- Email: [email protected]
Value of Feed Consumption Records to Predict Net Income in Layer-Type Chickens1
1912
W. A. BOUGH
Harris, C. E., and W. A. Moats, 1974. Recovery of egg solids from wastewater from egg grading and breaking plants. Poultry Sci. 53: 1933. Orr, M. L., and B. K. Watt, 1957. Amino acid content of foods. Home Economics Research Report No. 4. Supt. of Documents, U.S. Government Printing Office, Washington, D.C. Peniston, Q. P., and E. L. Johnson, 1970. Method for treating an aqueous medium with chitosan and
derivatives of chitin to remove an impurity. U.S. patent No. 3,533,940. Spackman, D. H., S. Moore and W. H. Stein, 1958. Automatic recording apparatus for use in chromatography of amino acids. Anal. Chem. 30: 1190-1206. Zall, R. R., and W. Siderewicz, 1974. Wastes from egg breaking plants. 34th Annual Inst, of Food Technologists Meeting, New Orleans, La. May 13. Abstract No. 12.
C. Y. L E E AND A. W. NORDSKOG
Department of Animal Science, Iowa State University, Ames, Iowa 50010 (Received for publication February 17, 1975)
ABSTRACT The importance of measuring feed consumption as a supplementary criterion of net performance was studied from two different sources of data: (1) random sample egg production tests conducted in the United States between 1949 and 1957 (USRST) and (2) a private test of experimental commercial strains conducted by the Goto Hatchery, Inc., Gifu City, Japan, between 1968 to 1971 (GOTO). Seven traits included in the study were net income (NI), egg rate (ER), egg weight (EW), body weight (BW), mortality (MORT), maturity (MAT) and feed consumption (FEED). Performance indexes were derived as multiple regression equations with NI dependent on the other traits. To test the value of feed consumption records, two performance indexes were compared in each data source. The first, 1(6), contained all 6 independent variables, and the second, 1(5), contained all variables except FEED. The correlations between NI and 1(6) were 0.820 and 0.824 and between NI and 1(5) were 0.818 and 0.822 in the USRST and GOTO data, respectively. The difference between the correlations with and without feed was nil in either data source. Thus, measuring FEED did not significantly improve the predictive value of a performance index when prior information was available on the other five variables. 1(5) was also compared with an empirical performance index, 1(E), defined as the ratio of egg mass output to the 3/4 power of body weight. The latter simulates a feed efficiency index where egg mass output is an estimator of feed utilized for egg production and the 3/4 power of body weight is an estimator of feed used in the maintenance of body tissues. The correlations between 1(E) and net income were 0.68 ond 0.74 in the USRST and GOTO data, respectively, which were only slightly lower than the corresponding correlations between net income and 1(5) when the latter was applied to the data source other than that from which it was derived. Thus, a rather simple empirical performance index of egg production efficiency, 1(E), was almost as accurate in predicting net income as a multiple regression-derived performance index. POULTRY SCIENCE 54: 1912-1918, 1975
INTRODUCTION ECAUSE feed is the major item in the cost of producing eggs, the question is whether feed consumption is worth including in an index of net income on a pen of birds.
B
1. Journal Paper No. J-8110of the Iowa Agriculture and Home Economics Experiment Station, Project 1711.
Nordskog (1960) reported that 93% of the variation in net income could be accounted for by differences in performance factors other than feed consumption in U.S. random sample egg production tests. The major portion of the variation was associated with hen-housed production (i.e., egg rate and mortality) and a minor portion with body size and egg size. McNally and Foster (1969)
Downloaded from http://ps.oxfordjournals.org/ at Univ of Iowa-Law Library on June 3, 2015
Value of Feed Consumption Records to Predict Net Income in Layer-Type Chickens*
F E E D CONSUMPTION AND INCOME
In the present study, performance records of laying hens were analyzed separately from each of two different sources of data: (1) random sample laying tests from the United States and (2) commercial experimental tests from the Goto Hatchery, Inc., Gifu City, Japan. By using the two independent sources of data, the accuracy of a performance index, formulated as a multiple regression equation from one set of data, was tested as a predictor of net income in the other set of data, with and without feed consumption as a variable. In addition, the multiple regression-derived performance indexes were compared with an empirical index of egg production efficiency, based on only egg mass output and body weight.
1949 to 1957. These data, except for the records on feed consumption, were used in a previous study (Nordskog, 1960). Included were 8 tests from California, 3 from Missouri, 7 from central New York and 3 from western New York. A test consisted of entries each of approximately 50 commercial egg-type pullets obtained from a random sample of hatching eggs or chicks from U. S. commercial breeders. The data were first adjusted for test location, year differences and number of pen replications per stock 2 before statistical analyses were made. The GOTO data consisted of experimental results on 161 commercial test crosses from the GOTO Hatchery Inc., a relatively large Japanese commercial breeding company. Various experimental strain crosses or "entries" were tested at nine different locations in the 1968-1971 period. In both sets of data, test stocks consisted of both Leghorn type and heavy-breed type layers maintained under either floor-type or cage management. Six independent traits (X) and one dependent trait (Y) were included in each set of data. These were, Symbols Y ~ NI
X, ~ ER
X 2 ~ EW MATERIALS AND METHODS The Data. For the first set of data, the records were taken from Official Random Sample Egg Production Tests conducted in the United States (USRST) and consisted of 360 entries from 21 separate random sample tests carried out over an 8-year period from
USRST—Income over feed and chick costs (dollars) GOTO—Income over feed costs (Yen) USRST—Egg production (number, hen-day) GOTO—Same (%, hen-day) USRST—Large and extralarge eggs (24 oz./doz. or over) GOTO—Average egg weight (g.)
