Genetica

, Volume 144, Issue 6, pp 651–664 | Cite as

Genome wide association study (GWAS) for grain yield in rice cultivated under water deficit

  • Gabriel Feresin Pantalião
  • Marcelo Narciso
  • Cléber Guimarães
  • Adriano Castro
  • José Manoel Colombari
  • Flavio Breseghello
  • Luana Rodrigues
  • Rosana Pereira Vianello
  • Tereza Oliveira Borba
  • Claudio Brondani
Article

Abstract

The identification of rice drought tolerant materials is crucial for the development of best performing cultivars for the upland cultivation system. This study aimed to identify markers and candidate genes associated with drought tolerance by Genome Wide Association Study analysis, in order to develop tools for use in rice breeding programs. This analysis was made with 175 upland rice accessions (Oryza sativa), evaluated in experiments with and without water restriction, and 150,325 SNPs. Thirteen SNP markers associated with yield under drought conditions were identified. Through stepwise regression analysis, eight SNP markers were selected and validated in silico, and when tested by PCR, two out of the eight SNP markers were able to identify a group of rice genotypes with higher productivity under drought. These results are encouraging for deriving markers for the routine analysis of marker assisted selection. From the drought experiment, including the genes inherited in linkage blocks, 50 genes were identified, from which 30 were annotated, and 10 were previously related to drought and/or abiotic stress tolerance, such as the transcription factors WRKY and Apetala2, and protein kinases.

Keywords

Oryza sativa L. SNPs Rice core collection Genotyping by sequencing 

Introduction

The impact of climate change on agricultural productivity is likely to be a major constraint to achieving increased food production (Henry 2014). Drought is a major problem for rainfed rice cultivation due to unevenly distributed rainfall patterns, and if it occurs in the reproductive stage, it can lead to a total production loss. The upland rice, although less productive than lowland rice, has the advantage of being cultivated in an aerobic environment, without flooding, thus saving a scarce resource. The large and variable area under rice cultivation as well as different methods of rice cultivation (direct-seeded upland, transplanted lowland) make the crop unique in terms of its inherent variability available for drought tolerance compared with other cereals (Kumar et al. 2014). Currently, improvement of tolerance to abiotic stresses in popular high-yielding varieties has been done by crossing with traditional varieties or wild relatives, for which the traditional varieties were used as donors and the high-yielding varieties provided the desired yield potential and preferred grain quality. However, abiotic stress tolerance genes may be related to unfavorable traits, such as low yield potential, excessive plant height, poor grain quality, leading to their elimination from varieties during the selection process (Vikram et al. 2015). Advances in molecular biology have provided new opportunities for breeders to identify such regions, refine these through fine mapping, and integrate them into drought-susceptible varieties, an opportunity that was not available a few years ago to break the yield improvement barrier under drought (Kumar et al. 2014). Using grain yield under reproductive-stage drought as a selection criterion, a number of large-effect QTLs for both upland and lowland conditions have been identified, including qDTY12.1, which was the first reported large-effect QTL for this trait (Bernier et al. 2007).

The rapid development of drought-tolerant versions of popular varieties could be one of the strategies to ensure rice production under reproductive-stage drought without compromising on yield potential and the preferences of both farmers and consumers. Because of the low positive phenotypic correlation between high yield potential and grain yield under reproductive-stage drought, marker-assisted breeding, using well-defined QTLs, allows for the precise combination of high yield potential and good yield under reproductive-stage drought (Vikram et al. 2011). Moreover, marker-assisted breeding also allows rapid product development with reduced efforts and with relatively smaller segregating populations. However, marker-assisted breeding for drought tolerance requires careful planning from the start of the QTL identification process (Kumar et al. 2014). As an example, consistent efforts have been made to introgress the identified qDTYs markers into drought-susceptible rice varieties through the marker-assisted breeding strategy (Shamsudin et al. 2016).

The Genome Wide Association Study (GWAS) methodology is based on the genotyping of hundreds of individuals with reduced genetic relationship, and the association of this data with the phenotype of interest. GWAS takes advantage of high density SNP markers, associating molecular polymorphisms with the phenotype, mostly using genetic diversity and recombination that evolved naturally through many generations and through directional selection. Compared with biparental QTL studies, it captures the available allelic diversity on the loci while achieving greater physical resolution due to lower linkage disequilibrium (LD) in diversity panels. GWAS is, therefore, an efficient way to dissect the genetic architecture of complex traits and a powerful tool for crop breeding (Rebolledo et al. 2015). Marker-assisted breeding will remain a valid option for major loci or QTL, while QTL cloning will provide novel opportunities for genetic engineering for abiotic stress tolerance and for a more targeted search for novel alleles in wild germplasm (Kole et al. 2015).

