Green Revolution (GR) rice varieties are high yielding but typically drought sensitive. This is partly due to the tight linkage between the loci governing plant height and drought tolerance. This linkage is illustrated here through characterization of qDTY1.1, a QTL for grain yield under drought that co-segregates with the GR gene sd1 for semi-dwarf plant height. We report that the loss of the qDTY1.1 allele during the GR was due to its tight linkage in repulsion with the sd1 allele. Other drought-yield QTLs (qDTY) also showed tight linkage with traits rejected in GR varieties. Genetic diversity analysis for 11 different qDTY regions grouped GR varieties separately from traditional drought-tolerant varieties and showed lower frequency of drought tolerance alleles. The increased understanding and breaking of the linkage between drought tolerance and undesirable traits has led to the development of high-yielding drought-tolerant dwarf lines with positive qDTY alleles and provides new hope for extending the benefits of the GR to drought-prone rice-growing regions.
The introduction of Green Revolution (GR) genes in rice and wheat is one of the most significant achievements in the history of agricultural research. The introduction of the dwarfing allele of the sd1 gene led to the development of dwarf, high-yielding, lodging-resistant, fertilizer-responsive rice varieties that provided yield gains and food security in rice. Breeding efforts since the GR era have continued to incorporate the dwarf sd1 allele and traditional tall varieties adapted to rainfed conditions have consequently largely been replaced by dwarf GR rice varieties1 that show substantial yield gain in irrigated areas2. Undeniably, the GR has enabled us to improve the food security of rice consumers in past decades. The high yield potential and preferred grain quality of some of these GR varieties have led to their cultivation on large areas in rainfed regions3. However, rainfed rice-growing areas prone to abiotic stresses such as drought continue to have low productivity and large yield losses are observed in GR varieties under drought because of their high sensitivity to drought3,4,5,6. Rainfed rice-growing areas comprise 45% of the total cultivated rice-growing area in the world, where drought in mild, moderate, or severe forms occurs in most years6. Given the large area under rainfed rice cultivation and the increasing incidence of drought stress in a range of cultivation systems because of receding water-table depths, improving the drought tolerance of GR varieties has a large potential for helping to achieve sustainable rice production. The annual yield gain of 1% for rice at the global level is far below the required 2% increase to meet rice demand in the coming decades7. Genetic improvement of rice for increased adaptation to rainfed environments can significantly help to fill this gap and improve food security in Asia and Africa.
To date, improvement of popular high-yielding GR varieties’ tolerance of abiotic stress has been achieved by crossing with traditional varieties or wild relatives8,9,10, for which the traditional varieties were used as donors and the high-yielding GR varieties provided the desired yield potential and preferred grain quality. Large segregating populations are needed in order to incorporate genes conferring tolerance of abiotic stresses in such crosses, which typically show poor recombination. One of the reasons for this could be the linkage/pleiotropic action of abiotic stress tolerance genes to traits that were considered undesirable, such as low yield potential, tall height, poor grain quality, etc., leading to their elimination from GR varieties during the selection process.
Because of the complexity of drought stress and the range of conditions to which rainfed rice crops are exposed, a multifaceted breeding strategy including marker-assisted approaches is necessary to improve drought tolerance in rice. In the past decade, the availability of molecular markers coupled with precision phenotyping has led to the identification of several quantitative trait loci (QTLs) for grain yield under drought (qDTY)11. Most of the qDTY were contributed by traditional varieties and are being deployed to develop drought-tolerant versions of popular GR varieties12. A major QTL, qDTY1.1, from the traditional variety N22 showed consistent effects in the backgrounds of popular GR varieties5,13. In addition to the drought tolerance allele of qDTY1.1, N22 possesses the tallness allele of the sd1 gene for gibberellic acid synthesis (OsGA20 oxidase). The GR varieties IR64, Swarna and MTU1010 possess the drought-susceptible allele of qDTY1.1 and the dwarfing allele of the sd1 gene, which has a 383-bp deletion causing loss of function of OsGA20 oxidase14. The qDTY1.1 region harboring sd1 showed a strong association with plant height, explaining around 50% of the phenotypic variation in two of the three mapping populations5. The qDTY1.1 effect on grain yield under drought was predicted to be either because of the pleiotropic effects of the sd1 gene itself or due to a separate locus within the qDTY1.1 region tightly linked to sd115,16,17,18. Both cases are a concern for breeding efforts to improve the drought tolerance of GR varieties, since linkage of qDTY1.1 and sd1 would indicate that the drought tolerance allele could have been lost in developing dwarf varieties and pleiotropy would indicate that drought susceptibility and dwarfness cannot be separated. Therefore, knowledge of the linkage or pleiotropic effect of qDTY1.1 and sd1 will help in future strategizing for breeding drought-tolerant rice varieties.
The tallness and drought tolerance of qDTY1.1 lines and traditional rice varieties as compared with the dwarf plant height and drought susceptibility of GR varieties underline the importance of this locus for drought response. Similar to qDTY1.1, other DTY QTLs such as qDTY6.2, qDTY3.1 and qDTY3.2 also showed associations with tall plant height, reduced yield under irrigated conditions and early days to flowering (DTF), respectively. The association of qDTY with such characteristics could be due to the close linkage of genes affecting these traits with genes for drought tolerance, or due to pleiotropic genes underlying these QTL regions. The rejection of these undesired traits during breeding since the GR era may have led to a loss of the positive qDTY alleles. We therefore hypothesized that the genomic alterations that occurred at the qDTY1.1 or sd1 locus and other qDTY loci (due to selection) have contributed to the increased drought susceptibility of post-GR rice varieties. Characterization of genetic phenomena such as linkage and pleiotropic gene action in relation to stress response will contribute to understanding the genetic control of complex traits, which can facilitate more efficient crop improvement for stress tolerance.
To determine whether the GR selected against drought tolerance in rice, the objective of the present study was to characterize the drought tolerance QTL qDTY1.1 in relation to the GR dwarfness gene sd1 by testing for linkage vs pleiotropy. The relationship between other qDTY with undesirable traits and the presence of qDTY in pre- and post-GR varieties was also investigated to determine the likelihood that qDTY were selected against because of their relationship with undesirable traits. Finally, we aimed to break the linkage between qDTY1.1 and the tall allele of sd1, as well as between other qDTY and undesirable traits, to develop high-yielding drought-tolerant rice varieties.
Analysis of pleiotropy vs tight linkage between sd1 and qDTY1.1
To determine whether the drought tolerance conferred by qDTY1.1 was due to a pleiotropic effect of the sd1 gene or whether there was a tight genetic linkage between qDTY1.1 and sd1 genes, a two-way approach was undertaken: (i) statistical analysis of multiple traits for linkage vs pleiotropy by saturation mapping in the recombinant inbred line (RIL) mapping populations used for the identification of qDTY1.1 and (ii) fine mapping of the qDTY1.1 locus using region-specific single nucleotide polymorphism (SNP) markers in a large backcross inbred line (BIL) population.