2. See Annual Report of Random Sample Egg Productive tests in United States and Canada, A.R.S., U.S.D.A.
Downloaded from http://ps.oxfordjournals.org/ at Univ of Iowa-Law Library on June 3, 2015
reported that 99% of the variation in net income could be accounted for by including feed consumption in a performance index based on data from the Gosford Random Sample Test in Great Britain. Kinney et al. (1969) developed a multiple regression performance index for predicting income by using7 independent variables: maturity, henhoused production, feed efficiency, egg weight, percentage laying mortality, mature body weight and percentage large and extralarge eggs. They concluded that percentage large and extra-large eggs and hen-housed egg production were the 2 major incomepredicting factors and that feed efficiency and egg weight were less important. Because mean egg weight and percentage large and extra-large eggs are, however, substantially equivalent traits, their results contain a redundancy.
1913
1914 X 3 ~ BW
X 4 ~ MORT
X 5 ~ MAT
X 6 ~ FEED
C. Y. LEE AND A. W. NORDSKOG
Because some of the measurements differed in the 2 sources of data, it was necessary to adjust for these before making comparisons. Net income in yen was converted to dollars in the GOTO test on the basis of the 1969 foreign exchange rate of 360 yen per dollar. Egg rate in the GOTO data was converted to number of eggs per bird on a hen-day basis. Egg size in the USRST data, measured as the percentage of large and extra-large eggs, was converted to grams of egg weight. This required the assumption that egg weight was normally distributed. Eggs classified as "large and extra-large" weighed 24 ounces per dozen or over or, equivalently, 57 g. or more per egg. The conversion of % large and extra-large to mean egg weight was based on the standard normal curve with mean fx and standard deviation, a as shown in Figure 1. Thus, if
FIG. 1. Standard normal frequency distribution with mean |j. and lower bound at D for a truncated distribution with area p.
a fraction, p, of eggs was classified as large and extra-large with lower bound D, and mean X of the truncated distribution then D = \x. - X/CT, is a function of the area p and X = jo. + Da is the estimated egg weight. For example, a typical value of % large and extra-large eggs in the USRST data was 70. The standard normal deviate, D, is 0.525 as given in any cumulative normal frequency distribution table for the area between 0.7 and 0.5. Then for \x. = 57 g. and a = 8.5 gX = 57 + (0.525) (8.51) = 61g. Thus, the estimated mean egg weight of any entry reporting 70% large and extra-large eggs is 61 g. Body weight, measured in pounds in the USRST data, was converted to kg. Mortality was expressed as the percentage of laying house mortality in both data sources. Feed consumption, measured in pounds per bird in the USRST data, was converted to kg. Multiple Regression Analysis. Multiple regression equations were constructed for the dependent variable Y, and the 6 independent variables XI, X2, X3, X4, X5 and X6. When used as a predictor of net income, the multiple regression equations are called performance indexes. From the multiple regression equations, the reduction in mean squares due to regression and multiple correlation coefficients were computed. Finally, correlations between the computed performance indexes and observed FEED were calculated. Efficiency Index. The main source of income from a laying hen operation is the egg mass output. A minor source of income is the salvage value of the bird at the end of the laying period. Egg mass per bird (EM) is the product of egg rate (ER) and mean
Downloaded from http://ps.oxfordjournals.org/ at Univ of Iowa-Law Library on June 3, 2015
USRST—Av. weight of birds alive at end of the test (lb.) GOTO—Same (kg.) USRST—Died after 150 days or after housing (%) GOTO—Same USRST—Age at 50% production (days) GOTO—Same USRST—Feed per bird for entire test period (lb.) GOTO—Feed per bird per day (g.)
1915
FEED CONSUMPTION AND INCOME
egg weight (EW). For this study, efficiency of feed for egg production is defined as the product of egg mass (g.) x survival rate (%) divided by "metabolic" body size, or in symbols, ER x EW x SR
KE) =
EM
3
BW 3 ' 4
BW /"
M = cW3'4 where c is a constant (Kleiber, 1936).
A summary of the means, standard deviations and ranges of each trait from the two different data sources is presented in Table 1. The net income, egg rate, and mortality means were greater in the GOTO data than in the USRST data, but egg weight, body weight, maturity, and feed consumption were greater in the USRST data. The mean differences between the two data sources were statistically significant for all traits (P < 0.01) except mortality. Table 2 presents the simple correlations of each trait with net income and with feed consumption. All traits except mortality were significantly correlated with net income and with feed consumption. Maturity was nega-
TABLE 1.—Summary of United States Random Sample Test data (USRST), and Goto Experimental Test Group data (GOTO) Unit
Trait NI
$
ER
No. of eggs
EW
g.
BW
kg.
MORT
%
MAT
Days at 50% egg prod.
