- Email: [email protected]
Two-Way Selection for Body Weight in Chickens
334
M. A. MALONEY, JR., J. C. GILBEEATH AND R. D. MORRISON
Robertson, F. W., 19SS. Selection response and the properties of genetic variation. Symp. Quant. Biol. 20: 166-177. Snedecor, G. W., 1959. Statistical Methods. Fifth
Edition, The Iowa State Univ. Press, Ames, Iowa. Waters, N. F., 1931. Inheritance of body weight in the domestic fowl. Rhode Island Agri. Expt. Sta. Bull. 228.
Two-Way Selection for Body Weight in Chickens 2. THE EFFECT OF SELECTION FOR BODY WEIGHT ON OTHER TRAITS
(Received for publication August 17, 1962)
S
ELECTION experiments for the improvement of certain traits have frequently resulted in the occurrence of a correlated response in an unselected trait (Mather and Harrison, 1949; Lerner, 1946, 1950; Falconer, 1954). Falconer (1954) discussed three main types of correlated responses. If two characters are uncorrelated genetically, then a correlated response would not be expected. A secondary character may nevertheless show an undirected departure from the original population following selection for the primary character. This type of correlated response was reported by Mather and Harrison (1949). If the two characters are correlated genetically, the secondary character will show a direct change following selection for the primary character. This is the commonly observed type of correlated response. Another type of correlated response exists if the secondary character forms an important component of the total fitness. Under these conditions the secondary character may be expected to decline, in response to selection, in either direction from the primary character. This type of response has been termed "genetic homeostasis" by Lerner (1950). Since the occurrence of a correlated re-
sponse is important in a selection program, it would be advantangeous to know the possible effect that long-term selection may have on other traits of economic importance. Results of the effects of selection for twelve-week body weight on other traits in the Silver Oklabar chicken are presented. EXPERIMENTAL PROCEDURE The data presented were collected during a study of selection for high and low body weight at twelve weeks of age and from a study of the effect of relaxed selection on twelve-week body weight following five generations of selection. At six and at twelve weeks of age, all birds were weighed to the nearest tenth of a pound. During the first eight generations, sex was determined at six weeks of age in both lines. In the ninth and tenth generations, however, the low-line progeny were sexed at twelve weeks of age, since the sex could not be determined accurately at six weeks of age. Individual egg production records were maintained through 500 days of age on all females selected as parental stock. Age at sexual maturity was recorded as the number of days from date of hatch to the date of the first egg laid. Per-
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
M. A. MALONEY, JR., J. C. GILBREATH AND R. D. MORRISON Poultry Science Department, Oklahoma State University, Agricultural Experimental Station, Stillwater, Oklahoma
SELECTION E F F E C T S ON N O N - S E L E C T E D T R A I T S
All pedigreed eggs were collected daily and held a t a temperature of 50 to 55 degrees Fahrenheit. T h e eggs were set at 14-day intervals to allow for the maxim u m number of chicks per hatch without a serious decline in hatchability due to the age of the eggs. T h e percentage of fertile eggs was determined b y candling the eggs on the 18th day of incubation. The infertile eggs were "broken o u t " to detect a n y sign of embryonic development, and those showing signs of development were recorded as fertile. T h e percentage hatch of fertile eggs was also determined in the study. Phenotypic correlations and linear regression analysis between twelve-week body weight and other traits were obtained only from the data collected during the last five generations. T h e authors were unable to obtain these estimates from the first five generations due to a loss of the necessary data. The I B M computer program, "Beaton Correlation Routine," was used to calculate correlations
in this study. This program was designed by Beaton (1957). Since the d a t a were obtained from a selected group of individuals, particularly in the high and low lines, the correlation coefficients were biased values. A regression analysis was also utilized in order to determine possible interrelationship between body weight at twelve-weeks of age and other traits. The correlation coefficients obtained between twelve-week body weight and six-week body weight as well as those values obtained from the relaxed-selected-lines are unbiased estimates since all individuals were utilized insofar as possible. Two methods were used to determine the overall correlation a n d regression coefficients. If a significant difference was not obtained between generations within each line, the coefficients were determined b y pooling the variance of each generation. This type of analysis was not used to determine the coefficients if the differences between generations within each line was significant. A group analysis was also utilized b y grouping the data for each generation within line. This procedure, however, does not consider any significant differences which m a y have occurred between generations. The layer-breeder ration fed during the laying and breeding seasons was changed a t the beginning of the sixth generation. This change consisted of a slight increase in the energy level a n d in the percentage of protein. All other management practices were the same, insofar as possible, for all lines within and between generations. RESULTS AND DISCUSSION Six-week Body Weight. Results of. the effect of continuous selection for twelveweek body weight on six-week body
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
centage egg production was determined as the average percentage egg production per production day. I n order to obtain this measurement the days to sexual maturity were subtracted from the smaller of two numbers, either the number of days lived or 500 days. This was designated as the number of production days for each hen. T h e percentage egg production for each hen was determined from the number of eggs laid during her production days. The average of these percentages determined the average percentage production for the line in question. During the first ten days of March, all pedigreed eggs were weighed. From these weights, the individual average egg weight was calculated. To obtain the average egg weight for comparison of lines, each hen's average was used as a single measurement.
335
336
M. A. MALONEY, JR., J. C. GILBREATH AND R. D. MORRISON
4 5 Generation
FIG. 1. The effect of continuous selection for 12-week body weight on 6-week body weight.
weight are shown in Figure 1. This figure shows the means of the six-week body weight of each generation as the arithmetic mean of the sexes combined. During the first six generations of selection, the high line increased steadily; however, during the last four generations the response became very irregular. The low line decreased irregularly during the ten generations of selection. In the last generation of low-line selection, the sixweek weight decreased to a value of 0.24 pound below the value of 0.81 pound for the original population. The linear regression analysis of sixweek body weight by generations is TABLE 1.—Linear regression analysis High line Six-week body weight Regression coefficient (b) 1 Standard error of (b) March egg weight Regression coefficient (b) 2 Standard error
Low line
0.088** + 0.015
-0.008 + 0.011
0.200 ±0.144
-0.632** + 0.149
1 Slope of regression line expressed in pounds change per generation. Shown graphically in Figure 1. 2 Slope of the regression line expressed in grams change per generation. Shown graphically in Figure 2. »* Significant (P<.01).
given in Table 1. The high line in this analysis exhibited a highly significant increase in six-week body weight per generation, since the slope of the regression line was 0.09 + 0.02 pound. The low line decreased 0.01 + 0.01 pound per generation. This rate of change per generation was not significant. The overall phenotypic correlation coefficients between weights at six weeks of age and weights at twelve weeks of age for each line are given in Table 2. When the last five generations were combined, the grouped overall correlation coefficients were 0.245 for the high line and 0.530 for the low line. Both of these correlations are statistically highly significant. The correlation of 0.488 obtained from pooling the within variance in the high line was highly significant. On this basis it may be concluded that there may exist a positive significant correlation between six-week and twelve-week body weight in the last five generations. Both of the linear regression coefficients obtained in the high line were very simi-
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
3
174 178
Pooled within Overall
0.006 0.024 0.015 0.022
0.018 0.018
2.127 2.125
0.456
177 181
d.f.
