Empirical Comparison of Tropical Maize Hybrids Selected Through Genomic and Phenotypic Selections

Genomic selection in plant breeding: Key factors shaping two decades of progress

Genomic selection, the application of genomic prediction (GP) models to select candidate individuals, has significantly advanced in the past two decades, effectively accelerating genetic gains in plant breeding. This article provides a holistic overview of key factors that have influenced GP in plant breeding during this period. We delved into the pivotal roles of training population size and genetic diversity, and their relationship with the breeding population, in determining GP accuracy. Special emphasis was placed on optimizing training population size. We explored its benefits and the associated diminishing returns beyond an optimum size. This was done while considering the balance between resource allocation and maximizing prediction accuracy through current optimization algorithms. The density and distribution of single-nucleotide polymorphisms, level of linkage disequilibrium, genetic complexity, trait heritability, statistical machine-learning methods, and non-additive effects are the other vital factors. Using wheat, maize, and potato as examples, we summarize the effect of these factors on the accuracy of GP for various traits. The search for high accuracy in GP-theoretically reaching one when using the Pearson's correlation as a metric-is an active research area as yet far from optimal for various traits. We hypothesize that with ultra-high sizes of genotypic and phenotypic datasets, effective training population optimization methods and support from other omics approaches (transcriptomics, metabolomics and proteomics) coupled with deep-learning algorithms could overcome the boundaries of current limitations to achieve the highest possible prediction accuracy, making genomic selection an effective tool in plant breeding.

Genomic selection for target traits in the Australian lentil breeding program

Genomic selection (GS) uses associations between markers and phenotypes to predict the breeding values of individuals. It can be applied early in the breeding cycle to reduce the cross-to-cross generation interval and thereby increase genetic gain per unit of time. The development of cost-effective, high-throughput genotyping platforms has revolutionized plant breeding programs by enabling the implementation of GS at the scale required to achieve impact. As a result, GS is becoming routine in plant breeding, even in minor crops such as pulses. Here we examined 2,081 breeding lines from Agriculture Victoria's national lentil breeding program for a range of target traits including grain yield, ascochyta blight resistance, botrytis grey mould resistance, salinity and boron stress tolerance, 100-grain weight, seed size index and protein content. A broad range of narrow-sense heritabilities was observed across these traits (0.24-0.66). Genomic prediction models were developed based on 64,781 genome-wide SNPs using Bayesian methodology and genomic estimated breeding values (GEBVs) were calculated. Forward cross-validation was applied to examine the prediction accuracy of GS for these targeted traits. The accuracy of GEBVs was consistently higher (0.34-0.83) than BLUP estimated breeding values (EBVs) (0.22-0.54), indicating a higher expected rate of genetic gain with GS. GS-led parental selection using early generation breeding materials also resulted in higher genetic gain compared to BLUP-based selection performed using later generation breeding lines. Our results show that implementing GS in lentil breeding will fast track the development of high-yielding cultivars with increased resistance to biotic and abiotic stresses, as well as improved seed quality traits.

Combination of linkage and association mapping with genomic prediction to infer QTL regions associated with gray leaf spot and northern corn leaf blight resistance in tropical maize

Among the diseases threatening maize production in Africa are gray leaf spot (GLS) caused by Cercospora zeina and northern corn leaf blight (NCLB) caused by Exserohilum turcicum. The two pathogens, which have high genetic diversity, reduce the photosynthesizing ability of susceptible genotypes and, hence, reduce the grain yield. To identify population-based quantitative trait loci (QTLs) for GLS and NCLB resistance, a biparental population of 230 lines derived from the tropical maize parents CML511 and CML546 and an association mapping panel of 239 tropical and sub-tropical inbred lines were phenotyped across multi-environments in western Kenya. Based on 1,264 high-quality polymorphic single-nucleotide polymorphisms (SNPs) in the biparental population, we identified 10 and 18 QTLs, which explained 64.2% and 64.9% of the total phenotypic variance for GLS and NCLB resistance, respectively. A major QTL for GLS, qGLS1_186 accounted for 15.2% of the phenotypic variance, while qNCLB3_50 explained the most phenotypic variance at 8.8% for NCLB resistance. Association mapping with 230,743 markers revealed 11 and 16 SNPs significantly associated with GLS and NCLB resistance, respectively. Several of the SNPs detected in the association panel were co-localized with QTLs identified in the biparental population, suggesting some consistent genomic regions across genetic backgrounds. These would be more relevant to use in field breeding to improve resistance to both diseases. Genomic prediction models trained on the biparental population data yielded average prediction accuracies of 0.66-0.75 for the disease traits when validated in the same population. Applying these prediction models to the association panel produced accuracies of 0.49 and 0.75 for GLS and NCLB, respectively. This research conducted in maize fields relevant to farmers in western Kenya has combined linkage and association mapping to identify new QTLs and confirm previous QTLs for GLS and NCLB resistance. Overall, our findings imply that genetic gain can be improved in maize breeding for resistance to multiple diseases including GLS and NCLB by using genomic selection.

Discovery of genomic regions associated with grain yield and agronomic traits in Bi-parental populations of maize (Zea mays. L) Under optimum and low nitrogen conditions

Low soil nitrogen levels, compounded by the high costs associated with nitrogen supplementation through fertilizers, significantly contribute to food insecurity, malnutrition, and rural poverty in maize-dependent smallholder communities of sub-Saharan Africa (SSA). The discovery of genomic regions associated with low nitrogen tolerance in maize can enhance selection efficiency and facilitate the development of improved varieties. To elucidate the genetic architecture of grain yield (GY) and its associated traits (anthesis-silking interval (ASI), anthesis date (AD), plant height (PH), ear position (EPO), and ear height (EH)) under different soil nitrogen regimes, four F3 maize populations were evaluated in Kenya and Zimbabwe. GY and all the traits evaluated showed significant genotypic variance and moderate heritability under both optimum and low nitrogen stress conditions. A total of 91 quantitative trait loci (QTL) related to GY (11) and other secondary traits (AD (26), PH (19), EH (24), EPO (7) and ASI (4)) were detected. Under low soil nitrogen conditions, PH and ASI had the highest number of QTLs. Furthermore, some common QTLs were identified between secondary traits under both nitrogen regimes. These QTLs are of significant value for further validation and possible rapid introgression into maize populations using marker-assisted selection. Identification of many QTL with minor effects indicates genomic selection (GS) is more appropriate for their improvement. Genomic prediction within each population revealed low to moderately high accuracy under optimum and low soil N stress management. However, the accuracies were higher for GY, PH and EH under optimum compared to low soil N stress. Our findings indicate that genetic gain can be improved in maize breeding for low N stress tolerance by using GS.