Multiple trait analysis and re-mapping of sd1/qDTY1.1 in N22-derived RIL populations
Multiple trait analysis in N22/IR64 and N22/Swarna RIL populations did not show any significant pleiotropic effects of qDTY1.1 on DTF, plant height (PH), or harvest index (HI) under reproductive-stage drought stress (RS) (Table 1). A re-mapping of the qDTY1.1 QTL interval (RM11943–RM431) in the three N22-derived RIL populations (N22/Swarna, N22/IR64 and N22/MTU1010) using three additional SNP markers in the region revealed that qDTY1.1 was located distal to the sd1 gene (Fig. 1, Table 2). The QTL intervals in the N22/Swarna, N22/IR64 and N22/MTU1010 populations were id1024366–id1024499, id1024499–RM431 and id1024366–id1024499, respectively. Further, sd1 gene-based marker analysis of 161 RILs (48 from N22/Swarna, 57 from N22/IR64 and 56 from N22/ MTU1010) based on flanking markers of the sd1 gene (Fig. 1) revealed the qDTY1.1 position to be below that of the sd1 gene, between interval id1024366−id1024499 in a combined analysis over both years (Table 2).
Fine mapping of the sd1/qDTY1.1 locus
In a recombination breakpoint analysis of the N22/2* Swarna BIL population with fixed DTF and sd1 alleles that segregated for markers in the qDTY1.1 interval only (Fig. 2, Fig. S1, Table 1), mean yield under drought was compared among different recombinant classes: classes A, B and C that possessed the N22 allele at nksdty1_1_30 and class D that lacked nksdty1_1_30 but possessed the N22 allele at adjacent marker nksdty1_1_34 and further down toward the telomere. Based on this analysis, there were two broad groups: group (1) with the N22 allele below the sd1 gene (classes A+B+C+D) and group (2) with the Swarna allele below the sd1 gene (classes E+F+G). The difference in mean grain yield (GY) under RS between these two groups was >1000 kg ha−1, confirming the presence of the qDTY1.1 gene below the sd1 gene. Results of fine mapping suggested the QTL location to be downstream of the sd1 gene, that is, between nksdty_1_34 and nksdty_1_38 (Fig. 3).
Selection of dwarf drought-tolerant lines from the BIL population
After analyzing the qDTY1.1/sd1 linkage and fine mapping, qDTY1.1 homozygous lines were developed with the dwarf allele of the sd1 gene from this BIL population. Plants with double crossover events on both sides of the sd1 gene were selected. Three dwarf qDTY1.1 homozygous lines (IR91659:41-95-5-B, IR91659:41-95-6-B and IR91659:54-36-9-B) with plant height similar to that of Swarna under irrigated conditions showed a significant yield advantage over Swarna in the severe field drought experiment (Table 3A).
Analysis of the sd1/qDTY1.1 region in traditional, pre-GR and GR varieties
Association analysis using a random variety set revealed a significant association of markers within qDTY1.1 with DTF, GY, HI and PH. The sd1 gene was found to be significantly associated with PH but not with GY under RS (Table 4). Four different varietal groups were identified among 123 random genotypes that were genotyped for sd1 and qDTY1.1 markers and compared for phenotypic variation for GY (Tables S2-S4) and related traits among medium- and late-duration groups (Fig. 4). Group 1 represents drought-tolerant genotypes that are either pre-GR varieties or landrace derivatives such as N22 and Vandana. This group has the tall allele of sd1 along with the partial/full region of qDTY1.1. Group 2 is composed of the GR varieties with the dwarf allele of sd1 and a Swarna-type qDTY1.1 region. Group 3 comprises lines that have the tallness allele of the sd1 gene but the drought-sensitive allele of qDTY1.1 (all three varieties have the Swarna allele at marker locus nks1_1_30). Group 4 represents GR varieties that had neither the sd1 tall allele nor the N22 allele of qDTY1.1. These are semi-dwarf varieties that show higher yield in rainfed areas5. The sd1 profile of genotypes in Groups 3 and 4 indicates that the sd1 gene is not responsible for high grain yield under drought and the performance of genotypes in Group 4 under drought is also not due to qDTY1.1. Though not absolutely, comparative analysis of all four groups presented in Fig. 4 indicates the responsiveness of qDTY1.1 and the non-responsiveness of sd1 to drought tolerance.
Analysis of qDTY1.1 and sd1 using 3K Rice Genome sequence data
A linkage between the sd1 allele and qDTY1.1 in the tall landraces, most of which are drought tolerant, is further supported through SNP analysis of the recently available 3K Rice Genome Project19. The relationship of culm length with type of cultivar revealed that GR advanced breeding lines and released varieties comprised 62% of the genotypes that have a culm length within 50 to 90 cm. In contrast, 90% of the genotypes with >90 cm culm length were landraces. Semi-dwarf plants in which one of the two well-known sd1 alleles (LOC_Os01g66100; L266F)20 was present were represented by only 0.1% of the genotypes. However, there was 100% association between dwarf genotypes (<50 cm) and a novel mutation (Y342*) that leads to a termination codon in the third exon of LOC_Os01g66100 encoding gibberellin-20 oxidase. Also, a haplotype with three mutations in intron 2 was significantly related to semi-dwarfism. The one known and two novel alleles of sd1, fully associated with the semi-dwarf phenotype, were not linked to the putatively drought-tolerant N22 haplotype of 58 downstream genes spanning qDTY1.1. In contrast, in 17% of the tall genotypes (WT gibberellin-20 oxidase; LOC_Os01g66100), the haplotype of the 58 downstream genes was significantly similar to the N22 haplotype, the putative source of drought tolerance (Fig. 5). These results reinforced the hypothesis that the sd1 gene was linked to drought-sensitive qDTY1.1 alleles.
Physiological basis of the drought tolerance effect of qDTY1.1
In detailed physiological studies on four drought-tolerant recombinant BILs in the Swarna background, the four recombinant BILs and Swarna showed a varying levels of flowering times under well-watered (NS) and drought stress (RS) conditions, with a significantly shorter flowering time observed in the qDTY1.1 BILs than in Swarna during the 2012 wet season (WS) and 2014 dry season (DS) under drought stress and in the 2013 DS and 2014 DS in the well-watered control (Fig. 6A). Drought stress was imposed early enough around 30 days after transplanting to minimize the effect of early DTF on drought stress. Swarna did not flower in the drought stress treatment in the 2013 DS. The aboveground shoot mass of qDTY1.1 BILs showed a dynamic response to the timing and severity of drought stress. In a greenhouse lysimeter study, qDTY1.1 BIL IR91659:41-95-B showed higher apparent leaf area than Swarna early in the drought stress treatment (Fig. 6B), but leaf area was similar among genotypes in the well-watered control (Figure S2-A). The height and plant type of the qDTY1.1 BILs were similar to those of Swarna (Fig. 6C). In the field, qDTY1.1 BILs showed a lower normalized difference vegetation index (NDVI) than Swarna when the stress treatment began early (2014 DS; Fig. 6D) and a higher NDVI than Swarna early in the drought stress treatment when the stress treatment began late (2013 DS; Figure S2-B). Starting from the reproductive stage, the qDTY1.1 BILs showed a greater allocation to stem mass as a proportion of the aboveground biomass in the field under drought and in the well-watered control (Figure S2-C, D) in the 2014 DS. The allocation to deep root growth was higher in the qDTY1.1 BILs than in Swarna. The qDTY1.1 BILs also showed higher maximum root depth in the drought stress treatment of the greenhouse lysimeter study (Fig. 6E) but not in the well-watered control (Figure S2-E). The BIL IR91659:41-95-B showed a higher proportion of root length at depth across seasons and treatments (Fig. 6F), except in the seasons when the drought stress was very severe (2012 WS and 2014 DS). The water uptake rates of qDTY1.1 BILs in the drought treatment of the greenhouse lysimeter study were closely related to leaf area patterns across the study (Figure S2-F).