The use of genetic and genomic analysis to identify DNA markers associated to stress tolerance can facilitate breeding strategies for crop improvement. This approach is particularly useful when target characters are controlled by several genes, as in the case of abiotic stress tolerance. The potential to map different markers contributing to a given agronomic trait will open up the possibility to simultaneously transfer several QTLs and to pyramid QTLs in one improved cultivar (Perez-Clemente et al. 2013). The GWAS analysis results may have a powerful application in genomic selection using the significantly identified SNPs as cofactors, increasing the accuracy of prediction. Such selected SNPs have great potential and are recommended for validation in future stages of assisted selection and genomic selection, and may be used into a routine of plant breeding programs (He et al. 2014; Zhang et al. 2014). This study aimed to identify markers and candidate genes associated with drought tolerance, in order to develop tools for use in rice breeding programs.

Materials and methods

Plant material

The panel for GWAS consisted of 175 upland rice (O. sativa ssp. japonica) accessions from the Embrapa Rice Core Collection (Abadie et al. 2005), and included 83 Brazilian landraces, 47 improved varieties from Brazil and 45 improved varieties from other countries (Supplemental Table 1).

Phenotyping

Two experiments were conducted to evaluate the 175 rice accessions, one with water deficit, hereafter named drought experiment, and another without water deficit, named control experiment. Both experiments were conducted in 2013, at the Drought Phenotyping Experimental Station, located in Porangatu, Goias State, Brazil (49°06′W, 13°18′S, 396 m of altitude). The experimental design was randomized complete blocks, with two replications. Plot size consisted of four lanes of 3 m with a density of 60 seeds per meter. The control experiment was monitored by tensiometers at a depth of 15 cm, in order to maintain suitable soil water conditions (−0.025 MPa). The drought experiment received the same amount of water until thirty days after emergence, when irrigation was reduced to approximately 50 % of the control until the end of the experiment. Statistical analysis of grain yield data (kg ha−1) was performed using the lme4 package from R software version 3.0.1 (The R Foundation for Statistical Computing 2015) using the mixed model analysis procedure. The treatment effects (Control experiment and drought experiment) were considered fixed because the treatments in the experiment are the only ones to which inference is to be made. Block effects were considered random because the blocks in the experiment are only a small subset of the larger set of blocks over which inference about treatment means is to be made. Estimates of Blups (best linear unbiased prediction) of the yield of each accession were used for the association analysis and are depicted in Fig. 1. The predicted values of the random effects (EBLUP) associated with each accession contain a part attributed to the constant \(\widehat{{\mu_{p} }}\) estimate, and another corresponding to the genotypic effect (\(\widetilde{{g _{i} }}\)) of each individual accession.
Fig. 1

Box plot of productivity values of the evaluated genotypes of rice in irrigated experiment (control) and water deficit experiment (drought)

A third experiment was carried out in an automated station for drought phenotyping (named Sitis), at the Experimental Station of Embrapa Rice and Beans, located in Goiânia, Goiás State, Brazil (49°17′W, 16°28′S, 779 m of altitude). The objective of this experiment was to validate the results obtained in the field using 20 genotypes common to both experiments (Supplementary Table 1), two irrigation levels (drought and control), with 6 repetitions each level, in the Lattice 5 × 5 experimental design. The water deficit consisted of the irrigation regime of 50 % of field capacity for 20 days, in the reproductive period (stage R1–R5). At the end of the water-restriction period, the stressed plants were irrigated again (80–95 % of field capacity) until the end of the cultivation cycle to determine their productivity (g grains pot−1) to evaluate the severity of water deprivation. Statistical analysis was performed as described in the field experiment.