Genomic prediction of optimal cross combinations to accelerate genetic improvement of soybean (Glycine max)

Improving yield is a primary soybean breeding goal, as yield is the main determinant of soybean's profitability. Within the breeding process, selection of cross combinations is one of most important elements. Cross prediction will assist soybean breeders in identifying the best cross combinations among parental genotypes prior to crossing, increasing genetic gain and breeding efficiency. In this study optimal cross selection methods were created and applied in soybean and validated using historical data from the University of Georgia soybean breeding program, under multiple training set compositions and marker densities utilizing multiple genomic selection models for marker evaluation. Plant materials consisted of 702 advanced breeding lines evaluated in multiple environments and genotyped using SoySNP6k BeadChips. An additional marker set, the SoySNP3k marker set, was tested in this study as well. Optimal cross selection methods were used to predict the yield of 42 previously made crosses and compared to the performance of the cross's offspring in replicated field trials. The best prediction accuracy was obtained when using Extended Genomic BLUP with the SoySNP6k marker set, consisting of 3,762 polymorphic markers, with an accuracy of 0.56 with a training set maximally related to the crosses predicted and 0.4 in a training set with minimized relatedness to predicted crosses. Prediction accuracy was most significantly impacted by training set relatedness to the predicted crosses, marker density, and the genomic model used to predict marker effects. The usefulness criterion selected had an impact on prediction accuracy within training sets with low relatedness to the crosses predicted. Optimal cross prediction provides a useful method that assists plant breeders in selecting crosses in soybean breeding.

Results from rapid-cycle recurrent genomic selection in spring bread wheat

Genomic selection (GS) in wheat breeding programs is of great interest for predicting the genotypic values of individuals, where both additive and nonadditive effects determine the final breeding value of lines. While several simulation studies have shown the efficiency of rapid-cycling GS strategies for parental selection or population improvement, their practical implementations are still lacking in wheat and other crops. In this study, we demonstrate the potential of rapid-cycle recurrent GS (RCRGS) to increase genetic gain for grain yield (GY) in wheat. Our results showed a consistent realized genetic gain for GY after 3 cycles of recombination (C1, C2, and C3) of bi-parental F1s, when summarized across 2 years of phenotyping. For both evaluation years combined, genetic gain through RCRGS reached 12.3% from cycle C0 to C3 and realized gain was 0.28 ton ha-1 per cycle with a GY from C0 (6.88 ton ha-1) to C3 (7.73 ton ha-1). RCRGS was also associated with some changes in important agronomic traits that were measured (days to heading, days to maturity, and plant height) but not selected for. To account for these changes, we recommend implementing GS together with multi-trait prediction models.

Detection of QTLs Regulating Six Agronomic Traits of Rice Based on Chromosome Segment Substitution Lines of Common Wild Rice (Oryza rufipogon Griff.) and Mapping of qPH1.1 and qLMC6.1

Wild rice is a primary source of genes that can be utilized to generate rice cultivars with advantageous traits. Chromosome segment substitution lines (CSSLs) are consisting of a set of consecutive and overlapping donor chromosome segments in a recipient's genetic background. CSSLs are an ideal genetic population for mapping quantitative traits loci (QTLs). In this study, 59 CSSLs from the common wild rice (Oryza rufipogon Griff.) accession DP15 under the indica rice cultivar (O. sativa L. ssp. indica) variety 93-11 background were constructed through multiple backcrosses and marker-assisted selection (MAS). Through high-throughput whole genome re-sequencing (WGRS) of parental lines, 12,565 mapped InDels were identified and designed for polymorphic molecular markers. The 59 CSSLs library covered 91.72% of the genome of common wild rice accession DP15. The DP15-CSSLs displayed variation in six economic traits including grain length (GL), grain width (GW), thousand-grain weight (TGW), grain length-width ratio (GLWR), plant height (PH), and leaf margin color (LMC), which were finally attributed to 22 QTLs. A homozygous CSSL line and a purple leave margin CSSL line were selected to construct two secondary genetic populations for the QTLs mapping. Thus, the PH-controlling QTL qPH1.1 was mapped to a region of 4.31-Mb on chromosome 1, and the LMC-controlling QTL qLMC6.1 was mapped to a region of 370-kb on chromosome 6. Taken together, these identified novel QTLs/genes from common wild rice can potentially promote theoretical knowledge and genetic applications to rice breeders worldwide.

Genome-wide association studies of grain yield and quality traits under optimum and low-nitrogen stress in tropical maize (Zea mays L.)

Genome-wide association study (GWAS) demonstrated that multiple genomic regions influence grain quality traits under nitrogen-starved soils. Using genomic prediction, genetic gains can be improved through selection for grain quality traits. Soils in sub-Saharan Africa are nitrogen deficient due to low fertilizer use and inadequate soil fertility management practices. This has resulted in a significant yield gap for the major staple crop maize, which is undermining nutritional security and livelihood sustainability across the region. Dissecting the genetic basis of grain protein, starch and oil content under nitrogen-starved soils can increase our understanding of the governing genetic systems and improve the efficacy of future breeding schemes. An association mapping panel of 410 inbred lines and four bi-parental populations were evaluated in field trials in Kenya and South Africa under optimum and low nitrogen conditions and genotyped with 259,798 SNP markers. Genetic correlations demonstrated that these populations may be utilized to select higher performing lines under low nitrogen stress. Furthermore, genotypic, environmental and GxE variations in nitrogen-starved soils were found to be significant for oil content. Broad sense heritabilities ranged from moderate (0.18) to high (0.86). Under low nitrogen stress, GWAS identified 42 SNPs linked to grain quality traits. These significant SNPs were associated with 51 putative candidate genes. Linkage mapping identified multiple QTLs for the grain quality traits. Under low nitrogen conditions, average prediction accuracies across the studied genotypes were higher for oil content (0.78) and lower for grain yield (0.08). Our findings indicate that grain quality traits are polygenic and that using genomic selection in maize breeding can improve genetic gain. Furthermore, the identified genomic regions and SNP markers can be utilized for selection to improve maize grain quality traits.

