University Microfilms, A XEROX. Company, Ann Arbor, Michigan

Transcription

1 SHIN, Han Foong, GENE ACTION IN THE INHERITANCE OF AGRONCMIC 'IRAITS IN INI'ERVARIETAL DIAILEL CROSSES.AND RELATIVE IMPORTANCE OF GENE EFFECTS FOR QUANTITATIVE ClIARACTERS IN ZEAMAYS L. University of Hawaii, Ph.D., 1972 Agronomy University Microfilms, A XEROX. Company, Ann Arbor, Michigan.--._ _--._ _.._ *----' _..- THIS DISSERTATION HAS BEEN MICROFILMED EXACTLY AS RECEIVED.

2 GENE ACTION IN THE INHERITANCE OF AGRONOMIC TRAITS IN INTERVARIETAL DIALLEL CROSSES AND RELATIVE IMPORTANCE OF GENE EFFECTS FOR QUANTITATIVE CHARACTERS IN ZEA MAYS L. A DISSERTATION SUBMITTED TO THE GRADUATE DIVISION OF THE UNIVERSITY OF HAWAII IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN AGRONOMY AND SOIL SCIENCE AUGUST 1972 By Han Poong Shin Dissertation Committee: Peter P. Rotar, Chairman James L. Brewbaker John R. Thompson James C. Gilbert Donald L. Plucknett Hans Meyer

3 PLEASE NOTE: Some pages may have indistinct print. Filmed as received. University Microfilms, A Xerox Education Company

4 ACKNOWLEDGEMENTS The author would like to acknowledge the Graduate Research Assistantship awarded him by the Department of Agronomy and Soil Science for the Ph. D. degree at the University of Hawaii.

5 ABSTRACT An investigation was undertaken to determine gene action in the inheritance of agronomic traits in a diallel set of eight inbred lines of sweet corn and all possible F l hybrids, including parents. Data were analyzed by Jinks and Hayman's diallel analysis for plant and ear height, weight with and without husk, shank and ear length, ear and cob diameter, kernel depth, and mid-silking days. Additive effects and environmental variations were significantly different from zero for April, June, and Combined dates of planting. Estimates of the component of variation due to dominance effects were significantly different from zero except for shank length and cob diameter in June and ear length in April planting. The parents carried an excess of dominant genes for plant height, ear length, weight with and without husk, mid-silking days, and an excess of recessive genes for shank length. Dominant and recessive alleles of each gene were distributed nearly equal among the parents for ear diameter, cob diameter, and kernel depth. Ear length, shank length, cob diameter, and mid-silking days for the June planting and ear height in Combined dates of planting were within the partial dominance range. Heritability estimates supported the conclusion that selection for weight with and without husk, mid-silking days, and ear height would be most effective and that selection for ear diameter, plant height, ear length, shank length, and kernel depth would be least effective. Eight inbred lines of sweet corn, including parents, Fl's, F 2 's, and first backcrosses were tested at two locations in one year. The population means obtained were used to estimate additive, dominance,

6 v additive x additive, additive x dominance, and dominance x dominance gene effects for nine quantitative agronomic traits. Additive gene effects appeared to be the most constant over locations for plant and ear height. Dominance gene effects for weight with and without husk, and ear length were more important than those of additive effects. The remaining types of gene effects indicated very little stability over locations for most of the agronomic traits studied. The relative magnitude of expected genetic gain expressed as a percentage F 2 mean would suggest that rapid progress should be accomplished by selecting and recombining in early generations for plant height in the crosses AA 11 x AA 18 and AA 8 x 190a. Slower progress should be expected from crosses AA 18 x 190a, AA 11 x 245, AA 2 x AA 11, and AA 20 x P 39. For ear height, rapid progress should be expected from crosses AA 18 x AA 20, AA 11 x AA 18, AA 8 x 190a, and AA 2 x AA 11. For weight with husk, good progress should be expected from crosses AA 11 x AA 18, AA 11 x 245, AA 8 x 190a, AA 2 x AA 11, and AA 20 x P 39. For ear length, rapid progress should be expected from crosses AA 18 x 190a, AA 11 x AA 18, and AA 20 x P 39. For the nine crosses considered, it was concluded that most of the variation was due to additive and dominance gene effects with epistasis being some importance.

7 TABLE OF CONTENTS Page ACKNOWLEDGEMENT ABSTRACT LIST OF TABLES LIST OF FIGURES INTRODUCTION REVIEW OF LITERATURE Inheritance of Quantitative Characters Dia11e1 Analysis Heritability. Inheritance of Agronomic Traits in Maize Genetic Variance iii iv viii xvii MATERIALS AND METHODS 32 Experimental Materials Parents Used to the Dia11el Analysis Parents Used in the Generation Mean Analysis Experimental Methods Field Design and Data Collection Source of Computer Program Used Presentation of Analysis Diallel Analysis The Statistics of the Diallel Table Components of Variation Tests for Validity of the Assumptions in the Diallel Generation Mean Analysis RESULTS AND DISCUSSION The Dia1lel Analysis Validity of Assumptions Least Square Estimators Genetic Ratios Heritability Graphical Analysis Generation Mean Analysis and Generation Mean Analysis Genetic Parameters Additive gene effects Genetic Parameters

8 vii TABLE OF CONTENTS (Contd.) Page Dominance gene effects Epistatic gene effects Partition. of Variance Comparison of additive and dominance variance Heritability Genetic Advance. Degree of Dominance SUMMARY AND CONCLUSIONS APPENDIX A APPENDIX B APPENDIX C LITERATURE CITED

