Association Mapping in Wheat: Issues and Trends Dr. Pawan L. Kulwal Mahatma Phule Agricultural University, Rahuri-413 722 (MS), India
Contents Status of AM studies in wheat Comparison with other important crops Issues and Trends Joint Linkage-LD mapping Ideal scenario
Status ~32 published reports so far. ~ 30% used less number of markers than genotypes used. 2 reports on candidate gene analysis. Regional AM also done (using markers of specific chromosomes). Genotypes used: 44-1055 Markers used: 16-1644 (mostly DArT and SSR).
Comparison between important crops Crop Wheat 170 276 Barley 148 768 Size of population No. of markers 1644 DArT +SSR 1230 DArT >3000 SNP 1536 SNP Reference Crossa et al. (2007) Yu et al. (2011) von Zitzewitz et al. (2011) Massman et al. (2011) Rice 950 >1.3 m SNPs Huang et al. (2010) Sorghum 336 ~265,000 SNPs Morris et al. (2013) Maize 5000 (NAM) 1.6 m SNPs Poland et al. (2011)
Traits studied and population used Traits studied Agronomic Quality Disease resistance Insect resistance Flowering time Plant height PHS Drought Seed longevity Population used Germplasm accessions Winter wheat Spring wheat Released varieties Core collection Breeding lines Diverse accessions Durum wheat Synthetic wheat
Some important QTLs identified Trait Marker R 2 (%) Chromosome Reference Grains per spike gwm294 11-13 2A Yao et al. 2009 FHB wmc18 ~12 2D Miedaner et al. 2011 Stem rust cssr2 wpt8171 BF145935 1000 Kernel wt. gwm160 orw6 Protein content Sedimentation volume Test weight wmc419 wmc18 gwm312 wmc419 gwm219 wmc415 ~21 18 18 15 21 11 10 19 21 34 20 3B 4A 7D 4A 7D 1B 2D 2A 1B 6B 5A Yu et al. 2011 Reif et al. 2011 PHS gwm610 ~12 4A Kulwal et al. 2012 Criteria: QTL Detected in majority of the test environments and R 2 10%; MLM results considered, but some may be GLM List is not comprehensive
Issues and Trends Results not used for MAS so far Estimates of LD decay Epistasis/ interaction effect Functional/ dynamic mapping
I. Results not used for MAS so far.. Due to Inappropriate choice of material Small variation explained by identified QTL Lack of validation reports Method of analysis False positives Role of interaction effects (g g; g e) Results are many a times hypothesis-driven and not data-driven
i) Choice of material Often, exotic germplasm or diverse materials from different geographic regions are used to minimize LD. Phenotype cannot be assayed appropriately in a diverse sample. The results from such studies may provide some insight about the diversity, it is not usually optimal for application in a breeding program. Same panel may not be suitable for all the traits. Breeding material: Ideal phenotypic data from multiple years readily available only genotyping needs to be done
For example Used 96 winter wheat accessions originating from 21 countries across five continents. Use of a diverse panel for adaptive trait like dormancy and PHS may not be appropriate; Phenotyping become difficult. The same panel was also used for AM study of number of agronomic characters (Neumann et al. 2011) and seed longevity (Rehman Arif et al. 2011).
ii) Other reasons Variation explained by majority of QTLs has been small (<10%). Few QTLs with major effects identified, but some of them based on inappropriate analysis method. Lack of validation studies. Although some of the identified QTLs through AM have also been identified earlier through IM (is it just by chance?)
iii) Method of analysis Various methods have been proposed for AM. Main emphasis has been to account for the population structure false positives and computational speed In the context of wheat, with lack of large data set, any method controlling population structure should prove effective (eg. K only or Q+K) if used correctly.
Stringency iv) False positives Are many of the studies underpowered? an arbitrary significance value eg. 0.05 has been used. leads to many false positive MTAs. Its important to account for false positive QTLs by using any of the available method Bonferroni Holm BH FDR q value, etc.
II. Estimates of LD Number of markers to be used for AM also depends on the LD decay (if LD decay is fast, more markers needed eg. maize). Different people have reported LD decay differently. e.g. 3.6 cm (Yu et al. 2011) to ~40 cm (Crossa et al. 2007). Both used elite spring wheat lines from CIMMYT. Hao et al. (2011) reported LD decay at >5 (landraces) ~25 cm (modern varieties) in Chinese mini core collection Coverage of markers across different chromosomes was not always uniform in the published reports. Rather than genome wide LD decay, chromosome wide LD decay is more important.
III. Interaction effect Amounts of genetic variation explained by the identified main effect QTLs through AM has been small. One of the important reasons attributed is the interaction effect (gene-gene/ gene-environment). In wheat, effect of g g, g e has been shown to be high in many IM studies. Only few studies have studied these interactions in wheat.
Nature Review Genetics, January 2013 ENVIRONMENT AND GENETIC BACKGROUND ARE MORE IMPORTANT THAN POPULATION STRUCTURE IN GWAS In wheat, we need larger data set to detect such interactions.
II. Dynamic trait mapping/ Functional mapping/ Conditional mapping During the life cycle, a plant passes through different developmental stages under different environmental conditions resulting in the end phenotype (eg. Plant height). QTLs/genes for these developmental traits express selectively at different developmental stages. Stage specific QTLs can be identified by analyzing the data of different developmental stages. MTMM can also be done with this data.
154 common wheat accessions used. Grown under drought stressed and well-watered condition. Plant height recorded after every 7 days from elongation till flowering (five observations; PH1 to PH4 and PHm). eg. PH1 represented the growth during the time interval between stage 1 and stage 2 (S2 S1). Most QTLs/genes controlling PH were influenced by the growing environment; none was expressed throughout the entire growth process.
How to increase utility of AM in breeding? Use of breeding population Joint Linkage-AM More replications; will give higher heritability Precise phenotyping Use of large number of markers Analysis in the sub-populations Validation studies
Three RIL populations and 305 diverse inbred lines used. Parallel mapping (independent linkage and LD analysis) and integrated mapping (combined analysis) carried out. 2,052 SNPs used. Integrated mapping identified 18 additional QTLs not detected by parallel mapping. Phenotypic variation explained (PVE) also increased.
What can be done in wheat? Many biparental/ MAGIC populations and genotypic data already available. Lot of historical data already available from different breeding programs. Needs to be organized in a systematic way (eg. flowering time, growth habit, etc.) Phenotyping of few of these populations with some breeding lines/ AM panel under common conditions.
contd... Genotyping of AM panel with common markers as used on biparental population. Use of already known QTL linked markers along with high throughput technique like GBS. Analysis of such data can give important insights and increase practical utility of results.
Conclusion Large number of studies highlights the importance of AM in wheat. Design of experiment is important. From practical point of view, use of breeding population is more rewarding. Technique like GBS provide more scope for AM in wheat on large scale. JL-LD mapping can be more effective. Data already available from past IM studies need to be used.
Acknowledgements Prof. PK Gupta, CCS University, Meerut
Thank you