FEED
g./bird/day
Data source USRST GOTO USRST GOTO USRST GOTO USRST GOTO USRST GOTO USRST GOTO USRST GOTO
Mean 2.63 2.84 211.20 225.50 61.73 61.01 2.40 2.19 23.24 24.08 188.72 164.23 120.00 118.05
Standard deviation 0.95 0.51 25.46 13.62 2.17 1.74 0.38 0.22 11.95 9.26 16.24 5.98 8.14 4.52
Minimum 0.31 1.06 101 167 52.83 54.40 1.63 1.69 2.00 4.80 147 145 90.72 105.00
Maximum 5.% 4.05 287 257 67.54 65.00 3.40 2.92 69.00 49.70 276 183 138.67 129.00
TABLE 2.—Correlations of performance traits with net income and feed consumption using data from U.S. Random Sample Egg Tests and the Goto Hatchery, Inc., Gifu City, Japan Trait Net income Egg rate Egg wt. Body wt.
L.H. mortality Maturity Feed consumption *P < .05. **P< .01.
U.S.
—
77** 14** .24** .61** .48** 22**
Net income
GOTO
Feed consumption U.S. GOTO .22**
55** .48** _ 22** — .45** _ 33** .19*
.14** .66** .02 -.18**
.19* .23** .41** 40** -.11 .05
—
—
]Q**
Downloaded from http://ps.oxfordjournals.org/ at Univ of Iowa-Law Library on June 3, 2015
where SR is the survival fraction (1-MORT). Energy for body maintenance can be estimated from the basal metabolic rate (M) which, in turn, can be estimated as a power function of body weight (W),
RESULTS
1916
C. Y. LEE AND A. W. NORDSKOG
TABLE 3.—Correlations between different performance indexes with net income and with feed consumption
Performance index Mult, regression1
Parameters from data source U.S.
Feed consumption (FEED) GOTO U.S. 0.224* • 0.287**
Net income (NI) GOTO U.S. _ '0.720 0.818-
tively correlated with feed consumption in the U.S. test but not in the GOTO test. From each set of data, multiple regression equations were calculated with net income as the dependent variable. In both sets, the multiple correlations, R, came out to be 0.82. When the multiple regression equations were calculated with feed consumption deleted, the R's were only slightly lower. By squaring R we can estimate how much of the variance in net income is associated with the independent variables. In the U.S. data, R 2 = 0.672 with feed in the index, and R 2 = 0.669 without feed in the index. In the GOTO data, R 2 = 0.680 with feed, and R 2 = 0.676 without feed. Thus, when information is available on the 5 traits (egg rate, egg size, body size, flock mortality and age at maturity) essentially nothing more is gained in predictive value by including feed consumption records in the index. An alternative to predicting income from a multiple regression performance index is the efficiency index 1(E), already defined. In Table 3, the efficiency index is contrasted with the multiple regression performance indexes derived from each data source. The basis for comparison is the degree of correlation between the index and net income. When the multiple regression index is constructed from the parameters estimated in one set of data (as US) and then fitted to the data in the other set (GOTO), and R drops from 0.82
to about 0.72. On the other hand, the 1(E) was only slightly less highly correlated with NI (r = 0.676 and r = 0.739 in the US and GOTO data, respectively). These results demonstrate that 1(E), as defined, is almost as good a predictor of performance as a multiple regression index with 5 or 6 variables. The total variance in dependent variable, Y ~ NI, can be partitioned into a fraction due to the direct effects of each independent variable, x, and to the indirect, or joint effects of each x{, xi pair of variables. Thus, in terms of standard partial regression coefficients, Pj = b i cr i /o- y , the total variance in y is partitionable as follows,
2
P?
+2
ESPipir« i=l
Direct effects
+Ee
Indirect effects
(
=1
j-1
Residual )
R
2
The quantity, 1 - R 2 is defined as the residual. The direct effects of each variable, the sum of the direct effects, the sum of the indirect effects and the residual effects on variance in net income are given in Table 4. The indirect response can be either negative
Downloaded from http://ps.oxfordjournals.org/ at Univ of Iowa-Law Library on June 3, 2015
Mult, regression1 Goto 0.732 "0.822 0.048 ns -0.305** Efficiency index, 1(E) Empirical 0.676 0.019 ns -0.077 ns 0.739 •Feed consumption is omitted in the index. Correlations between the performance index and the variable made within the same data source are connected by solid lines and by dotted lines not from the same data source.
1917
FEED CONSUMPTION AND INCOME
TABLE 4.—The partition of the variance influencing net income into direct and indirect effects associated with the independent variables in multiple regression (percentages) USRST Trait
5 variables 35.46 0.00 0.59 7.79 1.75 ~
or positive because r 12 can be either plus or minus. Likewise, it follows that the direct response could be greater than 1. The results however, show that in both sets of data the direct and indirect effects were positive, although the latter were considerably smaller in the GOTO data. Nevertheless, the sums of the direct and indirect effects, R 2 , were very much the same in both sets of data with or without FEED as a variable. Egg rate accounted for the largest fraction of variance for both sources: 31.4% in the USRST data and 30.6% in the GOTO data. Although the variance, in egg weight did not have an important effect on the variance in the net income in the USRST data, it accounted for 15.0% of the variance in the GOTO data. Mortality was the second most important influence on net income in both sources: the percentages of variance in net income attributable to mortality were 7.61 in the USRST data, and the 18.74 in the GOTO data. Compared with egg rate and mortality, body weight and maturity had relatively small effects on variance in net income in either data source. Feed consumption had only a slight effect on the variance in net income: 0.77 percent in the USRST data and 1.00 percent in the GOTO data.