9 LIST OF 'IABLES TABLE I II III V VI VII VIII IX LITERATURE REPORTS RELATING TO GENETIC VARIANCE ESTD1ATES FOR EAR HEIGHT OF MAIZE LITERATURE REPORTS CONCERNING NONALLELIC GENE INTERACTIONS FOR AGRONOMIC TRAITS IN MAIZE ANALYSES OF VARIANCE OF (Wr - Vr) VALUES FOR PLANT HEIGHT(X 1 ), EAR HEIGHT(~), EAR LENGTH (X 3 ), SHANK LENGTH(X 4 ), EAR DrAMETER (X 5 ), COB DIAMETER(~), KERNEL DEPTH(X 7 ), WEIGHT WITH HUSK(XS)' WEIGHT HUSKED(Xg), AND MID-SILKING DAYS(XlO) FOR APRIL, JUNE, AND COMBINED DATES MEAN ESTD1ATES OF GENETIC VARIANCE COMPONENTS AND ENVIRONMENTAL VARIANCES TOGETHER WITH THEIR STANDARD ERRORS FOR PLANT HEIGHT(X 1 ), EAR HEIGHT(X2)' EAR LENGTH(X3)' SHANK LENGTH(X 4 ), EAR DIAMETER(X 5 ), COB DIAMETER(X6)' KERNEL DEPTH(X 7 ), WEIGHT WITH HUSK(X S )' WEIGHT HUSKED (Xg), AND MID-SILKING DAYS(X10) FOR APRIL, JUNE AND COMBINED DATES RATIO OF COMPONENTS AND HERITABILITY FOR PLANT HEIGHT(Xl)' EAR HEIGHT(X2)' EAR LENGTH(X3), SHANK LENGTH(X4), EAR DIAMETER(X 5 ), COB DIAMETER(~), KERNEL DEPTH(X 7 ), WEIGHT WITH HUSK(XS), WEIGHT HUSKED(Xg), AND MID-SILKING DAYS(X lo ) FOR ~.PRIL, JUNE AND COMBINED DATES MEAN SQUARES FOR PLANT HEIGHT(Xl), EAR HEIGHT (X2), EAR LENGTH(X3), SHANK LENGTH(Xt.), EAR DIAMETER(X5), COB DIAMETER(~), KERNEL DEPTH (X7), WEIGHT WITH HUSK(X8), WEIGHT HUSKED(Xg), AND MID-SILKING DAYS(X lo ) FOR APRIL, JUNE AND COMBINED DATES VALUES FOR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES CALCULATED FROM APRIL, JUNE AND COMBINED DATES VALUES FOR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES CALCULATED FROM JUNE PLANTING PLANT AND EAR HEIGHT FOR DIFFERENT GENERATIONS, %2 GOODNESS OF FIT TO THE THREE GENETIC PARllMETERS FOR EACH OF NINE CROSSES OVER COMBINED LOCATIONS Page S 59 so

10 ix LIST OF TABLES (Contd.) TABLE X XI XII XIII XIV XVI XVII WEIGHT WITH.AND WITHOUT HUSK FOR DIFFERENT GENERATIONS. t,2 GOODNESS OF FIT TO THE THREE GENETIC PARAMETERS FOR EACH OF NINE CROSSES OVER COMBINED LOCATIONS SHANK.AND EAR LENGTH FOR DIFFERENT GENERATIONS. ~2 GOODNESS OF FIT TO THE THREE GENETIC PARAMETERS FOR EACH OF NINE CROSSES OVER COMBINED LOCATIONS EAR.AND COB DIAMETER FOR DIFFERENT GENERATIONS. 1,2 GOODNESS OF FIT TO THE THREE GENET:l:C PARAMETERS FOR EACH OF NINE CROSSES OVER COMBINED LOCATIONS KERNEL DEPTH FOR DIFFEREl-l"'T GENERATIONS. 1, 2 GOODNESS OF FIT TO THE THREE GENETIC PARAMETERS FOR EACH OF NINE CROSSES OVER COMBINED LOCATIONS ESTIMATES OF ADDITIVE (D) DOMINANCE(H).AND ENVIRONMENTAL VARIANCES(E) TOGETHER WITH HERITABILITIES(h 2 ), AVERAGE DEGREE OF DOMmANCE (H/D)~, GENETIC ADVANCE(Gs), AND GAIN IN RELATION TO F 2 FOR PLANT.AND EAR HEIGHT FOR EACH OF NINE CROSSES OVER COMBINED LOCATIONS ESTIMATES OF ADDITIVE (D), DOMINANCE (H),.AND ENVIRONMENTAL VARIANCES(E) TOGETHER WITH HERITABILITIES(h 2 ), AVERAGE DEGREE OF DOMmANCE (H/D)~. GENETIC ADVANCE(Gs), AND GAIN IN RELATION TO F2 FOR WEIGHT WITH.AND WITHOUT HUSK FOR EACH OF NINE CROSSES OVER COMBINED LOCATIONS ESTIMATES OF ADDITIVE (D), DOMINANCE(H),.AND ENVIRONMENTAL VARIANCES(E) TOGETHER WITH HERlTABILITIES(h 2 ), AVERAGE DEGREE OF DOMmANCE (H/D)~. GENETIC ADVANCE(Gs). AND GAIN IN RELATION TO F2 FOR SHANK.AND EAR LENGTH FOR EACH OF NINE CROSSES OVER COMBINED LOCATIONS ESTIMATES OF ADDITIVE (D) DOMINANCE(H). AND ENVIRONMENTAL VARIANCES (E) TOGETHER WITH HERITABILITIES(h 2 ) AVERAGE DEGREE OF DOMINANCE (H/D)~, GENETIC ADVANCE (Gs) AND GAIN IN RELATION TO F2 FOR EAR.AND COB DIAMETER FOR EACH OF NINE CROSSES OVER COMBINED LOCATIONS 0 Page

11 x LIST OF TABLES (Contd.) TABLE Page XVIII ESTIMATES OF ADDITIVE (D), DOMINANCE (H), AND ENVIRONMENTAL VARIANCES (E), TOGETHER WITH HERITABILITIES(h 2 ), AVERAGE DEGREE OF DOMINANCE (H/D)~, GENETIC ADVANCE (Gs), AND GAIN IN RELATION TO F FOR KERNEL DEPTH FOR EACH OF NINE CROSSES OVER COMBINED LOCATIONS 107 APPENDIX A I II III IV V VI VII VIII PLANT AND EAR HEIGHT FOR DIFFERENT GENERATIONS, GOODNESS OF FIT TO THE THREE AND SIX GENETIC PARAMETERS FOR EACH OF NINE CROSSES AT WAIMANALO WEIGHT WITH AND WITHOUT HUSK FOR DIFFERENT GENERATIONS, GOODNESS OF FIT TO THE THREE AND SIX GENETIC PARAMETERS FOR EACH OF NINE CROSSES AT WAIMANALO SHANK AND EAR LENGTH FOR DIFFERENT GENERATIONS, GOODNESS OF FIT TO THE THREE AND SIX GENETIC PARAMETERS FOR EACH OF NINE CROSSES AT WAIMANALO EAR Al\"D COB DIAMETER FOR DIFFERENT GENERATIONS, GOODNESS OF FIT TO THE THREE AND SIX GENETIC PARAMETERS FOR EACH OF NTh"E CROSSES AT WAIMANALO KERNEL DEPTH FOR DIFFERENT GENERATIONS, GOODNESS OF FIT TO THE THREE AND SIX GENETIC PARAMETERS FOR EACH OF NINE CROSSES AT WAIMANALO PLANT AND EAR HEIGHT FOR DIFFERENT GENERATIONS, GOODNESS OF FIT TO THE THREE AND SIX GENETIC PARAMETERS FOR EACH OF NINE CROSSES AT POAMOHO WEIGHT WITH AND WITHOUT HUSK FOR DIFFERENT GENERATIONS, GOODNESS OF FIT TO THE THREE AND SIX GENETIC PARAMETERS FOR EACH OF NINE CROSSES AT POAMOHO SHANK AND EAR LENGTH FOR DIFFERENT GENERATIONS, GOODNESS OF FIT TO THE THREE AND SIX GENETIC PARAMETERS FOR EACH OF NINE CROSSES AT POAMOHO