0.530**
0.131 0.158*
Age at sexual
1.211 0.996
March egg
0.047
0.049 0.048 226 230
% Egg Production 0.027 0.042 0.134 0.340
1.639 0.420 1.888 0.414
0.070 0.025
Sb
R.S.H. 1
0.001 0.011 0.002 0.042
226 -1.887 2.685 230 -1.261. 2.702
in days
226 230
in grams
226
Six-week body weight, in pounds
177 - 1 . 8 8 7 2.685 - 0 . 0 4 7 181 - 1 . 6 6 5 15.709 - 0 . 0 3 0
177 181
1,156
d.f.
Low line
3
2
Relaxed-selected-high line. Relaxed-selected-low line. Regression coefficient. 4 Standard deviation of the regression coefficient. 6 Correlation coefficient. 6 Estimated by pooling the variance within each generation, only if a significant difference did not occur. ' Correlation coefficients obtained from grouping five generations together. * Significant (P<.05). ** Significant (P<.01).
1
178 - 3 . 9 3 4 6.950 - 0 . 0 0 3
0.084 0.086
Pooled within Overall
0.193 0.073 0.103 0.054
174 178
0.448** 0.245**
Pooled within Overall
0.364 0.055 0.366 0.062
Sb4
1,294 1,298
b3
Pooled6 within Overall7
d.f.
High line
TABLE 2.—Regression coefficients and correlation coefficients between twelve-week body weights and other
rom http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
338
M. A. M A L O N E Y , J R . , J. C. G I L B R E A T H AND R. D.
After the initiation of relaxed selection in the high line, six-week weights remained a t about the same value as that of the fifth generation of high-line selection. During the fifth generation of relaxed selection, however, the relaxed-selectedhigh line declined below the fifth generation selected-high-line value of 1.34 pounds, but remained well above the value of 0.810 pound of the original population. The relaxed-selected-low line remained above the original mean value, with very little decrease during the last two generations. The decline noted during the fourth and the fifth generations in both the relaxed-selected-high line and the relaxed-selected-low line failed to return to the original mean values. This would seem to indicate that possibly the genetic makeup controlling six-week body weights was changed somewhat from t h a t of the original population, especially in the high line. The small b u t highly significant value of 0.250 for the correlation coefficient
(Table 2) obtained in the relaxed-selectedhigh line indicated a positive relationship between twelve-week body weight and six-week body weight. This was also inferred by the relatively small regression coefficient of 0.07 obtained for the grouped overall analysis (Table 2). The relaxed-selected-low line, however, indicated greater relationship since the highly significant correlation values of 0.304 from the pooled analysis and 0.407 from the grouped analysis were larger than those obtained from the relaxed-selected-high line. March Egg Weight. A similar correlated response was also noted in March egg weight. The mean egg weight, in grams, for each line by generations is shown in Figure 2. Even with the irregularity noted between generations, the correlated response to selection tended toward increased egg weight in the high line and decreased egg weight in the low line. The mean egg weight of the low line for the tenth generation was 5.8 grams below the original population mean of 54.3 grams. The high-line egg weight increased to 3.5 grams above the original mean value. A linear regression analysis (Table 1) of the mean egg weights of the high line indicated an average increase per generation of 0.20 + 0.14 gram, which was not statistically significant. The analysis of the low line, however, indicated t h a t the mean egg weight decreased at the rate of 0.60 + 0.15 gram per generation. This rate of decline was highly significant. The phenotypic correlation coefficients obtained between body weight a t twelve weeks of age and March egg weight are summarized in Table 2. The overall correlation coefficients of 0.086 and 0.084, determined by combining the last five generations, were not significant. These correlations indicate very little if any direct positive correlation between twelve-week
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
lar, 0.364 and 0.366. These coefficients give additional evidence of a linear relationship between body weight at twelveweeks of age and six-week body weight in the high line. The grouped overall regression coefficient of 0.456 in the low line also indicates the expected linear relationship. Nevertheless, an apparent cessation of the correlated response of the six-week weights in the high line after the sixth generation appears (Figure 1). This m a y suggest a disruption of the genetic correlation with twelve-week body weight. A similar disrupted response was reported by Falconer (1954) on the effect on tail length when selecting for large body size in mice. Since the correlation coefficients were not determined for the first five generations, assumptions relative to the change in this relationship are not warranted.
MORRISON
339
SELECTION EFFECTS ON NON-SELECTED TRAITS
59.6 58.2
HIOH
56.8
\R. S-H.
55.4
1
54.0
2
52.6
~^^RSL
.2*
51.2
•g
49.8
cn LU O
S
48.4 47.0
I 4 5 Generation
6
7
8
I 9
10
FIG. 2. The effect of continuous selection for 12-week body weight on March egg weight.
body weight and March egg weight in the high line. The correlation of 0.158 obtained from the grouped overall analysis was not significant. The low-line overall correlations suggest that a small but consistent positive relationship may exist between twelve-week body weight and March egg weight. The regression analysis of twelve-week body weight and March egg weight in both the high line and the low line also indicated slight linear relationship. The regression coefficients were 0.103 and 0.193 in the high line. These estimates obtained from the low line analysis were 2.125 and 2.127. Neither the relaxed-selected-high line nor the relaxed-selected-low line returned toward the original mean egg weight of 54.3 grams. The value for the relaxedselected-high line in the tenth generation was 2.0 grams above the original mean, and the relaxed-selected-low line was 2.7 grams below the original mean. The regression analysis and the correlation coefficients (Table 2) calculated for the relaxed-selected-high line indicated only a
slight relationship between March egg weight and twelve-week body weight. The correlation coefficients of 0.316 and 0.122 determined from the relaxed-selected-low line were significant. The regression coefficients also suggested that a possible linear relationship existed in the relaxedselected-low line. Even though the correlation coefficients obtained during the last five generations indicated very little relationshipbetween twelve-week body weight and March egg weight, the response to relaxed selection indicated that the genes which affect egg weight were recombined to some degree. Sexual Maturity. The mean age at sexual maturity by generations for each line is shown in Figure 3. Although the generation means fluctuate to a great degree, the low line was consistently higher than the high line. Neither of the two selected lines exhibited any directly correlated response to selection for twelveweek body weight. The difference obtained between the high line and the low line was not significant during the first four generations of selection. The dif-
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
%
340
M. A. M A L O N E Y , J R . , J. C. G I L B R E A T H AND R. D . M O R R I S O N
290 277 264 251
• High Line • Low Line • R. S. H. Line -R. S. L. Line
238 225
199 186 173 160
4 5 Generation
FIG. 3. The effect of continuous selection for 12-week body weight on age at sexual maturity.
