In our previous papers, we demonstrated that the inclusion of epistatic interactions in marker models improved prediction for corn (Zea mays L.) grain quality traits. The utility of pre-selecting markers for epistatic models was not reported. In papers by other researchers, including epistatic effects in a model did not improve prediction efficacy for whole genome selection. The objectives of this study were therefore to evaluate the value of: (1) pre-selecting markers and interactions at different type 1 error levels to predict performance; (2) adding epistatic interactions to models including all markers, and (3) using marker-based models to predict performance of kernel weight (KWT), flowering date (FDT), and plant height (PHT). Data for KWT, FDT, PHT, and oil and protein concentrations were obtained for 500 S2 lines and their testcrosses from the crosses of Illinois high oil × Illinois low oil and Illinois high protein × Illinois low protein corn strains. Pre-selection using an epistatic model including both single-locus and two-locus interaction effects significant at the P = 0.05 level significantly increased prediction efficacy over selection including all markers and epistatic interactions. Adding all epistatic interactions to a model including all markers did not improve prediction. For most traits, prediction based on the P = 0.05 epistatic pre-selection model was nearly as effective as prediction based on phenotype, suggesting subsequent marker-based selection would be effective.

In a previous paper, the utility of partial least squares as a tool for predicting performance using marker‐based models was demonstrated. Including interactions between markers in prediction models improved prediction efficacy. The objectives of this paper were to determine whether: (i) molecular marker models based on 1 yr's phenotypic data could be used to predict performance in a second year; (ii) models based on per se data would be useful in predicting testcross performance; and (iii) adding epistasis to a model would improve prediction in either case. Data for protein, oil, and starch were obtained from 500 S2 lines and their testcrosses from the crosses of Illinois High Oil (IHO) × Illinois Low Oil (ILO) and of Illinois High Protein (IHP) × Illinois Low Protein (ILP) corn (Zea mays L.) strains. Adding epistasis to a model significantly increased predictive power both between years and for testcross performance. The proportion of variability accounted for when predicting testcross performance from per se performance was lower than when predicting performance in different years. In all cases, observed vs. predicted correlations were high enough to suggest they would be useful in marker‐assisted, or marker‐based, breeding.

To be useful, adding epistasis to a prediction model must increase predictive power. The objectives of this study were to determine i) using partial least squares (PLS) techniques, whether the ability to predict performance can be increased by including epistasis in a prediction model; ii) whether relaxing the probability of preselecting a marker or interaction to include in the PLS analysis from 0.001 to 0.01 to 0.05 would increase predictive power; and iii) whether the proportion of variability accounted for could be raised to a level useful in breeding. Data for protein, oil, starch, and grain yield were obtained from 500 S 2 lines from the crosses of Illinois High Oil × Illinois Low Oil and of Illinois High Protein × Illinois Low Protein corn (Zea mays L.) strains. Lines per se and testcrosses were evaluated for oil, protein, and starch, and only testcrosses for grain yield. Adding epistasis to a model signifi cantly increased predictive power, as did increasing the probability level for inclusion of signifi cant markers and epistatic effects in the PLS analysis from 0.001 to 0.01 to 0.05. With epistasis in the model and P = 0.05, correlations of predicted and observed means were high enough to suggest they could be useful in breeding.

To be useful, adding epistatis to a prediction model must increase predictive power. The objectives of this study were to determine: 1) using partial least squares techniques, whether the ability to predict performance can be increased by including epistasis in a prediction model; 2) whether relaxing the probability level for inclusion of a marker or interaction in a model from 0.001 to 0.01 to 0.05 would increase predictive power; 3) whether the proportion of variability accounted for could be raised to a level useful in breeding; 4) whether molecular marker models based on one year’s phenotypic data could be used to predict performance in a second year; and 5) whether models based on per se data would be useful in predicting testcross performance. Data for protein, oil, and starch, were obtained from 500 S2 lines and their testcrosses from the crosses of Illinois High Oil (IHO) x Illinois Low Oil (ILO) and of Illinois High Protein (IHP) x Illinois Low Protein (ILP) corn (Zea mays L.) strains. Increasing the probability level for detection of significant markers and epistatic effects from 0.001 to 0.01 to 0.05 significantly increased predictive power. Adding epistasis to a model significantly increased predictive power when models were based on data over all year and locations, when models based on one year were used to predict a second year, or when per se data were used to predict testcross performance. With epistasis in the model and P=0.05, correlations of predicted and observed means were high enough to suggest they might be useful in breeding. Marker models based on one year’s data produced correlations nearly as high as phenotypic correlations between years. While epistasis significantly improved performance of models used to predict testcross performance from per se performance, the proportion of variability accounted for was somewhat lower than when predicting performance in different years.