Genetic trends in CIMMYT's tropical maize breeding pipelines

Fostering a culture of continuous improvement through regular monitoring of genetic trends in breeding pipelines is essential to improve efficiency and increase accountability. This is the first global study to estimate genetic trends across the International Maize and Wheat Improvement Center (CIMMYT) tropical maize breeding pipelines in eastern and southern Africa (ESA), South Asia, and Latin America over the past decade. Data from a total of 4152 advanced breeding trials and 34,813 entries, conducted at 1331 locations in 28 countries globally, were used for this study. Genetic trends for grain yield reached up to 138 kg ha^-1 yr^-1 in ESA, 118 kg ha^-1 yr^-1 South Asia and 143 kg ha^-1 yr^-1 in Latin America. Genetic trend was, in part, related to the extent of deployment of new breeding tools in each pipeline, strength of an extensive phenotyping network, and funding stability. Over the past decade, CIMMYT's breeding pipelines have significantly evolved, incorporating new tools/technologies to increase selection accuracy and intensity, while reducing cycle time. The first pipeline, Eastern Africa Product Profile 1a (EA-PP1a), to implement marker-assisted forward-breeding for resistance to key diseases, coupled with rapid-cycle genomic selection for drought, recorded a genetic trend of 2.46% per year highlighting the potential for deploying new tools/technologies to increase genetic gain.

Improving drought tolerance in maize: Tools and techniques

Drought is an important constraint to agricultural productivity worldwide and is expected to worsen with climate change. To assist farmers, especially in sub-Saharan Africa (SSA), to adapt to climate change, continuous generation of stress-tolerant and farmer-preferred crop varieties, and their adoption by farmers, is critical to curb food insecurity. Maize is the most widely grown staple crop in SSA and plays a significant role in food security. The aim of this review is to present an overview of a broad range of tools and techniques used to improve drought tolerance in maize. We also present a summary of progress in breeding for maize drought tolerance, while incorporating research findings from disciplines such as physiology, molecular biology, and systems modeling. The review is expected to complement existing knowledge about breeding maize for climate resilience. Collaborative maize drought tolerance breeding projects in SSA emphasize the value of public-private partnerships in increasing access to genomic techniques and useful transgenes. To sustain the impact of maize drought tolerance projects in SSA, there must be complementary efforts to train the next generation of plant breeders and crop scientists.

Empirical comparison of genomic and phenotypic selection for resistance to Fusarium ear rot and fumonisin contamination in maize

GS and PS performed similarly in improving resistance to FER and FUM content. With cheaper and faster genotyping methods, GS has the potential to be more efficient than PS. Fusarium verticillioides is a common maize (Zea mays L.) pathogen that causes Fusarium ear rot (FER) and produces the mycotoxin fumonisin (FUM). This study empirically compared phenotypic selection (PS) and genomic selection (GS) for improving FER and FUM resistance. Three intermating generations of recurrent GS were conducted in the same time frame and from a common base population as two generations of recurrent PS. Lines sampled from each PS and GS cycle were evaluated in three North Carolina environments in 2020. We observed similar cumulative responses to GS and PS, representing decreases of about 50% of mean FER and FUM compared to the base population. The first cycle of GS was more effective than later cycles. PS and GS both achieved about 70% of predicted total gain from selection for FER, but only about 26% of predicted gains for FUM, suggesting that heritability for FUM was overestimated. We observed a 20% decrease in genetic marker variation from PS and 30% decrease from GS. Our greatest challenge was our inability to quickly obtain dense and consistent set of marker genotypes across generations of GS. Practical implementation of GS in individual small-scale breeding programs will require cheaper and faster genotyping methods, and such technological advances will present opportunities to significantly optimize selection and mating schemes for future GS efforts beyond what we were able to achieve in this study.

Using an incomplete block design to allocate lines to environments improves sparse genome-based prediction in plant breeding

Genomic selection (GS) is a predictive methodology that trains statistical machine-learning models with a reference population that is used to perform genome-enabled predictions of new lines. In plant breeding, it has the potential to increase the speed and reduce the cost of selection. However, to optimize resources, sparse testing methods have been proposed. A common approach is to guarantee a proportion of nonoverlapping and overlapping lines allocated randomly in locations, that is, lines appearing in some locations but not in all. In this study we propose using incomplete block designs (IBD), principally, for the allocation of lines to locations in such a way that not all lines are observed in all locations. We compare this allocation with a random allocation of lines to locations guaranteeing that the lines are allocated to the same number of locations as under the IBD design. We implemented this benchmarking on several crop data sets under the Bayesian genomic best linear unbiased predictor (GBLUP) model, finding that allocation under the principle of IBD outperformed random allocation by between 1.4% and 26.5% across locations, traits, and data sets in terms of mean square error. Although a wide range of performance improvements were observed, our results provide evidence that using IBD for the allocation of lines to locations can help improve predictive performance compared with random allocation. This has the potential to be applied to large-scale plant breeding programs.