Linkage of other DTY QTLs with traits lost during the GR era
Similar to qDTY1.1, three other drought-yield QTLs were observed to show linkage with traits selected against during the GR. qDTY6.2 was linked with tall plant height, qDTY3.1 was linked with reduced grain yield under irrigated conditions and qDTY3.2 was linked with very early days to flowering. The presence of the qDTY6.2 donor (IR55419-04) allele at three marker loci (RM3, RM541 and RM275) within qDTY6.2 led to increased yield under drought. Recombinant classes identified from a mapping population showed an association of RM275 with tall plant height within this region (Figure S3-A). A height difference of up to 4 cm under non-stress conditions and up to 6 cm under stress was observed for lines with the IR55419-04 and TDK1 alleles at this marker locus. For qDTY3.1, three marker loci (RM520, RM416 and RM16030) were associated with increased yield under drought in an Apo/Swarna population4. However, class analysis based on the three marker loci within this QTL showed that the presence of the drought-tolerant Apo allele at RM16030 led to a decline in yield under irrigated conditions (Figure S3-B). The presence of the drought-tolerant qDTY3.2 allele from Vandana at three marker loci (RM7332, RM523 and RM545) within qDTY3.2 was related to a reduction in DTF (Figure S3-C), which led to early maturity. Although this earliness was advantageous under severe drought stress, a flowering time of less than 80 days decreases yield potential under well-watered conditions. This has been the main reason that the highest-yielding GR varieties were bred to mature after more than 110 days; thus, this QTL was rejected during the development of GR varieties. Class analysis revealed the association of RM523 with early DTF.
In this study, recombinant lines with the drought tolerance alleles of qDTY6.2, qDTY3.1 and qDTY3.2 with desired height, yield and DTF were identified. The dwarf lines with qDTY6.2 with plant height similar to that of TDK1 under irrigated conditions showed higher yield than TDK1 under severe field drought experiments (Table 3B). Lines with qDTY3.1 showed similar yield to Swarna under irrigated conditions and a significant yield advantage over Swarna under severe field drought (Table 3C). Similarly, lines with qDTY3.2 with later DTF than Vandana showed yield under drought similar to that of drought-tolerant cultivar Vandana, the donor of the qDTY3.2 QTL (Table 3D).
Proportion of qDTY alleles in traditional and GR varieties
Genetic diversity analysis for qDTY alleles conducted with 65 SSR markers in and around the 11 different DTY QTLs separated the drought-tolerant lines from the drought-susceptible lines into different clusters (Fig. 7). It also separated the traditional drought-tolerant donors (Cluster 1) with recently developed drought-tolerant varieties (Cluster 5) and further separated GR varieties into three distinct groups: early GR varieties represented by IR8 (Cluster 2), mid GR varieties represented by Jaya (Cluster 3) and late GR varieties represented by presently cultivated varieties IR36, IR64, Swarna, MTU1010 and Sambha Mahsuri (Cluster 4) (Fig. 7). Cluster 5 included the drought-tolerant varieties developed during the post-GR era: Apo, RD7, IR74371-46-1-1 and IR74371-70-1-1. The only exception in this group is Way Rarem, the cultivar which in itself is drought susceptible but has contributed the alleles for qDTY12.1. The proportion of qDTY tolerant, sensitive and other unknown alleles at 65 SSR marker loci of 11 qDTY QTLs showed the highest percentage of donor alleles in traditional upland drought-tolerant varieties (Cluster 1, 40%), followed by recently developed varieties (Cluster 5, 32%). Among the GR varieties, the proportion of drought-tolerant alleles declined to 21% in Cluster 2, 18% in Cluster 3 and 16% in Cluster 4 varieties (Fig. 8). Interestingly, the frequency of the recipient allele (contributed by drought-susceptible GR varieties) increased from 34% in traditional donor varieties to 45% and 49% in early and mid GR varieties (Clusters 2 and 3), respectively and to 58% in late GR varieties (Cluster 4). In recently developed drought-tolerant varieties, the frequency of the recipient allele decreased to 41% (Fig. 8).
The characterization of qDTY1.1 and sd1 conducted in this study in terms of genetic linkage, presence in varieties with known drought tolerance, association with dwarfing genes and physiological drought response, as well as the observed linkage of other qDTY with undesirable traits and the absence of qDTY alleles in GR varieties, strongly suggest that the GR did indeed select against drought tolerance in rice.
Multiple approaches confirmed that qDTY1.1 and sd1 are linked, indicating their distinct effects on plant performance (qDTY1.1 on drought tolerance and sd1 on plant height) and therefore the likelihood that semi-dwarf plant height could have been selected at the expense of drought tolerance in developing GR varieties. Testing the hypothesis on whether the sd1 gene and qDTY1.1 are linked required four variables (plant height, GY under drought, presence of the sd1 gene and markers for qDTY1.1) to segregate in a genetically defined population. In the first test for linkage vs pleiotropy conducted through multiple trait–multiple interval mapping (MT-MIM) analysis on the RIL populations in which qDTY1.1 was originally identified, the pleiotropic effect obtained for GY–PH was found to be non-significant, suggesting a possible linkage between qDTY1.1 and sd1 (Table 1). A similar approach has been followed previously by other researchers in testing the pleiotropic effects of co-located QTLs21. The re-mapping conducted in three N22-derived RIL populations in which qDTY1.1 was earlier identified5 suggested that qDTY1.1 is located downstream of the sd1 gene (Fig. 1, Table 2). These results suggested the QTL location to be between sd1 and RM431 (~500 kb). The position of qDTY1.1 was further validated to be distal to the sd1 locus in a BC3F4 N22/Swarna-derived population fixed for the sd1 dwarf allele, with similar PH and DTF and that had high genetic background similarity. Recombination breakpoint analysis using a high-density custom SNP assay confirmed that the qDTY1.1 region was below the sd1 gene, between nksdty1_1_34 and nksdty1_1_38 (Fig. 3). Association analysis conducted on a random set of varieties also revealed that the sd1 gene has a significant association with PH but not with GY under RS (Table 4). A number of previous studies have suggested the possible linkage of qDTY1.1 and the sd1 gene. Kumar et al.  identified a QTL in the qDTY1.1 region of a population that was not segregating for PH. Khowaja et al.  carried out a meta-QTL analysis that suggested linkage because the meta-QTL identified in that study was below the sd1 gene, which is in agreement with the present study. Venuprasad et al.  also suggested the possible linkage of qDTY1.1 and the sd1 gene in a study with five backcross-derived populations. In our study, re-mapping/fine mapping and association analysis as well as statistical analysis indicated that qDTY1.1 is located distal to the sd1 gene, is tightly linked to it and that the effect on increase in grain yield under drought contributed by this region is due to qDTY1.1 and not to sd1. The results from our study show a tight linkage between the tall allele of sd1 and qDTY1.1. We subsequently broke this linkage and developed a population of dwarf BILs that was used for fine mapping of qDTY1.1. The observation that most popular high-yielding GR varieties have the drought-susceptible qDTY1.1 allele along with the dwarf sd1 allele further suggests that the transfer of the dwarf allele of sd1 led to the loss of the drought-tolerant qDTY1.1 allele in these varieties, resulting in their increased susceptibility to drought.