Epistasis has been proposed as a possible explanation for the long continued progress from selection in the Illinois long‐term corn (Zea mays L.) selection strains. From the crosses of Illinois High Oil (IHO) × Illinois Low Oil (ILO) and of Illinois High Protein (IHP) × Illinois Low Protein (ILP), 500 S2 lines were developed. Each IHO×ILO line was genotyped for 479 single nucleotide polymorphism (SNP) markers, and each IHP×ILP line was genotyped for 499 SNP markers. Per se and testcross progenies of each S2 line were evaluated for oil, protein, and starch. A large number of QTL were found to control these traits. The objective of this paper is to report the results of analysis of these crosses for two‐way epistatic interactions. More epistatic interactions were significant than expected by chance. The proportion of significant interactions that involved a marker significant in a single marker analysis (SMA) was no greater than expected by chance. The number of markers associated only with significant epistatic effects ranged from 46.3 to 72.2% of the total number of markers significant for either an interaction effect or from SMA. The number of QTL controlling a trait is much greater than will be found by analyzing for significant QTL main effects. Thus, epistasis could contribute to the long continued response to selection in the Illinois long‐term selection strains and also may help explain the continued success of commercial corn breeding.

To identify and characterize quantitative trait loci (QTL) affecting kernel weight and concentrations of protein, oil, and starch in the corn (Zea mays L.) kernel, plants from generations 70 of the Illinois High Protein (IHP) and Illinois Low Protein (ILP) strains, previously developed by divergent selection for kernel protein concentration, were crossed. The cross was random mated (RM) seven generations and selfed twice to develop 500 F1RM7S2 lines. The lines per se were evaluated at three locations with two replications for 2 yr and testcrosses were evaluated at three locations in 1 yr. Genotypes were evaluated using 499 SNP markers on DNA from a bulk of leaf tissue from each line. As the parent plants used to make the original cross were not available for genotyping, previously reported multivariable and modified simple interval mapping (SIM) procedures were used. SIM identified more significant regions for all traits than did single marker analysis. Correlations and signs of QTL effects suggest development of high protein-high starch lines would be difficult but that it should be possible to develop high protein-high oil lines with minimal effects on kernel weight. The identification of a large number of QTL (at least 40 each for oil, protein, starch, and kernel weight) with small effects agrees in general with earlier estimates based on quantitative genetic theory and has implications for breeding strategies for improved corn kernel quality traits.

ABSTRACTDivergent selection for oil and protein concentration in the corn (Zea mays L.) kernel was initiated at the University of Illinois in 1896 by C.G. Hopkins. In 2005, 106 generations of selection had been completed for high oil and 105 for high protein. Limits to selection for low oil and low protein were reached but not for high oil or high protein. Over the more than 100 yr of the existence of the program a number of attempts have been made to analyze the experiment using quantitative genetic tools. The purpose of this paper is to trace the use of quantitative genetic techniques to analyze the results of divergent long‐term selection for oil and protein, to relate results to the question of the need for divergent parents for quantitative trait locus (QTL) analysis, and to provide a new look at the reasons for long‐continued progress from selection. Key findings include (i) progress from selection was much greater than could have been predicted; (ii) based on both classical quantitative genetic analysis and QTL studies, a large number of QTL are involved in control of the three traits; (iii) the number of QTL identified in a given study cannot be predicted by the magnitude of genetic variance or the divergence of the parents but is a function of the number of markers used and the number of lines evaluated; and (iv) epistasis may be an important factor in explaining long‐term response to selection.

Molecular breeding is defined as the application of molecular genetic tools to plant breeding. In this paper the application of population and quantitative genetics principles to plant breeding and the implications of those principles for the use of molecular genetic tools are discussed. The ability to transform plants and thus introduce genes from any species into a crop plant or to develop entirely new genes has greatly broadened the germplasm base for plant breeders. At the same time, this ability has created a set of problems, the solutions of which require application of population genetic theory. The availability of molecular markers has brought a rebirth of interest in quantitative genetics. It is now possible to identify chromosome segments which control quantitative traits and follow those traits in breeding. In addition, use of quantitative genetic principles provides a way of utilizing gene expression data as a plant breeding tool.