ference, however, was highly significant during the last five generations of selection. This would indicate that, even though the overall trend for both lines did not exhibit a directly correlated response to selection for twelve-week body weight, selection for small body size m a y increase the age at sexual maturity more rapidly than selection for high twelveweek body weight. The phenotypic correlation coefficients obtained in this study between twelveweek body weight and age at sexual maturity as shown in Table 2 indicate t h a t little if any correlation existed. The overall correlation coefficients were —0.003 in the high line and —0.030 in the low line. These correlations were not significant. This was further substantiated by the regression coefficients obtained for the two lines (Table 2). The relaxed-selected lines (Figure 3) indicated an almost immediate return toward the original population mean for age at sexual maturity, with the exception of the first generation of the relaxed-se-
lected-low line. The decrease in age a t sexual m a t u r i t y noted during the first three generations of relaxed selection was not sustained since both of the relaxedselected lines increased during the last two generations. The overall fluctuation of the selected lines, and the sudden increases observed in the relaxed-selected lines, indicate t h a t age at sexual m a t u r i t y may have been influenced to a great degree by nongenetic variation or by some type of natural selection in this experiment. The results obtained from the high line in this study do not concur directly with the results reported b y Godfrey and Jull (1935), Goodale (1936), and Lerner (1946). In these experiments, age a t sexual m a t u r i t y was increased as a result of selection for greater body size. Similar results were obtained in this study only during the first four generations of selection for high body weight at twelve weeks of age. The overall results and the low correlation coefficients obtained in this study tend to indicate t h a t age a t sexual
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
212
SELECTION E F F E C T S ON N O N - S E L E C T E D
341
TRAITS
54.0 52.0 50.0 48.0 46.0 44.0 42.0
38.0
—
36.0
— —
34.0
J 0
V L I
2
I 3
High Line Low Line R.S.H. Line R.S. L. Line
JL 4
I 5 6 Generation
J_
J_ 7
J 8
I 9
L_ 10
FIG. 4. The effect of continuous selection for 12-week body weight on % egg production. maturity is more highly influenced by environmental changes than by the change in body weight a t twelve weeks of age. These results are similar to those obtained b y Hays and Sanborn (1936) and Waters (1934), which indicated t h a t age at sexual maturity was influenced by genes a p a r t from those affecting body weight. Egg Production. Percentage egg production was utilized as a measurement of egg production, in order to correct for the significant difference noted between the selected lines in age a t sexual maturity. The mean percentage of egg production for each line by generation is shown in Figure 4. An analysis indicated t h a t a highly significant difference was obtained during the first, ninth and tenth generations. Since all other generations indicated little if any difference between the two selected lines, it m a y be possible t h a t the significant difference was due to sampling error,
especially in the first generation. This would tend to indicate t h a t selection for twelve-week body weight may have little influence on egg production directly. If selection for twelve-week body weight had exerted a permanent change in egg production, then the relaxed-selected lines would have remained subsequently unchanged from t h a t of the fifth generation of selection. Results as shown in Figure 4 suggest t h a t neither the relaxedselected-high line nor the relaxed-selectedlow line remained at the proposed consistent level of production. The fluctuation exhibited by the relaxed-selected lines was very similar to t h a t shown b y the growth-selected lines. The fluctuation m a y have been caused by non-genetic influences and therefore m a y overshadow any effects due to genetic sources. The analysis of the phenotypic correlations (Table 2) obtained between twelveweek body weight and percentage egg
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
40.0
342
M. A. M A L O N E Y , J R . , J. C. G I L B E E A T H AND R. D.
performance on\ the growth-selected lines
TABLE Z. i.—Reproductive
% Fertility
% Hatch of fertile eggs
Gen. Low
Difference
High
81.0 91.5 93.1 84.2 90.0 77.1 88.8 92.3 71.4 69.7 85.8
90.0 90.0 76.1 82.3 88.9 90.9 76.3 96.0 78.2 93.4
01.5 03.1 08.1* 07.7 11.8* 02.1 16.0* 24.6** 10.3 07.6
94.0 76.6 91.2 88.5 92.4 82.2 74.2 79.3 70.2 77.6 77.5
Low Difference
—
—
71.8 87.6 75.0 89.0 82.0 70.1 75.6 78.5 86.2 76.7
04.8 03.6 13.5* 03.4 00.2 04.1 03.7 08.3 08.6 00.8
* Significant (P<.05). ** Significant (P<.01).
production indicate t h a t little if any relationship existed. The grouped overall phenotypic correlation coefficients obtained in the high line and in the low line were 0.018 and 0.048, respectively. Similar results were also indicated by the regression analysis of the lines. Reproductive Performance. D a t a on reproductive performance of the two selected lines are summarized in Table 3. These data indicate t h a t a significant difference in fertility was obtained in the third, fifth, seventh and eighth generations of selection. I t should be noted, however, t h a t in considering all generations, neither line indicated a superiority of fertility over the other line. The average percentage fertility over all generations was 84.4 percent for the high line and 86.2 percent for the low line. In comparison to the original mean of 81.0 percent, the data for both lines indicated t h a t percentage fertility was increased. The increases, however, were not significant. An analysis of the percentage hatch of fertile eggs indicated no significant difference between the two selected lines, with the exception of the third generation.
The high line was consistently higher in percentage hatch of fertile eggs than the low line. The average percentage hatch of fertile eggs over all generations was 81.0 percent in the high line and 79.3 percent in the low line. In comparison to the original mean of 94.0 percent, both lines indicated that percentage hatch of fertile eggs was decreased. The decreases were not significant. D a t a on percentage fertility and percentage hatch of fertile eggs for the two relaxed-selected lines are summarized in Table 4. Due to the large a m o u n t of fluctuation in the relaxed-selected-high line, an accurate account of the influence of relaxed selection could not be determined. This fluctuation m a y possibly have been due to extreme environmental conditions exerted against reproductive performance of the relaxed lines. Although variation was not as great in the relaxedselected-low line, the difference between the two relaxed-selected lines was not significant, except in the ninth generation. The relaxed-selected-low line indicated t h a t percentage fertility remained above the original mean, with the exception of the second generation of relaxed selection. In comparing the average percentage fertility over all generations of the relaxed-selected-high line with the fifth generation of growth selection, a nonsignificant decrease was noted. TABLE 4.—Reproductive performance of the relaxed lines % Hatch of fertile eggs
% Fertility Gen. High 6 90.0 61.9 83.5 58.9 78.8
6 7 8 9 10 :
Low 82.3 66.3 84.8 85.2 83.2
Differ
" ence
High
07.7 04.4 01.3 26.3** 04.4
89.7 78.9 75.7 77.2 80.3
Significant (P<.01).
