the level prevailing when only 10 are tested. With respect to the first of these, it is necessary only to point out (a) that we use 40 to 50 daughters or more to evaluate each cockerel, (b) that these are all hatched within 15 days (sometimes within 8 days), and (c) that, as Mueller and Hutt (1946) have shown, larger families are not necessary to test a sire for viability and productivity of his daughters. Facilities for maintaining 1600 pullets under test are better utilized in providing tests for 40 sires than in testing only 16. Their second allegation is based on their assumption that selection pressure is exerted when the cockerels are selected for testing. To some extent this is true, but, when the cockerels for testing are raised only from previously proven dams and sires of high quality (as are ours), selection pressure based on partial sibtests is a minor consideration. Moreover, it is self-evident that whatever effective pressure the breeder may be able to exert before the cockerels are tested is insignificant in comparison with the selection pressure that can be exerted after they have been tested. The improvement in our resistant lines (Figure 1) has depended in part upon the selection of cockerels for testing, but infinitely more upon further selection among them (a year later) of the ones to be re-used as proven sires. As Table 2 shows, selection pressure at that point has been greatly increased since multiple shifts were adopted. Should Lerner and Dempster continue their statistical and theoretical analyses of the merits of multiple shifts it is to be hoped that they will consider some points overlooked in their first attempt. These include the advantages of testing dams in diallel crosses with 3 males rather than with one, the difficulty of selecting for 6 or 8 objectives at once, and the overhead cost of single-mating pens per cockerel when used to test 1, 2, or 3 shifts per pen. Another point to consider is that some poultry breeders who used multiple shifts were able to breed two strains of Leghorns

10 280 F. B. HUTT AND R. K. COLE highly resistant to lymphomatosis (Hutt and Cole, 1947, 1948, 1953), while others who did not use multiple shifts were unable to raise resistance to lymphomatosis in eight years of trying (Taylor, Lerner, De Ome and Beach, 1943). The Full-brother Fallacy. Lerner and Dempster concede that double shifts may have some advantages if two full brothers are selected for testing. Such a procedure would (in their opinion) increase the intensity of selection. During the past 11 years we have tested many pairs and trios of full brothers. This is done whenever certain combinations of good proven sires with proven dams suggest the desirability of testing several sons in the hope of finding a good one. In the 3 breeding seasons of 1951, '52, and '53, no fewer than 17 trios and 11 pairs of full brothers were tested. According to Lerner and Dempster, such full brothers represent equal intensity of selection, i.e. they are equally desirable. In our actual tests, however, one cockerel may transmit mostly good qualities but his brothers mostly bad ones. Trios and pairs in which two full brothers prove worthy of re-use are a small minority. Those predominate in which only one brother has been worth using again. To illustrate the extent to which full TABLE S. Differences in genotypes of three full brothers as revealed by progeny tests, when all three were mated successively with the same hens Number of daughters Performance of daughters to Jan. 1 Mortality, all causes, percent Mortality from neoplasm, percent Egg production, mean number Egg production, mean for shift* Performance of daughters to 500 days of age Mortality, all causes, percent Mortality from neoplasms, peroent Eggs per bird, mean number Eggs per bird, mean for shift* cf TJ2M <? U14M c? U26M * I.e. for all females of the same strain hatched concurrently in cockerel tests. brothers can differ in ability to transmit desirable genes to their daughters, a comparison is given (Table 5) of the records for 3 full brothers, all mated in successive shifts with the same pen of females in Of these brothers, U26M was eliminated early because of the high incidence of leucosis in his family, and U14M died, leaving behind him a record of mediocrity that would have disqualified him even if he had survived. In striking contrast, the other brother proved to be one of the best sires seen in 17 years of testing. Average production of his daughters to 500 days of age was 226 eggs. Their viability was good for a family severely exposed to leucosis but never culled. Equally valuable, this male, U2M, was highly fertile, a trait greatly needed in his strain. For three consecutive years he produced large families of consistently outstanding daughters. All three of these brothers would have had identical ratings with respect to pressure of selection as measured by Lerner and Dempster. The true differences in their genetic value were shown by their daughters, not by their sisters. Ratings of cockerels before tests may have some value, but ratings after tests are infinitely better. Sons of U2M were also tested. Among the first five of these, three proved to be outstanding, and they were used as proven sires for one or more years. Much of the improvement in the C line in the last 3 years must be credited to U2M and his sons. Without multiple shifts, he might never have been found. CRITICISM OF THE NEED FOR TESTING MANY SIRES Nordskog and Wyatt (1952) reported some statistical exercises from which they concluded that the number of breeding