12 xi LIST OF TABLES (Contd.) TABLE IX X XI XII XIII XIV EAR AND COB DIAMETER FOR DIFFERENT GENERATIONS, GOODNESS OF FIT TO THE THREE AIID SIX GENETIC PARAMETERS FOR EACH OF NINE CROSSES AT POAMOHO KERNEL DEPTH FOR DIFFERENT GENERATIONS, GOODNESS OF FIT TO THE THREE AND SIX GENETIC PARAMETERS FOR EACH OF NrnE CROSSES AT POAMOHO ESTIMATES OF ADDITIVE(D), DOMINANCE(H), AND ENVIRONMENTAL VARIANCES(E) TOGETHER WITH HERlTABn.ITIES(h 2 ), AVERAGE DEGREE OF DOMINANCE (H/D)~, GENETIC ADVANCE (Gs), AND GAIN IN RELATION TO F 2 FOR PLANT.AND EAR HEIGHT FOR EACH OF NINE CROSSES AT WAIMANALO ESTIMATES OF ADDITIVE(D), DOMINANCE(H),.AND ENVIRONMENTAL VARIANCES(E) TOGETHER WITH HERlTABn.ITIES(h 2 ), AVERAGE DEGREE OF DOMINANCE (H/D)~, GENETIC ADVANCE(Gs), AND GAIN IN RELATION TO F2 FOR WEIGHT WITH AND WITHOUT HUSK FOR EACH OF NINE CROSSES AT. WAIMANALO ESTIMATES OF ADDITIVE(D), DOMINANCE(H), AND ENVIRONMENTAL VARIANCES(E) TOGETHER WITH HERITABILITIES(h 2 ), AVERAGE DEGREE OF DOMINANCE (H/D)~, GENETIC ADVANCE(Gs), AND GAIN IN RELATION TO F2 FOR SHANK AND EAR. LENGTH FOR EACH OF NINE CROSSES AT. WAIMANALO ESTIMATES OF ADDITIVE(D), DOMINANCE(H), AND ENVIRONMENTAL VARIANCES(E) TOGETHER WITH HERlTABn.ITIES(h 2 ), AVERAGE DEGREE OF DOMINANCE (H/D)~, GENETIC ADVANCE (Gs), AND GAIN IN RELATION TO if2 FOR EAR. AND COB DIAMETER FOR EACH OF NINE CROSSES AT WAIMANALO ESTIMATES OF ADDITIVE(D), DOMINANCE (H), AND ENVIRONMENTAL VARIANCES(E) TOGETHER WITH HERITABILITIES(h 2 ), AVERAGE DEGREE OF DOMINANCE (H/D)~, GENETIC ADVANCE (Gs), AND GAIN IN RELATION TO if2 FOR KERNEL DEPTH FOR EACH OF NINE CROSSES AT WAIMANALO Page

13 xii LIST OF TABLES (Contd.) TABLE Page APPENDIX A XVI XVII XVIII XIX xx ESTIMATES OF ADDITIVE(D), DOMINANCE (H), AND ENVIRONMENTAL VARIANCES(E) TOGETHER WITH HERITABILITIES(h 2 ), AVERAGE DEGREE OF DOMINANCE (H/D)~, GENETIC ADVANCE (Gs), AND GAIN IN RELATION TO F2 FOR PLANT AND EAR HEIGHT FOR EACH OF NINE CROSSES AT POAMOHO ESTIMATES OF ADDITIVE(D), DOMINANCE (H), AND ENVIRONMENTAL VARIANCES(E) TOGETHER WITH HERITABILITIES(h 2 ), AVERAGE DEGREE OF DOMINANCE (H/D)~, GENETIC ADVANCE (Gs), AND GAIN IN RELATION TO F2 FOR WEIGHT WITH AND WITHOUT HUSK FOR EACH OF NINE CROSSES AT POAMOHO BSTIMATES OF ADDITIVE(D), DOMINANCE(H), AND ENVIRONMENTAL VARIANCES(E) TOGETHER WITH HERITABILITIES(h 2 ), AVERAGE DEGREE OF DOMINANCE (H/D)~, GENETIC ADVANCE (Gs), AND GAIN IN RELATION TO F2 FOR SHANK AND EAR LENGTH FOR EACH OF NINE CROSSES AT POAMOHO ESTIMATES OF ADDITIVE(D), DOMINANCE(H), AND ENVIRONMENTAL VARIANCES(E) TOGETHER WITH HERITABILITIES(h 2 ), AVERAGE DEGREE OF DOMINANCE (H/D)~, GENETIC ADVANCE (Gs), AND GAIN IN RELATION TO F2 FOR EAR. AND COB DIAMETER FOR EACH OF NINE CROSSES AT POAMOHO ESTIMATES OF ADDITIVE (D), DOMINANCE(H), AND ENVIRONMENTAL VARIANCES(E) TOGETHER WITH HERITABILITIES(h 2 ), AVERAGE DEGREE OF DOMINANCE (H/D)~, GENETIC ADVANCE (Gs), AND GAIN IN RELATION TO 'F2 FOR KERNEL DEPTH FOR EACH OF NINE CROSSES AT POAMOHO APPENDIX B I II FREQUENCY DISTRIBUTIONS FOR (3 INTERNODE BELOW/ 3 INTERNODE ABOVE) OF THE CROSSES CM 105 x 442a AND CM 104 x AA 8 FREQUENCY DISTRIBUTIONS FOR (3 INTERNODE BELOW/ 3 INTERNODE ABOVE) OF THE CROSSES CM 104 x 442a AND CM 105 x AA

