| Maize is one of the most important food crops in the world and a model system for the study of genetics, evolution and domestication. Drought is one of the most important limiting factors for maize production in many countries. Plant breeding is the most effective way to improve drought tolerance and reduce the yield loss. Collection and evaluation of maize germplasm has significant impact on maize genetics and breeding for agronomic traits and adaptability including drought tolerance. With genomics-based germplasm evaluation, drought tolerance related genes can be discovered. Furthermore, functional markers can be developed from these identified genes and then used in marker-assisted selection to improve the efficiency of drought tolerance breeding. Over 2000 lines representing temperate, tropical and subtropical maize germplasm have been collected from diverse global maize breeding programs including China, Brazil and International Maize and Wheat Improvement Center (CIMMYT) based in Mexico, Zimbabwe and Kenya. A total of 770 inbred lines were selected and used in this study for genetic diversity, population structure and allele frequency difference analyses. From these 770 lines,447 lines were further selected as a core-set for comparative genetic diversity and linkage disequilibrium (LD) analysis using intragenic and intergenic markers. In addition,550 lines were selected based on their biotic or abiotic stress tolerance, and then evaluated for drought tolerance under both well-watered and water-stressed environments. Genotypic analysis was done using single nucleotide polymorphism (SNP) markers developd for two 1536-SNP chips, one containing random SNPs (RA chip) and the other containing SNPs mostly developed from candidate genes involved in drought tolerance (DT chip). Using an integrated SNP map, quantitative trait loci (QTL) associated with plant height, normalized difference vegetation index (NDVI, representing biomass) measured at different stages and anthesis-silking interval (ASI) were identified using LD mapping and bi-parental population-based linkage mapping. Using ASI as an example, mapping results and efficiency were compared for LD mapping and joint linkage-LD mapping1. Molecular characterization of different maize breeding germplasms using genomewide SNP markers. A total of 770 maize inbred lines were evaluated using 1034 SNP markers from the RA chip that covered the whole maize genome and 449 SNP markers of high quality were identified as core markers. Pairwise comparisons across three distinct sets of germplasm, CIMMYT (394), China (282) and Brazil (94), showed that the elite lines from these diverse breeding pools have been developed with only limited utilization of genetic diversity available in the center of origin. Temperate and tropical/subtropical germplasm clearly clustered into two separate groups. The temperate germplasm could be further divided into six groups consistent with known heterotic patterns. The greatest genetic divergence was observed between temperate and tropical/subtropical lines, followed by the divergence between yellow and white kernel lines, whereas the least divergence was observed between dent and flint lines. Long-term selection for hybrid performance might have contributed to significant allele differentiation between heterotic groups at 20% of the SNP loci. There appeared to be substantial levels of genetic variation between different breeding pools as reveled by missing and unique alleles. Two SNPs, PZA03254.1 and PZA03587.1, developed from the same candidate gene were associated with the divergence between two opposite Chinese heterotic groups. Associated allele frequency change at these two SNPs and their allele missing in Brazilian germplasm indicated a LD block of 142 kb. These results confirm the power of SNP markers for diversity analysis, providing a feasible approach to unique allele discovery and the use of SNP markers in maize breeding programs.2. Drought tolerance evaluation and comparison of multiple selection criteria evaluated under two water regimes. A total of 550 inbred lines collected from global breeding programs were evaluated for drought tolerance under both well-watered and water-stressed environments at both vegetative and flowering stages. Vegetative stage evaluation was based on multiple measurements of biomass before and after the drought stress was applied using NDVI. Other drought tolerance-related selection criteria evaluated include anthesis-silking interval (ASI), leaf senescence, chlorophyll content, root capacitance, final grain yield and drought tolerance index (DTI, final grain yield under stressed condition compared with the well-watered). Kernel weight was the most stable traits under drought stress. Root capacitance and DTI had relatively low heritability and low correlation with other drought tolerance criteria, and thus are not recommended as drought tolerance criteria. High genetic correlation observed among NDVIs measured across different developmental stages indicates that plant growth can be measured only at one or two selected time points. Significant reduction of NDVI values measured in the afternoon when the leaves became rolling, compared to those measured in the morning when the leaves were open, provided a reliable index for leaving rolling caused by drought stress. NDVI showed the strongest correlation with other reliable criteria, which is now recommended as a new criterion for drought tolerance. Correlations between grain yield and other criteria, except root capacitance, were all significant. Regression models built based on non-yield drought criteria and yield components explained about 40% and 95% of the variation for the grain yield, respectively. Some maize lines developed in China for temperate regions showed strong drought tolerance comparable to tropical maize lines when tested under tropical condition, indicating that temperate lines with a wide adaptability can be used in drought tolerance breeding for both temperate and tropical environments.3. Genetic diversity and linkage disequilibrium analyses within and between genes in a core set of maize inbreds. Using a core set of maize inbreds, consisting of 447 lines that were selected from the 770 tested lines, and RA and DT chips, genetic diversity and LD were analyzed comparatively using intragenic and intergenic markers. The result indicated that the level of genetic diversity revealed by markers between genes was higher than that revealed by markers within genes, and that the level of genetic diversity revealed by intronic markers was higher than that revealed by exonic markers, based on the analysis of the 447 core inbreds. Haplotype analysis based on the markers from three