Altered concentrations of oil, protein or starch in maize (Zea mays L.) kernels can provide value-added products. Progeny developed from the cross between the long-term divergently selected strains, Illinois High Oil OHO) and Illinois Low Oil (ILO), cycle 70, were used to detect quantitative trait loci (QTL) for protein, starch, and oil concentration, and kernel weight. Recombinant inbred lines (RIL) were derived from an F2 created from one cycle of random mating the F1 (RM1) or after four cycles of random mating the RM1 (R-N45). Both per se and corresponding testcrossed lines (TC1 and TC5) were evaluated in four environments. Genetic variance was lower in the testcrosses than in the per se progenies for all traits. Random mating was associated with a reduction in genetic variance for oil concentration between TC1 and TC5. Recombination maps constructed for the RM1 and RM5 revealed an expansion of 2.05 fold after random mating, close to the expected 1.83 fold expansion. Fewer loci were significant in the random mated generations, reflecting recombination between the markets and the quantitative trait loci (QTL) or between linked QTL. Using composite interval mapping (CIM), kernel oil concentration QTL were detected in chromosomal bins 2.09, 3.04, 6.04, 8.04, 8.05, and 8.07 of RM1 or TC1. The oil QTL detected in bin 6.04 in RM1 was also detected in TCI and in previous research, but not in RM5 or TC5. This may be due to break up of marker-QTL associations and/or recombination between QTL.

To identify and characterize quantitative trait loci (QTL) affecting oil, protein and starch concentration in the corn (Zea mays L.) kernel, plants from Generation 70 of the Illinois High Oil (IHO) and Illinois Low Oil (ILO) populations, previously developed by divergent selection for kernel oil concentration, were crossed. The cross was random mated 10 generations and selfed two generations to develop 500 F1RM10S2 lines. The lines per se and their testcross progenies were evaluated at three locations with two replications for 2 yr. Genotypes were evaluated using 479 SNP markers on a bulk of kernels from each line. Since the parent plants used to make the original cross were not available for genotyping, a multivariable optimization procedure was developed to estimate parental population parameters required for QTL mapping. Simple interval mapping was performed with software specially developed to account for the complex mating structure and the fact that the initial cross was made between populations. Correlations and signs of QTL effects suggest development of high oil–high starch lines would be difficult but that it should be possible to develop high oil–high protein lines. The identification of a large number of QTL (at least 40 each for oil, protein, and starch) with small effects has implications for breeding for improved corn chemical composition.

In one of the longest-running experiments in biology, researchers at the University of Illinois have selected for altered composition of the maize kernel since 1896. Here we use an association study to infer the genetic basis of dramatic changes that occurred in response to selection for changes in oil concentration. The study population was produced by a cross between the high- and low-selection lines at generation 70, followed by 10 generations of random mating and the derivation of 500 lines by selfing. These lines were genotyped for 488 genetic markers and the oil concentration was evaluated in replicated field trials. Three methods of analysis were tested in simulations for ability to detect quantitative trait loci (QTL). The most effective method was model selection in multiple regression. This method detected ∼50 QTL accounting for ∼50% of the genetic variance, suggesting that >50 QTL are involved. The QTL effect estimates are small and largely additive. About 20% of the QTL have negative effects (i.e., not predicted by the parental difference), which is consistent with hitchhiking and small population size during selection. The large number of QTL detected accounts for the smooth and sustained response to selection throughout the twentieth century.

The Illinois Long-Term Selection Experiment for grain protein and oil concentration in maize (Zea mays) is the longest continuous genetics experiment in higher plants. A total of 103 cycles of selection have produced nine related populations that exhibit phenotypic extremes for grain composition and a host of correlated traits. The use of functional genomics tools in this unique genetic resource provides exciting opportunities not only to discover the genes that contribute to phenotypic differences but also to investigate issues such as the response of plant genomes to artificial selection, the genetic architecture of quantitative traits and the source of continued genetic variation within domesticated crop genomes.

The effects of random mating on population performance and identification of marker–quantitative trait loci (QTL) associations were studied in the cross of the Illinois High Protein (IHP) and Illinois Low Protein (ILP) maize (Zea mays L.) strains. The F1 was randomly mated to produce the Syn0 and Syn4 generations. Two hundred Syn0 S1 and 200 Syn4 S1 lines were crossed to two inbred testers. The S1 lines per se and the testcrosses (TCs) were evaluated for grain protein, starch, and oil concentrations in four environments. Genetic variance for protein and starch was reduced from the Syn0 to the Syn4 and genetic variance for the S1 lines per se was greater than for the TCs. Heritability was similar in the Syn0 and Syn4. With single factor (SF) analysis, the number of significant marker–QTL associations for protein and starch in the Syn4 was drastically reduced from the number found in the Syn0. Multiple regression (MR) analysis supported this result. Very few significant marker–QTL associations for oil were found in either the Syn0 or Syn4. Agreement between the testers and the S1 lines per se for chromosomal regions identified as associated with QTL for starch and protein was high. Genotypic correlations of starch with protein ranged from −0.96 to −0.98. However, three markers had the same sign and significant additive effects for both starch and protein in the S1 progeny and both TCs. Such markers should be useful for simultaneous selection for high starch and high protein.