Low
6
90.2 87.3 86.5 81.9 83.8
ence 00.5 08.4 10.8 04.7 03.5
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
0 1 2 3 4 5 6 7 8 9 10
High
MORRISON
SELECTION EFFECTS ON NON-SELECTED TRAITS
SUMMARY
The correlated response to selection for high and low twelve-week body weight in several traits was studied. Data were collected during ten generations of selection for high and low twelve-week body weight and five generations of relaxed selection for twelve-week body weight. The regression analysis of six-week body weight by generation indicated a highly significant increase of 0.09 pound per generation in the high line. The lowline six-week body weight decreased at a non-significant rate of 0.01 pound per generation. A correlated response was also noted in March egg weight. The regression analysis of March egg weight in the high-line showed an increase of 0.2 gram per generation. This rate of increase was not statistically significant. The analysis of data from the low line, however, indicated that March egg weight was decreased 0.6 gram per generation. This rate of decline was highly significant. It was further noted in this study that age at sexual maturity did not exhibit the definite correlated response as did March egg weight or the response of six-week body weight.
Highly significant phenotypic correlations were obtained between body weight at twelve weeks of age and body weight at six weeks of age in both growth-selected lines. These correlations were 0.245 in the high line and 0.530 in the low line. A significant phenotypic correlation of 0.158 was also obtained between twelveweek body weight and March egg weight in the low line. The phenotypic correlation coefficients obtained in this study between twelve-week body weight and age at sexual maturity were not significant in either of the growth-selected lines. Both the high line and the low line exhibited a small but nonsignificant increase in percentage fertile eggs when compared to the original stock. An analysis of the percentage hatch of fertile eggs indicated that the two growth-selected lines decreased, but not significantly, when compared to the mean of the original population. A significant difference did not occur between the two growth-selected lines in percentage egg production until the last two generations. The average percentage egg production was 47.0 percent in the high line and 43.0 percent in the low line. Results of relaxed selection, initiated after five generations of growth-selection, indicated that the subsequent performance of the relaxed-selected lines remained almost unchanged from that of the growthselected lines, the only exception being age at sexual maturity. ACKNOWLEDGMENTS
This investigation was conducted as a portion of the cooperative research under the Southern Regional Poultry Breeding Project S-41. The authors wish to acknowledge Dr. G. F. Godfrey (Present address: Honegger Farms Co., Inc.; Forrest, 111.) who initiated this study; Dr. Cletis Williams
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
Differences obtained in the percentage hatch of fertile eggs between the relaxedselected lines was not significant during the five generations of relaxed selection. The overall comparison of relaxedselected lines to original mean values of percentage fertility and percentage hatch of fertile eggs indicated a negative tendency to return toward the original mean. The relaxed-selected lines remained at about the same level as that of the fifth generation of growth selection. This would tend to indicate that relaxation of selection may not have changed the overall reproductive performance of these lines.
343
344
M. A. MALONEY, JR., J. C. GILBREATH AND R. D. MORRISON
(Present address: Honegger Farms Co., Inc.; Forrest, 111.), and Dr. B. L. Goodman (Present address: Southern Illinois University; Carbondale, Illinois) for their work and contributions during the earlier years of this study. REFERENCES
Hays, F. A., and R. Sanborn, 1939. Breeding for egg production. Massachusetts Agri. Expt. Sta. Bull. 307. Lerner, I. M., 1946. The effect of selection for shank length on sexual maturity and early egg weight in Single Comb White Leghorn pullets. Poultry Sci. 25:204-209. Lerner, I. M., 1950. Population Genetics and Animal Improvement. Cambridge Univ. Press, Cambridge, Mass. Mather, K., and B. J. Harrison, 1949. The manifold effect of selection. Heredity, 3: 1-52. Waters, N. F., 1934. Growth and sexual maturity in Brama and Leghorn fowl. Iowa State Coll. J. Sci. 8:367-384.
Evaluation of Albumen Quality in a PoultryBreeding Program J. RAFFA Poultry Division, Canada Department of Agriculture, Regina, Saskatchewan, Canada (Received for publication August 27, 1962)
INTRODUCTION
ALBUMEN quality as a factor in a -**• breeding program has received its greatest impetus from Random Sample Tests. These tests have demonstrated to breeders and specialized egg producers that variations in albumen quality, as measured by Haugh units (Haugh, 1939), do exist. Therefore this factor must be considered by the poultry breeder in order to satisfy a keener consumer demand for a quality egg. A review of the literature indicates that interior egg quality decreases during the course of the laying year. Hunter, Van Wagenen and Hall (1936) using albumen and yolk measurements as criteria concluded that there was a seasonal trend in interior quality decline beginning with March or April and continuing through
summer. Wilhelm and Heiman (1938) using albumen index and yolk index found a downward trend in albumen quality from October until July, after which the quality improved slightly. Knox and Godfrey (1938) reported a progressive decrease in percent of thick albumen during the laying year. That breeds and strains may vary in albumen quality was demonstrated by the work of Dawson et al. (1953), Gotterill and Winter, (1954), King and Hall (1955), Johnson and Gowe (1956) and Strain and Johnson (1956). Sauter et al. (1954) noted a consistent seasonal decline in physical egg quality. King and Hall (1955) suggested that selection for albumen quality may be more efficient when done in the fall of the first production year. The purpose of this paper is to propose
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
Beaton, A. E., 1957. Beaton correlation routine, IBM 650 program. Harvard Univ., Cambridge, Mass. Falconer, D. S., 1954. Asymmetrical responses in selection experiments. Symp. Genetics of Population Structure, Intern. Union Biol. Sci., Naples, Series B., No. 15:16-41. Godfrey, A. B., and M. A. Jull, 1935. Statistical studies on the inheritance of sexual maturity in White Leghorns and Rhode Island Reds. Poultry Sci. 14: 346-350.
Goodale, H. D., 1936. "Early egg size." Williamstown Mass. Mount Hope Farm. Reviewed by Hutt, F. B., 1949. Genetics of the Fowl. McGrawHill Book Co., Inc., New York, New York.
