THE EXPERIMENTAL BASIS FOR HYBRID MAIZE *

Summary The development of superior maize hybrids and their rapid acceptance by farmers constitutes one of the most important advances in American agriculture within the past century. In 1944 hybrid maize plantings occupied approximately 83 % of the maize acreage in the Corn Belt. The increase in yield resulting from the use of hybrid maize was estimated at 669,480,000 bushels in 1943. The production of hybrid seed involves four separate steps: (1) development of inbred lines and testing of these in hybrid combinations, (2) multiplication of seed of the component lines, (3) seed production of the foundation single crosses, and (4) production of double‐crossed seed used for commercial planting. The earlier methods used in maize improvement included mass selection, varietal hybridization and ear‐to‐row selections. These methods were of limited value and are now of interest largely from a historical standpoint. Inbred lines are developed by self‐fertilization accompanied by selection. Selection is practised for the various characters which are expressions of vigour. Positive and significant correlations have been reported between yields of the hybrids and yield, plant height, ear length, ear diameter, shelling percentage, number of ears per plant, leaf area, brace roots and root volume of the inbred parents. The final evaluation of an inbred line is determined by its performance in hybrid combinations. The top‐cross or inbred‐variety cross is commonly used for the preliminary evaluation. Subsequently the more promising lines are tested in single‐cross combinations. Evidence has been presented indicating that combining ability of inbred lines is heritable. In new lines isolated from single crosses, a greater proportion of high‐yielding combinations are obtained when the component lines are derived from unrelated parentage. Single‐cross performance data may be used to predict the performance of double‐crossed combinations. The most efficient method of estimation is based on the average performance of the four non‐parental single‐cross combinations. The order of pairing of the four component lines of a double cross may have a marked effect on variability and yield. In general the highest yielding double crosses are those that combine single crosses differing most widely in parentage. In the direct inbreeding of open‐pollinated varieties the great majority of lines are discarded after preliminary testing. The remaining lines produce desirable hybrids, and these are used as source material for a new cycle of inbreeding and testing. These cyclical repetitions of inbreeding and testing are designated as cumulative selection. Second generation seed is sometimes used as parent stock in crossing fields. If genetic identity has been maintained double‐crossed seed produced from F 2 or F 3 advanced generation single‐cross stocks will produce yields which are essentially similar to the same double crosses produced from F 1 single crosses. The use of F 2 seed for commercial planting invariably results in rather large decreases in yield. Synthetic varieties are advanced generations of the open‐pollinated seed mixtures of a number of inbred lines or of hybrids among them. When the component lines are selected on the basis of combining ability yields of the resulting synthetics approximate yields of standard double‐crossed combinations. Several new methods of isolating and testing inbred lines are under investigation. All of these involve some aspect of cumulative selection or early testing. Early testing is based on two assumptions: first, that there are marked differences in combining ability among the plants of a population selected for inbreeding, and secondly, that a selected sample based on tests of combining ability of S 0 plants provides a better sample for further inbreeding and selection than a more nearly random sample from the same population selected on visual appearance alone. Gametic selection, one aspect of early testing, is based on the assumption that if zygotes of a given combining ability occur with a frequency of 1 % then gametes having a corresponding combining ability should occur with a frequency of 10 %. The problem of identifying these superior gametes is discussed. Recurrent selection for specific combining ability involves selection for combining ability with a minimum of inbreeding. No experimental data are available to evaluate the method and possible advantages and limitations are presented. The general problem of heterosis is one of the most important problems in genetics at the present time. Various aspects of this problem are discussed. Recessive characters probably have little or no influence on heterosis. Data on growth rates have thrown some light on the amount of stimulation resulting from hybridization but have been of limited value in providing a physiological explanation of the nature of the stimulus involved. Convergent improvement studies indicate that the dominant genes from two lines may be concentrated in a single line by recurrent back pollination, accompanied by selection. The data available indicate that dominant favourable genes are more important in heterosis than physiological stimulation resulting from genetic diversity. Attempts to assess the relative importance of general and specific combining ability have little critical bearing on the heterosis problem at present. Inversions have been suggested as a possible tool for determining the chromosomal distribution of factors responsible for heterotic effects. Data are limited but suggest the method is feasible. Progress in the development of disease‐resistant lines and hybrids is limited by the lack of suitable techniques for inducing epidemics at will. Information on resistance is reviewed briefly. Relatively little work has been done in developing inbred strains resistant to insect pests except in the case of the European cornborer. The results obtained indicate ample genetic variation for progress when an adequate infestation can be obtained. Maize varieties are extremely variable, and striking responses to selection have been reported. The Illinois selection experiments for high and low oil and high and low protein are the most notable. After 29 years of selection for high and low values respectively oil percentage had been changed from 4.68 to 9.86 and 1.51. The corresponding modification in protein percentage was from 10.93 to 16.60 and 4.68. Further results and opportunities in breeding for chemical composition are mentioned.