Genomic Analysis of Resistance to Fall Armyworm (Spodoptera frugiperda) in CIMMYT Maize Lines

The recent invasion, rapid spread, and widescale destruction of the maize crop by the fall armyworm (FAW; Spodoptera frugiperda (J.E. Smith)) is likely to worsen the food insecurity situation in Africa. In the present study, a set of 424 maize lines were screened for their responses to FAW under artificial infestation to dissect the genetic basis of resistance. All lines were evaluated for two seasons under screen houses and genotyped with the DArTseq platform. Foliar damage was rated on a scale of 1 (highly resistant) to 9 (highly susceptible) and scored at 7, 14, and 21 days after artificial infestation. Analyses of variance revealed significant genotypic and genotype by environment interaction variances for all traits. Heritability estimates for leaf damage scores were moderately high and ranged from 0.38 to 0.58. Grain yield was negatively correlated with a high magnitude to foliar damage scores, ear rot, and ear damage traits. The genome-wide association study (GWAS) revealed 56 significant marker–trait associations and the predicted functions of the putative candidate genes varied from a defense response to several genes of unknown function. Overall, the study revealed that native genetic resistance to FAW is quantitative in nature and is controlled by many loci with minor effects.

Multi-generation genomic prediction of maize yield using parametric and non-parametric sparse selection indices

Genomic prediction models are often calibrated using multi-generation data. Over time, as data accumulates, training data sets become increasingly heterogeneous. Differences in allele frequency and linkage disequilibrium patterns between the training and prediction genotypes may limit prediction accuracy. This leads to the question of whether all available data or a subset of it should be used to calibrate genomic prediction models. Previous research on training set optimization has focused on identifying a subset of the available data that is optimal for a given prediction set. However, this approach does not contemplate the possibility that different training sets may be optimal for different prediction genotypes. To address this problem, we recently introduced a sparse selection index (SSI) that identifies an optimal training set for each individual in a prediction set. Using additive genomic relationships, the SSI can provide increased accuracy relative to genomic-BLUP (GBLUP). Non-parametric genomic models using Gaussian kernels (KBLUP) have, in some cases, yielded higher prediction accuracies than standard additive models. Therefore, here we studied whether combining SSIs and kernel methods could further improve prediction accuracy when training genomic models using multi-generation data. Using four years of doubled haploid maize data from the International Maize and Wheat Improvement Center (CIMMYT), we found that when predicting grain yield the KBLUP outperformed the GBLUP, and that using SSI with additive relationships (GSSI) lead to 5-17% increases in accuracy, relative to the GBLUP. However, differences in prediction accuracy between the KBLUP and the kernel-based SSI were smaller and not always significant.

Genomic Selection in Sugarcane: Current Status and Future Prospects

Sugarcane is a C4 and agro-industry-based crop with a high potential for biomass production. It serves as raw material for the production of sugar, ethanol, and electricity. Modern sugarcane varieties are derived from the interspecific and intergeneric hybridization between Saccharum officinarum, Saccharum spontaneum, and other wild relatives. Sugarcane breeding programmes are broadly categorized into germplasm collection and characterization, pre-breeding and genetic base-broadening, and varietal development programmes. The varietal identification through the classic breeding programme requires a minimum of 12-14 years. The precise phenotyping in sugarcane is extremely tedious due to the high propensity of lodging and suckering owing to the influence of environmental factors and crop management practices. This kind of phenotyping requires data from both plant crop and ratoon experiments conducted over locations and seasons. In this review, we explored the feasibility of genomic selection schemes for various breeding programmes in sugarcane. The genetic diversity analysis using genome-wide markers helps in the formation of core set germplasm representing the total genomic diversity present in the Saccharum gene bank. The genome-wide association studies and genomic prediction in the Saccharum gene bank are helpful to identify the complete genomic resources for cane yield, commercial cane sugar, tolerances to biotic and abiotic stresses, and other agronomic traits. The implementation of genomic selection in pre-breeding, genetic base-broadening programmes assist in precise introgression of specific genes and recurrent selection schemes enhance the higher frequency of favorable alleles in the population with a considerable reduction in breeding cycles and population size. The integration of environmental covariates and genomic prediction in multi-environment trials assists in the prediction of varietal performance for different agro-climatic zones. This review also directed its focus on enhancing the genetic gain over time, cost, and resource allocation at various stages of breeding programmes.

Application of Genomic Selection at the Early Stage of Breeding Pipeline in Tropical Maize

In maize, doubled haploid (DH) line production capacity of large-sized maize breeding programs often exceeds the capacity to phenotypically evaluate the complete set of testcross candidates in multi-location trials. The ability to partially select DH lines based on genotypic data while maintaining or improving genetic gains for key traits using phenotypic selection can result in significant resource savings. The present study aimed to evaluate genomic selection (GS) prediction scenarios for grain yield and agronomic traits of one of the tropical maize breeding pipelines of CIMMYT in eastern Africa, based on multi-year empirical data for designing a GS-based strategy at the early stages of the pipeline. We used field data from 3,068 tropical maize DH lines genotyped using rAmpSeq markers and evaluated as test crosses in well-watered (WW) and water-stress (WS) environments in Kenya from 2017 to 2019. Three prediction schemes were compared: (1) 1 year of performance data to predict a second year; (2) 2 years of pooled data to predict performance in the third year, and (3) using individual or pooled data plus converting a certain proportion of individuals from the testing set (TST) to the training set (TRN) to predict the next year's data. Employing five-fold cross-validation, the mean prediction accuracies for grain yield (GY) varied from 0.19 to 0.29 under WW and 0.22 to 0.31 under WS, when the 1-year datasets were used training set to predict a second year's data as a testing set. The mean prediction accuracies increased to 0.32 under WW and 0.31 under WS when the 2-year datasets were used as a training set to predict the third-year data set. In a forward prediction scenario, good predictive abilities (0.53 to 0.71) were found when the training set consisted of the previous year's breeding data and converting 30% of the next year's data from the testing set to the training set. The prediction accuracy for anthesis date and plant height across WW and WS environments obtained using 1-year data and integrating 10, 30, 50, 70, and 90% of the TST set to TRN set was much higher than those trained in individual years. We demonstrate that by increasing the TRN set to include genotypic and phenotypic data from the previous year and combining only 10-30% of the lines from the year of testing, the predicting accuracy can be increased, which in turn could be used to replace the first stage of field-based screening partially, thus saving significant costs associated with the testcross formation and multi-location testcross evaluation.

