Molecular Plant Breeding 2024, Vol.15, No.1, 15-26 http://genbreedpublisher.com/index.php/mpb 19 These studies represent a sample of innovative research in the field of genomic prediction and selection in plant breeding, showcasing various approaches and their impact on improving breeding programs. 3.2 Application of high-throughput phenotyping In Breeding 4.0, there have been significant advancements in the application of high-throughput phenotyping. High-throughput phenotyping utilizes advanced sensor technology, imaging technology, and automated systems to efficiently measure multiple traits of crops. Through large-scale, high-resolution trait data, breeders can gain a more accurate understanding of the phenotypic characteristics of crops, guiding breeding decisions and strategy formulation. The application of high-throughput phenotyping makes crop assessment more comprehensive and accurate, accelerating the screening and promotion of superior varieties. Recent research indicated that in the field of plant breeding, High-throughput phenotyping (HTP) technology is rapidly advancing, particularly in the high-throughput phenotyping of canopy-based traits for major crops in field environments. Here are some key findings: Kuroki et al. (2022) developed a ground-based high-throughput phenotyping rover for use in size-limited breeding fields. This rover is suitable for small-scale breeding fields commonly found in Japan and other Asian countries. The device, an open-source hardware, can be constructed at a low cost, effectively improving phenotyping efficiency. To enhance the phenotyping efficiency of small plants, Wu et al. (2022) developed a miniaturized phenotyping platform, MVS-Pheno V2, based on multi-view stereo 3D reconstruction. This platform is suitable for low-stature plants, particularly in breeding and management research related to canopy structure. Combining high-throughput phenotyping with spatial dependency analysis, Jang et al. (2023) demonstrated its potential application in revealing hidden heterogeneity in breeding fields, which could be valuable for precision agriculture in field management. Tayade et al. (2022) employed spectral index analysis methods by using various vegetation indices, combined with artificial intelligence and various remote sensing applications. This provides essential tools for high-throughput phenotyping in precision agriculture. These studies indicate the rapid development of high-throughput phenotyping technology in the field of plant breeding, particularly in crop phenotyping under field conditions. These technologies not only enhance the efficiency of collecting phenotypic data but also provide more accurate data support for breeding programs. 3.3 Application of artificial intelligence and machine learning in Breeding 4.0 Another innovation in Breeding 4.0 is the application of artificial intelligence (AI) and machine learning (ML) in breeding. By employing AI and ML methods, breeders can handle and interpret large-scale genetic and phenotypic data, discovering hidden patterns and correlations. Algorithms of AI and ML can learn from massive datasets and predict the breeding potential of crops, suggesting optimal selection strategies. This data-driven approach provides more accurate and intelligent support for breeding decisions, fostering improvement and innovation in crop varieties. The latest literature showcases the recent applications and advancements of artificial intelligence and machine learning in the field of plant breeding. These studies not only encompass genome selection and phenotype analysis but also involve the investigation of traits in various crops, demonstrating the significant role of artificial intelligence and machine learning in modern plant breeding. The study “Machine Learning Applied to the Search for Nonlinear Features in Breeding Populations” indicated that deep learning methods can better identify differences between positive alleles and genetic backgrounds. By employing machine learning methods, the understanding of non-linear interactions in plant breeding datasets is enhanced, leading to improved prediction accuracy, significant reduction in computation time, and enhanced detection of important alleles related to qualitative or quantitative traits (Gabur et al., 2022).
RkJQdWJsaXNoZXIy MjQ4ODYzMg==