45.59 21.30 66.89 33.11 100.00
GOTO 6 variables 5 variables 30.64 25.44 15.02 12.31 0.81 2.25 18.74 18.71 0.06 0.00 1.00 ~ 66.87 ~~ 58.71 1.18 8.91 68.05 67.62~ " 31.95 32.38 100.00 l6o.OO~ " DISCUSSION
Because feed is the largest cost item of egg production, it would seem that information on feed consumption would be needed to accurately predict net income. Feed consumption can be measured directly by pen groups of birds or by experimental crosses as was done for this study. This requires extra labor and special facilities. An alternative is to estimate feed consumption from the ratio of egg mass output to body weight. Earlier Nordskog et al. (1969) tentatively concluded that the slightly greater accuracy in estimating net performance by measuring feed consumption on pen groups of birds does not justify the additional work required. The results of the present study support this conclusion. The efficiency index, defined as the ratio of egg mass output to the 3/4 power of body weight, is an indirect expression of feed conversion. It evidently predicts net income almost as well as a multiple regression index used to predict income in an independent set of data. Kinney et al. (1969) also showed that the predictive value of a performance index was much lower when applied to an independent set of data, but this is a completely expected consequence of the property of fitting data by least squares.
Downloaded from http://ps.oxfordjournals.org/ at Univ of Iowa-Law Library on June 3, 2015
ER EW BW MORT MAT FEED fdirecreffects (a) X indirect effects (b) a~+ b~ = ib0R~ Residual = 100(1-R 2 ) Toral
6 variables 31.49 0.00 0.00 7.63 1.41 0.77 41.30~ 25.86 67 J 6 32.84 100.00
1918
C. Y. L E E AND A. W. NORDSKOG
ACKNOWLEDGMENT We are indebted to Goto Hatchery, Inc. of Gifu City, Japan for their kind financial support in carrying out this study and for making the GOTO performance test data available to us.
REFERENCES
Reproductive Response to Intermittent Light Regimens in Coturnix coturnix japonica WAYNE L . BACON AND KARL E . NESTOR
Department of Poultry Science, Ohio Agricultural Research and Development Center, Wooster, Ohio 44691 (Received for publication February 17, 1975)
ABSTRACT The effect of intermittent light regimens of one hour of light interspaced with dark periods of one, two or three hours, repeated three, four, five or six times per day was determined for age to first egg, hen day percent production and egg production for a 126 day experimental period in females, and the age to initial response of the cloacal gland in males. Three broad classes of response were noted based on the equivalent light period (ELP), which is defined as the interval from the beginning of the first intermittent light period to the end of the last intermittent light period each 24 hour period, or the length of the single six and 14 hour light periods in the control groups. Group 1 was non-stimulated and had an ELP of <10 hours. There was an intermediately stimulated group with an ELP of 11 hours, and a stimulated group with an ELP of s l 3 hours. The responses in males and females were parallel. Light intensity of either 460 lumens (average) or 2.6 lumens (average) intensity had little effect on the responses in females or males. Single, long duration exposures to light of <10 hours gave responses similar to intermittent regimens with ELP of s 10 hours, while those with durations of 11 and 12 hours were intermediate and those of >13 hours were equivalent to those with ELP of &13 hours. POULTRY SCIENCE 54: 1918-1926, 1975
INTRODUCTION
T
HE reproductive systems of quail are stimulated by photoperiods of 14 or more
hours duration (14L:10D) per 24 hour day (Wilson et al., 1962). An intermittent light Approved for publication as Journal Article No. 11-75 of the Ohio Agricultural Research and Development Center, Wooster, Ohio.
regimen of one hour of light followed by 3 of darkness, repeated 6 times per day (1L:3D, 6 x ) was also stimulatory. Birds held under 6L:18D were not stimulated at 49 days of age, but when exposed to 14L:10D, the testes and ovarian weights increased within 9 and 14 days respectively. The first egg was recorded 36 days after exposure to 14L:10D. Male quail were successfully reared to adult
Downloaded from http://ps.oxfordjournals.org/ at Univ of Iowa-Law Library on June 3, 2015
Kinney, T. B., Jr., W. R. Harvey, G. E. Dickerson and L. H. Baker, 1969. The value of a performance index for predicting income in random sample egg production tests. Poultry Sci. 48: 1371-1390.
Kleiber, M., 1936. Problems involved in breeding for efficiency of food utilization. Amer. Soc. Animal Production Proc. 1936: 247-258. McNally, N. H., and W. H. Foster, 1969. The relationship between financial margin and biological traits for nationally and internationally known entries in the 4th and 5th Gosford, R. S. tests. British Poultry Sci. 10: 191-201. Nordskog, A. W., 1960. Importance of egg size and other factors in determining net time in random sample tests. Poultry Sci. 39: 327-338. Nordskog, A. W., H. French, and S. L. Balloun, 1969. Direct versus indirect estimation of feed efficiency as a measure of performance. Poultry Sci. 48: 1303-1310.