different window sizes,3-SNP slide window,10-kb window (all the markers within 10 kb) and gene window (all markers within a gene), indicated that a higher level of genetic diversity was revealed by the haplotypes constructed from markers within larger windows (bigger genetic distances covered by the markers). The highest genetic diversity was revealed by the haplotypes from the 3-SNP slide window with an average PIC of 0.566, followed by the haplotypes from the 10kb window with an avereag PIC of 0.437, and then the haplotypes from the gene window with an average PIC of 0.291. However, the levels of genetic diversity revealed by different types of haplotypes were all higher than that revealed by individual SNPs. Genomewide LD analysis indicated that 33.97% of the pairwise SNP markers showed significant LD at the 0.01 level when tested with 447 core-set lines. When tested using two subsets from the 447 core-set lines, however,34.14% of the pairwise SNPs showed significant LD with 160 temperate maize lines, while only 5.55% of the pairwise SNPs showed significant LD with 287 tropical maize lines. Genomewide LD decay analysis indicated that on the average and at R2>0.1, LD decay happened within 5-10 kb for both the core-set lines and the tropical maize lines, while it did within 10-100 kb for the temperate lines. Therefore, the LD level was higher in the temperate maize lines, compared to those in the core-set and tropical lines. As the physical distances between pairwise SNPs increased, average R2 decreased, which was consistent across all the tests with three different groups of inbred lines. 4. Construction of integrated SNP linkage map and detection of genomewide genetic recombination frequency variation and segregation distortion. Variation in genetic recombination frequency and segregation distortion regions (SDRs) were investigated through three genetic linkage maps. An integrated SNP map also constructed using the three RIL populations with parental genotypes from Africa, Asia and Latin America, i.e., XB (X178 x B73), C5 (Ac7643 x Ac7729/TZSRW) and C6 (CML444 x Malawi). The integrated map consisted of 1442 molecular markers including 1155 SNPs, spanning 1346 cM. Among 1697 informative SNPs scored from the two SNP chips each containing 1536 SNPs,446 (26%) showed segregation distortion atα=5% across the three populations. A total of 13 independent SDRs were identified, one of which co-located with gametophyte factor ga2 within a region of 2.2 Mbp. A 4.5 fold difference in recombination frequency for the same marker interval in the three populations was observed. In addition, a 100-fold difference in recombination frequency was observed across chromosomal regions, ranging from an average of 0.09 cM/Mb for pericentromeric regions to 7.08 cM/Mb for telomeric regions across the three populations. Based on recombination suppression, three bins (1.05, 3.04 and'8.03) were confirmed in this study as centromeric regions. Recombination suppression in non-centromic regions identified 11 regions likely to contain condensed heterochromatin (knobs) in the XB population. This integrated map, along with the genomewide information on recombination frequency variation and segregation distortion generated in this study, will greatly facilitate genetic studies, map-based cloning and marker-assisted plant breeding.5. QTL analysis for plant height and the NDVIs measured at different stages under two water regimes. A total of 2052 informative and high-quality SNP markers were selected from the RA and DT chips, from which 386 haplotype loci were constructed and used for QTL analysis of plant height and the NDVIs measured at different stages under well watered and water-stressed conditions, with the three RIL populations and 305 inbred lines as mapping populations. Interval mapping identified 60 unique marker intervals that were associated with NDVIs measured at different stages under the two water regimes. Linkage mapping using single SNPs and haplotypes identified 50 SNPs and 14 haplotype loci, respectively. Some QTL intervals were shared across different stages while others were unqiue to specific stages. The numbers of QTL identified increased with the advance of stages and more QTL were identified at later stages. Three mapping methods, i.e., linkage mapping, single-SNP based LD mapping and haplotype-based LD mapping, identified 15,13 and 5 QTL, respectively, for plant height under two water regimes. Three QTL for plant height and four QTL for NDVI were identified by all the three methods. In addition,13 identified QTL were shared between plant height and DNVI.6. Joint linkage-linkage disequilibrium mapping improves quantitative trait locus analysis for complex traits:a case study of anthesis-silking interval for improving drought tolerance in maize. The power of QTL mapping has been limited by population size, marker allele diversity and the inherent properties of specific mapping methodologies. Several of these rate limiting factors have been addressed in this study by joint linkage-LD mapping using 2052 SNP markers, which has been achieved through parallel linkage-LD mapping and integrated linkage-LD mapping, by using three biparental RIL populations and 305 inbred lines independently and combined. ASI, an important trait for maize drought tolerance, was measured under both well watered and water stressed conditions. Compared to single SNP-based analysis, the use of haplotypes significantly improved the mapping efficiency with P-value decreased from 5.8×10-4 to 7.59×10-8 and with combined single SNP- and haplotype-based analysis the sum of PVE increased to 28.63% from 5.38%. Parallel linkage-LD mapping complemented and testified each other and identified 13 QTL. Integrated linkage-LD mapping identified 17 QTL additional to the parallel mapping, mainly attributed to genetic effect, background effect, or both contributed by RIL populations and the inbreds. The allele combination with multiple SNPs showed a much higher PVE ranging from 5.18 to 18.2%, compared to 1.38 to 1.80% for single markers. Allele frequencies for 113 of 227 (40.8%) SNPs with minor allele frequency (<5%) in 305 lines were recovered to the normal level in at least one of the three RIL populations, three of which were significantly associated with ASI in XB population. The candidate genes identified by two significant haplotype loci included one for a SET domain protein involved in the control of flowering time, and the other encoding aldo/keto reductase associated with detoxification pathways that contribute to cellular damages caused by environmental stresses. Joint linkage-LD mapping is a powerful approach to detect QTL underlying complex traits including drought tolerance.
|
PDF Full Text Request