Genotyping

The DNA was extracted using the DNeasy 96 Plant Kit (QIAGEN) following the manufacturer’s instructions. For each accession, a single individual plant was used. The SNP markers were obtained by GBS methodology (Elshire et al. 2011) at the Genomic Diversity Institute at Cornell University (USA). Data generation was carried out in a Genome Analyzer II platform (Illumina, Inc., San Diego, CA) and sequencing was the single-end type with 96-plex samples. DNA digestion was performed using ApeKI, which was shown to cut every 1 kb on average in an in silico digestion of the Nipponbare reference genome. Raw sequence data filtering, sequence alignment to the rice reference genome (Os-Nipponbare-Reference-IRGSP-1.0) and SNP calling from low-coverage GBS genotyping, were carried out using the TASSEL GBS pipeline v5.0 provided by the Buckler Lab for Maize Genetics and Diversity. The basic bioinformatics and filtering of the raw data were based on previous estimates of linkage disequilibrium and inbreeding rates obtained for rice. After anchoring the reads in the reference genome (Nipponbare, available at the rice genome annotation project—RGAP), SNPs were identified in each accession, with a minor allele frequency (MAF) set as 0.01. Default memory size parameters were modified according to data size. An inbreeding coefficient value equal to 0.9 was also considered, with minimum locus coverage equal to 0.1, which in turn corresponds to the proportion of accessions with at least one tag in a locus. Sequences were aligned to the genome with the “Burrows-Wheeler Aligner” (BWA) using default parameters for genomes smaller than 2 G. The imputation of missing genotype data was performed by the software FastPHASE 1.3 (Scheet and Stephens 2006).

Population structure

Population structure was estimated using the Bayesian model of the Markov chain Monte Carlo (MCMC) implemented in STRUCTURE program (Pritchard et al. 2000). Five iterations were performed for each number of populations (k) tested from 1 to 10. Burn-in value and number of replications of MCMC were set at 50,000 and 100,000, respectively. The K value was determined by the data log likelihood [LNP (D)] and delta K, based on the change rate of [LNP (D)] between successive values of k. These analyzes were performed using the Structure Harvester program (Earl and Vonholdt 2011).

Genome wide association study analysis

From the population structure and genetic relatedness data, we used the Compressed Mixed Linear Model module (MLM) statistical model (Zhang et al. 2010) with Q and K matrices, to perform association analysis between grain yield and SNP data. The analysis of genome-wide association was conducted with the TASSEL version 5.0 software (Bradbury et al. 2007) using the default settings for analysis of variance. The structuring data was obtained by the STRUCTURE software and the relationship matrix (K matrix or kinship) was obtained by the TASSEL 4.0 software and GAPIT R package. For better reliability of GWAS, rare alleles were removed by filtering the imputed SNP data with a minimum frequency allele (MAF) 0.05. The markers were defined as being significantly associated to yield based on p < 0.05. The Manhattan plot distribution chart for each experiment (drought and control) was obtained by the R software. After the GWAS analysis, the SNPs associated to yield in both experiments were submitted to a regression analysis, in order to select informative markers that could be used in marker assisted selection. The association between SNP genotypes and grain yield phenotypes was tested fitting one SNP at a time in a mixed linear model of the following form (in matrix notation):
$$y = 1\mu + Xb + Za + e$$
where y is the vector of phenotypic observations; μ is the overall mean; b is the vector of SNP effects, with the corresponding matrix X of SNP genotypes (either 0 or 1 for the two homozygous classes—AA, BB); a is the vector of polygenic effects with the related incidence matrix Z; e is the vector of residuals.

SNP effect

For each significant SNP associated to grain yield, regression beta values were obtained (SNP effect) from equation \(y = 1\mu + Xb + Za + e\), which is related to the minor allele of each SNP. This SNP effect aims to highlight which allele has a positive effect on the grain yield trait.

Stepwise regression

A stepwise regression analysis (forward and reverse) was performed from the method for conditional and joint analysis using summary statistics from GWAS implemented in the GCTA tool in UNIX (Yang et al. 2011), in order to remove SNPs with overlapping effect in grain yield. The proportion of the phenotypic variance explained by SNPs was obtained by R2 values before and after the removal of a SNP in the regression model. Then, an analysis of variance was performed to identify significant differences between these R2 values obtained before and after the stepwise regression analysis.

Haplotype analysis

Haplotype blocks were calculated for grain yield for drought and control experiments using the programs Haploview (Barrett et al. 2005) and SNP & Variation Suite v8 (Golden Helix, Inc., Bozeman, USA). The input genotype file contained all SNPs that passed through a MAF filter of 0.05. The algorithm used to define the haplotypic blocks and calculate the haplotype frequencies was described by Gabriel et al. (2002).

Development of SNP marker set

To validate a SNP set for marker-assisted selection, TaqMan Real-Time PCR assays were developed (Thermo Fisher Scientific, USA) from the DNA sequence adjacent to each SNP associated with drought tolerance. These assays were used to genotype the 20 upland rice genotypes evaluated in both drought experiments. The PCR reactions were carried out in a QuantStudio 7 Flex Real-Time PCR System (Applied Biosystems, USA) using the TaqMan master mix GTXpress kit, according to the manufacturer’s recommendations.