W. A. BOUGH
Harris, C. E., and W. A. Moats, 1974. Recovery of egg solids from wastewater from egg grading and breaking plants. Poultry Sci. 53: 1933. Orr, M. L., and B. K. Watt, 1957. Amino acid content of foods. Home Economics Research Report No. 4. Supt. of Documents, U.S. Government Printing Office, Washington, D.C. Peniston, Q. P., and E. L. Johnson, 1970. Method for treating an aqueous medium with chitosan and
derivatives of chitin to remove an impurity. U.S. patent No. 3,533,940. Spackman, D. H., S. Moore and W. H. Stein, 1958. Automatic recording apparatus for use in chromatography of amino acids. Anal. Chem. 30: 1190-1206. Zall, R. R., and W. Siderewicz, 1974. Wastes from egg breaking plants. 34th Annual Inst, of Food Technologists Meeting, New Orleans, La. May 13. Abstract No. 12.
C. Y. L E E AND A. W. NORDSKOG
Department of Animal Science, Iowa State University, Ames, Iowa 50010 (Received for publication February 17, 1975)
ABSTRACT The importance of measuring feed consumption as a supplementary criterion of net performance was studied from two different sources of data: (1) random sample egg production tests conducted in the United States between 1949 and 1957 (USRST) and (2) a private test of experimental commercial strains conducted by the Goto Hatchery, Inc., Gifu City, Japan, between 1968 to 1971 (GOTO). Seven traits included in the study were net income (NI), egg rate (ER), egg weight (EW), body weight (BW), mortality (MORT), maturity (MAT) and feed consumption (FEED). Performance indexes were derived as multiple regression equations with NI dependent on the other traits. To test the value of feed consumption records, two performance indexes were compared in each data source. The first, 1(6), contained all 6 independent variables, and the second, 1(5), contained all variables except FEED. The correlations between NI and 1(6) were 0.820 and 0.824 and between NI and 1(5) were 0.818 and 0.822 in the USRST and GOTO data, respectively. The difference between the correlations with and without feed was nil in either data source. Thus, measuring FEED did not significantly improve the predictive value of a performance index when prior information was available on the other five variables. 1(5) was also compared with an empirical performance index, 1(E), defined as the ratio of egg mass output to the 3/4 power of body weight. The latter simulates a feed efficiency index where egg mass output is an estimator of feed utilized for egg production and the 3/4 power of body weight is an estimator of feed used in the maintenance of body tissues. The correlations between 1(E) and net income were 0.68 ond 0.74 in the USRST and GOTO data, respectively, which were only slightly lower than the corresponding correlations between net income and 1(5) when the latter was applied to the data source other than that from which it was derived. Thus, a rather simple empirical performance index of egg production efficiency, 1(E), was almost as accurate in predicting net income as a multiple regression-derived performance index. POULTRY SCIENCE 54: 1912-1918, 1975
INTRODUCTION ECAUSE feed is the major item in the cost of producing eggs, the question is whether feed consumption is worth including in an index of net income on a pen of birds.
B
1. Journal Paper No. J-8110of the Iowa Agriculture and Home Economics Experiment Station, Project 1711.
Nordskog (1960) reported that 93% of the variation in net income could be accounted for by differences in performance factors other than feed consumption in U.S. random sample egg production tests. The major portion of the variation was associated with hen-housed production (i.e., egg rate and mortality) and a minor portion with body size and egg size. McNally and Foster (1969)
Downloaded from http://ps.oxfordjournals.org/ at Univ of Iowa-Law Library on June 3, 2015
Value of Feed Consumption Records to Predict Net Income in Layer-Type Chickens*
F E E D CONSUMPTION AND INCOME
In the present study, performance records of laying hens were analyzed separately from each of two different sources of data: (1) random sample laying tests from the United States and (2) commercial experimental tests from the Goto Hatchery, Inc., Gifu City, Japan. By using the two independent sources of data, the accuracy of a performance index, formulated as a multiple regression equation from one set of data, was tested as a predictor of net income in the other set of data, with and without feed consumption as a variable. In addition, the multiple regression-derived performance indexes were compared with an empirical index of egg production efficiency, based on only egg mass output and body weight.
1949 to 1957. These data, except for the records on feed consumption, were used in a previous study (Nordskog, 1960). Included were 8 tests from California, 3 from Missouri, 7 from central New York and 3 from western New York. A test consisted of entries each of approximately 50 commercial egg-type pullets obtained from a random sample of hatching eggs or chicks from U. S. commercial breeders. The data were first adjusted for test location, year differences and number of pen replications per stock 2 before statistical analyses were made. The GOTO data consisted of experimental results on 161 commercial test crosses from the GOTO Hatchery Inc., a relatively large Japanese commercial breeding company. Various experimental strain crosses or "entries" were tested at nine different locations in the 1968-1971 period. In both sets of data, test stocks consisted of both Leghorn type and heavy-breed type layers maintained under either floor-type or cage management. Six independent traits (X) and one dependent trait (Y) were included in each set of data. These were, Symbols Y ~ NI
X, ~ ER
X 2 ~ EW MATERIALS AND METHODS The Data. For the first set of data, the records were taken from Official Random Sample Egg Production Tests conducted in the United States (USRST) and consisted of 360 entries from 21 separate random sample tests carried out over an 8-year period from
USRST—Income over feed and chick costs (dollars) GOTO—Income over feed costs (Yen) USRST—Egg production (number, hen-day) GOTO—Same (%, hen-day) USRST—Large and extralarge eggs (24 oz./doz. or over) GOTO—Average egg weight (g.)
