| Flowering time(Heading date)is an important trait in rice(Oryza sativa L.)which determines the planting time and regional adaptability of rice.Meanwhile,a suitable flowering time is also a guarantee for stable and high yield of rice.The flowering time of rice is mainly determined by genetics,but it is also susceptible to environmental conditions such as photoperiod and temperature.Therefore,constructing environmentally-adaptive prediction models for flowering time by the combination of genomic data and environmental information is the key and difficult part of breeding design and developing variety promotion and planting plans,and it is also important for elucidating the mechanisms of genotype-environment interactions.In this study,based on 422 hybrid indica rice genotypes containing three ecotypes and their multi-year multi-pilot phenotypic data and their meteorological information for the whole growth period,an environmentallyadaptive flowering time prediction study was conducted and the main results obtained were as follows:1.The Accumulated Temperature Index(ATI)that can better predict the flowering time in rice was redefined,and an environmentally-adaptive prediction model based on ATI for predicting the flowering time was developed.Based on the flowering time and mean daily temperature information from 178 filed trials,it was found that the accumulated temperature from 1 days after sowing to 26 days before flowering had a high heritability and highest prediction accuracy for predicting flowering time.The ATI prediction model was constructed based on the assumption of constant accumulated temperature required for a specific period for each variety,and a stable and accurate prediction performance was obtained.The average Pearson correlation coefficient between the predicted flowering time and the observed flowering time under the three application scenarios: tested hybrid ×untested environment,untested hybrid × tested environment and untested hybrid × untested environment were 0.895,0.842,and 0.843,respectively.And the error between the predicted flowering time and the observed flowering time was no more than 5 days under the three application scenarios were 83.16%,68.96% and 65.88% of the samples,respectively.2.The genetic basis of ATI and ecotypes were dissected.Genome-wide association study identified a total of 13 significant intervals associated with ATI,and the significant intervals included cloned genes such as Ef-cd,Ghd8 and DEP1.The results of functional typing analysis of the cloned flowering time genes Ef-cd,Hd1,Ghd7 and Ghd8 showed that although there was no obvious subgroup structure among ecotypes,the frequency of functional types of key flowering time genes differed significantly among ecotypes and thus influenced the regional adaptability of different ecotypes.3.The prediction performance of the ATI prediction model was compared with a published flowering time prediction model.The results show that the ATI prediction model has better predictive ability in all three application scenarios compared with the published reaction norm flowering time prediction model when faced with a more complex genetic structure of the variety population,especially in the untested hybrid × untested environment scenario.In addition,the ATI prediction model has fewer and more biologically meaningful parameter,and only a small amount of training data(e.g.,planting the untested varieties in local)is required to obtain the parameter for new varieties,which can be used to predict flowering time under different environments.4.Proposed a feature-refined ATI prediction model scheme.The 13 most significant variant loci identified by genome-wide association study with ATI as phenotype,and 18 cloned key variant loci for flowering time genes were selected,and the functional typing results of Ef-cd,Hd1,Ghd7,and Ghd8 were integrated,for a total of 35 features were used to construct the refined ATI prediction model.The performance of the feature-refined ATI prediction model is comparable to that of genome-wide ATI prediction models,while requiring genotype input for only a small number of loci,which greatly reduces the cost and therefore is more likely to be used in breeding.In summary,this study constructs an environmentally universal flowering time prediction model by redefining ATI,with the expectation that this study can provide new ideas for phenotype prediction and help production practices and crop breeding. |
