Genetics and Plant Breeding ISBN No.

- 978-93-48217-69-1
for Crop Improvement Textify Publishers

Development of
Dissects complex
Integrates genomics trait-specific
traits, accelerates
Advantages with breeding markers for early
breeding cycles,
strategy generation
enhances precision
selection

Genomic Selection and Prediction of Quantitative Traits

A. Genomic Selection
Genomic selection (GS) is a modern breeding approach that uses genome-wide
molecular markers to predict the breeding values of individuals for complex
quantitative traits. Unlike traditional marker-assisted selection, which relies on
identifying and using significant QTLs, GS considers the cumulative effect of
thousands of markers across the genome, whether or not they are statistically
significant. The concept was introduced by Meuwissen, Hayes, and Goddard in
2001 as a strategy to overcome the limitations of QTL mapping for polygenic traits.
GS uses dense marker data to estimate genomic estimated breeding values
(GEBVs), which are then used to select individuals with superior performance
before phenotyping. GS is especially useful for traits with low heritability, traits
measured late in the crop cycle, or traits expensive to phenotype. It reduces
dependency on multi-season phenotyping and accelerates selection cycles.
B. Statistical Models Used (e.g., GBLUP, Bayesian Approaches)
Several statistical models have been developed to perform genomic prediction.
These models use training population data comprising genotype and phenotype to
train algorithms that can predict the genetic merit of untested individuals.
1. GBLUP (Genomic Best Linear Unbiased Prediction)
GBLUP is one of the most widely used models. It assumes equal variance across
markers and estimates breeding values by fitting a genomic relationship matrix.

Page | 59
Genetics and Plant Breeding ISBN No.- 978-93-48217-69-1
for Crop Improvement Textify Publishers

GBLUP performs well for traits governed by many small-effect loci and is
computationally efficient for large datasets.
2. Bayesian approaches (BayesA, BayesB, BayesC, Bayesian LASSO)
These models incorporate different assumptions about marker effect distributions.
 BayesA assumes all markers have non-zero effects with unique variances.
 BayesB assumes that only a subset of markers have effects, useful when
some major QTLs exist.
 Bayesian LASSO uses shrinkage techniques to penalize less-informative
markers.
Bayesian models often outperform GBLUP for traits with a few large-effect QTLs
and many minor-effect loci, such as disease resistance or flowering time.
3. Machine learning models
Advanced methods such as Random Forests, Support Vector Machines, and deep
learning neural networks are being explored for genomic prediction. These models
can capture complex, nonlinear relationships but require careful tuning and large
datasets.
C. Benefits Over Marker-Assisted Selection
Genomic selection offers several advantages when compared to marker-assisted
selection (MAS), particularly for complex traits controlled by many loci with small
effects.
1. Prediction without identifying QTLs
GS does not require prior identification of significant QTLs. Instead, it uses the
combined effect of all markers to estimate breeding values. This avoids the
limitations of QTL detection thresholds and missed minor-effect loci.
2. Greater accuracy for polygenic traits
GS has higher predictive accuracy for traits such as grain yield, biomass, and
drought tolerance, which are influenced by multiple genes and environmental
interactions.
3. Early selection and reduced breeding cycle time
By enabling selection based on genotypic data alone, GS allows breeders to make
selection decisions early, even before phenotypes are expressed. This is particularly
useful for perennial crops, long-duration cereals, or traits measured at maturity.

Page | 60
Genetics and Plant Breeding ISBN No.- 978-93-48217-69-1
for Crop Improvement Textify Publishers