2. See Annual Report of Random Sample Egg Productive tests in United States and Canada, A.R.S., U.S.D.A.
Downloaded from http://ps.oxfordjournals.org/ at Univ of Iowa-Law Library on June 3, 2015
reported that 99% of the variation in net income could be accounted for by including feed consumption in a performance index based on data from the Gosford Random Sample Test in Great Britain. Kinney et al. (1969) developed a multiple regression performance index for predicting income by using7 independent variables: maturity, henhoused production, feed efficiency, egg weight, percentage laying mortality, mature body weight and percentage large and extralarge eggs. They concluded that percentage large and extra-large eggs and hen-housed egg production were the 2 major incomepredicting factors and that feed efficiency and egg weight were less important. Because mean egg weight and percentage large and extra-large eggs are, however, substantially equivalent traits, their results contain a redundancy.
1913
1914 X 3 ~ BW
X 4 ~ MORT
X 5 ~ MAT
X 6 ~ FEED
C. Y. LEE AND A. W. NORDSKOG
Because some of the measurements differed in the 2 sources of data, it was necessary to adjust for these before making comparisons. Net income in yen was converted to dollars in the GOTO test on the basis of the 1969 foreign exchange rate of 360 yen per dollar. Egg rate in the GOTO data was converted to number of eggs per bird on a hen-day basis. Egg size in the USRST data, measured as the percentage of large and extra-large eggs, was converted to grams of egg weight. This required the assumption that egg weight was normally distributed. Eggs classified as "large and extra-large" weighed 24 ounces per dozen or over or, equivalently, 57 g. or more per egg. The conversion of % large and extra-large to mean egg weight was based on the standard normal curve with mean fx and standard deviation, a as shown in Figure 1. Thus, if
FIG. 1. Standard normal frequency distribution with mean |j. and lower bound at D for a truncated distribution with area p.
a fraction, p, of eggs was classified as large and extra-large with lower bound D, and mean X of the truncated distribution then D = \x. - X/CT, is a function of the area p and X = jo. + Da is the estimated egg weight. For example, a typical value of % large and extra-large eggs in the USRST data was 70. The standard normal deviate, D, is 0.525 as given in any cumulative normal frequency distribution table for the area between 0.7 and 0.5. Then for \x. = 57 g. and a = 8.5 gX = 57 + (0.525) (8.51) = 61g. Thus, the estimated mean egg weight of any entry reporting 70% large and extra-large eggs is 61 g. Body weight, measured in pounds in the USRST data, was converted to kg. Mortality was expressed as the percentage of laying house mortality in both data sources. Feed consumption, measured in pounds per bird in the USRST data, was converted to kg. Multiple Regression Analysis. Multiple regression equations were constructed for the dependent variable Y, and the 6 independent variables XI, X2, X3, X4, X5 and X6. When used as a predictor of net income, the multiple regression equations are called performance indexes. From the multiple regression equations, the reduction in mean squares due to regression and multiple correlation coefficients were computed. Finally, correlations between the computed performance indexes and observed FEED were calculated. Efficiency Index. The main source of income from a laying hen operation is the egg mass output. A minor source of income is the salvage value of the bird at the end of the laying period. Egg mass per bird (EM) is the product of egg rate (ER) and mean
Downloaded from http://ps.oxfordjournals.org/ at Univ of Iowa-Law Library on June 3, 2015
USRST—Av. weight of birds alive at end of the test (lb.) GOTO—Same (kg.) USRST—Died after 150 days or after housing (%) GOTO—Same USRST—Age at 50% production (days) GOTO—Same USRST—Feed per bird for entire test period (lb.) GOTO—Feed per bird per day (g.)
1915
FEED CONSUMPTION AND INCOME
egg weight (EW). For this study, efficiency of feed for egg production is defined as the product of egg mass (g.) x survival rate (%) divided by "metabolic" body size, or in symbols, ER x EW x SR
KE) =
EM
3
BW 3 ' 4
BW /"
M = cW3'4 where c is a constant (Kleiber, 1936).
A summary of the means, standard deviations and ranges of each trait from the two different data sources is presented in Table 1. The net income, egg rate, and mortality means were greater in the GOTO data than in the USRST data, but egg weight, body weight, maturity, and feed consumption were greater in the USRST data. The mean differences between the two data sources were statistically significant for all traits (P < 0.01) except mortality. Table 2 presents the simple correlations of each trait with net income and with feed consumption. All traits except mortality were significantly correlated with net income and with feed consumption. Maturity was nega-
TABLE 1.—Summary of United States Random Sample Test data (USRST), and Goto Experimental Test Group data (GOTO) Unit
Trait NI
$
ER
No. of eggs
EW
g.
BW
kg.
MORT
%
MAT
Days at 50% egg prod.