Identification of transcripts in haplotype blocks

The significant SNPs from the GWAS analysis were positioned in haplotypic blocks, followed by the identification of SNPs located in transcripts from the Rice Genome Annotation Project databank (Kawahara et al. 2013). Aminoacid sequences of genes that still were not annotated were used as templates to look for homologous genes in Arabidopsis by Blast analysis at the TAIR website (www.arabidopsis.org).

Results

Field experiments

Mean grain yield estimated by Blups in the drought experiment ranged from 570.9 kg ha−1 (Farroupilha landrace) to 2969.2 kg ha−1 (Aimoré variety) (Table 1), with an overall mean of 1459.8 kg ha−1. In the control experiment, yield ranged from 799.3 kg ha−1 (Pingo de Ouro landrace) to 4476.9 kg ha−1 (inbred line CT11891) (Table 1), with an overall mean of 2727.1 kg ha−1, i.e., a 53.5 % increase in relation to the average yield of the drought experiment. The yield boxplot chart showed a greater dispersion of data in the control experiment (Fig. 1).
Table 1

List of genotypes with higher (10+) and lower (10−) productivity from the experiments with (drought) and without (control) water deficit

Drought experiment

Control experiment

10+ genotypes

Code

Country

Yield (kg ha−1)

10+ genotypes

Code

Country

Yield (kg ha−1)

AIMORÉ

CNA0007119

Brazil

2969.2

CT11891

CNA0009123

Colombia

4476.9

GUARANI

CNA0004121

Brazil

2872.9

AIMORÉ

CNA0007119

Brazil

4433.3

CARAJÁS

CNA0006701

Brazil

2852.5

7 SEMANAS

CA790116

Brazil

4330.2

URUÇUÍ

CNA0005901

Brazil

2811.4

CASCA BRANCA

CA870160

Brazil

4226.8

CAROLINO

CA820103

Brazil

2589.5

URUÇUÍ

CNA0005901

Brazil

4210.9

IRAT 112

CNA0006574

France

2540.2

LS 85-158

CNA0006174

Brazil

4158.4

4 MESES

CA870158

Brazil

2538.5

A12-286

CNA0006666

Brazil

4083.2

DOURADÃO

CA800015

Brazil

2486.6

3 MESES BRANCO

CA780021

Brazil

4073.8

TANGARÁ

CNA0005180

Brazil

2389.9

IAC 165

CNA0002075 

Brazil

4058.7

CASCA BRANCA

CA870160

Brazil

2386.3

ITA 150

CNA0006034

Nigeria

4016.9

YN1906-UUL 65

CNA0010506

Philippines

715.5

BICO GANGA

CA870142

Brazil

1373.6

CHATÃO BURITI

CA800173

Brazil

686.6

CHATAO AMARELO

CA820092

Brazil

1293.0

PRETO

CA210006

Brazil

680.4

L 285

CNA0008093

France

1228.4

BR4742

CNA0010527

Philippines

674.1

TOX 1871

CNA0005334

Nigeria

1218.7

IREM 238

CNA0003288

France

634.3

CANA ROXA

CA870186

Brazil

1190.7

ROXO

CA870177

Brazil

633.4

MEARIN

CNA0003490

Brazil

1188.2

BICO PRETO

CA820083

Brazil

629.4

IAC 81

CNA0005673

Brazil

1066.7

TB47H

CNA0010469

Philippines

626.8

JAGUARIZINHO

CA820040

Brazil

991.1

BLUEBELLE

CNA0008411

USA

580.8

CATETINHO

CA800092

Brazil

926.5

FARROUPILHA

CA940008

Brazil

570.9

PINGO DE OURO

CA840128

Brazil

799.3

Considering the 10 most productive accessions from the drought experiment, nine are from Brazil (five varieties and four landraces), and considering the 10 less productive materials, five are foreign varieties and five are Brazilian landraces (Table 1). In the control experiment, eight of the 10 most productive accessions are from Brazil (five varieties and three landraces), and considering the 10 least productive, eight are from Brazil (two varieties and six landraces).