4. Applicability to low-heritability traits

For traits with high environmental influence and low heritability, traditional
selection methods are inefficient. GS enhances selection response in such cases by
capturing genetic signal masked by environmental noise.
D. Role in Accelerating Breeding Cycles
Reducing the breeding cycle time is critical for improving the rate of genetic gain
per unit time. GS facilitates this through early-generation selection and elimination
of the need for extensive phenotyping in each generation.
1. Rapid generation advancement
By selecting individuals based on GEBVs in early generations (e.g., F2 or F3),
breeders can advance lines quickly and reduce time per selection cycle.
2. Integration with doubled haploid (DH) technology
In maize and wheat, GS is combined with doubled haploids to produce genetically
fixed lines in a single generation. This accelerates the development of pure lines and
their evaluation for hybrid production.
3. Increased selection intensity
With high-throughput genotyping and GS models, breeders can evaluate large
populations at reduced cost and select the top-performing individuals with high
precision.
4. Genomic recurrent selection
In crops like maize, genomic recurrent selection involves repeated cycles of
prediction and recombination. Studies have shown that this approach can double the
rate of gain compared to traditional recurrent selection.
E. Case Studies in Major Crops
1. Maize (Zea mays)
Genomic selection has been widely adopted in maize for improving yield under
both optimal and stress conditions. Studies using GBLUP and Bayesian models
reported predictive accuracies ranging from 0.4 to 0.8 depending on trait and
environment. Commercial programs have used GS to accelerate hybrid
development and improve drought tolerance.
2. Rice (Oryza sativa)
GS is applied for traits such as yield, flowering time, and grain quality. The
International Rice Research Institute (IRRI) developed GS models that achieved

Page | 61
Genetics and Plant Breeding ISBN No.- 978-93-48217-69-1
for Crop Improvement Textify Publishers

prediction accuracies of 0.6–0.75 for traits like panicle number and plant height. GS
is being implemented in rapid cycling schemes using early-generation prediction.
3. Wheat (Triticum aestivum)
GS has been applied to improve traits including grain yield, disease resistance, and
quality parameters such as grain protein content. The CIMMYT wheat program
reported 25–35% increase in selection gain using GS compared to traditional
selection. Multi-environment GS models incorporating G×E interaction have further
improved prediction accuracies.
4. Soybean (Glycine max)
In soybean, GS has been used to predict seed oil and protein content, flowering
time, and resistance to soybean cyst nematode. Accuracy estimates ranged from
0.45 to 0.70 depending on the trait. Early-cycle genomic prediction has accelerated
the release of superior breeding lines.
5. Barley and chickpea
GS models in barley have shown high accuracy for malting quality traits and
disease resistance. In chickpea, GS has been integrated with speed breeding and
trait-specific genotyping to shorten cycle time and enhance resistance to drought
and Fusarium wilt.
Applications in Crop Improvement Programs
A. Trait-Based Breeding Approaches
Trait-based breeding focuses on selecting for specific physiological, morphological,
or molecular traits that are directly associated with performance or stress adaptation.
This strategy complements traditional yield-based selection and enhances genetic
gain through more targeted decisions. Trait-based breeding relies on well-
characterized genetic variation and a strong understanding of trait heritability and
correlation with target outcomes. For example, selecting for canopy temperature
depression and stomatal conductance in wheat has led to improved drought
resilience. In rice, breeding lines with increased harvest index and deeper roots have
been prioritized for water-limited environments.

 Selection indices
Trait-based selection often uses indices that combine multiple traits, weighted by
their economic importance or genetic correlation with yield. These indices help
prioritize genotypes that offer balanced performance across traits like maturity,
plant height, disease resistance, and abiotic stress tolerance.
Examples

Page | 62
Genetics and Plant Breeding ISBN No.- 978-93-48217-69-1
for Crop Improvement Textify Publishers

In maize, trait-based breeding has improved stay-green characteristics under

terminal drought. In chickpea, genotypes with early vigor, compact canopy, and
rapid pod filling have shown better yield stability in dry conditions.
B. Integration of Phenotyping and Genotyping
The integration of high-resolution phenotyping with dense genotypic data has
transformed modern crop breeding. This combination supports precise selection,
rapid trait discovery, and improved prediction accuracy.
1. Phenotyping platforms
Digital tools, such as UAVs with multispectral cameras, thermal sensors, and
LiDAR, provide rapid assessment of biomass, chlorophyll content, canopy
temperature, and plant architecture. These phenotypic traits serve as proxies for
physiological performance, enabling selection without waiting for yield data.
2. Genotyping platforms
High-throughput genotyping platforms such as Genotyping-by-Sequencing (GBS),
SNP arrays, and DArTseq enable genome-wide marker coverage at low cost.
Marker data from thousands of loci provide insights into genetic diversity,
population structure, and loci associated with traits.
3. Data integration and analysis
Advanced software and machine learning tools analyze large-scale phenotypic and
genotypic datasets (Bayat et.al., 2020). Integration models such as genomic
prediction (using GBLUP or Bayesian methods) and multi-trait models allow
simultaneous selection for multiple complex traits. In sorghum, combining
phenotypic data from managed drought trials with genome-wide SNP markers has
led to identification of superior lines with stable yield.
C. Breeding for Complex Traits (e.g., Yield Under Drought)
Complex traits such as yield under drought are influenced by many genes, each with
small effect, and are highly sensitive to environmental variation. These traits require
specialized breeding strategies that combine field selection, physiological
understanding, and molecular tools.
1. Component trait dissection
Dissecting yield into its components such as tiller number, grain size, biomass, and
harvest index enables indirect selection for contributing traits with higher
heritability. For example, in pearl millet, selecting for panicle length and grain
filling rate has improved yield under water stress.