FEED
g./bird/day
Data source USRST GOTO USRST GOTO USRST GOTO USRST GOTO USRST GOTO USRST GOTO USRST GOTO
Mean 2.63 2.84 211.20 225.50 61.73 61.01 2.40 2.19 23.24 24.08 188.72 164.23 120.00 118.05
Standard deviation 0.95 0.51 25.46 13.62 2.17 1.74 0.38 0.22 11.95 9.26 16.24 5.98 8.14 4.52
Minimum 0.31 1.06 101 167 52.83 54.40 1.63 1.69 2.00 4.80 147 145 90.72 105.00
Maximum 5.% 4.05 287 257 67.54 65.00 3.40 2.92 69.00 49.70 276 183 138.67 129.00
TABLE 2.—Correlations of performance traits with net income and feed consumption using data from U.S. Random Sample Egg Tests and the Goto Hatchery, Inc., Gifu City, Japan Trait Net income Egg rate Egg wt. Body wt.
L.H. mortality Maturity Feed consumption *P < .05. **P< .01.
U.S.
—
77** 14** .24** .61** .48** 22**
Net income
GOTO
Feed consumption U.S. GOTO .22**
55** .48** _ 22** — .45** _ 33** .19*
.14** .66** .02 -.18**
.19* .23** .41** 40** -.11 .05
—
—
]Q**
Downloaded from http://ps.oxfordjournals.org/ at Univ of Iowa-Law Library on June 3, 2015
where SR is the survival fraction (1-MORT). Energy for body maintenance can be estimated from the basal metabolic rate (M) which, in turn, can be estimated as a power function of body weight (W),
RESULTS
1916
C. Y. LEE AND A. W. NORDSKOG
TABLE 3.—Correlations between different performance indexes with net income and with feed consumption
Performance index Mult, regression1
Parameters from data source U.S.
Feed consumption (FEED) GOTO U.S. 0.224* • 0.287**
Net income (NI) GOTO U.S. _ '0.720 0.818-
tively correlated with feed consumption in the U.S. test but not in the GOTO test. From each set of data, multiple regression equations were calculated with net income as the dependent variable. In both sets, the multiple correlations, R, came out to be 0.82. When the multiple regression equations were calculated with feed consumption deleted, the R's were only slightly lower. By squaring R we can estimate how much of the variance in net income is associated with the independent variables. In the U.S. data, R 2 = 0.672 with feed in the index, and R 2 = 0.669 without feed in the index. In the GOTO data, R 2 = 0.680 with feed, and R 2 = 0.676 without feed. Thus, when information is available on the 5 traits (egg rate, egg size, body size, flock mortality and age at maturity) essentially nothing more is gained in predictive value by including feed consumption records in the index. An alternative to predicting income from a multiple regression performance index is the efficiency index 1(E), already defined. In Table 3, the efficiency index is contrasted with the multiple regression performance indexes derived from each data source. The basis for comparison is the degree of correlation between the index and net income. When the multiple regression index is constructed from the parameters estimated in one set of data (as US) and then fitted to the data in the other set (GOTO), and R drops from 0.82
to about 0.72. On the other hand, the 1(E) was only slightly less highly correlated with NI (r = 0.676 and r = 0.739 in the US and GOTO data, respectively). These results demonstrate that 1(E), as defined, is almost as good a predictor of performance as a multiple regression index with 5 or 6 variables. The total variance in dependent variable, Y ~ NI, can be partitioned into a fraction due to the direct effects of each independent variable, x, and to the indirect, or joint effects of each x{, xi pair of variables. Thus, in terms of standard partial regression coefficients, Pj = b i cr i /o- y , the total variance in y is partitionable as follows,
2
P?
+2
ESPipir« i=l
Direct effects
+Ee
Indirect effects
(
=1
j-1
Residual )
R
2
The quantity, 1 - R 2 is defined as the residual. The direct effects of each variable, the sum of the direct effects, the sum of the indirect effects and the residual effects on variance in net income are given in Table 4. The indirect response can be either negative
Downloaded from http://ps.oxfordjournals.org/ at Univ of Iowa-Law Library on June 3, 2015
Mult, regression1 Goto 0.732 "0.822 0.048 ns -0.305** Efficiency index, 1(E) Empirical 0.676 0.019 ns -0.077 ns 0.739 •Feed consumption is omitted in the index. Correlations between the performance index and the variable made within the same data source are connected by solid lines and by dotted lines not from the same data source.
1917
FEED CONSUMPTION AND INCOME
TABLE 4.—The partition of the variance influencing net income into direct and indirect effects associated with the independent variables in multiple regression (percentages) USRST Trait
5 variables 35.46 0.00 0.59 7.79 1.75 ~
or positive because r 12 can be either plus or minus. Likewise, it follows that the direct response could be greater than 1. The results however, show that in both sets of data the direct and indirect effects were positive, although the latter were considerably smaller in the GOTO data. Nevertheless, the sums of the direct and indirect effects, R 2 , were very much the same in both sets of data with or without FEED as a variable. Egg rate accounted for the largest fraction of variance for both sources: 31.4% in the USRST data and 30.6% in the GOTO data. Although the variance, in egg weight did not have an important effect on the variance in the net income in the USRST data, it accounted for 15.0% of the variance in the GOTO data. Mortality was the second most important influence on net income in both sources: the percentages of variance in net income attributable to mortality were 7.61 in the USRST data, and the 18.74 in the GOTO data. Compared with egg rate and mortality, body weight and maturity had relatively small effects on variance in net income in either data source. Feed consumption had only a slight effect on the variance in net income: 0.77 percent in the USRST data and 1.00 percent in the GOTO data.