Scalable Sparse Testing Genomic Selection Strategy for Early Yield Testing Stage

To enable a scalable sparse testing genomic selection (GS) strategy at preliminary yield trials in the CIMMYT maize breeding program, optimal approaches to incorporate genotype by environment interaction (GEI) in genomic prediction models are explored. Two cross-validation schemes were evaluated: CV1, predicting the genetic merit of new bi-parental populations that have been evaluated in some environments and not others, and CV2, predicting the genetic merit of half of a bi-parental population that has been phenotyped in some environments and not others using the coefficient of determination (CDmean) to determine optimized subsets of a full-sib family to be evaluated in each environment. We report similar prediction accuracies in CV1 and CV2, however, CV2 has an intuitive appeal in that all bi-parental populations have representation across environments, allowing efficient use of information across environments. It is also ideal for building robust historical data because all individuals of a full-sib family have phenotypic data, albeit in different environments. Results show that grouping of environments according to similar growing/management conditions improved prediction accuracy and reduced computational requirements, providing a scalable, parsimonious approach to multi-environmental trials and GS in early testing stages. We further demonstrate that complementing the full-sib calibration set with optimized historical data results in improved prediction accuracy for the cross-validation schemes.

Beat the stress: breeding for climate resilience in maize for the tropical rainfed environments

Intensive public sector breeding efforts and public-private partnerships have led to the increase in genetic gains, and deployment of elite climate-resilient maize cultivars for the stress-prone environments in the tropics. Maize (Zea mays L.) plays a critical role in ensuring food and nutritional security, and livelihoods of millions of resource-constrained smallholders. However, maize yields in the tropical rainfed environments are now increasingly vulnerable to various climate-induced stresses, especially drought, heat, waterlogging, salinity, cold, diseases, and insect pests, which often come in combinations to severely impact maize crops. The International Maize and Wheat Improvement Center (CIMMYT), in partnership with several public and private sector institutions, has been intensively engaged over the last four decades in breeding elite tropical maize germplasm with tolerance to key abiotic and biotic stresses, using an extensive managed stress screening network and on-farm testing system. This has led to the successful development and deployment of an array of elite stress-tolerant maize cultivars across sub-Saharan Africa, Asia, and Latin America. Further increasing genetic gains in the tropical maize breeding programs demands judicious integration of doubled haploidy, high-throughput and precise phenotyping, genomics-assisted breeding, breeding data management, and more effective decision support tools. Multi-institutional efforts, especially public-private alliances, are key to ensure that the improved maize varieties effectively reach the climate-vulnerable farming communities in the tropics, including accelerated replacement of old/obsolete varieties.

Genetic dissection of Striga hermonthica (Del.) Benth. resistance via genome-wide association and genomic prediction in tropical maize germplasm

KEY MESSAGE

Genome-wide association revealed that resistance to Striga hermonthica is influenced by multiple genomic regions with moderate effects. It is possible to increase genetic gains from selection for Striga resistance using genomic prediction. Striga hermonthica (Del.) Benth., commonly known as the purple witchweed or giant witchweed, is a serious problem for maize-dependent smallholder farmers in sub-Saharan Africa. Breeding for Striga resistance in maize is complicated due to limited genetic variation, complexity of resistance and challenges with phenotyping. This study was conducted to (i) evaluate a set of diverse tropical maize lines for their responses to Striga under artificial infestation in three environments in Kenya; (ii) detect quantitative trait loci associated with Striga resistance through genome-wide association study (GWAS); and (iii) evaluate the effectiveness of genomic prediction (GP) of Striga-related traits. An association mapping panel of 380 inbred lines was evaluated in three environments under artificial Striga infestation in replicated trials and genotyped with 278,810 single-nucleotide polymorphism (SNP) markers. Genotypic and genotype x environment variations were significant for measured traits associated with Striga resistance. Heritability estimates were moderate (0.42) to high (0.92) for measured traits. GWAS revealed 57 SNPs significantly associated with Striga resistance indicator traits and grain yield (GY) under artificial Striga infestation with low to moderate effect. A set of 32 candidate genes physically near the significant SNPs with roles in plant defense against biotic stresses were identified. GP with different cross-validations revealed that prediction of performance of lines in new environments is better than prediction of performance of new lines for all traits. Predictions across environments revealed high accuracy for all the traits, while inclusion of GWAS-detected SNPs led to slight increase in the accuracy. The item-based collaborative filtering approach that incorporates related traits evaluated in different environments to predict GY and Striga-related traits outperformed GP for Striga resistance indicator traits. The results demonstrated the polygenic nature of resistance to S. hermonthica, and that implementation of GP in Striga resistance breeding could potentially aid in increasing genetic gain for this important trait.

Genetic Dissection of Quantitative Resistance to Common Rust (Puccinia sorghi) in Tropical Maize (Zea mays L.) by Combined Genome-Wide Association Study, Linkage Mapping, and Genomic Prediction

Common rust is one of the major foliar diseases in maize, leading to significant grain yield losses and poor grain quality. To dissect the genetic architecture of common rust resistance, a genome-wide association study (GWAS) panel and a bi-parental doubled haploid (DH) population, DH1, were used to perform GWAS and linkage mapping analyses. The GWAS results revealed six single-nucleotide polymorphisms (SNPs) significantly associated with quantitative resistance of common rust at a very stringent threshold of P-value 3.70 × 10^-6 at bins 1.05, 1.10, 3.04, 3.05, 4.08, and 10.04. Linkage mapping identified five quantitative trait loci (QTL) at bins 1.03, 2.06, 4.08, 7.03, and 9.00. The phenotypic variation explained (PVE) value of each QTL ranged from 5.40 to 12.45%, accounting for the total PVE value of 40.67%. Joint GWAS and linkage mapping analyses identified a stable genomic region located at bin 4.08. Five significant SNPs were only identified by GWAS, and four QTL were only detected by linkage mapping. The significantly associated SNP of S10_95231291 detected in the GWAS analysis was first reported. The linkage mapping analysis detected two new QTL on chromosomes 7 and 10. The major QTL on chromosome 7 in the region between 144,567,253 and 149,717,562 bp had the largest PVE value of 12.45%. Four candidate genes of GRMZM2G328500, GRMZM2G162250, GRMZM2G114893, and GRMZM2G138949 were identified, which played important roles in the response of stress resilience and the regulation of plant growth and development. Genomic prediction (GP) accuracies observed in the GWAS panel and DH1 population were 0.61 and 0.51, respectively. This study provided new insight into the genetic architecture of quantitative resistance of common rust. In tropical maize, common rust could be improved by pyramiding the new sources of quantitative resistance through marker-assisted selection (MAS) or genomic selection (GS), rather than the implementation of MAS for the single dominant race-specific resistance gene.

