Rice (Oryza sativa L.) is the staple food for more than half of the population worldwide. However, sub-optimal climatic conditions can significantly reduce rice yield. For example, temperature stress during reproductive development can induce sterility in flower spikelets (Yoshida 1976) with consequent losses in grain yield. However, increasing the frequency and intensity of heat events due to global climate change (IPCC 2022) would impose environmental challenges on rice crop productivity, but little has been explored regarding grain quality (Ray et al., 2013). Heat stress during late-grain filling significantly reduces the head rice yield in Australian rice cultivars, but this response was genotype-dependent (Ali et al., 2019a). Therefore, we require a genetic model to identify the best-performing genotypes following heat stress during late-grain filling periods with higher HRY and reduced milling losses to mitigate the economic losses. Reduction in HRY following heat stress events is generally attributed to increased fissuring or grain cracking, which may be caused by grain chalkiness contributing to structural weakness (Cooper et al., 2006), but contradictory findings were found (Ali et al., 2023). Also, it was recently found that grain breakage during the de-husking or polishing milling operations was also genotype-dependent (Ali et al. 2019b). The degree of the economic worth to which each loss process (husk loss, broken brown rice loss and broken white rice loss) contributes to overall HRY reduction following heat stress during the late grain filling period (10–20 DAA) has not been fully resolved yet.
Quantitative trait loci (QTL) for milling yield and head rice have been identified (Aluko et al., 2004; Mei H et al., 2002; Nelson et al., 2012; Nelson et al., 2011; Ren et al., 2016; Septiningsih et al., 2003; Tan et al., 2001; Xu et al., 2015; Wang et al., 2017), although no QTL for HRY has been mapped under heat stress condition. In addition, several studies mapped QTL in populations derived from crosses between long-grain and short-grain cultivars with the grains milled under uniform conditions for phenotyping (Nelsen et al., 2011; Bazrkar-Khatibani et al., 19). Given that grain breakage during milling is highly dependent on grain length (Nelsen et al., 2012; Deng et al., 2022) and genetic background (Ali et al., 2019b), QTL mapped for grain breakage, or HRY, may have been associated with differences in grain shape or genetic architecture in the mapping population. Further, all previous studies so far have mapped HRY as a single trait without considering its relation to other loss-related traits. The complexity of HRY yields is associated with different pre-and-post-harvest factors as well as multiple physiological factors beyond maturity (Ali et al., 2023) that contribute to the overall reduction in the yields of HRY. Given that, there is genetic variation in the contribution of these de-husking and polishing loss processes to the overall HRY and to the reduction in HRY under heat stress (Ali et al. 2019b). Mapping HRY as a single trait may mask loci associated with individual loss processes contributing to HRY.
There may be a value in using genomic data for identifying the individual with higher genetic worth for HRY with reduced associated losses (husk loss, broken brown rice loss and broken white rice loss). Notably, best linear unbiased prediction (BLUP) fitted by residual maximum likelihood methods (REML) based on a linear mixed model is a powerful tool to predict genetic worth based on trait heritability across the population (Jahufer and Luo, 2018; Jahufer et al., 2021; Ali et al., 2022). The BLUP values are calculated using the phenotypic variance-covariance matrix and the shrunken factor that involves the heritability to adjust the means across the population. However, BLUP values coupled with kinship matrix or G-matrix (GBLUP) could increase the power of selection where the heritability of each trait is estimated based on genetic markers. Further, If the traits are ranked based on their economic importance, consider 10% selection intensity (Smith-Hazel index) across the population; the bioeconomic selection indexes (I) for each genotype can then be calculated and used to rank the genetic worth of each genotype. The selection of best-performing individuals based on ‘I’ values will fulfil the industry/grower’s requirements and demands and can further streamline the selection process in the breeding program.
The present study aimed to look at three methods: (i) Estimating genetic effects on HRY% and associated traits that were responsible for reductions in overall HRY%, (ii) estimating the predicted genetic gain (ΔG) of targeted traits across the RIL population using Smith Hazel’s selection index-based approach based on either the molecular variance-covariance matrix (G-matrix) or the phenotypic variance-covariance matrix, and (iii) building a bioeconomic genetic model by integrating bioeconomic selection indices (I) with genomic best linear unbiased prediction (GBLUP) to predict the top-performing individuals that had higher HRY% and reduced associated losses.