SNP genotyping

The total number of SNPs was 150,325, distributed across the 12 rice chromosomes. An average of 402 SNPs/Mbp was found, ranging from a minimum of 327 SNPs/Mbp on chromosome 5 and a maximum of 463 SNPs/Mbp on chromosome 11. Chromosome 1 showed the highest number of SNPs (19,323), while chromosome 9 showed the smallest number of SNPs (8729), with an average of 12,527 SNPs per chromosome (Table 2).
Table 2

SNP distribution and frequency from GBS analysis of 175 rice accessions

Chromosome

Length (bp)

No. of SNPs

SNPs/Mb

1

43,270,923

19,323

446.56

2

35,937,250

14,510

403.76

3

36,413,819

15,579

427.83

4

35,502,694

13,304

374.73

5

29,958,434

9819

327.75

6

31,248,787

12,572

402.32

7

29,697,621

12,094

407.24

8

28,443,022

11,292

397.00

9

23,012,720

8729

379.31

10

23,207,287

10,518

453.22

11

29,021,106

13,465

463.97

12

27,531,856

9120

331.25

Total

373,245,519

150,325

402.75

Genetic structure and GWAS

The structure model showed that all individuals were distributed in two sub-populations (k = 2), structured according to the origin of accesses (access from Brazil and outside access). The Kinship matrix was obtained previously for the GWAS analysis. From the 150,325 SNPs, 13 SNPs were significantly associated with yield in the drought experiment and 13 SNPs in the control experiment (Figs. 2, 3; Table 3).
Fig. 2

Manhattan Plot of the association analysis in irrigated experiment (control) using the MLM model. The y-axis is the association value of the SNPs and the x-axis corresponds to the position on the chromosomes

Fig. 3

Manhattan Plot of the association analysis in water deficit experiment (drought) using the MLM model. The y-axis is the association value of the SNPs and the x-axis corresponds to the position on the chromosomes

Table 3

Rice SNP markers related to yield (Kg ha−1) from drought and control field experiments

SNP

p value

R2

Stepwise

Block

Bl. Size

Gene/block

Ma. Al.

Mi. Al.

Mi. Al. Ef.

Yield Ma.

Yield Mi.

Gain (Mi.–Ma.)

p value

  

Drought experiment

C1_245695

7.31E−05

0.1

*

No

LOC_Os01g43060

A

G

−459.25

1521.44

1196.42

−325.02

2.23E−04

C1_288166

3.06E−05

0.11

*

No

No gene

G

A

512.91

1365.62

1972.54

606.92

8.33E−07

C2_265606

2.12E−04

0.08

*

4674

7651

No gene

T

C

652.84

1412.64

2158.64

746.00

3.41E−04

C2_49592

2.41E−04

0.08

*

No

LOC_Os02g09650

T

G

612.43

1424.07

2114.87

690.80

1.37E−02

C3_98563

2.41E−04

0.08

*

5989

1569

LOC_Os03g17710

T

G

−413.97

1525.09

1119.37

−405.71

1.27E−07

C3_98564

2.41E−04

0.08

 

5989

1569

LOC_Os03g17710

C

T

−413.97

1525.09

1119.37

−405.71

1.27E−07

C3_288777

1.24E−04

0.09

*

7065

123,479

18 genes

T

C

−357.91

1556.71

1244.45

−312.26

3.95E−05

C6_249141

2.66E−04

0.08

 

12816

122,200

23 genes

G

A

−345.35

1592.23

1214.48

−377.75

8.27E−07

C6_249165

1.77E−04

0.09

 

12816

122,200

23 genes

G

A

−352.46

1597.23

1211.54

−385.68

4.23E−07

C6_249167

1.77E−04

0.09

 

12816

122,200

23 genes

C

A

−352.46

1597.23

1211.54

−385.68

4.23E−07

C6_250235

1.80E−04

0.08

*

No

No gene

G

A

−356.39

1612.62

1215.74

−396.89

1.84E−07

C6_250764

1.80E−04

0.08

 

12824

22,735

5 genes

C

A

−356.39

1612.62

1215.74

−396.89

1.84E−07

C11_187804

2.06E−04

0.08

*

No

LOC_Os11g31950

A

C

416.69

1442.55

1503.81

61.26

5.32E−01

  

Control experiment

C2_59400

2.41E−04

0.08

*

No

No gene

T

A

1104.29

2670.20

3666.11

995.90

4.01E−07

C3_288777

1.50E−05

0.11

*

7065

123,479

18 genes

T

C

−702.61

2890.36

2351.33

−539.03

7.29E−05

C4_204118

1.48E−04

0.09

*

No

LOC_Os04g33710

C

G

−906.21

2789.25

1952.76

−836.49

2.77E−04

C6_241880

8.36E−05

0.09

*

No

LOC_Os06g40570

A

G

−664.57

3036.99

2359.13

−677.87

2.49E−07

C6_249141

2.27E−04

0.08

 

