Doubled haploid (DH) technology facilitates rapid development of homozygous inbred lines for hybrid breeding in maize {Zea mays L.). However, the required artificial chromosome duplication step, which commonly involves toxic and costly chemicals, represents a bottleneck. Exploiting the haploids' natural fertility may enable abolishment of artificial treatments and enhance efficiency of line development. We screened haploid populations derived from U.S. Corn Belt and tropical germplasm for the proportion of fertile haploids (FP) and the number of intact seeds (IS) on selfed ears and examined the effects of environments, heterotic groups, maturity groups, and population types on FP and IS. The FP ranged from 0 to 20% under field conditions and from 0 to 70% under greenhouse conditions. Tropical elite germplasm had higher median FP and mean IS than tropical landrace accessions. The Corn Belt heterotic group Stiff Stalk had higher niedian FP than lodent and Lancaster while early germplasm showed higher median FP than the other maturity groups. Significant (p > 0.01) genetic variance for FP was observed among elite Corn Belt materials and heritability was 0.79, indicating that recurrent selection to increase FP is promising. We propose that artificial chromosome duplication is not necessary for DH line production from germplasm with high FP. This seems particularly relevant to enable small maize breeding programs in developing countries to adopt the DH technology for line development.

Published information is lacking on whether genomewide selection, based on a single tester and a single year of testing, can identify maize (Zea mays L.) lines that would perform well in multiple subsequent years and with multiple testers. Our objectives were to determine (i) if phenotypic selection or genomewide selection is more predictive of maize performance in future environments and with different testers, (ii) if combining both marker and phenotypic information is advantageous in selection, and (iii) the upward bias in correlations between marker‐predicted values and phenotypic values (rMP) when cross‐validation across individuals and across environments is not performed. We evaluated four elite populations, each with 150 or 250 doubled haploid (DH) lines, in 18 environments in the US Corn Belt. The DH lines were genotyped with 3072 single nucleotide polymorphism markers. The rMP values were 0.14 to 0.66 for grain yield and 0.49 to 0.66 for moisture. Phenotypic selection was always as effective as or more effective than genomewide selection. The rMP was lower when different testers were used for the training and test populations. Selection based on marker and phenotypic information was slightly more effective than genomewide selection or phenotypic selection alone. The correlation between predicted and observed performance was higher when cross‐validation across individuals and years was not performed than when cross‐validation across individuals and years was performed. For genomewide selection to be superior to phenotypic selection, the gains must not be measured in terms of the per‐generation response with equal population sizes and selection intensities.

Doubled haploid (DH) lines produced via in vivo haploid induction have become an indispensable tool in maize (Zea mays L.) breeding and research. To determine the predictive value of the per se performance of haploid and DH lines in population and hybrid breeding, genetically balanced sets of haploid and DH lines along with testcrosses of the DH lines were evaluated in field trials across four locations over 2 yr in Germany. Suitable material sets were provided by three collaborating breeders. Each set comprised 54 to 58 DH lines developed from a proprietary elite dent single cross. These DH lines were crossed with one or two flint testers and subjected to haploid induction for production of corresponding haploid versions. Haploid lines, DH lines, and testcrosses were grown in separate but adjacent blocks. Haploid lines were surrounded by a mixture of inbred lines functioning as pollen source. Most haploid plants were male sterile but showed a certain degree of female fertility. Highly significant genetic variation reflected by high heritability coefficients existed in all material sets for all traits. Genetic correlations between haploid and DH lines were moderate to strong (0.5 < genetic correlation coefficient [rg] < 0.9) for early vigor, silking date, plant height, and stover weight per plant. Correlations between DH lines and testcrosses varied from nonsignificant to moderately strong for grain yield but were strong for silking date, plant height, stover yield, and grain moisture content. Silking date, early vigor, and plant height of haploid lines were moderately but significantly associated with grain yield of testcrosses. Somewhat higher estimates were obtained for the corresponding correlations between DH lines and testcrosses. In conclusion, selection for silking date, early vigor, plant height, and stover weight at the haploid level is expected to result in positive correlated genetic gain for various traits not only at the DH but also at the testcross level. Likewise, selection at the DH level may substantially speed up progress in combining ability.

Recurrent testcross selection based on doubled haploid lines can be highly effective in hybrid maize (Zea mays L.) breeding. As a consequence, however, the genetic variance of the breeding population is reduced more rapidly. To limit the reduction of variance, a minimum effective population size (Ne) has to be ensured. However, maintaining sufficient Ne restricts the achievable selection intensity, in particular when crosses are made every year to start a new selection program such that the hybrid breeding population representing one heterotic group is divided into multiple, timely staggered subpopulations. Intercrossing lines from different subpopulations, defined herein as interlinking, allows the breeder to alleviate this restriction. In this study, the formula for predicting the genetic gain from selection is extended to account for interlinking of subpopulations. The optimum allocation of test resources was determined for two interlinking strategies and a control mating scheme (without interlinking) under one‐, two‐, and three‐stage testcross selection. Our results show that optimum interlinking of staggered subpopulations provides faster breeding progress than advancing non‐interlinked subpopulations separately. Moreover, the manner in which subpopulations are interlinked and the proportion of intercrossed lines from different subpopulations is fundamental in the design of efficient recurrent selection plans.

Doubled haploid (DH) lines are increasingly being used in commercial hybrid maize (Zea mays L.) breeding. They allow for various quantitative genetic and logistic advantages provided that they are implemented in efficient and optimally allocated breeding procedures. In the present study, a new software was applied to optimize two recurrent selection (RS) schemes for hybrid maize breeding based on DH lines under a restricted annual budget and an upper limit for the relative annual loss of genetic variance. This software maximizes the expected gain from selection in general combining ability by means of quantitative genetic model calculations. Optimization results are compared for one, two, and three stages of testcross selection under different assumptions regarding the evaluation of lines per se and the annual budget. Results show that the optimum allocation of technical and budget resources to the individual steps of an RS program and the efficiency of alternative RS procedures are decisively determined by the number of selection stages. Under standard assumptions, one-stage selection was superior to two-and three-stage selection. Thus, shortening the length of an RS scheme considerably increases its efficiency. By intercrossing a reduced number of selected lines for starting a new RS cycle, the short-term response to selection may be increased, but the population size and, thus, the selection limits in the long run are diminished. Therefore, fair comparisons of alternative RS procedures require to define the intended time span for maximizing the genetic gain from RS and to restrict the relative annual loss of genetic variance accordingly.