Breeding for the semi-dwarf trait that led to introgression of drought susceptibility was also revealed by the analysis of qDTY1.1 and sd1 using the 3K Rice Genome sequence data. This susceptibility could be noticed only in recent years when such GR varieties showed a high reduction in yield due to increased occurrence of drought. Nevertheless, there are sufficient exceptions to the tall phenotype being linked to the N22 drought tolerance haplotype but very few exceptions to the short phenotype being linked to the drought susceptibility haplotype. Moreover, all genotypes analyzed for sd1 linkage to the drought tolerance haplotype of N22 have not yet been phenotyped for response to drought. Therefore, the value of the data analyzed is restricted to addressing the hypothesis that the sd1 gene is not causal to the tolerant haplotype of N22 genes comprising qDTY1.1. In that respect, our data support this hypothesis with 99.9% of the sd1 dwarf genotypes not being linked to the drought-tolerant N22 qDTY1.1 haplotype and 17% of the sd1 tall genotypes being linked to it (Fig. 5).
The physiological mechanisms related to the presence of qDTY1.1 further demonstrate that drought tolerance by this QTL is independent of plant height. It is likely that the mechanisms observed in the qDTY1.1 BILs – slightly earlier flowering time, a plastic shoot biomass response to drought and the ability to increase root length at depth – act concertedly to confer higher yield under drought. The dynamic shoot mass response may be a growth regulation response to conserve water.
Of the candidate genes identified within the qDTY1.1 region (Table S5), NAM (no apical meristem protein) is one of the components of the drought-responsive NAC gene complex. The SNAC1 gene in rice is a typical example of this family reported to increase spikelet fertility and seed setting rate under severe drought stress22. Another gene of this complex, OsNAC10, is reported to be responsible for root-specific expressions imparting drought tolerance and enhancing grain yield under drought23. One of the serine/threonine protein kinases in this fine-mapped region (LOC_Os01g66860) has been reported to show a differential response under drought stress in cultivar N22, which was used as the drought-tolerant parent in the present study24. A drought-responsive zinc finger protein factor has been isolated from rice cultivar N2225. Other examples of zinc finger proteins related to drought tolerance in rice are ZFP252, Zat10/STZ and WRKY genes26,27,28. STZ was reported to increase spikelet fertility and grain yield under drought27. Further work in terms of differential expression analysis and/or transgenic validation of the candidate genes is necessary to pinpoint the gene(s) responsible for the increased drought tolerance conferred by qDTY1.1.
In addition to qDTY1.1, numerous other drought grain yield QTLs have been identified at the International Rice Research Institute (IRRI) during the past few years4,5,12,13,29,30,31. Some of these are reported to be associated with undesirable traits such as tallness (qDTY1.1 and qDTY6.2) and low yield under irrigated conditions (qDTY3.1). The qDTY3.2 QTL is associated with earliness, which often results in low yield potential. In all four of these cases, the characteristic traits of landraces such as tall plant height, low yield potential and earliness are closely associated with the drought-tolerant DTY allele. In addition to studying the linkage between sd1 and qDTY1.1, we investigated the association of qDTY6.2, qDTY3.1 and qDTY3.2 with the respective undesirable traits. The positive alleles of qDTY6.2, qDTY3.1 and qDTY3.2 were responsible for tall plant height, low yield under irrigated conditions and very early maturity, respectively4,5,30,32. Indeed, most of the known drought-tolerant donors are intermediate/tall (more than 110–130 cm in lowland, 90–125 cm in upland) to tall (more than 130 cm in lowland, more than 125 cm in upland) in height, have low yield potential (less than 3.0 t ha−1) and mature early (80–100 days) compared with the GR varieties that are semi-dwarf (less than 110 cm in lowland, less than 90 cm in upland), have higher yield potential (more than 5.0 t ha−1) and show medium maturity duration (110–130 days). Therefore, the probable reason for traditional drought-tolerant varieties having characteristics such as tall plant height, lower yield potential and early maturity is their linkages with QTLs for grain yield under drought (qDTY).
An allelic analysis with 11 DTY QTLs revealed a genetic shift in modern varieties with respect to these loci (Fig. 7). A sharp decrease in the frequency of drought-tolerant alleles in GR varieties as compared with the traditional drought-tolerant varieties/donors was observed (Fig. 8). Interestingly, the loss of drought-tolerant alleles continued as breeding for semi-dwarf varieties proceeded from IR8 in the 1960s to Jaya in the 1970s to IR36, IR64, Swarna and Samba Mahsuri in the 1980s and beyond (Fig. 8). This trend appears to have resulted from the replacement of drought-tolerant alleles with alleles that were adapted to irrigated ecosystems through intercrossing among GR varieties, an approach that plant breeding has followed from 1966 until now. The linkage of drought-tolerant alleles with unfavorable traits/traits rejected for in the GR period (tall plant height, lower yield potential and very early maturity) is likely the key factor for their loss during selection for semi-dwarf height, high yield potential and medium maturity duration during the GR era.
In addition to the loss of alleles affecting grain yield under drought, another challenge to be addressed in improving the drought tolerance of modern rice varieties is the loss of alleles showing interactions with major alleles that either activate the drought tolerance response or increase the response of major drought tolerance alleles30,33. A significantly more detailed understanding of these interactions will be required before these aspects of drought tolerance can be improved.
The identification of undesirable linkages between alleles governing drought tolerance and tall plant height, lower yield under favorable conditions, or very early maturity was possible because of the strategic shift to work on QTLs for grain yield under drought from the previous strategy of identifying QTLs for secondary traits. Fortunately, the effect under drought is not a pleiotropic effect of the sd1 gene or other such alleles as reported earlier15 but is the tight linkage between alleles imparting drought tolerance and alleles governing traits rejected for in developing GR varieties. This understanding provides a chance to eliminate these detrimental effects through careful breeding and preserving the drought tolerance imparted by these loci. The present study shows an example where the linkage between sd1 and qDTY1.1 was systematically broken through the identification of double crossover recombinants to generate a large recombinant population and dwarf drought-tolerant lines possessing high yield potential were developed. Semi-dwarf drought-tolerant genotypes homozygous for qDTY1.1, dwarf plant height and grain yield under favorable conditions similar to Swarna were successfully developed (Table 3). Recombinant lines without undesirable linkages related to qDTY3.1, qDTY3.2 and qDTY6.2 were also identified from the populations in which these were detected. In all four cases, plants with a significant yield advantage under drought over the recipient parents were developed without any compromise in yield potential under favorable conditions.
Conserved synteny predicts that, like rice, wheat and other cereals may also have lost important drought tolerance alleles because of similar undesirable linkages. The GR gene in wheat Rht-b1 is located inside qDSI4B.1, a major drought QTL34. Several QTLs were reported in the same region of wheat chromosome 4B35,36,37,38. The importance of the co-localization of drought tolerance QTLs with plant height loci was also revealed in sorghum39. Within qDTY1.1, a meta-QTL, MqTL1.1 was reported to show synteny with the chromosome 3 region near msu 2 in maize, with chromosome 4B near Rht-b1 in wheat and with the chromosome 6H region near Bmac0316 in barley. The markers are reported to be linked to grain yield under drought in all three crops40. Loss of alleles adaptable to drought could also have occurred in other cereal crops wherein similar breeding procedures of selection for traits/alleles to increase yield under favorable conditions have been followed over decades. A comprehensive analysis involving all major cereal crops is required to investigate the allelic shift in modern breeding programs.
In conclusion, this work on fine mapping, physiological characterization and linkage breaking of qDTY1.1 provides strong evidence that the tall sd1 allele and qDTY1.1 loci are tightly linked and that the increased grain yield under drought conferred by qDTY1.1 is independent of sd1 and plant height. The argument that GR variety development selected against drought tolerance in rice is further supported by the tight linkage between traits rejected for in the GR era and other qDTY, as well as by genetic diversity analyses. New drought-tolerant dwarf lines have been developed through the successful breakage of linkages between loci for drought tolerance and undesirable traits. This work provides new hope for developing high-yielding drought-tolerant dwarf varieties of rice and other cereal crops in order to extend the benefits of the GR to drought-prone regions.