14 xiii LIST OF TABLES (Contd.) TABLE Page APPENDIX B III V VI VII VIII IX FREQUENCY DISTRIBUTIONS FOR (3 INTERNODE BELOW/ 3 INTERNODE ABOVE) OF THE CROSSES CM 104 x Oh 43.AND CM 105 x Oh 43 FREQUENCY DISTRIBUTIONS FOR (3 INTERNODE BELOW/ 3 INTERNODE ABOVE) OF THE CROSSES CM 104 x CM 105 AND Oh 43 x 442a FREQUENCY DISTRIBUTIONS FOR PLANT HEIGHT OF THE CROSSES CM 105 x 442a AND CM 104 x AA 8 FREQUENCY DISTRIBUTIONS FOR PLANT HEIGHT OF THE CROSSES CM 104 x 442a AND CM 105 x AA 8 FREQUENCY DISTRIBUTIONS FOR PLANT HEIGHT OF THE CROSSES CM 104 x Oh 43 AND CM 105 x Oh 43 FREQUENCY DISTRIBUTIONS FOR PLANT HEIGHT OF THE CROSSES CM 104 x CM 195 AND Oh 43 x 442 a FREQUENCY DISTRIBUTIONS FOR EAR HEIGHT OF THE CROSSES CM 105 x 442a AND CM 104 x AA x FREQUENCY DISTRIBUTIONS FOR EAR HEIGHT OF CROSSES CM 104 x 442a.AND CM 105 x AA 8 THE 144 XI XII XIII XIV FREQUENCY DISTRIBUTIONS FOR EAR HEIGHT OF THE CROSSES CM 104 x Oh 43 AND CM 105 x Oh 43 FREQUENCY DISTRIBUTIONS FOR EAR HEIGHT OF THE CROSSES CM 104 x CM 105 AND Oh 43 x 442a FREQUENCY DISTRIBUTIONS FOR HEIGHT ABOVE EAR TO BASE OF TASSEL OF THE CROSSES CM 105 x 442a AND CM 104 x AA 8 FREQUENCY DISTRIBUTIONS FOR HEIGHT ABOVE EAR TO BASE OF TASSEL OF THE CROSSES CM 104 x 442a AND CM 105 x AA 8 FREQUENCY DISTRIBUTIONS FOR HEIGHT ABOVE EAR TO BASE OF TASSEL OF THE CROSSES CM 104 x Oh 43 AND CM 105 x Oh

15 xiv LIST OF TABLES (Contd.) TABLE Page APPENDIX B XVI XVII XVIII XIX xx FREQUENCY DISTRIBUTIONS FOR HEIGHT ABOVE EAR TO BASE OF TASSEL OF THE CROSSES CM 104 x CM 105 AND Oh 43 x 442a FREQUENCY DISTRIBUTIONS FOR TASSEL LENGTH OF THE CROSSES CM 104 x 442a AND CM 104 x AA 8 FREQUENCY DISTRIBUTIONS FOR TASSEL LENGTH OF THE CROSSES CM 104 x 442a AND CM 105 x AA 8 FREQUENCY DISTRIBUTIONS FOR TASSEL LENGTH OF THE CROSSES CM 104 x Oh 43 AND CM 105 x Oh 43 FREQUENCY DISTRIBUTIONS FOR TASSEL LENGTH OF THE CROSSES CM 104 x CM 105 AND Oh 43 x CM APPENDIX C I MEAN PLANT HEIGHT (em) FOR EIGHT PARENTAL DIALLEL CROSSES, TOTAL OF TWO REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES OF APRIL PLANTING 156 II. MEAN PLANT HEIGHT (em) FOR EIGHT PARENTAL DIALLEL CROSSES, TOTAL OF TWO REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES OF JUNE PLANTING 156 III IV v MEAN PLANT HEIGHT (em) FOR EIGHT PARENTAL DIALLEL CROSSES, TOTAL OF FOUR REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES OF COMBINED DATES MEAN EAR HEIGHT (em) FOR EIGHT PARENTAL DIALLEL CROSSES, TOTAL OF TWO REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES OF APRIL PLANTING MEAN EAR HEIGHT (em) FOR EIGHT PARENTAL DIALLEL CROSSES, TOTAL OF TWO REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, M,'T]) THEIR DIFFERENCES OF JUNE PLANTING

16 xv LIST OF TABLES (Contd.) TABLE Page APPENDIX C VI VII VIII IX X XI XII XUI XIV MEAN EAR HEIGHT (em) FOR EIGHT PARENTAL DIALLEL CROSSES, TOTAL OF FOUR REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES OF COMBINED DATES MEAN EAR LENGTH (em) FOR EIGHT PARENTAL DIALLEL CROSSES, TOTAL OF TWO REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES OF APRIL PLANTING MEAN EAR HEIGHT (em) FOR EIGHT PARENTAL DIALLEL CROSSES, TOTAL OF TWO REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES OF COMBINED DATES MEAN EAR LENGTH (em) FOR EIGHT PARENTAL DIALLEL CROSSES, TOTAL OF FOUR REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES OF COMBINED DATES MEAN WEIGHT WITH HUSK (kg) FOR EIGHT PARENTAL DIALLEL CROSSES, TOTAL OF TWO REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES OF JUNE PLANTING MEAN WEIGHT WITHOUT HUSK (kg) FOR EIGHT PARENTAL CROSSES, TOTAL OF TWO REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES OF JUNE PLANTING MEAN MID-SILKING DAYS FOR EIGHT PARENTAL CROSSES, TOTAL OF TWO REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES OF JUNE PLANTING MEAN SHANK LENGTH (em) FOR EIGHT PARENTAL DIALLEL CROSSES, TOTAL OF TWO.REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES OF JUNE PLANTING MEAN EAR DIAMETER (em) FOR EIGHT PARENTAL DIALLEL CROSSES, TOTAL OF TWO REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES OF JUNE PLANTING

17 xvi LIST OF TABLES (Contd.) TABLE Page APPENDIX C xv MEAN COB DIAMETER (em) FOR EIGHT PAREN!'AL DIALLEL CROSSES, TOTAL OF TWO REPLICATIONS, WITH THEIR COVARIANCES, VARIANCES, AND THEIR DIFFERENCES OF JUNE PLANTING 160

