Browsing by Person "Chang, Che-Wei"

Now showing 1 - 1 of 1

Exploring adaptive genetic variation in exotic barley germplasm with landscape genomics
(2025) Chang, Che-Wei; Schmid, Karl
Understanding genetic variation underlying local adaptation is essential for improving crop resilience to address challenges posed by climate change. Barley (Hordeum vulgare L. ssp. vulgare), one of the most important crops, is suitable for studying local adaptation due to its remarkable adaptability. This PhD dissertation investigated adaptive genetic variation in exotic barley germplasm, including wild barley (Hordeum vulgare ssp. spontaneum) and barley landraces, from diverse environments and explored strategies to improve the use of genebank accessions for harnessing valuable genetic variants. In the first study, local adaptation in wild barley populations from the Southern Lev- ant was explored using landscape genomics approaches, combining genomic data with the climatic and soil properties of geographical origins. Through redundancy analysis (RDA), we found spatial autocorrelation explained 45% of genomic variation, and environmental factors accounted for 15%. Adaptive signatures were identified in the pericentromeric regions by the population-genetics-based scans and genome- environment association (GEA) scans, but they mostly disappeared when the population structure was considered. Our findings overall highlighted the role of nonselective forces in shaping the genetic variation of wild barley even in divergent environments. The second study addressed challenges in passport data quality control for large- scale samples, such as germplasm collections in genebanks. The R package GGoutlieR was developed in this work to tackle the shortcomings of traditional manual data cleaning. It efficiently detects and visualizes samples with unusual geo-genetic patterns by characterizing geography-genotype associations with distance-based statis- tics via K-nearest neighbors and calculating empirical p-values accordingly. By stream- lining data cleaning and quality control, GGoutlieR supports more reliable landscape genomics studies, which is crucial for studying loci involved in local adaptation. The third study explored the use of neural networks to predict geographical origins for genebank accessions lacking passport data, enabling their integration into genome- environment association (GEA) analyses. Neural network models demonstrated high prediction accuracy in cross-validation. Incorporating imputed environmental data (N = 11,032) into GEA, using barley flowering time genes as benchmarks, revealed complementary detection of genomic regions near flowering time genes compared to regular GEA (N = 1,626). Furthermore, simulations of polygenic local adaptation in selfing species showed that GEA power is insensitive to large sample sizes. These findings suggest that GEA with imputed environmental data can be a complementary approach for uncovering novel adaptive loci that might remain undetected in conventional GEA, rather than improving the statistical power of GEA. Overall, this dissertation contributes to understanding the adaptive genetic variation in barley and expanding methodologies in landscape genomics, providing a direction for the future development of GEA approaches to better support allele mining for prebreeding to enhance crop resilience.