Statistical modeling of phenotypic, pedigree and genomic information for improved genetic evaluation in modern plant breeding : a case study with sorghum

Abstract

Global climate change and food insecurity are major concerns of the 21st century. Agricultural production should increase by 60–110% to meet the projected food demands of the word population by 2050 (McGuire 2012). However, the rates of global crop production are still far below the mentioned requirements and most studies predict a future decline in grain yield of major crops due to climate change (Ray et al. 2013; Wiltshire et al. 2013). Rainfed farming systems are drastically affected by climatic conditions, with water scarcity and increasing temperature being the most important limiting factors for crop productivity and, ultimately, for food security worldwide (Daryanto et al. 2013). Efforts to ensure food supply will require accelerating the development of climate resilient crop varieties. This is particularly necessary for crops that provide staple food grain in developing countries and semi-arid regions of the world. Plant breeding can play a crucial role in enhancing crop productivity and adaptation to climate change. The main goal of breeding programs is to efficiently identify and select the best-performing genotypes as potential cultivars or as parental material to improve crop performance in future generations (Falconer and Mackay 1996; Bernardo 2010). For this, new selection techniques based on modern approaches to quantitative genetics have to be adopted by breeding programs in order to accelerate genetic progress. Advances in high-throughput genotyping technologies and the increasing cost-effective access to high-density genomic data have facilitated the adoption of a novel form of marker-assisted selection known as genomic selection (GS). This genetic evaluation method has already revolutionized animal breeding over the past decade and is gaining momentum in crop breeding. In GS, phenotypic and genome-wide marker data from a reference (or training) population is used to predict genetic merit of selection candidates that have only been genotyped but not phenotyped (Meuwissen et al. 2001). As a result, selection efficiency can potentially increase, reducing phenotyping costs and generation interval. Moreover, additional opportunities for GS in crops are provided by current developments in high throughput phenotyping technologies. The success in the incorporation of genomics as breeding tool depends on an appropriate statistical analysis of the phenotypic and genetic data generated in crop breeding programs.