45.59 21.30 66.89 33.11 100.00
GOTO 6 variables 5 variables 30.64 25.44 15.02 12.31 0.81 2.25 18.74 18.71 0.06 0.00 1.00 ~ 66.87 ~~ 58.71 1.18 8.91 68.05 67.62~ " 31.95 32.38 100.00 l6o.OO~ " DISCUSSION
Because feed is the largest cost item of egg production, it would seem that information on feed consumption would be needed to accurately predict net income. Feed consumption can be measured directly by pen groups of birds or by experimental crosses as was done for this study. This requires extra labor and special facilities. An alternative is to estimate feed consumption from the ratio of egg mass output to body weight. Earlier Nordskog et al. (1969) tentatively concluded that the slightly greater accuracy in estimating net performance by measuring feed consumption on pen groups of birds does not justify the additional work required. The results of the present study support this conclusion. The efficiency index, defined as the ratio of egg mass output to the 3/4 power of body weight, is an indirect expression of feed conversion. It evidently predicts net income almost as well as a multiple regression index used to predict income in an independent set of data. Kinney et al. (1969) also showed that the predictive value of a performance index was much lower when applied to an independent set of data, but this is a completely expected consequence of the property of fitting data by least squares.
Downloaded from http://ps.oxfordjournals.org/ at Univ of Iowa-Law Library on June 3, 2015
ER EW BW MORT MAT FEED fdirecreffects (a) X indirect effects (b) a~+ b~ = ib0R~ Residual = 100(1-R 2 ) Toral
6 variables 31.49 0.00 0.00 7.63 1.41 0.77 41.30~ 25.86 67 J 6 32.84 100.00
1918
C. Y. L E E AND A. W. NORDSKOG
ACKNOWLEDGMENT We are indebted to Goto Hatchery, Inc. of Gifu City, Japan for their kind financial support in carrying out this study and for making the GOTO performance test data available to us.
REFERENCES
Reproductive Response to Intermittent Light Regimens in Coturnix coturnix japonica WAYNE L . BACON AND KARL E . NESTOR
Department of Poultry Science, Ohio Agricultural Research and Development Center, Wooster, Ohio 44691 (Received for publication February 17, 1975)
ABSTRACT The effect of intermittent light regimens of one hour of light interspaced with dark periods of one, two or three hours, repeated three, four, five or six times per day was determined for age to first egg, hen day percent production and egg production for a 126 day experimental period in females, and the age to initial response of the cloacal gland in males. Three broad classes of response were noted based on the equivalent light period (ELP), which is defined as the interval from the beginning of the first intermittent light period to the end of the last intermittent light period each 24 hour period, or the length of the single six and 14 hour light periods in the control groups. Group 1 was non-stimulated and had an ELP of <10 hours. There was an intermediately stimulated group with an ELP of 11 hours, and a stimulated group with an ELP of s l 3 hours. The responses in males and females were parallel. Light intensity of either 460 lumens (average) or 2.6 lumens (average) intensity had little effect on the responses in females or males. Single, long duration exposures to light of <10 hours gave responses similar to intermittent regimens with ELP of s 10 hours, while those with durations of 11 and 12 hours were intermediate and those of >13 hours were equivalent to those with ELP of &13 hours. POULTRY SCIENCE 54: 1918-1926, 1975
INTRODUCTION
T
HE reproductive systems of quail are stimulated by photoperiods of 14 or more
hours duration (14L:10D) per 24 hour day (Wilson et al., 1962). An intermittent light Approved for publication as Journal Article No. 11-75 of the Ohio Agricultural Research and Development Center, Wooster, Ohio.
regimen of one hour of light followed by 3 of darkness, repeated 6 times per day (1L:3D, 6 x ) was also stimulatory. Birds held under 6L:18D were not stimulated at 49 days of age, but when exposed to 14L:10D, the testes and ovarian weights increased within 9 and 14 days respectively. The first egg was recorded 36 days after exposure to 14L:10D. Male quail were successfully reared to adult
Downloaded from http://ps.oxfordjournals.org/ at Univ of Iowa-Law Library on June 3, 2015
Kinney, T. B., Jr., W. R. Harvey, G. E. Dickerson and L. H. Baker, 1969. The value of a performance index for predicting income in random sample egg production tests. Poultry Sci. 48: 1371-1390.
Kleiber, M., 1936. Problems involved in breeding for efficiency of food utilization. Amer. Soc. Animal Production Proc. 1936: 247-258. McNally, N. H., and W. H. Foster, 1969. The relationship between financial margin and biological traits for nationally and internationally known entries in the 4th and 5th Gosford, R. S. tests. British Poultry Sci. 10: 191-201. Nordskog, A. W., 1960. Importance of egg size and other factors in determining net time in random sample tests. Poultry Sci. 39: 327-338. Nordskog, A. W., H. French, and S. L. Balloun, 1969. Direct versus indirect estimation of feed efficiency as a measure of performance. Poultry Sci. 48: 1303-1310.