M. A. MALONEY, JR., J. C. GILBEEATH AND R. D. MORRISON
Robertson, F. W., 19SS. Selection response and the properties of genetic variation. Symp. Quant. Biol. 20: 166-177. Snedecor, G. W., 1959. Statistical Methods. Fifth
Edition, The Iowa State Univ. Press, Ames, Iowa. Waters, N. F., 1931. Inheritance of body weight in the domestic fowl. Rhode Island Agri. Expt. Sta. Bull. 228.
Two-Way Selection for Body Weight in Chickens 2. THE EFFECT OF SELECTION FOR BODY WEIGHT ON OTHER TRAITS
(Received for publication August 17, 1962)
S
ELECTION experiments for the improvement of certain traits have frequently resulted in the occurrence of a correlated response in an unselected trait (Mather and Harrison, 1949; Lerner, 1946, 1950; Falconer, 1954). Falconer (1954) discussed three main types of correlated responses. If two characters are uncorrelated genetically, then a correlated response would not be expected. A secondary character may nevertheless show an undirected departure from the original population following selection for the primary character. This type of correlated response was reported by Mather and Harrison (1949). If the two characters are correlated genetically, the secondary character will show a direct change following selection for the primary character. This is the commonly observed type of correlated response. Another type of correlated response exists if the secondary character forms an important component of the total fitness. Under these conditions the secondary character may be expected to decline, in response to selection, in either direction from the primary character. This type of response has been termed "genetic homeostasis" by Lerner (1950). Since the occurrence of a correlated re-
sponse is important in a selection program, it would be advantangeous to know the possible effect that long-term selection may have on other traits of economic importance. Results of the effects of selection for twelve-week body weight on other traits in the Silver Oklabar chicken are presented. EXPERIMENTAL PROCEDURE The data presented were collected during a study of selection for high and low body weight at twelve weeks of age and from a study of the effect of relaxed selection on twelve-week body weight following five generations of selection. At six and at twelve weeks of age, all birds were weighed to the nearest tenth of a pound. During the first eight generations, sex was determined at six weeks of age in both lines. In the ninth and tenth generations, however, the low-line progeny were sexed at twelve weeks of age, since the sex could not be determined accurately at six weeks of age. Individual egg production records were maintained through 500 days of age on all females selected as parental stock. Age at sexual maturity was recorded as the number of days from date of hatch to the date of the first egg laid. Per-
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
M. A. MALONEY, JR., J. C. GILBREATH AND R. D. MORRISON Poultry Science Department, Oklahoma State University, Agricultural Experimental Station, Stillwater, Oklahoma
SELECTION E F F E C T S ON N O N - S E L E C T E D T R A I T S
All pedigreed eggs were collected daily and held a t a temperature of 50 to 55 degrees Fahrenheit. T h e eggs were set at 14-day intervals to allow for the maxim u m number of chicks per hatch without a serious decline in hatchability due to the age of the eggs. T h e percentage of fertile eggs was determined b y candling the eggs on the 18th day of incubation. The infertile eggs were "broken o u t " to detect a n y sign of embryonic development, and those showing signs of development were recorded as fertile. T h e percentage hatch of fertile eggs was also determined in the study. Phenotypic correlations and linear regression analysis between twelve-week body weight and other traits were obtained only from the data collected during the last five generations. T h e authors were unable to obtain these estimates from the first five generations due to a loss of the necessary data. The I B M computer program, "Beaton Correlation Routine," was used to calculate correlations
in this study. This program was designed by Beaton (1957). Since the d a t a were obtained from a selected group of individuals, particularly in the high and low lines, the correlation coefficients were biased values. A regression analysis was also utilized in order to determine possible interrelationship between body weight at twelve-weeks of age and other traits. The correlation coefficients obtained between twelve-week body weight and six-week body weight as well as those values obtained from the relaxed-selected-lines are unbiased estimates since all individuals were utilized insofar as possible. Two methods were used to determine the overall correlation a n d regression coefficients. If a significant difference was not obtained between generations within each line, the coefficients were determined b y pooling the variance of each generation. This type of analysis was not used to determine the coefficients if the differences between generations within each line was significant. A group analysis was also utilized b y grouping the data for each generation within line. This procedure, however, does not consider any significant differences which m a y have occurred between generations. The layer-breeder ration fed during the laying and breeding seasons was changed a t the beginning of the sixth generation. This change consisted of a slight increase in the energy level a n d in the percentage of protein. All other management practices were the same, insofar as possible, for all lines within and between generations. RESULTS AND DISCUSSION Six-week Body Weight. Results of. the effect of continuous selection for twelveweek body weight on six-week body
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
centage egg production was determined as the average percentage egg production per production day. I n order to obtain this measurement the days to sexual maturity were subtracted from the smaller of two numbers, either the number of days lived or 500 days. This was designated as the number of production days for each hen. T h e percentage egg production for each hen was determined from the number of eggs laid during her production days. The average of these percentages determined the average percentage production for the line in question. During the first ten days of March, all pedigreed eggs were weighed. From these weights, the individual average egg weight was calculated. To obtain the average egg weight for comparison of lines, each hen's average was used as a single measurement.
335
336
M. A. MALONEY, JR., J. C. GILBREATH AND R. D. MORRISON
4 5 Generation
FIG. 1. The effect of continuous selection for 12-week body weight on 6-week body weight.
weight are shown in Figure 1. This figure shows the means of the six-week body weight of each generation as the arithmetic mean of the sexes combined. During the first six generations of selection, the high line increased steadily; however, during the last four generations the response became very irregular. The low line decreased irregularly during the ten generations of selection. In the last generation of low-line selection, the sixweek weight decreased to a value of 0.24 pound below the value of 0.81 pound for the original population. The linear regression analysis of sixweek body weight by generations is TABLE 1.—Linear regression analysis High line Six-week body weight Regression coefficient (b) 1 Standard error of (b) March egg weight Regression coefficient (b) 2 Standard error
Low line
0.088** + 0.015
-0.008 + 0.011
0.200 ±0.144
-0.632** + 0.149
1 Slope of regression line expressed in pounds change per generation. Shown graphically in Figure 1. 2 Slope of the regression line expressed in grams change per generation. Shown graphically in Figure 2. »* Significant (P<.01).
given in Table 1. The high line in this analysis exhibited a highly significant increase in six-week body weight per generation, since the slope of the regression line was 0.09 + 0.02 pound. The low line decreased 0.01 + 0.01 pound per generation. This rate of change per generation was not significant. The overall phenotypic correlation coefficients between weights at six weeks of age and weights at twelve weeks of age for each line are given in Table 2. When the last five generations were combined, the grouped overall correlation coefficients were 0.245 for the high line and 0.530 for the low line. Both of these correlations are statistically highly significant. The correlation of 0.488 obtained from pooling the within variance in the high line was highly significant. On this basis it may be concluded that there may exist a positive significant correlation between six-week and twelve-week body weight in the last five generations. Both of the linear regression coefficients obtained in the high line were very simi-
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
3
174 178
Pooled within Overall
0.006 0.024 0.015 0.022
0.018 0.018
2.127 2.125
0.456
177 181
d.f.