12816

122,200

23 genes

G

A

−610.00

2952.44

2316.43

−636.01

3.46E−06

C6_249638

2.50E−04

0.08

 

12820

29491

4 genes

A

C

−587.96

2959.89

2323.39

−636.50

3.54E−06

C6_249715

3.50E−05

0.10

*

12820

29,491

4 genes

T

C

−652.11

2991.20

2311.57

−679.63

4.03E−07

C6_250235

1.66E−04

0.09

 

No

No gene

G

A

−625.84

2974.53

2337.79

−636.74

2.16E−06

C6_250764

1.66E−04

0.09

 

12824

22,735

5 genes

C

A

−625.84

2974.53

2337.79

−636.74

2.16E−06

C8_79148

2.77E−04

0.08

 

15514

52,833

8 genes

A

G

−788.45

2981.65

2171.75

−809.90

6.54E−09

C8_79932

2.19E−04

0.08

 

15520

2580

2 genes

A

G

−806.14

2985.20

2192.82

−792.38

9.58E−09

C8_80559

1.60E−04

0.09

*

No

LOC_Os08g13560

T

C

−812.24

2985.10

2178.89

−806.21

7.31E−09

C8_82588

2.42E−04

0.08

 

No

LOC_Os08g13840

A

C

−785.32

2982.32

2212.29

−770.04

1.26E−08

R2, phenotypic variation explained by SNPs; Stepwise, SNP markers selected from stepwise regression analysis; Ma. Al., major allele; Mi. Al., minor allele

From the significant SNP markers, we performed a regression analysis to identify the minimum number of markers that could potentially be used in a marker-assisted selection, and found eight SNPs for the drought experiment, and six SNPs for the control experiment (Table 3). The model with 13 and 8 SNPs showed the same value of accumulated R2 = 0.34 (p < 0.001) for the drought experiment, while for the control experiment, the model with 13 and 6 markers showed the same value of accumulated R2 = 0.37 (p < 0.001). Four SNPs were significant for both the drought and control conditions, however, only the marker C3_288777 was included in both regression models.

Considering the 13 SNPs from drought experiment, two yield-associated SNPs were located in intergenic regions, five SNPs were located in genes (LOC_Os03g17710 had two SNPs) and six SNPs were located in linkage blocks, i.e., the haplotype analysis showed that genes belonging to the same linkage blocks are inherited together, so it is not possible to indicate which gene of the linkage block is associated with grain yield (Table 3). In one of these blocks (12816), the SNPs C6_249141, C6_249165 and C6_249167 were detected; the last two located on the gene LOC_Os06g41580, and the remaining SNP in an intergenic region (Table 3 and Supplemental Table 2). From the five gene-located SNPs, C2_49592 was found in the gene LOC_Os02g09650 (Apetala2 gene).

In the drought experiment, the minor allele frequencies (alleles with lower frequencies in a specific SNP locus) of the SNPs C1_288166, C2_265606, C2_49592 and C11_187804 showed a positive effect on yield (Table 3). Two of these SNPs were located on the genes LOC_Os02g09650, and LOC_Os11g31950, an unknown protein in rice, but homologous to Arabidopsis AT3G48860, which is the stomatal cytokinesis defective 2 gene (SCD2, E-value = 10−6) (Supplemental Table 3).

Considering the 13 SNPs from control experiment, three SNPs were located in intergenic regions, four SNPs were located in genes and six SNPs were located in linkage blocks (Table 3). From these SNPs, four were directly related to genes, and from them, the SNP C8_82588 was located on LOC_Os08g13840, which is the WRKY transcription factor. In the control experiment, just one minor allele (SNP C2_59400) was able to produce a positive effect in the trait, but it was located in an intergenic region.