18 LIST OF FIGURES Figure 1 Plant height (April), arrays 1 to 8; Vr, Wr graph Page 2 Plant height (June), arrays 1 to 8; Vr, Wr graph 62 3 Plant height (Comb ined), arrays 1 to 8, Vr, Wr graph 63 4 Ear height (April), arrays 1 to 8; Vr, Wr graph 65 5 Ear height (June), arrays 1 to 8; Vr, Wr graph 66 6 Ear height (Combined), arrays 1 to 8, Vr, Wr graph 67 7 Ear length (April), arrays 1 to 8; Vr, Wr graph 69 8 Ear length (June), arrays 1 to 8-, Vr. Wr graph 70 9 Ear length (Combined), arrays 1 to 8; Vr, Wr graph Shank length (June), arrays 1 to 8; Vr, Wr graph Ear diameter (June), arrays 1 to 8; Vr, Wr graph Cob diameter (June), arrays 1 to 8; Vr, Wr graph Weight with husk (June), arrays 1 to 8, Vr, Wr graph Weight husked (June), arrays 1 to 8; Vr, Wr graph Mid-si1king days (June), arrays 1 to 8-, Vr, Wr graph 78 61

19 Introduction Individual effects of genes controlling quantitative traits can not ordinarily be distinguished from one another. Consequently, it is not possible to determine the mode of inheritance for single genes. By studying their combined effects in segregating generations, however, one can gain some insight into their behavior and can make inferences about their average action involved in the expression of a particular agronomic traits. Hull (1945) has considered some aspects of diallei cross method of investigating the genetical properties of a groups of homozygous lines of maize. A short summary of a more general approach and its application to several set of maize data has also appeared, Jinks and Hayman (1953). Griffing's (1956) diallel analysis indicated the potential success of selection for the traits being considered but did not measure the type of prevalent gene action but only the average as expressed as variances. Hayman's (1960) generation mean analysis measured the average gene action expressed as effects. A knowledge of the relative contributions of the various types of gene actions present for quantitative traits in the populations undergoing selection is basic to all plant breeding programs. Estimates of heritability, prediction of response to selection, and the design of the most effective breeding schemes are dependent upon the relative proportions of additive, dominance and environmental variances. If the estimates of genetic variance indicate that the additive genetic variance is of major importance, heritability would be expected to be rel~tively high, and single plant selection (mass selection) or selection among progenies should be effective in improving the traits under

20 2 consideration. If a major portion of the genetic variation is nonadditive, an inbreeding and hybridization program may be the most effective alternate breeding scheme. In either case, the most effective breeding scheme is determined by the type of gene action present in the population under consideration; the lack of this basic information results in a breeding program that is largely empirical. The plant breeder is interested in the estimation of additive, dominance, and digenic epistatic gene effects in order to formulate the most advantageous breeding procedures for improvement of the traits. This information is to be of value to the plant breeder 1) as an indication of the relative importance of gene effects in the basic genetic mechanism, 2) for the best plant breeding procedures to follow in a breeding program, and 3) for a measure of the bias present in estimates of gene effects obtained from models assuming no epistatic gene effects. The investigation reported herein was to extend the studies of general and specific combining abilities in sweet corn (Shin 1970) to make it applicable for genetic variance components and variance and covariance analyses, special attention was given to the genetic statistics from which least square estimators of genetic parameters are examined in theory as described by Jinks and Hayman. The primary purpose in connection with these was to provide the relative importance of additive, dominance and digenic epistatic gene effects involved with respect to plant and ear height, weight with and without husk, shank and ear length, ear and cob diameter, and kernel depth in the nine selected crosses from eight inbred lines of corn.

21 Review of Literature Inheritance of Quantitative Characters The rediscovery of Mendel's laws was followed by a series of investigations extending Mendelian concepts to characteristics showing continuous variation. Johanssen (1909) demonstrated with beans that environmental influence contributed to phenotypic variation and that variation in segregating generations was due to both heredity and environmental causes. Variations in pure lines was considered to be entirely environmental. Nilsson-Ehle (1909) sh~ed that intensity of seed-coat color in wheat was dependent on the number of genes conditioning color. In the absence of information on gene number obtained from controlled pollination, the total range of coloration simulated a continuous distribution of seed-coat color. East (1907, 1910, 1913) working with maize and tobacco, concluded that continuous variation could be explained on the basis of Mendelian concepts of heredity. When the number of factors influencing a character is fairly large, an analysis based on single genes can not be made; but the variation observed in different generations is entirely compatible with expectations based on Mendelian premises. Emerson and East (1913), in the most extensive report on the quantitative inheritance of the period, showed that variations with FIls were comparable to that of the parents and that F 2 generation was more variable than the parental F l or F 3 generations. Individual F 3 families tended to show a within-family reduction in variability, which was entirely consistent with Mendelian concepts.

22 4 Hays (1912), extending a Mendelian mechanism to explain the occu=rence of transgressive segregation leaf number in tobacco, pointed out the difficulty of reconciling the occurrence of offspring more extreme than either parental type with any theory of heredity not based on particulate genetic factors. Following the cited demonstrations of the operation of Mendelian m~chanisms in the inheritance of quantitative characters, the investigation of this aspect of genetics followed several different trends. Shull's (1908, 1911) work was one major trend to investigate the effects of inbreeding and hybridization in maize. His work led to a vigorous interest in formulating an explanation for hybrid vigor in maize (Shull, 1911; East and Hays, 1912; Bruce, 1910; Jones, 1917; Hull, 1945, 1952; Crow, 1948). Much of the experimentation in maize after Shull's work dealt with some aspects of the utilization of hybrid vigor and with evolving a rational breeding procedure for reaching the objective. The other major trend in the investigation of the inheritance of quantitative characters was initiated by Fisher (1918) which dealt with statistical analysis of the nature of phenotypic variation. Fisher (1918) derived statistical methods for partitioning phenotypic variation into its genetic and environmental components placing emphasis on second degree statistics and correlation between relatives. Fisher et al. (1932) illustrated this approach of partitioning phenotypic variance and also indicated the possible importance that third degree statistics might have in analysis of quantitative inheritance. Mather (1949) reviewed the hypothesis of Fisher and his co-workers and discussed the effects of data transformation and linkage