Maximizing efficiency of genomic selection in CIMMYT's tropical maize breeding program

Historical data from breeding programs can be efficiently used to improve genomic selection accuracy, especially when the training set is optimized to subset individuals most informative of the target testing set. The current strategy for large-scale implementation of genomic selection (GS) at the International Maize and Wheat Improvement Center (CIMMYT) global maize breeding program has been to train models using information from full-sibs in a "test-half-predict-half approach." Although effective, this approach has limitations, as it requires large full-sib populations and limits the ability to shorten variety testing and breeding cycle times. The primary objective of this study was to identify optimal experimental and training set designs to maximize prediction accuracy of GS in CIMMYT's maize breeding programs. Training set (TS) design strategies were evaluated to determine the most efficient use of phenotypic data collected on relatives for genomic prediction (GP) using datasets containing 849 (DS1) and 1389 (DS2) DH-lines evaluated as testcrosses in 2017 and 2018, respectively. Our results show there is merit in the use of multiple bi-parental populations as TS when selected using algorithms to maximize relatedness between the training and prediction sets. In a breeding program where relevant past breeding information is not readily available, the phenotyping expenditure can be spread across connected bi-parental populations by phenotyping only a small number of lines from each population. This significantly improves prediction accuracy compared to within-population prediction, especially when the TS for within full-sib prediction is small. Finally, we demonstrate that prediction accuracy in either sparse testing or "test-half-predict-half" can further be improved by optimizing which lines are planted for phenotyping and which lines are to be only genotyped for advancement based on GP.

Genetic Dissection of Grain Yield of Maize and Yield-Related Traits Through Association Mapping and Genomic Prediction

High yield is the primary objective of maize breeding. Genomic dissection of grain yield and yield-related traits contribute to understanding the yield formation and improving the yield of maize. In this study, two genome-wide association study (GWAS) methods and genomic prediction were made on an association panel of 309 inbred lines. GWAS analyses revealed 22 significant trait-marker associations for grain yield per plant (GYP) and yield-related traits. Genomic prediction analyses showed that reproducing kernel Hilbert space (RKHS) outperformed the other four models based on GWAS-derived markers for GYP, ear weight, kernel number per ear and row, ear length, and ear diameter, whereas genomic best linear unbiased prediction (GBLUP) showed a slight superiority over other modes in most subsets of the trait-associated marker (TAM) for thousand kernel weight and kernel row number. The prediction accuracy could be improved when significant single-nucleotide polymorphisms were fitted as the fixed effects. Integrating information on population structure into the fixed model did not improve the prediction performance. For GYP, the prediction accuracy of TAMs derived from fixed and random model Circulating Probability Unification (FarmCPU) was comparable to that of the compressed mixed linear model (CMLM). For yield-related traits, CMLM-derived markers provided better accuracies than FarmCPU-derived markers in most scenarios. Compared with all markers, TAMs could effectively improve the prediction accuracies for GYP and yield-related traits. For eight traits, moderate- and high-prediction accuracies were achieved using TAMs. Taken together, genomic prediction incorporating prior information detected by GWAS could be a promising strategy to improve the grain yield of maize.

Combining Genomic and Phenomic Information for Predicting Grain Protein Content and Grain Yield in Spring Wheat

Genomics and high throughput phenomics have the potential to revolutionize the field of wheat (Triticum aestivum L.) breeding. Genomic selection (GS) has been used for predicting various quantitative traits in wheat, especially grain yield. However, there are few GS studies for grain protein content (GPC), which is a crucial quality determinant. Incorporation of secondary correlated traits in GS models has been demonstrated to improve accuracy. The objectives of this research were to compare performance of single and multi-trait GS models for predicting GPC and grain yield in wheat and to identify optimal growth stages for collecting secondary traits. We used 650 recombinant inbred lines from a spring wheat nested association mapping (NAM) population. The population was phenotyped over 3 years (2014-2016), and spectral information was collected at heading and grain filling stages. The ability to predict GPC and grain yield was assessed using secondary traits, univariate, covariate, and multivariate GS models for within and across cycle predictions. Our results indicate that GS accuracy increased by an average of 12% for GPC and 20% for grain yield by including secondary traits in the models. Spectral information collected at heading was superior for predicting GPC, whereas grain yield was more accurately predicted during the grain filling stage. Green normalized difference vegetation index had the largest effect on the prediction of GPC either used individually or with multiple indices in the GS models. An increased prediction ability for GPC and grain yield with the inclusion of secondary traits demonstrates the potential to improve the genetic gain per unit time and cost in wheat breeding.