Materials and Methods
Plant material and genotyping
Seven sets of plant materials were used
The previously identified qDTY1.1 region in three populations with 292 lines (N22/Swarna), 289 lines (N22/IR64) and 362 lines (N22/MTU1010) was genotyped with SNP markers underlying qDTY1.1 using the Fluidigm SNP genotyping platform (K-Biosciences Ltd.) for improved resolution of the map of the qDTY1.1/sd1 region. RIL recombinants from the three populations were also genotyped with SNP markers flanking qDTY1.1 and with an sd1 gene-specific marker.
An N22/4*Swarna backcross population was used for the qDTY1.1/sd1 linkage and fine-mapping study. To develop this population, 217 semi-dwarf plants with height similar to that of Swarna were selected from 3000 BC3F1 plants. The selected plants were genotyped with qDTY1.1 markers RM11943, RM431, RM12023, RM12091 and RM12146. Two plants segregating for qDTY1.1 were identified and analyzed with the sd1 gene-based marker to confirm the dwarf allele and were selfed. The resulting 180 BC3F2:3 N22/Swarna plants were analyzed with the qDTY1.1 markers. From the 180 BC3F2:3 N22/Swarna plants, 20 heterozygote recombinants for flanking markers for the sd1 gene as well as qDTY1.1 were identified. These heterozygotes were selfed to produce ~1200 BC3F4 plants, which were then genotyped to select 160 recombinants based on the two flanking markers of qDTY1.1. In this BC3F4 N22/Swarna population, DTF, PH and the sd1 allele were similar in all selected lines. Background QTL effects were also neutralized in this population (Figure S1). Another QTL, qDTY3.2, identified in the N22/Swarna population5, was also fixed. Along with qDTY3.2, the five introgressed regions (RM246, RM278, RM335, RM434 and RM551) in the recombinant lines were fixed for the N22 allele and the population was segregating for only one (RM411) of the 120 markers used to test the background effect, which did not show significance for any of the tested drought-related traits. The fine-mapping study was conducted on these 160 recombinants (BC3F4:5 and BC3F4:6). Phenotypic variation for GY and related traits in the recombinant BIL population is presented in Table S1.
Four qDTY1.1 near-isogenic lines (NILs), IR91659-41-95-B, IR91659-41-36-B, IR91659-41-39-B and IR91659-41-59-B, along with parents N22 and Swarna, were characterized in lysimeter studies as well as in field studies from 2012 to 2014 to understand the physiological mechanism of the drought tolerance conferred by qDTY1.1.
The effect of the sd1 gene was analyzed in a set of 123 rice genotypes (Table S2), including landraces, traditional varieties and cultivated varieties. These lines were genotyped with a panel of 29 SNP markers underlying qDTY1.1 with the Sequenom platform (described below). A gel-based analysis with the sd1 gene-specific marker was carried out. SNP markers showing spurious calls were omitted and a total of 22 markers (21 SNPs + sd1) were used for analysis. This set of diverse genotypes was screened for GY under RS at IRRI in the 2012 DS and 2013 DS following the protocol described by Venuprasad et al. .
Culm length data were extracted from the SNP-Seek database42 for 3000 rice genotypes used to study the linkage pattern between the sd1 allele and qDTY1.1 N22 allele (N22 being one of the 3000 genotypes sequenced). Varieties with missing culm length were eliminated from the analysis.
A BC1F3-derived IR55419-04/TDK1 (365 lines) population, a BC1F4-derived Apo/Swarna (490 lines) population and an F3-derived Vandana/Way Rarem (242 lines) population were used to understand the linkage in the qDTY6.2, qDTY3.1 and qDTY3.2 regions, respectively. Genotypic data available for these populations4,5,33 were used to understand the linkages and to identify recombinants free from linkage drag within these populations (Figure S3).
In order to understand the allelic diversity present for 11 qDTY loci, a set of 132 diverse genotypes, including short and tall plant types of traditional drought-tolerant varieties, traditional drought-susceptible varieties and modern rice varieties cultivated in rainfed and irrigated rice ecosystems, was used.
Phenotyping under drought stress and irrigated conditions
The N22/Swarna BIL population was screened under lowland reproductive-stage drought stress (RS) and irrigated non-stress (NS) conditions in the 2012 DS and 2013 DS, whereas qDTY1.1 homozygote plants (N22/Swarna, N22/IR64 and N22/MTU1010) were screened in the 2012 DS at IRRI. The Apo/Swarna BC1-derived population was screened in the 2006 DS and 2007 DS under RS and NS conditions, respectively; Vandana/Way Rarem lines were screened under upland RS and NS conditions in the 2005 DS and 2006 DS; and the IR55419-04/TDK1 population was screened in the 2011 DS and 2012 DS under lowland RS and NS conditions, respectively. The RS and NS experiments were laid out in an alpha lattice design with two replications and managed as per the protocol described by Venuprasad et al.41 and Kumar et al.43.
For the physiological study, four BILs and parents N22 and Swarna were grown in a greenhouse lysimeter experiment from May to July 2013 in large lysimeters (95 cm tall × 20 cm diameter), as described by44. Five replicates of the two treatments were included in a completely randomized design. The soil in the lysimeters was maintained flooded until 20 days after planting, after which the lysimeters in the RS treatment were drained. The four qDTY1.1 BILs and parents were also characterized in the field during the 2012 WS (June 2012–October 2012), 2013 DS (December 2012–April 2013), 2013 WS (June 2013–October 2013) and 2014 DS (December 2013–April 2014) under transplanted lowland conditions, including NS and RS treatments with four replicates per treatment. Experimental plots were maintained flooded until 62, 75, 58 and 60 days after sowing (DAS) in the 2013 WS, 2013 DS, 2013 WS and 2014 DS, respectively, after which irrigation in the RS treatment was stopped and rain was excluded using an automatic rainout shelter. The RS experiments were re-watered periodically as needed to sustain the plants until harvest. Soil water potential was monitored with one tensiometer per replicate installed at a depth of 30 cm in each RS experiment. The drought stress was most severe in the 2012 WS and 2014 DS, as indicated by the more negative soil water potential values for longer periods (Figure S4). Observations of the number of days to 50% flowering, plant height, biomass (BIO), grain yield and harvest index were recorded as per the protocol described by Venuprasad et al.4.
All lysimeters were weighed three times per week to determine water uptake rates, at which time images of the shoot were acquired to measure apparent leaf area in Image J until plants were harvested at 60 days after sowing. Maximum root depth was determined as the distance from the base of the plant to the deepest root in the lysimeter. In the field, shoot growth was monitored by NDVI measured around mid-day using a GreenSeeker Handheld Sensor (NTech Industries, CA, USA) that was carried through the field (during the 2013 DS) or mounted about 1 m above the soil from a rack that rolled along the tracks of the rainout shelter (during the 2014 DS). Aboveground biomass partitioning was monitored throughout the 2014 DS, in which two hills per plot were harvested at each sampling. The leaf blades were separated from the sheaths and culms (the stems) and dry weight was determined. Root samples were taken at 128, 119, 113 and 101 DAS following re-watering of the RS treatment in the 2013 WS, 2013 DS, 2013 WS and 2014 DS, respectively. Root samples were acquired using a 4-cm-diameter core sampler (fabricated at IRRI) to a depth of 60 cm (divided into 15-cm segments) and root length determined according Henry et al.45. ‘Percent deep roots’ was calculated as the root length below the soil depth of 30 cm as a percent of the total root length in the soil core.
