Elucidating the genetic basis of natural variation in grain size and weight among rice varieties can help breeders develop high-yielding varieties. We identified a novel gene, GW3a (Grain Weight 3a) (LOC_Os03g27350), that affects rice grain size and weight. gw3a mutants showed higher total starch content and dry matter accumulation than the wild type (WT), Nipponbare, suggesting that GW3a negatively regulates grain size and weight. Moreover, our study found that GW3a interacted with OsATG8 by cleaving it, suggesting that GW3a may be involved in the assembly of autophagosomes and starch degradation in plants. The haplotype analysis of GW3a showed functional differences between indica and japonica rice. Taken together, we conclude that GW3a is expressed in the autophagosome pathway regulating starch metabolism in rice, affecting yield-related traits, such as grain size, grain weight and thousand grain weight (TGW). Our findings also shed new light on autophagy-mediated yield trait regulation, proposing a possible strategy for the genetic improvement of high-yield germplasm in rice.

The environmental conditions in customered speed breeding practice are, to some extent, empirical and, thus, can be further optimized. Crop and plant models have been developed as powerful tools in predicting growth and development under various environments for extensive crop species. To improve speed breeding, crop models can be used to predict the phenotypes resulted from genotype by environment by management at the population level, while plant models can be used to examine 3-dimensional plant architectural development by microenvironments at the organ level. By justifying the simulations via numerous virtual trials using models in testing genotype × environment × management, an optimized combination of environmental factors in achieving desired plant phenotypes can be quickly determined. Artificial intelligence in assisting for optimization is also discussed. We admit that the appropriate modifications on modeling algorithms or adding new modules may be necessary in optimizing speed breeding for specific uses. Overall, this review demonstrates that crop and plant models are promising tools in providing the optimized combinations of environment factors in advancing crop growth and development for speed breeding.