Genetic Dissection of Resistance to Gray Leaf Spot by Combining Genome-Wide Association, Linkage Mapping, and Genomic Prediction in Tropical Maize Germplasm

Gray leaf spot (GLS) is one of the major maize foliar diseases in sub-Saharan Africa. Resistance to GLS is controlled by multiple genes with additive effect and is influenced by both genotype and environment. The objectives of the study were to dissect the genetic architecture of GLS resistance through linkage mapping and genome-wide association study (GWAS) and assessing the potential of genomic prediction (GP). We used both biparental populations and an association mapping panel of 410 diverse tropical/subtropical inbred lines that were genotyped using genotype by sequencing. Phenotypic evaluation in two to four environments revealed significant genotypic variation and moderate to high heritability estimates ranging from 0.43 to 0.69. GLS was negatively and significantly correlated with grain yield, anthesis date, and plant height. Linkage mapping in five populations revealed 22 quantitative trait loci (QTLs) for GLS resistance. A QTL on chromosome 7 (qGLS7-105) is a major-effect QTL that explained 28.2% of phenotypic variance. Together, all the detected QTLs explained 10.50, 49.70, 23.67, 18.05, and 28.71% of phenotypic variance in doubled haploid (DH) populations 1, 2, 3, and F3 populations 4 and 5, respectively. Joint linkage association mapping across three DH populations detected 14 QTLs that individually explained 0.10-15.7% of phenotypic variance. GWAS revealed 10 significantly (p < 9.5 × 10-6) associated SNPs distributed on chromosomes 1, 2, 6, 7, and 8, which individually explained 6-8% of phenotypic variance. A set of nine candidate genes co-located or in physical proximity to the significant SNPs with roles in plant defense against pathogens were identified. GP revealed low to moderate prediction correlations of 0.39, 0.37, 0.56, 0.30, 0.29, and 0.38 for within IMAS association panel, DH pop1, DH pop2, DH pop3, F3 pop4, and F3 po5, respectively, and accuracy was increased substantially to 0.84 for prediction across three DH populations. When the diversity panel was used as training set to predict the accuracy of GLS resistance in biparental population, there was 20-50% reduction compared to prediction within populations. Overall, the study revealed that resistance to GLS is quantitative in nature and is controlled by many loci with a few major and many minor effects. The SNPs/QTLs identified by GWAS and linkage mapping can be potential targets in improving GLS resistance in breeding programs, while GP further consolidates the development of high GLS-resistant lines by incorporating most of the major- and minor-effect genes.

Genomic prediction across years in a maize doubled haploid breeding program to accelerate early-stage testcross testing

Genomic selection with a multiple-year training population dataset could accelerate early-stage testcross testing by skipping the first-stage yield testing, which significantly saves the time and cost of early-stage testcross testing. With the development of doubled haploid (DH) technology, the main task for a maize breeder is to estimate the breeding values of thousands of DH lines annually. In early-stage testcross testing, genomic selection (GS) offers the opportunity of replacing expensive multiple-environment phenotyping and phenotypic selection with lower-cost genotyping and genomic estimated breeding value (GEBV)-based selection. In the present study, a total of 1528 maize DH lines, phenotyped in multiple-environment trials in three consecutive years and genotyped with a low-cost per-sample genotyping platform of rAmpSeq, were used to explore how to implement GS to accelerate early-stage testcross testing. Results showed that the average prediction accuracy estimated from the cross-validation schemes was above 0.60 across all the scenarios. The average prediction accuracies estimated from the independent validation schemes ranged from 0.23 to 0.32 across all the scenarios, when the one-year datasets were used as training population (TRN) to predict the other year data as testing population (TST). The average prediction accuracies increased to a range from 0.31 to 0.42 across all the scenarios, when the two-years datasets were used as TRN. The prediction accuracies increased to a range from 0.50 to 0.56, when the TRN consisted of two-years of breeding data and 50% of third year's data converted from TST to TRN. This information showed that GS with a multiple-year TRN set offers the opportunity to accelerate early-stage testcross testing by skipping the first-stage yield testing, which significantly saves the time and cost of early-stage testcross testing.

Combination of Linkage Mapping, GWAS, and GP to Dissect the Genetic Basis of Common Rust Resistance in Tropical Maize Germplasm

Common rust (CR) caused by Puccina sorghi is one of the destructive fungal foliar diseases of maize and has been reported to cause moderate to high yield losses. Providing CR resistant germplasm has the potential to increase yields. To dissect the genetic architecture of CR resistance in maize, association mapping, in conjunction with linkage mapping, joint linkage association mapping (JLAM), and genomic prediction (GP) was conducted on an association-mapping panel and five F₃ biparental populations using genotyping-by-sequencing (GBS) single-nucleotide polymorphisms (SNPs). Analysis of variance for the biparental populations and the association panel showed significant genotypic and genotype x environment (GXE) interaction variances except for GXE of Pop4. Heritability (h²) estimates were moderate with 0.37-0.45 for the individual F₃ populations, 0.45 across five populations and 0.65 for the association panel. Genome-wide association study (GWAS) analyses revealed 14 significant marker-trait associations which individually explained 6-10% of the total phenotypic variances. Individual population-based linkage analysis revealed 26 QTLs associated with CR resistance and together explained 14-40% of the total phenotypic variances. Linkage mapping revealed seven QTLs in pop1, nine QTL in pop2, four QTL in pop3, five QTL in pop4, and one QTL in pop5, distributed on all chromosomes except chromosome 10. JLAM for the 921 F3 families from five populations detected 18 QTLs distributed in all chromosomes except on chromosome 8. These QTLs individually explained 0.3 to 3.1% and together explained 45% of the total phenotypic variance. Among the 18 QTL detected through JLAM, six QTLs, qCR1-78, qCR1-227, qCR3-172, qCR3-186, qCR4-171, and qCR7-137 were also detected in linkage mapping. GP within population revealed low to moderate correlations with a range from 0.19 to 0.51. Prediction correlation was high with r = 0.78 for combined analysis of the five F₃ populations. Prediction of biparental populations by using association panel as training set reveals positive correlations ranging from 0.05 to 0.22, which encourages to develop an independent but related population as a training set which can be used to predict diverse but related populations. The findings of this study provide valuable information on understanding the genetic basis of CR resistance and the obtained information can be used for developing functional molecular markers for marker-assisted selection and for implementing GP to improve CR resistance in tropical maize.