Genotyping with SSR markers of DTY QTLs
SSR genotyping of the N22-derived RILs and BILs was carried out for qDTY1.1. DNA of the populations was extracted from freeze-dried leaf samples that were cut in Eppendorf tubes and ground with a GENO grinder. Extraction was carried out by the modified CTAB method46. DNA samples were stored in 2-mL deep-well plates (Axygen Scientific, California, USA). DNA samples were quantified on 0.8% agarose gel and concentration adjusted to approximately 25 ng μL−1. For SSR analysis, PCR amplification was carried out with a 15-μL reaction mixture having 50 ng DNA, 1 × PCR buffer, 100 μM dNTPs, 250 μM primers and 1 unit Taq polymerase enzyme. To resolve the PCR products, 8% non-denaturing polyacrylamide gels (PAGE) were used47. The set of 123 random genotypes was genotyped with 65 SSR markers across 11 DTY loci.
Genotyping for the alleles of the sd1 gene
Apart from phenotypic measurements for the presence of the sd1 gene, the dwarfing allele of sd1, characterized by a 383-bp deletion spanning the first intron and second exon of the gene, was assayed by PCR amplification using primers flanking this deletion. The primers used were sd1-Forward: 5′-CACGCACGG GTTCTTCCAGGTG-3′ and sd1-Reverse: 5′-AGG AGAATA GGA GAT GGT TTA CC- 3′48. Product sizes for the dwarfing allele and the tallness allele were 348 and 731 bp, respectively. This region of the sd1 gene is difficult to amplify consistently because of its high GC content; thus the PCR reaction was optimized for consistent amplification with minor modifications to the protocol of Masouleh49. For a 20-μl PCR reaction, 2 μl of 10x Enhancer buffer from Invtrogen (50 mM ), 1.2 μl of MgCl2 (50 mkM ), 3.0 μl of primers (5 pM each), 1 μl of dNTPs (2.5 mM each), 6 μl of 2X Enhancer, 0.2 μl of Platinum® Taq polymerase (5 U/μl) and 2.0 μl (20 ng/μl) of template were used. The thermocycling program consisted of initial denaturation at 94 °C for 5 min, followed by 45 cycles of amplification at 94 °C for 30 s, 55 °C for 30 s and 72 °C for 1 min and a final extension at 72 °C for 3 min. The PCR product was visualized by electrophoresis in 2% agarose gel.
Sequenom MassARRAY assay for fine mapping of qDTY1.1
A 29-plex Sequenom SNP assay was designed and validated for fine mapping of qDTY1.1 located between the 36.04- and 40.70-Mb region on the long arm of rice chromosome 1 (Fig. 2, Table S6). A total of 46 SNPs located in the 36.36- to 40.06-Mb region on chromosome 1 and polymorphic between parents N22 and Swarna (cf. OryzaSNP database; http://oryzasnp.plantbiology.msu.edu/) were taken for assay design using the Sequenom MALDI-TOF MassARRAY system50. The multiplex assay was designed using MassARRAY assay design 4.0 software and the reaction for a single tube multiplex containing 29 assays was optimized. The assay also included the sd1 deletion locus with three possible SNP alleles, in addition to 28 SNPs located in 28 different genes in the QTL region (Fig. 2). After standardization of the primer concentrations for MALDI-TOF and validation of the assays in test samples, DNA samples of BILs showing recombination between qDTY1.1 and QTL flanking markers in a large N22/Swarna fine-mapping population and 46 random varieties were analyzed. All SNP calls worked well, except for one designed for the sd1 del, for which only one of the alleles (T), representing the wild-type (tall) allele was called because of the high GC content of the amplicon resulting in poor amplification.
The primers were procured from IDT (Heverlee, Belgium). The iPLEX GOLD SNP genotyping was performed as per the manufacturer’s protocols and the genotype calls were analyzed using Sequenom Typer 4.0 software. The SNP calls in ACGT format generated for all the samples were converted to A, B and H format for easy visualization of the recombinant break points. In cases of heterozygote calls (due to incomplete fixing of the inbred lines), a manual inspection of peak intensity was carried out to cross-check the validity of the calls. Data were analyzed using graphical genotyping of the qDTY1.1 region of the SNP calls from recombinant lines to identify the recombination break points and to find association with the phenotyping data.
Statistical analysis was conducted using CROPSTAT v.4.2.3 (available at www.irri.org). Phenotypic means of entries were estimated using the following linear mixed model for the analysis of variance:
where Pijk is the measurement recorded on a plot, M is the mean over all plots and R, B, L and e are replications, blocks, lines and error, respectively. For estimating the entry means, replications and blocks within replicates were considered random whereas entries were fixed. While estimating the entry means across years, season effects were also taken as random.
Statistical analyses for the physiology experiments were performed in R v. 2.15.251 by analysis of variance (ANOVA) (aov script) test for genotype and replication and least significant difference (LSD) mean comparison was used as the post-hoc test. For measurements conducted on multiple dates, a repeated measures analysis was conducted with the mixed model ASREML using Wald’s test in R, with ‘genotype’ and ‘days after sowing’ as fixed variables and ‘replicate’ as a random variable.
Multiple trait analysis
Multiple trait−multiple interval mapping (MT-MIM) analysis in the N22/IR64 and N22/Swarna RIL population was carried out as described by21. The analysis was performed with QGene software version 4.3.10 with 1000 permutations at significance level P < 0.0152. Combined mean phenotypic data of two years were used for the analysis. The model used for MT-MIM was
where q is the number of QTLs being fitted simultaneously, t is the number of analyzed traits, p is the number of non-genetic fixed factors, a is the additive effect, d is the dominant effect, X is the non-genetic fixed effect, B is the incidence matrix that links observation of the data with fixed effects, E is random error and i = 1 − d.
QTL analysis using additional markers in the qDTY1.1 region
Genetic map distances between markers were estimated according to their position on the physical map in reference to rice variety Nipponbare. One mega base pair (Mb) was assumed to be equivalent to ~3.92 cM53. The QTL analysis in different populations was performed using Inclusive Composite Interval Mapping (ICIM) software54. Results were also confirmed with QTL Network software v2.1 following a mixed model-based composite interval mapping method described by Yang et al.55. Candidate intervals were first selected, the significance of the association of intervals with the trait of interest was analyzed and then the additive effect explained by the significantly associated interval was estimated. The significance level for determination of the candidate intervals, detection of putative QTLs and their additive effects was P < 0.01.The F-value threshold for significance of the QTL was determined using 1000 permutation tests. The window size and walk speed used for the genome scan were 10 cM and 1 cM, respectively.
