**4. Genomic selection**

The GS represents an advanced iteration of marker-assisted selection, enhancing the efficiency of selection and expediting the progress in selective breeding within a shorter timeframe. It achieves this by employing markers across the entire genome to predict the impact of quantitative gene loci, subsequently calculating genomic estimated breeding values (GEBVs) [58, 59]. The GS breeding approach was initially proposed by Meuwissen et al. [60]. GS can swiftly and cost-effectively forecast the yield potential of individual plants, ultimately leading to a reduction in both the time and expenses associated with a breeding cycle. In contrast to GWAS (genome-wide association studies) and linkage analysis, the primary goal of GS is not to pinpoint specific QTLs but rather to make predictions performance of offspring based on the DNA information gathered in the present. In GS, breeders can predict a plant's breeding value by utilizing data from all markers without the need for direct phenotype evaluation. This prediction relies on statistical models developed using a "training population" where both genotypes and phenotypes have been recorded. In mixed model analysis for genomic selection, markers are treated as random factors. This approach is necessary because the number of markers often exceeds the number of individuals in the training population, making it impractical to estimate the effect of each marker individually due to limited degrees of freedom [35, 61]. MAS proves valuable when choosing traits such as grain yields, flower colors, seed characteristics, and others that manifest primarily during the later reproductive stages. Through the application of GS, these traits can be detected by employing DNA markers in a genotype even at the preliminary stages of plant development as highlighted by Madhusudhana [62]. It offers numerous advantages compared to traditional MAS. Notably, it does not require QTL mapping, as it efficiently estimates breeding values using a comprehensive set of molecular markers that ideally spans the whole genome [63]; at early selection, it is more precise as it estimates all QTLs effects by utilizing high-density molecular markers and explains genetic variance for desirable traits. This stands in contrast to MAS, which relies on a limited number of markers for trait selection [64]; it shortens generation intervals, accelerates genetic progress (4–25% farther up phenotypic selection), and reduces breeding costs (26–56% lesser traditional methods) [65]; it exhibits superior efficiency in selecting traits with low heritability compared to MAS; in GS, breeding values serve as the selection criteria which are the sum of all allele genetic

*Enhancing Maize (*Zea mays *L.) Crop through Advanced Techniques: A Comprehensive Approach DOI: http://dx.doi.org/10.5772/intechopen.114029*

effects for each individual. This approach is known for its superior accuracy because it assesses the average performance of the offspring, rather than relying solely on the parents' performance [66]. In crops like maize, research conducted by the CIMMYT (International Maize and Wheat Improvement Center) [67], suggests that the breeding interval could be shortened to as little as half of the conventional timeframe. GS has been effectively applied to enhance various traits in maize, including shelling percentage, grain yield, grain moisture, ear length, ear width, ear rows, tassel branch number, kernels per spike, hundred-grain weight, kernel number per ear, and kernel depth [68, 69]. Challenges in GS arise from factors including the training population's size and variability, as well as the heritability of the traits being predicted. The statistical intricacies in GS are connected to the vast amount of marker data, where the number of markers far exceeds the number of observations [33]. In conclusion, GS has proven to be a valuable tool for the improvement of multiple important traits in maize, offering promising prospects for enhancing the overall performance and quality of maize crops.