Training Population Optimization for Genomic Selection in Miscanthus

Miscanthus is a perennial grass with potential for lignocellulosic ethanol production. To ensure its utility for this purpose, breeding efforts should focus on increasing genetic diversity of the nothospecies Miscanthus × giganteus (M×g) beyond the single clone used in many programs. Germplasm from the corresponding parental species M. sinensis (Msi) and M. sacchariflorus (Msa) could theoretically be used as training sets for genomic prediction of M×g clones with optimal genomic estimated breeding values for biofuel traits. To this end, we first showed that subpopulation structure makes a substantial contribution to the genomic selection (GS) prediction accuracies within a 538-member diversity panel of predominately Msi individuals and a 598-member diversity panels of Msa individuals. We then assessed the ability of these two diversity panels to train GS models that predict breeding values in an interspecific diploid 216-member M×g F2 panel. Low and negative prediction accuracies were observed when various subsets of the two diversity panels were used to train these GS models. To overcome the drawback of having only one interspecific M×g F2 panel available, we also evaluated prediction accuracies for traits simulated in 50 simulated interspecific M×g F2 panels derived from different sets of Msi and diploid Msa parents. The results revealed that genetic architectures with common causal mutations across Msi and Msa yielded the highest prediction accuracies. Ultimately, these results suggest that the ideal training set should contain the same causal mutations segregating within interspecific M×g populations, and thus efforts should be undertaken to ensure that individuals in the training and validation sets are as closely related as possible.

Development and Proof-of-Concept Application of Genome-Enabled Selection for Pea Grain Yield under Severe Terminal Drought

Terminal drought is the main stress limiting pea (Pisum sativum L.) grain yield in Mediterranean environments. This study aimed to investigate genotype × environment (GE) interaction patterns, define a genomic selection (GS) model for yield under severe drought based on single nucleotide polymorphism (SNP) markers from genotyping-by-sequencing, and compare GS with phenotypic selection (PS) and marker-assisted selection (MAS). Some 288 lines belonging to three connected RIL populations were evaluated in a managed-stress (MS) environment of Northern Italy, Marchouch (Morocco), and Alger (Algeria). Intra-environment, cross-environment, and cross-population predictive ability were assessed by Ridge Regression best linear unbiased prediction (rrBLUP) and Bayesian Lasso models. GE interaction was particularly large across moderate-stress and severe-stress environments. In proof-of-concept experiments performed in a MS environment, GS models constructed from MS environment and Marchouch data applied to independent material separated top-performing lines from mid- and bottom-performing ones, and produced actual yield gains similar to PS. The latter result would imply somewhat greater GS efficiency when considering same selection costs, in partial agreement with predicted efficiency results. GS, which exploited drought escape and intrinsic drought tolerance, exhibited 18% greater selection efficiency than MAS (albeit with non-significant difference between selections) and moderate to high cross-population predictive ability. GS can be cost-efficient to raise yields under severe drought.

Strategies for Effective Use of Genomic Information in Crop Breeding Programs Serving Africa and South Asia

Much of the world's population growth will occur in regions where food insecurity is prevalent, with large increases in food demand projected in regions of Africa and South Asia. While improving food security in these regions will require a multi-faceted approach, improved performance of crop varieties in these regions will play a critical role. Current rates of genetic gain in breeding programs serving Africa and South Asia fall below rates achieved in other regions of the world. Given resource constraints, increased genetic gain in these regions cannot be achieved by simply expanding the size of breeding programs. New approaches to breeding are required. The Genomic Open-source Breeding informatics initiative (GOBii) and Excellence in Breeding Platform (EiB) are working with public sector breeding programs to build capacity, develop breeding strategies, and build breeding informatics capabilities to enable routine use of new technologies that can improve the efficiency of breeding programs and increase genetic gains. Simulations evaluating breeding strategies indicate cost-effective implementations of genomic selection (GS) are feasible using relatively small training sets, and proof-of-concept implementations have been validated in the International Maize and Wheat Improvement Center (CIMMYT) maize breeding program. Progress on GOBii, EiB, and implementation of GS in CIMMYT and International Crops Research Institute for the Semi-Arid Tropics (ICRISAT) breeding programs are discussed, as well as strategies for routine implementation of GS in breeding programs serving Africa and South Asia.

Genetic Dissection of Nitrogen Use Efficiency in Tropical Maize Through Genome-Wide Association and Genomic Prediction

In sub-Saharan Africa, one of the major challenges to smallholder farmers is soil with low fertility and inability to apply nitrogen fertilizer externally due to the cost. Development of maize hybrids, which perform better in nitrogen depleted soils, is one of the promising solutions. However, breeding maize for nitrogen use efficiency (NUE) is hindered by expensive phenotypic evaluations and trait complexity under low N stress. Genome-wide association study (GWAS) and genomic prediction (GP) are promising tools to circumvent this interference. Here, we evaluated a mapping panel in diverse environments both under optimum and low N management. The objective of this study was to identify SNPs significantly associated with grain yield (GY) and other traits through GWAS and assess the potential of GP under low N and optimum conditions. Testcross progenies of 411 inbred lines were planted under optimum and low N conditions in several locations in Africa and Latin America. In all locations, low N fields were previously depleted over several seasons, and no N fertilizer was applied throughout the growing season. All inbred lines were genotyped with genotyping by sequencing. Genotypic and GxE interaction variances were significant, and heritability estimates were moderate to high for all traits under both optimum and low N conditions. Genome-wide LD decay at r ² = 0.2 and r ² = 0.34 were 0.24 and 0.19 Mbp, respectively. Chromosome-specific LD decays ranged from 0.13 to 0.34 Mbps with an average of 0.22 Mbp at r ² = 0.2. GWAS analyses revealed 38 and 45 significant SNPs under optimum and low N conditions, respectively. Out of these 83 significant SNPs, 3 SNPs on chromosomes 1, 2, and 6 were associated either with different traits or the same trait under different management conditions, suggesting pleiotropic effects of genes. A total of 136 putative candidate genes were associated with the significant SNPs, of which seven SNPs were linked with four known genes. Prediction accuracies were moderate to high for all traits under both optimum and low N conditions. These results can be used as useful resources for further applications to develop hybrids or lines with better performance under low N conditions.