Mapping of qDTY1.1 for improving resolution
A three-way approach was followed for the fine mapping of qDTY1.1 through N22-derived RILs, RIL recombinants and N22/4*Swarna BIL populations. (i) In order to identify the break-point events between sd1 and qDTY1.1, three SNP markers were added in and around sd1. One SNP was between RM11943 and sd1 (id1024167) and two SNPs were between sd1 and RM431 (id1024366 and id1024499). These markers were based on a panel of 44,000 SNPs published by Zhao et al.56. Genotyping was carried out using the Fluidigm SNP genotyping platform by K-Biosciences Ltd. QTL analysis was conducted on the phenotypic data from two field seasons. (ii) These additional SNPs were used to test the recombination between sd1 and qDTY1.1. Recombinant lines among RILs between markers RM11943 and RM431 flanking both the sd1 gene and qDTY1.1 locus were identified. The tight linkage between sd1 and qDTY1.1 reduced the probability of double recombination in this interval; therefore, only the recombinant lines between the flanking markers were analyzed to find lines with break points between sd1 and qDTY1.1. Further, because of a low number of recombinant lines in the individual populations, a pooled analysis of the 161 recombinant lines (48 from N22/Swarna, 57 from N22/IR64 and 56 from N22/MTU1010) from the three populations was carried out. Segregation for the sd1 gene was analyzed using a functional marker as described above. (iii) The dwarf N22/Swarna RIL population was genotyped with SSR markers of qDTY1.1 (RM11943, RM431, RM12023, RM12091 and RM12146), SNP markers of qDTY1.1, as well as with the sd1 gene-based indel marker. This population was homozygous for the dwarf sd1 allele.
Association analysis in a random set of varieties
Genotyping of 123 random varieties was conducted with genome-wide SNP markers within the qDTY1.1 region and an sd1 gene-based functional marker. A genome-wide SNP panel was used to determine the population structure. A Sequenom MassARRAY multiplex assay was designed with 72 SNPs, including two wells of 36 plexi PLEX gold chemistry. These 72 SNPs represented 72 conserved single-copy rice genes and six genes/chromosomes, two genes each for all the telomeric and centromeric regions57. All SNPs were used for the population structure analysis as determined by STRUCTURE v.2.3.3 Software58. A burn-in period of 100000 with 10000 MCMC was followed. The qDTY1.1-specific SNP assay (Fig. 2) was used for marker-trait associations. Association of qDTY1.1 SNP markers as well as the sd1 gene with drought-related traits was analyzed using population structure file, kinship matrix and genotypic and drought-related phenotypic data of the 123 rice genotypes (Table S7) through TASSEL59 using the mixed linear model (MLM).
Genetic diversity analysis of a set of diverse varieties
The genetic diversity analysis on the set of diverse varieties was conducted using the software PowerMarker V3.2560. Allele frequency was obtained using the bootstrap method with bootstrap number 10,000 and confidence interval 0.950. The frequency-based distance was obtained by the C.S. Chord method61. The neighbor-joining tree was constructed according to frequency-based distances obtained from PowerMarker using the software MEGA 5.262.
QTL class analysis
Recombinant classes were identified within qDTY3.1, qDTY3.2, and qDTY6.2 based on the SSR marker data available for the three QTLs. Class means calculated based on the mean data of different recombinant classes were used to estimate the class effect on the respective phenotypic traits.
How to cite this article: Vikram, P. et al. Drought susceptibility of modern rice varieties: an effect of linkage of drought tolerance with undesirable traits. Sci. Rep. 5, 14799; doi: 10.1038/srep14799 (2015).
Frei, M. & Becker, K. Agro-biodiversity in subsistence-oriented farming systems in a Philippine upland region: nutritional considerations. Biodivers. Conserv. 13, 1591−1610 (2004).
Hazell, P. B. R. “The Asian Green Revolution, 2009.” International Food Policy Research Institute, Washington, DC: 34 (2009).
Kanyama-Phiri, G., Wellard, K. & Snapp, S. in Agricultural systems: agroecology and rural innovation for development (eds Snapp, S. & Pound, B. ) 1–24 (Academic Press, 2008).
Venuprasad, R. et al. Identification and characterization of large-effect quantitative trait loci for grain yield under lowland drought stress in rice using bulk-segregant analysis. Theor. Appl. Genet. 120, 177−190 (2009).
Vikram, P. et al. qDTY1.1, a major QTL for rice grain yield under reproductive-stage drought stress with a consistent effect in multiple elite genetic backgrounds. BMC Genet. 12, 89 (2011).
Kumar, A. et al. High-yielding, drought-tolerant, stable rice genotypes for the shallow rainfed lowland drought-prone ecosystem. Field Crops Res. 133, 37−47 (2012).
Ray, D., Mueller, N., West, P. & Foley, J. Yield trends are insufficient to double global crop production by 2050. PLoS One 8(6), e66428 (2013).
Neeraja, C. N. et al. A marker-assisted backcross approach for developing submergence-tolerant rice cultivars. Theor. Appl. Genet. 115, 767–776 (2007).
Chin, J. H. et al. Development and application of gene-based markers for the major rice QTL Phosphorus uptake 1. Theor. Appl. Genet. 120, 1073–1086 (2010).
Linh, L. H. et al. Molecular breeding to improve salt tolerance of rice (Oryza sativa L.) in the Red River Delta of Vietnam. Int. J. Plant Genomics Volume 2012, Article ID 949038, 10.1155/2012/949038 (2012).
Kumar, A. et al. Breeding high-yielding drought-tolerant rice: genetic variations and conventional and molecular approaches. J. Exp. Bot. 10.1093/jxb/eru363 (2014) (Accessed on 26/03/2015).
Swamy, B. P. M. et al. Genetic, physiological and gene expression analyses reveal that multiple QTL enhance yield of rice mega-variety IR64 under drought. PLoS ONE 8, e62795 (2013).
Ghimire, K. H. et al. Identification and mapping of a QTL (qDTY1.1) with a consistent effect on grain yield under drought. Field Crops Res. 131, 88−96 (2012).
Monna, L. et al. Positional cloning of rice semidwarfing gene, sd-1: rice “Green Revolution gene” encodes a mutant enzyme involved in gibberellin synthesis. DNA Res. 9, 11–17 (2002).
Babu, R. C. et al. Genetic analysis of drought resistance in rice by molecular markers: association between secondary traits and field performance. Crop Sci. 43, 1457–1469 (2003).
Kumar, R. et al. Genetic analysis of rainfed lowland rice drought tolerance under naturally-occurring stress in eastern India: heritability and QTL effects. Field Crops Res. 103, 42–52 (2007).
Khowaja, F. S., Norton, G. J., Courtois, B. & Price, A. H. Improved resolution in the position of drought-related QTLs in a single mapping population of rice by meta-analysis. BMC Genomics 10, 276 (2009).
Venuprasad, R. et al. A large-effect QTL for rice grain yield under upland drought stress on chromosome 1.Mol. Breed. 30, 535−547 (2012).
3K RGP: The 3,000 rice genomes project. GigaScience 3, 7 (2014).
Spielmeyer, W. et al. Semidwarf (sd-1), “green revolution” rice, contains a defective gibberellin 20-oxidase gene. Proc. Natl. Acad. Sci. USA 99, 9043–9048 (2002).
Kumawat, G. et al. Molecular mapping of QTLs for plant type and earliness traits in pigeonpea (Cajanus cajan L. Millsp.). BMC Genet. 13, 84 (2012).
Hu, H. et al. Overexpressing a NAM, ATAF and CUC (NAC) transcription factor enhances drought resistance and salt tolerance in rice. Proc. Natl. Acad. Sci. USA 103, 12987–12992 (2006).
Jeong, J. S. et al. Root-specific expression of OsNAC10 improves drought tolerance and grain yield in rice under field drought conditions. Plant Physiol. 153, 185–197 (2010).
Gorantla, M. et al. Identification of stress-responsive genes in an indica rice (Oryza sativa L.) using ESTs generated from drought stressed seedlings. J. Exp. Bot. 58, 253–265 (2007).
Soren, K. R., Ali, K. & Tyagi, A. Cloning and characterization of gene encoding novel zinc finger protein transcription factor induced under water deficit stress from rice (Oryza sative) cv. N22. Indian J. Biochem. Biophysiol. 49, 36−41 (2011).