0.530**
0.131 0.158*
Age at sexual
1.211 0.996
March egg
0.047
0.049 0.048 226 230
% Egg Production 0.027 0.042 0.134 0.340
1.639 0.420 1.888 0.414
0.070 0.025
Sb
R.S.H. 1
0.001 0.011 0.002 0.042
226 -1.887 2.685 230 -1.261. 2.702
in days
226 230
in grams
226
Six-week body weight, in pounds
177 - 1 . 8 8 7 2.685 - 0 . 0 4 7 181 - 1 . 6 6 5 15.709 - 0 . 0 3 0
177 181
1,156
d.f.
Low line
3
2
Relaxed-selected-high line. Relaxed-selected-low line. Regression coefficient. 4 Standard deviation of the regression coefficient. 6 Correlation coefficient. 6 Estimated by pooling the variance within each generation, only if a significant difference did not occur. ' Correlation coefficients obtained from grouping five generations together. * Significant (P<.05). ** Significant (P<.01).
1
178 - 3 . 9 3 4 6.950 - 0 . 0 0 3
0.084 0.086
Pooled within Overall
0.193 0.073 0.103 0.054
174 178
0.448** 0.245**
Pooled within Overall
0.364 0.055 0.366 0.062
Sb4
1,294 1,298
b3
Pooled6 within Overall7
d.f.
High line
TABLE 2.—Regression coefficients and correlation coefficients between twelve-week body weights and other
rom http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
338
M. A. M A L O N E Y , J R . , J. C. G I L B R E A T H AND R. D.
After the initiation of relaxed selection in the high line, six-week weights remained a t about the same value as that of the fifth generation of high-line selection. During the fifth generation of relaxed selection, however, the relaxed-selectedhigh line declined below the fifth generation selected-high-line value of 1.34 pounds, but remained well above the value of 0.810 pound of the original population. The relaxed-selected-low line remained above the original mean value, with very little decrease during the last two generations. The decline noted during the fourth and the fifth generations in both the relaxed-selected-high line and the relaxed-selected-low line failed to return to the original mean values. This would seem to indicate that possibly the genetic makeup controlling six-week body weights was changed somewhat from t h a t of the original population, especially in the high line. The small b u t highly significant value of 0.250 for the correlation coefficient
(Table 2) obtained in the relaxed-selectedhigh line indicated a positive relationship between twelve-week body weight and six-week body weight. This was also inferred by the relatively small regression coefficient of 0.07 obtained for the grouped overall analysis (Table 2). The relaxed-selected-low line, however, indicated greater relationship since the highly significant correlation values of 0.304 from the pooled analysis and 0.407 from the grouped analysis were larger than those obtained from the relaxed-selected-high line. March Egg Weight. A similar correlated response was also noted in March egg weight. The mean egg weight, in grams, for each line by generations is shown in Figure 2. Even with the irregularity noted between generations, the correlated response to selection tended toward increased egg weight in the high line and decreased egg weight in the low line. The mean egg weight of the low line for the tenth generation was 5.8 grams below the original population mean of 54.3 grams. The high-line egg weight increased to 3.5 grams above the original mean value. A linear regression analysis (Table 1) of the mean egg weights of the high line indicated an average increase per generation of 0.20 + 0.14 gram, which was not statistically significant. The analysis of the low line, however, indicated t h a t the mean egg weight decreased at the rate of 0.60 + 0.15 gram per generation. This rate of decline was highly significant. The phenotypic correlation coefficients obtained between body weight a t twelve weeks of age and March egg weight are summarized in Table 2. The overall correlation coefficients of 0.086 and 0.084, determined by combining the last five generations, were not significant. These correlations indicate very little if any direct positive correlation between twelve-week
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
lar, 0.364 and 0.366. These coefficients give additional evidence of a linear relationship between body weight at twelveweeks of age and six-week body weight in the high line. The grouped overall regression coefficient of 0.456 in the low line also indicates the expected linear relationship. Nevertheless, an apparent cessation of the correlated response of the six-week weights in the high line after the sixth generation appears (Figure 1). This m a y suggest a disruption of the genetic correlation with twelve-week body weight. A similar disrupted response was reported by Falconer (1954) on the effect on tail length when selecting for large body size in mice. Since the correlation coefficients were not determined for the first five generations, assumptions relative to the change in this relationship are not warranted.
MORRISON
339
SELECTION EFFECTS ON NON-SELECTED TRAITS
59.6 58.2
HIOH
56.8
\R. S-H.
55.4
1
54.0
2
52.6
~^^RSL
.2*
51.2
•g
49.8
cn LU O
S
48.4 47.0
I 4 5 Generation
6
7
8
I 9
10
FIG. 2. The effect of continuous selection for 12-week body weight on March egg weight.
body weight and March egg weight in the high line. The correlation of 0.158 obtained from the grouped overall analysis was not significant. The low-line overall correlations suggest that a small but consistent positive relationship may exist between twelve-week body weight and March egg weight. The regression analysis of twelve-week body weight and March egg weight in both the high line and the low line also indicated slight linear relationship. The regression coefficients were 0.103 and 0.193 in the high line. These estimates obtained from the low line analysis were 2.125 and 2.127. Neither the relaxed-selected-high line nor the relaxed-selected-low line returned toward the original mean egg weight of 54.3 grams. The value for the relaxedselected-high line in the tenth generation was 2.0 grams above the original mean, and the relaxed-selected-low line was 2.7 grams below the original mean. The regression analysis and the correlation coefficients (Table 2) calculated for the relaxed-selected-high line indicated only a
slight relationship between March egg weight and twelve-week body weight. The correlation coefficients of 0.316 and 0.122 determined from the relaxed-selected-low line were significant. The regression coefficients also suggested that a possible linear relationship existed in the relaxedselected-low line. Even though the correlation coefficients obtained during the last five generations indicated very little relationshipbetween twelve-week body weight and March egg weight, the response to relaxed selection indicated that the genes which affect egg weight were recombined to some degree. Sexual Maturity. The mean age at sexual maturity by generations for each line is shown in Figure 3. Although the generation means fluctuate to a great degree, the low line was consistently higher than the high line. Neither of the two selected lines exhibited any directly correlated response to selection for twelveweek body weight. The difference obtained between the high line and the low line was not significant during the first four generations of selection. The dif-
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
%
340
M. A. M A L O N E Y , J R . , J. C. G I L B R E A T H AND R. D . M O R R I S O N
290 277 264 251
• High Line • Low Line • R. S. H. Line -R. S. L. Line
238 225
199 186 173 160
4 5 Generation
FIG. 3. The effect of continuous selection for 12-week body weight on age at sexual maturity.