11 MULTIPLE SHIFTS TOR TESTING COCKERELS 281 pens (i.e. males) has only a slight influence on possible genetic gains from selection unless the number is very small. They considered that the advantage obtained from using many pens rapidly diminishes after the number reaches 20 or 30. While we can agree with their suggestion that "even 8 to 10 pedigree breeding pens, or possibly fewer if a multiple shift of males is employed, would adequately take care of the inbreeding problem," we can hardly concur in the implication that the same number would suffice for improvement of a strain. It is obvious that of two breeders, A and B, testing respectively 10 and 100 males a year, A will gain relatively more than B by testing an additional 10. However evidence has not yet been adduced to show that even 20 pens (i.e. males) will suffice to ensure rapid improvement of a strain with respect to all the many traits that the modern poultry breeder must consider. Certainly any breeder having only 20 pens would be well advised to use 14 of them to test 42 cockerels annually, the remainder for proven sires. Nordskog and Wyatt based their calculations on only a single objective, but showed no actual improvement in any economic character with any number of pens whatever. They overlooked the mortality and other wastage that inevitably reduces the sires tested in one generation before the best of them can be used to beget the next. Their views are contrary to the beliefs on which the desirability of multiple shifts is based, namely, that progress in poultry breeding results from the use of superior proven sires, that these are rare and that the only way to find a few is to test many. In the absence of supporting evidence, the theory of Nordskog and Wyatt merits little comment. It is to be hoped that they will themselves test its validity. There are now available many highlyimproved strains of egg-producing breeds. It would be interesting to see how much two of these, if allowed only 10 singlemale breeding pens per strain per year, and no multiple shifts, could be simultaneously improved with respect to fertility, hatchability, viability, resistance to leucosis, uniformity of body size, early attainment of large egg size, egg production, and freedom from undesirable defects including such defects of eggs as blood spots, watery albumen, and the thin shells of summer. Most poultry breeders would also require that the two strains give good results when crossed, with a minimum of broodiness in the F x hybrids. Such a program, if carried out for about 10 years, would provide a fair test of the theory of Nordskog and Wyatt, but, until convincing evidence from such a test is provided, their theory remains a theory, and nothing more. COCKERELS FOR TESTING In our breeding program only cockerels from proven sires and proven dams are tested. All cockerel chicks from cockereltest matings are eliminated. Each proven sire's genotype, or his ability to transmit desirable genes, has been appraised from his matings with several females. Similarly, most of the proven dams have been appraised (in cockerel-test matings) by their daughters from three different males. In both sexes these diallel matings have presumably revealed the true breeding value. This situation is less likely to occur when hens are mated only with one male throughout the season, since effects of nicking in such matings might make some hens appear to be much better than they really are. For this reason, we consider sons of our proven sires and dams to be the best pros-

12 282 F. B. HUTT AND R. K. COLE pects for testing, and that, when facilities or need for holding surplus cockerels are limited, one is not justified in holding sons of cockerels undergoing test. It is possible, however, that some poultry breeders who also operate hatcheries could use extra males for the flocks that supply eggs to the hatchery. If enough of these could not be obtained from the proven-sire matings, there might be justification for holding some cockerels from the cockerel-test matings, particularly from Shifts I and II. If some of the males under test should seem (at the time of evaluation by partial test) to be better than any of the proven sires, their sons could then be retrieved for testing. Theoretically this would "gain a generation" for these few birds. In our opinion, the advantages of thus "gaining a generation" have been somewhat over-rated. This is partly because of the difficulty and expense of holding surplus males, but also because the proportion of cockerels worth testing can at best be only about 20 percent among sons of males under test, but is 100 percent for sons of proven sires and dams. SUMMARY A method is described by which three cockerels are tested each season in every pen utilized for cockerel tests. After removal of the males of Shift I, eggs are credited to them for 5 days. Replacing males of Shift II go