23 5 on the parameters computed, presenting a number of equations for computing the magnitude of dominance and additive gene action and environmental effects for given crosses and generations. Wright (1912, 1922, 1928, 1935) independently derived methods and formulae for studying variance components. These methods have been used mainly in the field of animal genetics and breeding. In his 1922 paper Wright introduced a concept of epistasis into the discussion of quantitative genetics. He partitioned total heritable variances into three components, each component arising from a specific type of genotypic-phenotypic relationship. The phenotypic manifestation of the genotype was classified into (1) additive gene effect, (2) dominance deviations from additivity, and (3) epistasis or inter-allelic deviations from additivity. Wright's (1932, 1949) concern with epistatic gene effects resulted in his view of the genetic structure of natural populations. He visualized populations as composed of arrays of genotypes resembling "a rugged field with many peaks." The peaks are genotypic combinations with high selective advantages which in a random breeding population would approach some genotype with optimum adaptation. If only additive gene effects were present, selection would tend to shift gene frequencies to one peak in the field with a resultant loss of genetic variability in the population. The two concepts of nonlinear gene effects; namely, dominance and epistasis, were taken from more easily visualized situations in the study of qualitative character. The observation of the dominance of one allele over the other; e.g., the equality of the phenotypes AA and Aa has its origin with Mendel's work. Epistasis as a nonlinear effect of

24 6 different nonallelic combinations was defined by Bateson (1909). The statistical consideration of these two types of nonlinear gene effects is inevitable in any discussion of the inheritance of quantitative characters. The nature of these phenomena (dominance and epistasis) and their evolution has attracted a great deal of speculation and controversy, particularly involving Fisher (1928, 1929, 1931) and Wright (1929, 1934). 3riefly, Fisher's view was that dominance of a given allele was affected by the selection of "modifying" genes which, in the course of evolution, rendered the more favorable allele dominant to the unfavorable. Wright countered that mutation rates and selection intensities for "modifiers" would have to be of an enormous intensity to be effective and held to the view that dominance is a consequence of rates of gene action and physiological pathways between immediate gene products and final phenotype. Charles and Smith (1939) approached nonlinearity of gene effects from a statistical point of view. They postulated that additive gene action can occur in geometrical, as well as arithmetical progression and presented methods for distinguishing between the two types of additivity. Mather (1949), Wright (1952) and Powers et al. (1950) have discussed the need for scaling and transformation of data. Powers et al. (1950) developed a method for partitioning the frequency distributions of the segregating populations on the basis of provisional hypotheses concerning the number of loci in which the two parents differ. These wo~~~~s and Powers (1941, 1944) in earlier papers laid great stress on treating a final morphological character by means

25 7 of its logical components. Weight of tomato fruit was considered in terms of the genetic systems operating on the number of locules per fruit and weight per locule. Leng (1956), in studying the effects of heterosis on the yield of grain per plant in corn, subdivided this character into fq~r primary components, each component being a morphologically distinct entity which could not be further subdivided readily. He observed intermediate dominance, or heterotic effects, in the inheritance, depending on the component of yield that was examined. Fisher (1928) pointed out that analysis of variance can be used for indicating significant differences between treatment means. A further use of the technique was indicated that if the observed statistical variate is assumed to be a sum of several separate effects, the total variance can be partitioned into portions attributable to these various identifiable separate effects. By using the analysis of variance concept, Comstock and Robinson (1948, 1952) have assigned additive and dominance components of variance to mean square estimates obtained from either biparental of backcross progenies. In each of the three experiments proposed (Design I, II and III), the unique features of their technique is the availability of an estimate of the nonadditive parameter from data obtained in one segregating generation. For Design I and II the experimental populations consist of progenies from random mating among plants of the F 2 generation of a cross of two inbred lines. Design I consists of sets of m males each mated with n different females. Design II consists of sets of mn crosses, all m males crossed

26 8 to n females within each set and is adapted to species with multiple inflorescens. In Design III experimental material is produced from backcross matings of F 2 plants to the two inbred lines from which the F 2 is derived. The F 2 plants are used as pollen parents. Each of these designs is based on genetic models assuming (a) regular diploid meiosis, (b) no multiple alleles, (c) no interaction of nonallelic genes and (d) no linkage. Horner et al. (1955) examined the effect of different types of epistasis for the three designs. They found that generally Design I and II were more likely to yield biased estimates of variance components than Design III. The problem of nonallelic gene interaction has been considered by Anderson and Kempthorne (1954), Cockerham (1954), Hayman and Mather (1955), Hayman (1958b, 1960b), Gamble (1962a, 1962b) and Eberhart and Gardner (1966a) have partitioned the genotypic value into additive, dominance and epistatic gene effects. Cockerham (1954) and Hayman and Mather (1955) have devised techniques to partition the genetic variance into additive, dominance and epistatic components. Anderson and Kempthorne's model employs the means of populations obtained from crossing two homozygous lines followed by subsequent crossing and selfing to produce backcrosses and advanced filial generations. Six parameters, K 2, E, F, G, I and M, are derived where K 2 represents mean effects, E and F represent nonepistatic effects and G, I and M represent epistatic effects. Some of these parameters are not easily interpretable due to pooled gene effects in the parameters.

27 9 Hayman (1958a) described parameters related to those of Anderson and Kempthorne's (1954) which permit estimation of the additive, dominance, additive x additive, additive x dominance and dominance x dominance gene effects with less difficulty in their interpretation. However, additive and dominance gene effects can not be uniquely measured when significant epistasis is present. According to Hayman (1960a), the relative contributions of the type of gene action to various genetic phenomena such as heterosis can not be ascertained by this partitioning method. On the other hand, estimates of the parameters do provide an indication of the relative importance of the various types of gene effects affecting the total genetic variation of a plant attribute. Lush (1949) discussed heritability in the broad sense and in its narrow aspects. In the broad sense it is a ratio of genotypic to total variance. In the narrow sense it is the proportion of genotypic variance that can be fixed by mass selection or the ratio of additive variance to genotypic variance. The method of computation of this parameter differs with different workers. Mahmud and Kramer (1951) working with yield and height of soybeans computed heritability by two different methods i.e., regressions of F 3 lines on F 2 plants and regression of F 4 lines on F 3 lines and by the use of variance components. Heritability was higher for both characters (yield and height) when the estimates were obtained in the same year on the basis of variance components. Heritability was negligible for yield and about one-half smaller for height when computed on the basis of parental offspring regression. Smith (1952) computed heritability estimates based on the methods of Mather (1949), in tobacco for leaf