ference, however, was highly significant during the last five generations of selection. This would indicate that, even though the overall trend for both lines did not exhibit a directly correlated response to selection for twelve-week body weight, selection for small body size m a y increase the age at sexual maturity more rapidly than selection for high twelveweek body weight. The phenotypic correlation coefficients obtained in this study between twelveweek body weight and age at sexual maturity as shown in Table 2 indicate t h a t little if any correlation existed. The overall correlation coefficients were —0.003 in the high line and —0.030 in the low line. These correlations were not significant. This was further substantiated by the regression coefficients obtained for the two lines (Table 2). The relaxed-selected lines (Figure 3) indicated an almost immediate return toward the original population mean for age at sexual maturity, with the exception of the first generation of the relaxed-se-
lected-low line. The decrease in age a t sexual m a t u r i t y noted during the first three generations of relaxed selection was not sustained since both of the relaxedselected lines increased during the last two generations. The overall fluctuation of the selected lines, and the sudden increases observed in the relaxed-selected lines, indicate t h a t age at sexual m a t u r i t y may have been influenced to a great degree by nongenetic variation or by some type of natural selection in this experiment. The results obtained from the high line in this study do not concur directly with the results reported b y Godfrey and Jull (1935), Goodale (1936), and Lerner (1946). In these experiments, age a t sexual m a t u r i t y was increased as a result of selection for greater body size. Similar results were obtained in this study only during the first four generations of selection for high body weight at twelve weeks of age. The overall results and the low correlation coefficients obtained in this study tend to indicate t h a t age a t sexual
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
212
SELECTION E F F E C T S ON N O N - S E L E C T E D
341
TRAITS
54.0 52.0 50.0 48.0 46.0 44.0 42.0
38.0
—
36.0
— —
34.0
J 0
V L I
2
I 3
High Line Low Line R.S.H. Line R.S. L. Line
JL 4
I 5 6 Generation
J_
J_ 7
J 8
I 9
L_ 10
FIG. 4. The effect of continuous selection for 12-week body weight on % egg production. maturity is more highly influenced by environmental changes than by the change in body weight a t twelve weeks of age. These results are similar to those obtained b y Hays and Sanborn (1936) and Waters (1934), which indicated t h a t age at sexual maturity was influenced by genes a p a r t from those affecting body weight. Egg Production. Percentage egg production was utilized as a measurement of egg production, in order to correct for the significant difference noted between the selected lines in age a t sexual maturity. The mean percentage of egg production for each line by generation is shown in Figure 4. An analysis indicated t h a t a highly significant difference was obtained during the first, ninth and tenth generations. Since all other generations indicated little if any difference between the two selected lines, it m a y be possible t h a t the significant difference was due to sampling error,
especially in the first generation. This would tend to indicate t h a t selection for twelve-week body weight may have little influence on egg production directly. If selection for twelve-week body weight had exerted a permanent change in egg production, then the relaxed-selected lines would have remained subsequently unchanged from t h a t of the fifth generation of selection. Results as shown in Figure 4 suggest t h a t neither the relaxedselected-high line nor the relaxed-selectedlow line remained at the proposed consistent level of production. The fluctuation exhibited by the relaxed-selected lines was very similar to t h a t shown b y the growth-selected lines. The fluctuation m a y have been caused by non-genetic influences and therefore m a y overshadow any effects due to genetic sources. The analysis of the phenotypic correlations (Table 2) obtained between twelveweek body weight and percentage egg
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
40.0
342
M. A. M A L O N E Y , J R . , J. C. G I L B E E A T H AND R. D.
performance on\ the growth-selected lines
TABLE Z. i.—Reproductive
% Fertility
% Hatch of fertile eggs
Gen. Low
Difference
High
81.0 91.5 93.1 84.2 90.0 77.1 88.8 92.3 71.4 69.7 85.8
90.0 90.0 76.1 82.3 88.9 90.9 76.3 96.0 78.2 93.4
01.5 03.1 08.1* 07.7 11.8* 02.1 16.0* 24.6** 10.3 07.6
94.0 76.6 91.2 88.5 92.4 82.2 74.2 79.3 70.2 77.6 77.5
Low Difference
—
—
71.8 87.6 75.0 89.0 82.0 70.1 75.6 78.5 86.2 76.7
04.8 03.6 13.5* 03.4 00.2 04.1 03.7 08.3 08.6 00.8
* Significant (P<.05). ** Significant (P<.01).
production indicate t h a t little if any relationship existed. The grouped overall phenotypic correlation coefficients obtained in the high line and in the low line were 0.018 and 0.048, respectively. Similar results were also indicated by the regression analysis of the lines. Reproductive Performance. D a t a on reproductive performance of the two selected lines are summarized in Table 3. These data indicate t h a t a significant difference in fertility was obtained in the third, fifth, seventh and eighth generations of selection. I t should be noted, however, t h a t in considering all generations, neither line indicated a superiority of fertility over the other line. The average percentage fertility over all generations was 84.4 percent for the high line and 86.2 percent for the low line. In comparison to the original mean of 81.0 percent, the data for both lines indicated t h a t percentage fertility was increased. The increases, however, were not significant. An analysis of the percentage hatch of fertile eggs indicated no significant difference between the two selected lines, with the exception of the third generation.
The high line was consistently higher in percentage hatch of fertile eggs than the low line. The average percentage hatch of fertile eggs over all generations was 81.0 percent in the high line and 79.3 percent in the low line. In comparison to the original mean of 94.0 percent, both lines indicated that percentage hatch of fertile eggs was decreased. The decreases were not significant. D a t a on percentage fertility and percentage hatch of fertile eggs for the two relaxed-selected lines are summarized in Table 4. Due to the large a m o u n t of fluctuation in the relaxed-selected-high line, an accurate account of the influence of relaxed selection could not be determined. This fluctuation m a y possibly have been due to extreme environmental conditions exerted against reproductive performance of the relaxed lines. Although variation was not as great in the relaxedselected-low line, the difference between the two relaxed-selected lines was not significant, except in the ninth generation. The relaxed-selected-low line indicated t h a t percentage fertility remained above the original mean, with the exception of the second generation of relaxed selection. In comparing the average percentage fertility over all generations of the relaxed-selected-high line with the fifth generation of growth selection, a nonsignificant decrease was noted. TABLE 4.—Reproductive performance of the relaxed lines % Hatch of fertile eggs
% Fertility Gen. High 6 90.0 61.9 83.5 58.9 78.8
6 7 8 9 10 :
Low 82.3 66.3 84.8 85.2 83.2
Differ
" ence
High
07.7 04.4 01.3 26.3** 04.4
89.7 78.9 75.7 77.2 80.3
Significant (P<.01).
