| Rice is one of the most important cereal crops,which is more sensitive to drought stress than other crops,such as maize,wheat.Drought resistance(DR)is the final result of dynamic and complex drought responses.With the help of plant phenomics and natural variation,we studied drought response of rice and identified a large number of DR-related loci,which were validated by other experiments.Our study provided new insights into genetic basis of DR and a new approach for genetic research of DR.Our work included two sections:(1)combination of plant phenomics and genome-wide association study(GWAS)to reveal natural genetic variation;(2)genetic research of DR with the help of phenomics and GWAS.We collected phenotypic data of a natural population containing 529 rice accessions using phenomic platform,which includes three sections:RAP(Rice automatic phenotyping platform),HLS(High-throughput leaf scorer),and YTS(Yield traits scorer).Plant height,tiller number,green leaf area,shoot weight were collected by RAP,which were significantly correlated with the traits measured by standard manual methods.A total of 29 leaf traits were collected by HLS,such as leaf size,shape,and color.Nine yield-related traits were collected by YTS,such as grain projected area which is hard to measure by manual methods.Besides,the phenomic platform can identify a few new traits with biological significance,such as PC6 which reflects the level of plant compactness.Based on the phenotypic data mentioned above and genotype data from the next-generation sequencing,we performed GWAS using linear mixed model.A total of 49loci were associated with the traits by RAP and some priori genes were located close to GWAS peaks,such as green revolution gene SD1.The loci associated with some traits were co-localized due to pleiotropic gene or genes in linkage.The GWAS results of traits derived from RAP and standard manual measurement were similar.GWAS of the new trait PC6 also identified some new association signals.GWAS of leaf traits by HLS identified 291 loci and the differentiation of genetic basis of leaf traits among different growth and developmental stages was observed.Many subpopulation-specific association signals were observed,indicating the existence of subpopulation-derived genetic heterogeneity.GWAS of yield-related traits by YTS scanned 89 genetic loci,in which some known yield-related genes were located,such as GS3 and q SW5.With the help of RAP,HLS,and YTS,we performed drought experiment using the association mapping population mentioned above.The drought stress was initiated when plant entered the booting stage.RAP was used to dynamically collect the phenotypic data before drought stress,under drought stress,and after re-watering recovery;HLS was used to measure stay-green under stress;YTS was used to measure yield traits under drought and normal conditions.The symbol"＿B","＿D","＿Re","＿R"represented before drought stress,under drought stress,after re-watering recovery,and ratio traits(under drought stress/before drought stress),respectively.Based on RAP,51 drought-related image features were extracted(referred to as"I-traits"hereinafter),which can monitor the dynamic drought responses of rice.The relationship of the I-traits changed under drought condition.The large variation of I-traits and drought responses of different rice accessions and the significant difference of drought responses of different subpopulations were observed.Our work demonstrated significant correlations between I-traits and traditional DR-related traits,such as relative yield traits,stay-green,and leaf moisture content under stress.While the broad-sense heritability(H~2)of 21 I-traits was larger than 0.5,H~2 of all the traditional DR-related traits was smaller than 0.5.The DR-related I-traits with high heritability are very useful for the following genetic dissection of DR.Among the 51I-traits,We paid more attentions to GPAR＿R and TBR＿R since they reflected stay-green and leaf rolling at the whole plant level,representative of drought tolerance and avoidance,respectively,and they had high heritability(H~2=0.80,0.77).We performed GWAS of all the traits mentioned above.To better reflect drought response or drought resistance,we focused on the GWAS results of ratio traits.In total,470 loci associated with ratio traits were identified,of which 443 loci were identified by I-traits.Some known DR-related genes were located in these loci,such as WRKY13 and Os DREB2A.The reported DR-related QTLs were retrieved from Trop Gene,QTARO,and Pub Med.93%(437/470)of DR-related loci identified by GWAS were co-localized with these DR-related QTLs.A new DR-related gene associated with I-trait"T6＿R"was identified and validated by haplotype analysis and genetic transformation.In order to validate our results,300 diverse accessions from the association population were selected for drought phenotyping again using the phenomic platform in2016.Although the population size was reduced,a total of 313 out of 470 loci identified in 2013 were re-identified in 2016.Our work demonstrated I-traits were also useful for linkage analysis.After comparing the GWAS results and QTLs by linkage analysis for the I-traits＿R,we found 69 i-trait-locus associations were identified by both GWAS and linkage analyses.To further study the relevance of the I-traits to DR performance in the field,the same association population was evaluated for DR performance in the field facilitated with a movable rain-off shelter in the years 2011 and 2016 using relative yield(stress/non-stress)and leaf rolling-related traits.We propose that the I-traits are very useful to predict DR performance in the field based on the results of phenotypic correlation and co-localized loci,and the genetic detection power of I-traits is higher than that of traditional traits in the field. |