in the pens late on the fourth day. Eggs are not credited to them until 8 days after their introduction, which is 12 days after removal of Shift I. Thus eggs are discarded for 6 days. Prior to adoption of multiple shifts in 1942, among 70 cockerels tested in one strain of White Leghorns by singlemale matings, wastage to January 1 was 36 percent, and 40 percent of those then available were used as proven sires. In the first 10 years of multiple shifts, among 172 cockerels tested in the same strain, wastage to January 1, was 34 percent, and only 24 percent of those still available (16 percent of those started on test), were used as proven sires. In the 5 years before multiple shifts were begun, improvement in viability and productivity was practically nil. With multiple shifts, progress was resumed and has been continued over an 11- year period. This is attributed chiefly to the increase in selection pressure made possible by testing annually two or three times the number of cockerels previously tested. Methods for evaluating sires tested in 3 successive shifts are given. One advantage of multiple shifts is that for cockerels used in Shifts II and III, their partial sister-tests are lengthened by one and two months, respectively, over the test period available when cockerels of Shift I are selected. Examples are given to show how the testing of females in diallel matings with 3 successive sires permits a better appraisal than is possible with hens mated only to a single male. Various criticisms of multiple shifts are refuted and, after 12-years' experience with it, the system is recommended to other poultry breeders. REFERENCES Carey, W. M., and G. F. Godfrey, Errors in paternity resulting from the double shift of males. Poultry Sci. 31: Cole, R. K., Differences in familial incidence of mortality from "blue-comb" disease. Poultry Sci. 29: 398^04. Cole, R. K., and F. B. Hutt, Paternity following the replacement of breeding cockerels. Poultry Sci. 34: Crew, F. A. E., On fertility in the domestic fowl. Proc. Roy. Soc. Edinburgh, 46 (Pt. 2): Hutt, F. B., Genetics of the Fowl, xi+590 pp. McGraw-Hill Book Co., Inc., New York. Hutt, F. B., and R. K. Cole, Genetic control

13 PATERNITY FOLLOWING REPLACEMENT OF BREEDING COCKERELS 283 of lymphomatosis in the fowl. Science, 106: Hutt, F. B., and R. K. Cole, The development of strains genetically resistant to avian lymphomatosis. Proc. Eighth World's Poultry Congr. (Copenhagen) I: Hutt, F. B., and R. K. Cole, The interaction of genetic and environmental influences affecting the incidence of avian leucosis. Science, 117: Lerner, I. M., and E. R. Dempster, The theory of multiple shifts. Poultry Sci. 30: THE use of multiple shifts for testing large numbers of cockerels has proven of value as a method for recognizing superior sires (Hutt and Cole, 1955). Males proven to be superior by performance of approximately 50 daughters are then used in subsequent years throughout an entire breeding season of 8 to 9 weeks and produce families two to four times as large. The chief objection to the shifting or replacing of males in the breeding pens is that for a period of time the paternity of the chicks hatched may be in doubt. In practice, this uncertainty need not be a deterrent to the use of multiple shifts. The actual error in assignment of paternity is so low that it would not influence the evaluation of either parent. PREVIOUS REPORTS Crew (1926) exchanged males in two breeding pens and measured the speed with which the semen of the replacing male superseded that of the previous sire. The first Leghorn-sired chicks came from eggs laid two days after the exchange and by the fifth day all chicks were sired by the Leghorn. In the reciprocal exchange, Mueller, C. D., and F. B. Hutt, The numbers of daughters necessary for progeny tests in the fowl. Poultry Sci. 25: Nordskog, A. W., and A. J. Wyatt, Genetic improvement as related to size of breeding operations. Poultry Sci. 31: Taylor, L. W., I. M. Lerner, K. B. DeOme and J. R. Beach, Eight years of progeny-test selection for resistance and susceptibility to lymphomatosis. Poultry Sci. 22: Warren, D. C, and L. Kilpatrick, Fertilization in the domestic fowl. Poultry Sci. 8: Paternity Following the Replacement of Breeding Cockerels R. K. COLE AND F. B. HUTT Department of Poultry Husbandry, Cornell University Agricultural Experiment Station, Ithaca, New York (Received for publication June ) the Redcap sire was slower in effecting fertility, requiring eight days for the first Redcap-sired chick, but apparently entirely supplanting the Leghorn by the eleventh day. Warren and Kilpatrick (1929) alternated, at" 20 and 21 day intervals, Black Minorca and White Leghorn males in two pens of black females and found that the replacing male soon became the sire of a majority of the chicks. Only rarely was a chick from the previous sire found subsequent to a chick sired by the replacing male. Their results, in terms of the error that would have occurred if all chicks hatched from eggs laid starting 6 or 7 days after the introduction of the replacing sire had been assigned to him, were as follows: Starting with the 6th day 7th day 1926 Leghorn 1.8% 1.3% Minorca 13.3% 11.3% 1927 Leghorn 31.0% 29.1% Minorca 2.5% 1.3%