SNPs for marker assisted selection

The selected SNP markers from the drought (8 SNPs) and control (6 SNPs) experiments were used to check the SNP profile of the most and least productive accessions from both experiments. From the 10 most productive accessions from the drought experiment, we observed a variation from 4 (50 %, found in one accession) to 8 (100 %, also in one accession) of SNP alleles with a positive effect on yield, while for the 10 least productive accessions, the positive alleles varied from 1 (12.5 %, two accessions) to four (50 %, two accessions) SNP markers. The use of the 8 SNP marker sets could identify all accessions with more than 50 % SNP markers with positive alleles as being drought tolerant. Accessions with 50 % or less of positive SNP alleles were considered drought susceptible (Table 4). In relation to the 10 most productive accessions from the control experiment, the number of SNPs with positive alleles varied from 3 (50 %, one accession) to 5 (83.3 %, seven accessions), while for the 10 least productive accessions, the SNPs with positive alleles varied from 2 (33.3 %, nine accessions) to 3 (50 %, one accession). Then, as observed in the set of markers from drought experiment, these markers could be used to select the accessions with greater yield potential without water restriction. In both sets of SNP markers, one high yielding accession would not be selected, since they showed 50 % positive alleles (Table 4). Then, these eight SNP markers were used to genotype, via real time-PCR, the 20 rice genotypes phenotyped to drought in field and Sitis platform experiments. Of these eight SNP markers, two (SNPs 1 and 8) were monomorphic for this group of genotypes and were removed from the analysis. The genotypes which had simultaneously the allele A (SNP 2) and allele G (SNP 7) showed higher productivity average, while those with the profile G and G, G and A, or A and A (SNPs 2 and 7, respectively), had lower productivity (Table 5).
Table 4

In silico marker-assisted selection using the SNP sets selected from drought and control experiments

SNP(*)

Alleles

GUARANI

TANGARÁ

URUÇUÍ

IRAT 112

CARAJÁS

AIMORÉ

DOURADÃO

CAROLINO

4 MESES

CASCA BRANCA

10+ genotypes (drought experiment)

C1_245695

A/G

A+

A+

A+

A+

A+

A+

A+

A+

A+

A+

C1_288166

A/G

G

G

A+

A+

A+

A+

G

G

A+

A+

C2_49592

T/G

T

G+

T

T

G+

G+

T

G+

T

T

C2_265606

T/C

T

T

C+

T

C+

T

T

T

C+

C+

C3_98563

T/G

T+

T+

T+

T+

T+

T+

T+

T+

T+

T+

C3_288777

T/C

T+

T+

T+

T+

T+

T+

T+

T+

T+

T+

C6_250235

A/G

G+

G+

G+

G+

G+

G+

G+

G+

G+

G+

C11_187804

A/C

A

A

A

C+

C+

C+

C+

C+

A

A

% Superior Al.

 

50

62.5

75

75

100

87.5

62.5

75

75

75

SNP(*)

Alleles

BLUEBELLE

TB47H

YN1906-UUL 65

BR4742

PRETO

CHATÃO BURITI

BICO PRETO

ROXO

FARROUPILHA

IREM 238

10genotypes (drought experiment)

C1_245695

A/G

A+

A+

A+

A+

G

G

A+

A+

A+

G

C1_288166

A/G

G

G

G

G

G

G

G

G

G

G

C2_49592

T/G

T

T

T

T

T

T

T

T

T

T

C2_265606

T/C

T

T

T

T

T

T

T

T

T

T

C3_98563

T/G

T+

T+

T+

T+

T+

T+

T+

T+

G

T+

C3_288777

T/C

T+

C

C

C

C

C

T+

T+

C

T+

C6_250235

A/G

A

A

A

G+

G+

A

A

A

A

G+

C11_187804

A/C

A

C+

C+

C+

A

A

A

C+

A

A

% Superior Al.

 

37.5

37.5

37.5

50

25

12.5

37.5

50

12.5

37.5

SNP

Alleles

IAC 165

URUÇUÍ

ITA 150

LS 85-158

A12-286

AIMORÉ

CT11891

3 MESES BRANCO

7 SEMANAS

CASCA BRANCA

10+ genotypes (control experiment)

C2_59400

A/T

T

T

T

T

T

T

T

T

T

T

C3_288777

T/C

T+

T+

T+

T+

C

T+

T+

T+

T+

T+

C4_204118

G/C

C+

C+

C+

C+

C+

C+

C+

C+

C+

C+

C6_241880

A/G

A+

A+

A+

G

A+

A+

A+

A+

A+

A+

C6_249715

T/C

T+

T+

T+

C

T+

T+

T+

T+

T+

T+

C8_80559

T/C

T+

T+

T+

T+

T+

T+

T+

C

T+

T+

% Superior Al.