28 10 length, plant height and node number. He compared the observed gains with heritability estimates obtained and found a very poor agreement between the observed and the expected gains for the three characters. Warner (1952) proposed the use of only the F and first backcross 2 generations for the computation of heritability based on the relations derived by Mather (1949). This method does not separate dominance from environmental variance and the magnitude of these effects can not be estimated. Robinson et ale (1949), Comstock et ale (1952) and Gardner et ale (1953) have presented estimates of dominance and expected gains due to selection in three open pollinated varieties of maize. Moll et ale (1960) estimated heritabilities in biparental offspring of a variety cross and concluded that the cross is a better source of material for selection than either of the va~ieties. There is a good deal of literature discussing heterosis and hybrid vigor. In maize breeding, voluminous literature pertaining to this subject is available. Hutchinson et ale (1938) suggested that epistatic along with dominance might be involved in heterosis. No attempt is made here to review the heterosis thoroughly. The importance of non-allelic gene interaction (epistasis) in heterosis and in quantitative inheritance is subject to controversy. Complementary and duplicate factor type interactions have been reported for a number of qualitative characters. Therefore, it might reasonably be theorized that such gene action is involved in quantitative inheritance. The statement by Crow (1948) that no single explanation is yet satisfactory or responsible for all the effects observed in heterosis is as good now as when it was made.

29 11 Following the inbreeding experiments of Shull (1908, 1911), Jennings (1941) expressed the rate of approach to homozygosity due to inbreeding as (1/2)n per generation in relation to the original number of heterozygous loci and where n is the number of generations selfed. Wright (1922) showed that in the absence of epistasis or selection, a random breeding population derived from n lines will have lin less heterozygosity than the average of the F l crosses over their respective inbred parents. With dominance and no epistasis, the performance of a hybrid in an advanced generation can be expressed as where n is the number of lines, P the average of the parents, F l the average of all possible F l crosses and F i represents the estimated performance of any advanced generation. According to Hardy-Weinberg's law, there would be no further decline in heterozygosity after the F 2 generation provided mating is completely at random and there is no differential selection. Kiesselbach (1933) found good correspondence between Wright's formula and yield of 2, 4, 8 and 16 lines hybrids tested over several years. Neal (1935) compared the parental F l and F 2 generations of a number of single, three-way and double crosses and an F generation in single and three-way crosses and reached the same 3 conclusion as Kiesselbach. Neal's (1935) results were in close agreement with the theoretical expectation that yield decreases in linear order from the F l hybrids to the inbred lines as heterozygosis decreases. This is in accordance with the formula proposed by Wright (1922).

30 12 Stringfield (1950) tested the linearity of yield, time of silking and height of the ear-bearing node under four levels of heterozygosity. Four inbred lines of maize were combined into crosses so that all lines contributed in equal to (1) crosses of zero heterozygosis, sibling x sibling within lines, (2) crosses of 50 per cent heterozygosis, F 2 generations of single crosses and backcrosses. (3) crosses of the type (AxB) (AxC) having 75 per cent heterozygosis and (4) crosses of 100 per cent of attainable heterozygosis represented by the F l generations of single crosses, three-way crosses and double crosses. Results indicated that the expression of plant vigor for all the three characters studied as a function of added and equal proportions of heterozygosis is arithmetic rather than geometric. However, nonlinear increments in the area between 50 per cent and 100 per cent heterozygosis led to the conclusion that law of diminishing return operated as the levels of heterozygosity increased. Sentz et al. (1954) tested the relation between levels of heterozygosis and performance of a number of quantitative characters in Maize. such as yield, number of ears per plant, ear length, ear diamete:t, maturity, plant height and ear height. The five levels of heterozygosis developed were 0 per cent, 25 per cent, 50 per cent. 75 per cent and 100 per cent in two populations, CI 21 x NC 7 and NC 16 x NC 18. Results obtained were similar to those of Stringfield. Diallel Analysis The diallel cross appears to have been first proposed by Yates (1947). HaYman (1954a. 1954b), and Jinks (1954) and Jinks and Hayman (1953) outlined the analysis. Sprague and T.atum (1942), Henderson (1952). Griffing (1950, 1956a, 1956b), and Matzinger, Sprague and

31 13 Cockerham (1959) have considered the utility of diallel crosses in investigation of the notions of general and specific combining ability in plant and animal materials. Another application to a practical problem - the early generation evaluation of parental materials in breeding programs - has been discussed by Jinks (1955), Allard (1956b, 1956c), and Whitehouse, Thompson, and Valle Ribeiro (1958). The application to still another problem - the investigation of genotypicenvironmental interactions - has been considered by Rojas and Sprague (1952), Matzinger and Kempthorne (1956), Allard (1956a), and Crumpacker and Allard (1962). The theory of diallel crosses, and procedures for estimating certain genetic parameters in terms of gene models in varying degrees of complexity, have been discussed by Hull (1952), Griffing (1950, 1956a, 1956b), Hayman (1954a, 1954b, 1957, 1958a, 1960a), Jinks (1954, 1956), Dickinson and Jinks (1956) and Kempthorne (1956). Diallel analyses have been described which allow for every variation in experimental design including the presence and absence of parental means and of reciprocal crosses and differing relative degrees of replication of diagonal (parental) and off-diagonal (F l ) entries in the diallel table (Jones 1965). There is a corresponding range of alternative methods of partitioning the total variation and methods of deriving the variance components from the mean squares according to whether maternal or reciprocal effects are present or not (Griffing 1956; Wearden 1964) and whether the parental lines are fixed sample or a random sample of a population of inbred lines (Griffing 1956; Wearden 1964; Hayman 1960b). Littlewood, Carmer, and Hittle (1964) developed a computer program for analysis of diallei crosses for the four methods

32 14 and two models proposed by Griffing (1956). Diallel analysis has been widely used in studies on a nature of gene action in cross-pollinated crops like maize (Matzinger 1958; Moll et al. 1962) and self-pollinated ones like Nicotiana (Jinks 1954) where a sufficient number of pollinations can be made with ease. However, a full diallel set becomes unmanageable with an increase in number of parents, particularly in crops like wheat where the number of seeds per reproductive unit is very low. The reports on partial diallel analysis (Kemp thorne and Curnow 1961; Hinkelman and Kempthorne 1963; Curnow 1963; Fyfe and Gilbert 1963) have shown that selection can be made among crosses from a wider range of parents and the additive gene effects of a larger number of parents can be estimated although with a certain loss of precision which is compensated by the greater intensity of selection that can be applied to the parents. Kearsey (1965), in comparing different experimental designs, has found that the partial diallel appears to yield no more information that the two North Carolina designs. Arunachalam (1967) developed a computer program for partial diallel crosses. Lee and Kaltsikes (1971) developed a computer program to perform the Jinks-HaYman method for analysis in relation to dia11e1 crossing systems and a computer program to perform the general weighted least square analysis. Heritability The study of hereditary and environmental components of variation had its beginnings in the work of Johannsen (1909), East (1916), and Nilsson-Ehle (1909). The former demonstrated that both heritable and nonheritable agencies contributed to somatic variation in segregating