Page | 63
Genetics and Plant Breeding ISBN No.- 978-93-48217-69-1
for Crop Improvement Textify Publishers

2. Marker-assisted and genomic selection

Marker-assisted backcrossing (MABC) has been used to in Progress major QTLs
such as qDTY12.1 (drought yield in rice) into elite backgrounds. Genomic selection
accelerates improvement by predicting yield performance based on genome-wide
markers, without direct measurement of yield under every environment.
3. Managed stress environments
Breeding for drought tolerance often involves evaluation under managed stress
environments that simulate realistic field conditions. This improves selection
accuracy and ensures trait expression is consistent with target production areas. In
wheat, managed terminal stress trials have identified lines with sustained
photosynthesis and better grain filling under heat and water deficit.
D. Role in Hybrid and Synthetic Variety Development
Quantitative genetics plays a critical role in the development of hybrids and
synthetic varieties by exploiting additive and non-additive gene action, particularly
heterosis and combining ability.
1. Hybrid breeding
In maize and pearl millet, hybrid breeding has led to significant yield gains through
exploitation of heterosis. General combining ability (GCA) and specific combining
ability (SCA) are estimated from diallel or line × tester crosses to select parental
combinations. Hybrids developed using genomic selection and trait-based indices
have shown 15–25% higher yield over conventionally bred varieties.
2. Synthetic varieties
In cross-pollinated crops like alfalfa and ryegrass, synthetic varieties are developed
by intercrossing multiple selected parents with desirable traits and then allowing
open pollination. These varieties provide yield stability and adaptability over
seasons.
3. Heterotic grouping and prediction
Use of SNP-based genetic distance and genomic prediction models aids in
identifying heterotic groups and predicting hybrid performance. In sorghum,
genomic prediction has accurately forecasted hybrid vigor based on parental
genotypic profiles, reducing time and cost in hybrid development.
E. Participatory and Farmer-Centric Approaches
Participatory breeding involves collaboration with farmers during the selection and
evaluation process, ensuring the developed varieties align with local preferences,
practices, and constraints.

Page | 64
Genetics and Plant Breeding ISBN No.- 978-93-48217-69-1
for Crop Improvement Textify Publishers

1. On-farm selection
Farmers evaluate breeding lines under their own management systems, which helps
in identifying genotypes suited to real-world conditions. In rice, participatory
varietal selection has led to higher adoption rates of stress-tolerant varieties such as
Sahbhagi Dhan and DRR Dhan 42.
2. Trait prioritization by farmers
Farmers often prioritize traits like grain quality, maturity duration, tillering ability,
and straw yield, which may not be emphasized in station-based trials. Integrating
farmer feedback helps refine selection indices and breeding goals.
3. Gender-responsive breeding
Participatory programs often include gender-focused evaluation to account for the
distinct preferences and roles of women and men in farming systems. In legumes,
women’s preference for ease of cooking, seed size, and color has influenced
breeding targets in common bean and cowpea.
4. Scaling through community seed systems
Linking participatory breeding with community seed production accelerates
dissemination and uptake. Farmers trained in seed selection and quality control help
maintain the purity and availability of improved varieties.
Future and Emerging Technologies
A. Integration of Omics (Transcriptomics, Metabolomics)
The integration of omics technologies transcriptomics, metabolomics, proteomics,
and epigenomics is revolutionizing the understanding of complex traits at multiple
biological levels. These platforms complement traditional genomics and offer
deeper insights into gene function, regulation, and trait expression under various
conditions.
1. Transcriptomics
Transcriptomics refers to the study of RNA transcripts to assess gene expression
patterns under specific conditions. RNA sequencing (RNA-Seq) provides
quantitative data on the expression levels of thousands of genes simultaneously. In
maize, transcriptomic profiling under drought stress revealed differential expression
of over 3,000 genes associated with ABA signaling, cell wall modification, and
osmotic regulation. These candidate genes are now targets for validation and
editing. In rice, transcriptome studies during reproductive stage drought have
identified key regulators such as DREB, NAC, and bZIP transcription factors, which
coordinate stress response pathways.

