| As the largest major grain crop in China,the phenotypic and genetic research of maize traits are of great significance for the stable and incremental yield.Stomatal phenotype in the ear leaf is an important microscopic character.Adjusting stomata in maize leaves can improve photosynthetic carbon assimilation and water use efficiency.The update and development of image technology provide feasibility for the accurate phenotypic identification of maize,and it provides a possibility for further analysis of the association between phenotype and genotype.In this study,486 maize inbred lines were obtained by rapid scanning electron microscopy,and 13 phenotypic traits including stomatal number,stomatal size and stomatal shape were obtained by YOLOv5 s and UNet.The phenotypic traits were accurately identified and comprehensively evaluated by descriptive statistical analysis,correlation analysis,hierarchical clustering,principal component analysis and heritability analysis.The differences of stomatal density in leaf shape and yield-related traits and the differences of stomatal phenotype among different subgroups of maize inbred lines were analyzed.Genome-wide association study was used to identify candidate genes,and the maize inbred lines were selected based on the haplotype analysis for field phenotypic verification.The main results were as follows:1.Stomatal image acquisition: 486 ear leaves from the maize natural population were obtained in 2021,and 350× and 900× abaxial stomatal images were obtained by rapid scanning electron microscopy.Among them,6497 350× images were used to extract stomatal number traits,and 6460 900× images were used to extract stomatal size and stomatal shape traits.2.Model training and stomatal traits extraction: YOLOv5 s model was used for target detection of 350× images.The UNet model was used to segment 900× images of stomatal depth traits,and eleven phenotypic traits were obtained,including stomatal width,stomatal length,stomatal area,stomatal perimeter,guard cell and pore width,guard cell and pore area,subsidiary cell area,stomatal length and width ratio,stomatal roundness,stomatal eccentricity and proportion of guard cell and pore area.These 13 traits were divided into three types: stomatal number,stomatal size and stomatal shape.Three representative traits were used for manual verification: stomatal number(R2=0.990,RMSE=1.100),stomatal width(R2=0.949,RMSE=2.693),and stomatal length(R2=0.981,RMSE=2.613).In detection efficiency,the average efficiency of stomata extracted by YOLOv5 s was about 0.009s/ picture,and the average efficiency of other characters obtained by UNet was 0.815s/ picture.3.Analysis of stomatal phenotype: Descriptive statistical analysis showed that the coefficient of variation of stomatal number and stomatal density was the largest,while the coefficient of variation of the proportion of guard cells and pore area was the smallest.And the heritability of 13 traits was above average.The correlation analysis showed that there was a significant negative correlation between stomatal number and stomatal size,but no significant correlation between stomatal number and stomatal shape.And there is a very significant positive correlation between stomatal complex shape and size.The results of principal component analysis showed that the first two principal components explained 72.3% of the stomatal traits.The study further reflected the three types of phenotypic traits of stomata,and it was consistent with the previous classification.The results of variance analysis and multiple comparison between the TST subgroup and the other two subgroups(NSS subgroup and SS subgroup)showed significant differences.And TST subgroup had high stomatal density,small stomatal size,narrow and long stomatal shape,large subsidiary cells and small guard cells.4.Genome-wide association study: A total of 116 SNPs were screened and 44 candidate genes were annotated by GWAS of 13 traits of stomata.They mainly encode proteins related to organic metabolic processes,membrane-bound organelle structures and plant hormone signal transduction.Candidate genes were enriched to 2 KEGG pathways(P<0.05)and 41 GO terms(P<0.05).Gene expression analysis showed that there were differences in gene expression levels during growth and development of plant.And there were 17 genes had higher expression levels in vegetative growth stage.From vegetative growth stage to reproductive growth stage,the expression of 16 genes was higher.The expression levels of 8 genes were higher in the reproductive growth stage.Among the 44 candidate genes,31 were specific to phenotypic traits,and 13 were shared by two or more traits.The candidate genes GRMZM2G154447 encode casein kinase 1,GRMZM2G102601 encode ethylene receptor1-25 and GRMZM2G405064 encode PYL5.Casein kinase 1,an important regulator of hormone signaling,is involved in stomatal opening and closing by ethylene and abscisic acid.5.Haplotype analysis and field experiment of stomatal density haplotype: 8 SNPs of stomatal traits were closely linked to form haplotype.Among them,haplotype analysis of stomatal density identified significant SNPs on chromosome 5,and found that 13 SNPs were closely and formed 49 k blocks.Using the Solid Spine of LD method,only two haplotypes were found.78.8% of the materials carried the AATCGAATTCGGC haplotype and 21.0% of the materials carried the CGCTAGGCCAAAT haplotype.The result of field experiment showed that there were significant differences in stomatal density between the two haplotypes,and the stomatal density of CGCTAGGCCAAAT was significantly higher than that of AATCGAATTCGGC.The photosynthesis,stomatal conductance and transpiration of CGCTAGGCCAAAT was significantly increased compared with AATCGAATTCGGC.In conclusion,this study provides important phenotypic basis and theoretical basis for improving the systematic and accurate identification of stomatal traits in maize leaves,high-throughput screening of key functional genes,and promoting the breeding of new varieties with high yield and quality. |
PDF Full Text Request