New Guinea impatiens (Impatiens hawkeri) is an economically important floral crop, however, little work has been conducted to further our understanding of the genetics of this crop. In this study, we used amplified fragment length polymorphism (AFLP) technology to investigate the level of polymorphism present among 41 commercial cultivars of New Guinea impatiens, study their genetic relatedness, and assess the genetic diversity in this material. An efficient DNA extraction protocol was developed, and a total of 48 EcoRI and MseI primer combinations were used for PCR amplification. Amplification products were then subjected to polyacrylamide gel electrophoresis. The AFLP analysis showed that all 41 cultivars generated between 73 and 130 scoreable polymorphic bands per primer combination. Gower's Genetic Dissimilarity estimates for the entire set of cultivars ranged between 0.940 and 0.488. A dendogram was generated from these dissimilarity data that revealed four groupings among these 41 cultivars. The implications of these results on genotypic variation, genetic relationships, and genetic diversity in New Guinea impatiens will be discussed.

Our objectives were to: i) identify quantitative trait loci (QTL) associated with grain yield, moisture, plant height and ear height in two maize (Zea-mays L.) crosses; ii) compare results from different methods of mapping QTL; iii) evaluate QTL x environment (QTLxE) interaction; iv) determine effect of identified QTL on multiple traits; and v) com pare results with those in the Maize database. The genetic material consisted of 122 random S 5 lines from Va26xMo17 testcrossed to B73 and 95 random S 5 lines from B84xB73 testcrossed to Mo17. Testcrosses were evaluated for grain yield, plant height, ear height, and grain moisture in 6 environments. Restriction fragment length polymorphism (RFLP) data were obtained for 75 DNA probes for Va26xMo17 and 45 for B84xB73. Methods used to detect quantitative trait loci (QTL) included simple mapping using single factor analysis in SAS (SM). stepwise multiple regression (MR). simple interval mapping (SIM). and composite interval mapping (CIM). More QTL were identified in the Va26xMo17 CTOSS than in B84xB73. This difference may have resulted from Va26xMo17 having more probes and more phenotypic data available and/or because of greater genetic diversity in the Va26xMo17 cross. For yield CIM and SIM both identified 6 significant QTL. Five of the six QTL identified by CIM were also identified by SIM. Using MR, 9 QTL were identified, 5 of which were in the same regions as those identified by CIM Similar agreement between methods was found for plant and ear height. However. for grain moisture CIM identified nearly twice as many QTL as SIM or MR. More QTL were detected for plant height. ear height. and grain moisture than for grain yield. Some QTL for grain yield were also associated with QTL for moisture. plant height and ear height. Only one QTL for yield was in common between the Va26xMo17 cross and the B73xB84 cross A small number of markers found associated with QTL were listed as associated with QTL for the same trait in the Maize database records. Although agreement between individual environments for identification of QTL was poor, QTL x environment interaction measured by CIM was not important.

A possible use of nonelite germplasm is as a source of alleles for disease resistance. Our objective was to determine the value of tropical and temperate maize (Zea mays L.) germplasm accessions as sources of alleles to improve disease resistance of the Corn Belt hybrid FR1064 × LH185. A group of tropical populations and hybrids from the Germplasm Enhancement of Maize Project (GEM) crossed to either Mo17 or B73 (temperate inbreds), was studied. In addition, a group of temperate accessions was evaluated. Reaction to southern corn leaf blight (SCLB) [Bipolaris maydis (Nisikado & Miyake) Shoemaker = Helminthosporium maydis Nisikado & Miyake], northern corn leaf spot (NCLS) [Bipolaris zeicola (G. L. Stout) Shoemaker = Helminthosporium carbonum Ullstrup Races 2 and 3], gray leaf spot (GLS) (Cercospora zeae‐maydis Tehon & E.Y. Daniels), northern corn leaf blight (NCLB) [Exserohilum turcicum (Pass.) K. J. Leonard & E. G. Suggs = Helminthosporium turcicum Pass., Races 1, 1, 2, 23N and three unidentified races] and common rust (Puccinia sorghi Schwein) was studied using Dudley's method for identifying populations with favorable alleles. All accessions had favorable dominant alleles for resistance not present in FR1064 × LH185 for common rust, GLS, and SCLB. Four accessions had a number of favorable alleles for resistance not significantly different from the best accession for all five diseases. Crosses of the accessions containing Mo17 to FR1064 and LH185 were less susceptible to SCLB, NCLB, and rust than crosses of accessions containing B73. At least seven of the best 10 populations were tropical × Mo17 crosses for SCLB, NCLB, and rust when populations were ranked for presence of favorable alleles for resistance not present in either LH185 or FR1064. Net value statistics for tropical × B73 and tropical × Mo17 accessions differed in indicating whether backcrossing or selfing from the F1 is more desirable. Therefore, if tropical populations are crossed to Corn Belt inbreds to increase adaptation, comparisons among tropical populations should only be made when all populations being compared have been crossed to the same Corn Belt inbred.