Page | 65
Genetics and Plant Breeding ISBN No.- 978-93-48217-69-1
for Crop Improvement Textify Publishers

2. Metabolomics
Metabolomics involves the profiling of primary and secondary metabolites that
reflect biochemical activity. It helps link genotype to phenotype through metabolic
pathways. In wheat, comparative metabolomics under heat stress identified
accumulation of protective compounds such as proline, glycine betaine, and
trehalose. These metabolites correlate with stay-green phenotypes and grain filling
duration. Metabolomic markers can be used as indirect selection tools when they
exhibit stable associations with yield or stress tolerance across environments.
3. Multi-omics integration
Combining transcriptomics, metabolomics, and proteomics allows researchers to
map regulatory networks, identify biomarkers, and validate functional gene-trait
relationships. These approaches enhance the discovery of trait-specific targets for
genomic selection and gene editing.
B. Artificial Intelligence and Data-Driven Breeding
Artificial Intelligence (AI), including machine learning and deep learning, is
emerging as a powerful tool for managing, analyzing, and predicting breeding
outcomes from large and complex datasets.
1. Machine learning in genomic prediction
Machine learning algorithms such as Random Forests, Gradient Boosting Machines,
and Support Vector Machines can capture nonlinear marker-trait associations,
epistasis, and G×E interactions. These models often outperform traditional GBLUP
in prediction accuracy, particularly for low-heritability traits. In sorghum, machine
learning models have been used to predict biomass and flowering time with over
20% improvement in accuracy compared to linear models. In chickpea, AI has
enabled the identification of drought-tolerant lines using image-based phenotyping
and genotype data.
2. Computer vision in phenotyping
Convolutional Neural Networks (CNNs) are used in image classification for disease
scoring, seed counting, and canopy structure analysis. UAV imagery analyzed
through AI can classify plant health, vigor, and nutrient status without manual
scoring.
3. Automated decision support systems
AI-based systems integrate historical performance data, weather forecasts, and
genotype data to recommend optimal crosses, select breeding lines, or manage trial
logistics. These tools are being tested in rice and wheat breeding platforms for real-
time data interpretation.

Page | 66
Genetics and Plant Breeding ISBN No.- 978-93-48217-69-1
for Crop Improvement Textify Publishers

C. Climate-Smart Trait Prediction

Breeding programs are increasingly focused on climate-resilient traits such as heat
tolerance, water-use efficiency, and carbon assimilation under elevated CO₂ (Kundu
et.al., 2023). Predictive tools and environment-specific trait modeling are essential
for sustaining yield gains under climate variability.
1. Trait prioritization under climate risk
Crop simulation models such as APSIM and DSSAT are used to simulate genotype
performance under future climate scenarios. These models integrate physiological
parameters with climate data to identify key adaptive traits. In maize, traits like
stay-green, anthesis-silking interval (ASI), and root angle have been modeled for
yield advantage under projected temperature rise and erratic rainfall patterns.
2. Climate-resilient genomic selection
Genomic prediction models incorporating weather covariates and genotype-specific
climate responses are now being developed. These models enhance prediction
accuracy for traits sensitive to environmental fluctuation, enabling climate-smart
cultivar deployment. In wheat, genomic × environment models have successfully
identified genotypes with superior yield stability under high-temperature stress
conditions using 20+ year historical climate data.
D. Phenomic and Genomic Data Integration
The convergence of phenomics and genomics is foundational to next-generation
breeding. High-throughput phenotyping (HTP) platforms produce massive datasets
capturing dynamic plant responses, while genomic data provides a blueprint of
underlying genetic potential.
1. Phenomic data layers
Phenomic data includes time-series images, sensor-based canopy traits, thermal
response, and spectral reflectance indices. These are collected using drones, fixed
platforms, or ground robots.
2. Genomic integration
Genomic selection pipelines incorporate phenomic traits as input variables or as
target traits. In maize, vegetative indices (e.g., NDVI, canopy temperature) collected
at flowering stage have been used to predict grain yield with high correlation to
actual performance.

Page | 67