Xu, D. Q. et al. Overexpression of a TFIIIA-type zinc finger protein gene ZFP252 enhances drought and salt tolerance in rice (Oryza sativa L.). FEBS Lett. 582, 1037–1043 (2008).
Xiao, B. Z. et al. Evaluation of seven function-known candidate genes for their effects on improving drought resistance of transgenic rice under field conditions. Mol. Plant. 2, 73–83 (2009).
Wu, X., Shiroto, Y., Kishitani, S., Ito, Y. & Toriyama, K. Enhanced heat and drought tolerance in transgenic rice seedlings overexpressing OsWRKY11 under the control of HSP101 promoter. Plant Cell Rep. 28, 21–30 (2009).
Mishra, K. K. et al. qDTY12.1: a locus with a consistent effect on grain yield under drought in rice. BMC Genet. 14, 12 (2013).
Bernier, J., Kumar, A., Venuprasad, R., Spaner, D. & Atlin, G. N. A large-effect QTL for grain yield under reproductive-stage drought stress in upland rice. Crop Sci. 47, 507–516 (2007).
Yadaw, R. et al. A QTL for high grain yield under lowland drought in the background of popular rice variety Sabitri from Nepal. Field Crops Res. 144, 281–287 (2013).
Dixit, S. et al. Increased drought tolerance and wider adaptability of qDTY12.1 conferred by its interaction with qDTY2.3 and qDTY3.2. Mol. Breed. 30, 1767−1779 (2012).
Dixit, S. et al. Multiple major QTL lead to stable yield performance of rice cultivars across varying drought intensities. BMC Genet. 15, 16 (2014).
Kadam, S. et al. Genomic associations for drought tolerance on the short arm of wheat chromosome 4B. Funct. Integr. Genomics 12, 447–464 (2012).
McCartney, C. A. et al. Mapping quantitative trait loci controlling agronomic traits in the spring wheat cross RL4452 9 ‘AC Domain’. Genome 48, 870–883 (2005).
Dashti, H. et al. QTL analysis for drought resistance in wheat using doubled haploid lines. Int. J. Agric. Biol. 9, 98–101 (2007).
Mathews, K. L. et al. Multi-environment QTL mixed models for drought stress adaptation in wheat. Theor. Appl. Genet. 117, 1077–1091 (2008).
Pinto, R. S. et al. Heat and drought adaptive QTL in a wheat population designed to minimize confounding agronomic effects. Theor. Appl. Genet. 121, 1001–1021 (2010).
Sabadin, P. K. et al. Studying the genetic basis of drought tolerance in sorghum by managed stress trials and adjustments for phenological and plant height differences. Theor. Appl. Genet. 124, 1389−1402 (2012).
Swamy, B. P. M. et al. Meta-analysis of grain yield QTL identified during agricultural drought in grasses showed consensus. BMC Genomics 12, 319 (2011).
Venuprasad, R. et al. Response to direct selection for grain yield under drought stress in rice. Crop Sci. 47, 285–293 (2007).
Alexandrov N. et al. SNP-Seek database of SNPs derived from 3000 rice genomes. Nucl. Acids Res. 43, D1023−D1027 (2015).
Kumar, A. et al. Breeding for drought tolerance: direct selection for yield, response to selection and use of drought-tolerant donors in upland and lowland-adapted populations. Field Crops Res. 107, 221–231 (2008).
Kijoji, A. A. et al. Water extraction and root traits in Oryza sativa × Oryza glaberrima introgression lines under different soil moisture regimes. Funct. Plant Biol. 40, 54−66 (2013).
Henry, A. et al. Genetic variation in root architecture and drought response in Oryza sativa: rainfed lowland field studies of the Oryza SNP panel. Field Crops Res. 120, 205−214 (2011).
Murray, M. G. & Thompson, W. F. Rapid isolation of high molecular weight plant DNA. Nucl. Acids Res. 8, 4321–4326 (1980).
Sambrook, J., Fritsch, E. F. & Maniatis. T. Molecular cloning: a laboratory manual, 2nd edn. Cold Spring Harbor, New York (1989).
Ellis, M. & Spielmeyer, W. “Perfect” markers for the rice sd1 semi-dwarfing gene. Int. Rice Res. Notes 27(2), 13−14 (2002).
Masouleh, A., Waters, D. L. E., Reinke, R. F.. & Henry, R. J. A high throughput assay for rapid and simultaneous analysis of perfect markers for important quality and agronomic traits in rice using multiplexed MALDI-TOF mass spectrometry. Plant Biotechnol. J. 7, 355−363 (2009).
www.sequenom.com. Sequenom Inc. (SQNM) 2015. Date of access: 26/03/2015.
R Foundation for Statistical Computing, R version 2.15.2 (2012).
Joehanes, R. & Nelson, J. C. QGene 4.0, an extensible java QTL-analysis platform. Bioinformatics 24, 2788–2789 (2008).
IRGSP. The map-based sequence of the rice genome. Nature 436, 793–800 (2005).
Li, H., Ye, G. Y. & Wang, J. A modified algorithm for the improvement of composite interval mapping. Genetics 175, 1–14 (2007).
Yang, J. et al. QTL Network: mapping and visualizing genetic architecture of complex traits in experimental populations. Bioinformatics 24, 721–723 (2008).
Zhao, K. et al. Genome-wide association mapping reveals a rich genetic architecture of complex traits in Oryza sativa. Nat. Commun. 2, 467 (2011).
Singh, A. et al. SNP haplotypes of the BADH1 gene and their association with aroma in rice (Oryza sativa L.). Mol. Breed. 26, 325–338 (2010).
Pritchard, J. K., Stephens, M. & Donnelly, P. Inference of population structure using multilocus genotype data. Genetics 155, 945–959 (2000).
Bradbury, P. J., Zhang, Z., Kroon, D. E., Casstevens, T. M., Ramdoss, Y. & Buckler, E. S. TASSEL: software for association mapping of complex traits in diverse samples. Bioinformatics 23, 2633–2635 (2007).
Liu, K. & Muse, S. V. Powermarker: integrated analysis environment for genetic marker data. Bioinformatics 21, 2128–2129 (2005).
Cavalli-Sforza, L. L. & Edwards, A. W. F. Phylogenetic analysis: models and estimation procedures. Am. J. Human Genet. 19, 233–257 (1967).
Tamura, K. et al. MEGA 5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance and maximum parsimony methods. Mol. Biol. Evol. 28, 2731–2739 (2011).
The authors thank the Generation Challenge Programme (GCP) and the Department of Biotechnology (DBT), Government of India, for providing financial support for this study.
The authors declare no competing financial interests.
Electronic supplementary material
About this article
Cite this article
Vikram, P., Swamy, B., Dixit, S. et al. Drought susceptibility of modern rice varieties: an effect of linkage of drought tolerance with undesirable traits. Sci Rep 5, 14799 (2015). https://doi.org/10.1038/srep14799
In silico identification and evaluation of Bacillus subtilis cold shock protein B (cspB)-like plant RNA chaperones
Journal of Biomolecular Structure and Dynamics (2020)
Genetic Mapping Identifies Consistent Quantitative Trait Loci for Yield Traits of Rice under Greenhouse Drought Conditions
Comparative Transcriptomics and Co-Expression Networks Reveal Tissue- and Genotype-Specific Responses of qDTYs to Reproductive-Stage Drought Stress in Rice (Oryza sativa L.)
Journal of the Science of Food and Agriculture (2020)
Trends in Ecology & Evolution (2020)