Low
6
90.2 87.3 86.5 81.9 83.8
ence 00.5 08.4 10.8 04.7 03.5
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
0 1 2 3 4 5 6 7 8 9 10
High
MORRISON
SELECTION EFFECTS ON NON-SELECTED TRAITS
SUMMARY
The correlated response to selection for high and low twelve-week body weight in several traits was studied. Data were collected during ten generations of selection for high and low twelve-week body weight and five generations of relaxed selection for twelve-week body weight. The regression analysis of six-week body weight by generation indicated a highly significant increase of 0.09 pound per generation in the high line. The lowline six-week body weight decreased at a non-significant rate of 0.01 pound per generation. A correlated response was also noted in March egg weight. The regression analysis of March egg weight in the high-line showed an increase of 0.2 gram per generation. This rate of increase was not statistically significant. The analysis of data from the low line, however, indicated that March egg weight was decreased 0.6 gram per generation. This rate of decline was highly significant. It was further noted in this study that age at sexual maturity did not exhibit the definite correlated response as did March egg weight or the response of six-week body weight.
Highly significant phenotypic correlations were obtained between body weight at twelve weeks of age and body weight at six weeks of age in both growth-selected lines. These correlations were 0.245 in the high line and 0.530 in the low line. A significant phenotypic correlation of 0.158 was also obtained between twelveweek body weight and March egg weight in the low line. The phenotypic correlation coefficients obtained in this study between twelve-week body weight and age at sexual maturity were not significant in either of the growth-selected lines. Both the high line and the low line exhibited a small but nonsignificant increase in percentage fertile eggs when compared to the original stock. An analysis of the percentage hatch of fertile eggs indicated that the two growth-selected lines decreased, but not significantly, when compared to the mean of the original population. A significant difference did not occur between the two growth-selected lines in percentage egg production until the last two generations. The average percentage egg production was 47.0 percent in the high line and 43.0 percent in the low line. Results of relaxed selection, initiated after five generations of growth-selection, indicated that the subsequent performance of the relaxed-selected lines remained almost unchanged from that of the growthselected lines, the only exception being age at sexual maturity. ACKNOWLEDGMENTS
This investigation was conducted as a portion of the cooperative research under the Southern Regional Poultry Breeding Project S-41. The authors wish to acknowledge Dr. G. F. Godfrey (Present address: Honegger Farms Co., Inc.; Forrest, 111.) who initiated this study; Dr. Cletis Williams
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
Differences obtained in the percentage hatch of fertile eggs between the relaxedselected lines was not significant during the five generations of relaxed selection. The overall comparison of relaxedselected lines to original mean values of percentage fertility and percentage hatch of fertile eggs indicated a negative tendency to return toward the original mean. The relaxed-selected lines remained at about the same level as that of the fifth generation of growth selection. This would tend to indicate that relaxation of selection may not have changed the overall reproductive performance of these lines.
343
344
M. A. MALONEY, JR., J. C. GILBREATH AND R. D. MORRISON
(Present address: Honegger Farms Co., Inc.; Forrest, 111.), and Dr. B. L. Goodman (Present address: Southern Illinois University; Carbondale, Illinois) for their work and contributions during the earlier years of this study. REFERENCES
Hays, F. A., and R. Sanborn, 1939. Breeding for egg production. Massachusetts Agri. Expt. Sta. Bull. 307. Lerner, I. M., 1946. The effect of selection for shank length on sexual maturity and early egg weight in Single Comb White Leghorn pullets. Poultry Sci. 25:204-209. Lerner, I. M., 1950. Population Genetics and Animal Improvement. Cambridge Univ. Press, Cambridge, Mass. Mather, K., and B. J. Harrison, 1949. The manifold effect of selection. Heredity, 3: 1-52. Waters, N. F., 1934. Growth and sexual maturity in Brama and Leghorn fowl. Iowa State Coll. J. Sci. 8:367-384.
Evaluation of Albumen Quality in a PoultryBreeding Program J. RAFFA Poultry Division, Canada Department of Agriculture, Regina, Saskatchewan, Canada (Received for publication August 27, 1962)
INTRODUCTION
ALBUMEN quality as a factor in a -**• breeding program has received its greatest impetus from Random Sample Tests. These tests have demonstrated to breeders and specialized egg producers that variations in albumen quality, as measured by Haugh units (Haugh, 1939), do exist. Therefore this factor must be considered by the poultry breeder in order to satisfy a keener consumer demand for a quality egg. A review of the literature indicates that interior egg quality decreases during the course of the laying year. Hunter, Van Wagenen and Hall (1936) using albumen and yolk measurements as criteria concluded that there was a seasonal trend in interior quality decline beginning with March or April and continuing through
summer. Wilhelm and Heiman (1938) using albumen index and yolk index found a downward trend in albumen quality from October until July, after which the quality improved slightly. Knox and Godfrey (1938) reported a progressive decrease in percent of thick albumen during the laying year. That breeds and strains may vary in albumen quality was demonstrated by the work of Dawson et al. (1953), Gotterill and Winter, (1954), King and Hall (1955), Johnson and Gowe (1956) and Strain and Johnson (1956). Sauter et al. (1954) noted a consistent seasonal decline in physical egg quality. King and Hall (1955) suggested that selection for albumen quality may be more efficient when done in the fall of the first production year. The purpose of this paper is to propose
Downloaded from http://ps.oxfordjournals.org/ at University of Hawaii - Manoa on June 12, 2015
Beaton, A. E., 1957. Beaton correlation routine, IBM 650 program. Harvard Univ., Cambridge, Mass. Falconer, D. S., 1954. Asymmetrical responses in selection experiments. Symp. Genetics of Population Structure, Intern. Union Biol. Sci., Naples, Series B., No. 15:16-41. Godfrey, A. B., and M. A. Jull, 1935. Statistical studies on the inheritance of sexual maturity in White Leghorns and Rhode Island Reds. Poultry Sci. 14: 346-350.
Goodale, H. D., 1936. "Early egg size." Williamstown Mass. Mount Hope Farm. Reviewed by Hutt, F. B., 1949. Genetics of the Fowl. McGrawHill Book Co., Inc., New York, New York.