33 15 populations and that variation in pure lines was en~irely environmental. East and Nilsson-Ehle (op. cit.) further confirmed the work of Johannsen and demonstrated how such results conformed with the concept of Mendelian Genetics. Fisher (1918) first separated genetic variance into three components; that due to additive effects of genes, that due to dominance deviations from the additive scheme, and that due to deviations from the additive scheme attributable to inter-allelic interactions. Charles and Smith (1939) and Powers (1942, 1950) separated genetic from total variance by use of estimates of environmental variance based on nonsegregating populations and possible relations between means and variances. Fisher, Immer, and Tedin (1932), Mather (1949), Lush (1945). Smith (1936), Panse (1940), and others have studied heritable variation and have further subdivided it into that portion attributable to additive genetic effects and that due to deviations from additive scheme. Lush and Panse (op. cit.) proposed the use of the ratio of the additive genetic components of variance to total variance as a measure of the degree of heritability. Robinson, Comstock, and Harvey (1949) used a method involving estimates of components of variance through study of bi-parental progenies to measure heritability in corn. They further compared this method with heritability as calculated from parent-offspring regressions. Techniques for estimating the degree of heritability in crop plants previously reported in the literature fall into three main categories: those based on a) parent-offspring regression b) variance components from an analysis of variance, and

34 16 c) approximation of nonheritab1e variance from genetically uniform populations to estimate total genetic variance. Warner (1952) further outlined a method to estimate the heritability by using the F 2 and the backcross generations. This method is presented below which offers the following advantages: 1) The estimate is made entirely on the basis of the F2 and the backcross of the F 1 to each parent 2) The estimating nonheritab1e variance is unnecessary. The second advantage listed above is of particular importance in such crops as corn where an estimate of heritable variance contingent on measuring the variance of nonsegregating populations as an approximation of nonheritab1e variance is unsatisfactory. Presumably because of greatly reduced vigor in inbred lines, the total variance in such populations measure not only the nonheritab1e effects comparable to more normal populations, but, in addition, nonheritab1e variation which is present only in populations of low vigor. These weak plants apparently tend to be much more susceptible to variations in environment than are plants with more normal vigor. Conversely the use of the F1 to estimate nonheritab1e variance may prove too small a value to be comparable to that present in segregating populations if the vigorous genotype tends to be less sensitive to environmental fluctuations, or too large a value if environmental variance is highly correlated with the means. Inheritance of Agronomic Traits in Maize Investigations relating to inheritance of ear height and plant height in maize can be broadly classified as those which deal with

35 17 (1) number of genes and (2) gene action. Yang (1949) suggested that only a small number of genes were involved controlling plant height in maize, probably two to three. Gierbrecht (1961) reported that perhaps six genetic factors were segregating in the F Z and backcrosses of two inbred lines of maize, V 3 and B 14 for ear height. His conclusions have limited application, however, because the estimates of number of genetic factors were based on a method which gives only a probable estimate of number of genetic factors. He did not fit observed means, variances and frequency distribution to their expected parameters on the bases of genetic models. Weijer (195Z) has catalogued maize genetic types. According to him at least eight recessive genes controlling plant height have been reported and assigned various linkage groups. These genes are d l, d Z ' d 3, d 4, d 5, d 6, d 7, and br. The presence of some of the dwarf series genes (d l,,d 7 ) results poor pollen shedding and poor viability along with certain other anomalies like stamen in ears. The presence of br (brachytic) genes does not result in poor viability or pollen shedding. All these different height-reducing or dwarfing genes can be grouped into three basic types. These types are (1) the "true dwarfs" which are highly abnormal in appearance with fair to poor viability, (2) the "compact dwarfs" in which all plant parts are proportionally reduced in size and (3) the "brachytic dwarfs" which are similar in size and appearance to the normal types except for marked shortening of the stalk below the ear.

36 18 Leng (1956) reported on the performance of dwarf hybrids obtained by introducing dwarfing gene brachytic-2 into the parental inbreds. Results showed that dwarf hybrids as single crosses yield as much grain as their normal counterpart while producing only about two-thirds as much dry stover. Resistance to stalk-lodging was good. Anderson and Chow (1963) discussed the possibility of using br-2 gene in the development of shorter, lower-eared varieties. Their studies indicated that the ear-height of br-2 hybrid was positively correlated to the ear height of the normal counterpart converted to brachytic type. They concluded that hybrid with higher ear placement in the normal type also produced the higher ear placement in its corresponding brachytic hybrid. They confirmed the conclusion made by Thompson et al. (~960) in a previous study. There is a good deal of evidence that plant height and ear height are highly correlated characters. It seems probable that certain of the genes are common for both these characters. Reports of high correlation between ear height and plant height have come from Robinson et al. (1951), r = 0.84; Lindsey et al. (1962), r = 0.78 and 0.88; Stuber et al. (1966b), r = 0.69 and 0.82; and Hallauer and Wright (1967), r = These correlation estimates are based upon large number of observations. Literature reports relating to quantitative genetic est~~tes for ear height in maize are presented in two tables and are listed ~1 chronological order. Table I lists the studies reporting est.es for dominance and additive variance and degree of dominance. Table II lists those studies reporting nonallelic interactions and the techniques utilized in obtaining the estimates.

37 TABLE I LITERATURE REPORTS RELATING TO GENETIC VARIANCE ESTIMATES FOR EAR HEIGHT OF MAIZE - a Author Material Generation Sample Dominance Additive Degree of Ratio of Dom. tested No variance variance dominance to add. var. ctd 2 f1'a 2 0'\ 2/(1:2 a D A Robinson CI 21 X NC 7 F 2 et a NC 16 X NC 18 F NC 34 X NC 45 F2 Gardner NC 7 X CI 21 F et a F NC 33 X K 64 F F Robinson Jarvis BIP et a BIP Weekly BIP BIP b Indian Chief BIP b Gardner et a M 14 X CI 21 F F F F Lindsey NC 7 X CI 21 F F F t-' F \0