 

83.3

83.3

83.3

50

66.6

86.3

83.3

66.6

83.3

83.3

SNP

Alleles

MEARIN

TOX 1871

IAC 81

L 285

CATETINHO

JAGUARIZINHO

CHATAO AMARELO

PINGO DE OURO

BICO GANGA

CANA ROXA

10− genotypes (control experiment)

C2_59400

A/T

T

T

T

T

T

T

T

T

T

T

C3_288777

T/C

C

C

C

T+

T+

T+

T+

T+

T+

T+

C4_204118

G/C

C+

C+

C+

C+

C+

G

C+

C+

C+

C+

C6_241880

A/G

G

A+

G

G

G

G

G

G

G

G

C6_249715

T/C

T+

C

T+

C

C

T+

C

C

C

C

C8_80559

T/C

C

T+

C

C

C

C

C

C

C

C

% Superior Al.

 

33.3

50

33.3

33.3

33.3

33.3

33.3

33.3

33.3

33.3

10+ and 10−: best and least productive genotypes; N+: superior allele for drought tolerance

* C1_245695 (SNP1); C1_288166 (SNP2); C2_49592 (SNP3); C2_265606 (SNP4); C3_98563 (SNP5); C3_288777 (SNP6); C6_250235 (SNP7); C11_187804 (SNP8)

Table 5

Average productivity of genotypes evaluated under drought in two experiments (Sitis and Porangatu)

 

Number of genotypes

Sitis productivity (g pot−1)

Porangatu productivity (Kg ha−1)

Experiment

20

59.7

2047.1

SNP2 (A) SNP7 (G)

6

66.5

2477.7

SNP2 (G) SNP7 (A)

6

60.4

1379.3

SNP2 (G) SNP7 (G)

7

54.2

2270

The subgroups of genotypes were formed from genotyping data of the SNP2 and SNP7

Annotation of SNPs associated to yield

From the drought experiment, including the genes inherited in linkage blocks, 50 genes were identified, from which 30 were annotated, and 10 were previously related to drought and/or abiotic stress tolerance (Supplemental Table 2). The remaining 20 genes had no predicted function.

From the control experiment, 64 genes associated to yield were identified, from which 24 had no predicted function in rice, and 11 were previously related to drought and/or abiotic stress tolerance (Supplemental Table 2). From 27 rice genes of unknown function, 15 had putative homologous genes in Arabidopsis, and from these, only LOC_Os06g41630 was homologous to an Arabidopsis gene (AT4G30960, E-value 5 × 10−5) previously related to drought tolerance (Supplemental Table 3).

Discussion

Several experiments to standardize the procedures for the uniform screening of segregating populations for grain yield under reproductive-stage drought showed moderate heritability of this trait, thereby confirming the suitability of grain yield as a selection criterion (Kumar et al. 2014). The water deficit applied in the experiment was adequate to assess the potential of the genotypes to react to drought, since there was a strong reduction (46.5 % in average) in productivity as well as variations in yield range among accessions. Three genotypes have stood out among the top-10 yields in the experiments with and without water deficit: the cultivar Aimoré, released in 2000, the cultivar Uruçuí, released in 1993, and the landrace Casca Branca, all from Brazil. These genotypes are indicated as being of great interest to the breeding programs for being potential parent for the development of higher yielding rice cultivars for cultivation in both drought and irrigated environments.

SNP genome distribution

Compared to previous studies, this study generated a higher density of SNP markers, as well as a higher number of SNPs after filtering by MAF and after the imputation of missing data (150,325 SNPs), enough to cover the entire genome of rice, and consequently enabling to perform a GWAS analysis (Courtois et al. 2013; Rebolledo et al. 2015). In addition, this study also showed a distribution of 402.75 SNPs/Mb, demonstrating the great genetic variability of the 175 rice accessions. This was expected considering that those entries are a subset of the Brazilian core collection of rice germplasm. In a similar study, Biscarini et al. (2016), who worked with a panel of 391 temperate japonica rice accessions widely distributed in Europe, obtained 57,179 SNPs after data filtering, with a high efficiency in the use of genotyping platform GBS for genome-wide association studies and its subsequent application in marker assisted selection.

Drought GWAS: potential impact for molecular breeders

After the stepwise analysis, the number of SNP markers in the model was reduced while maintaining the values of the explained phenotypical variation in both experiments. In the drought experiment, from the eight SNPs associated to yield, four of the minor alleles showed a positive effect in this trait, and would be the target on MAS. Of the markers that were kept in the model from the drought, just the AP2 domain gene was previously related to drought. However, according to Kang et al. (2015), due to the existence of LD and the imperfection in data collection (persistence of data error), the most significant SNPs may not be the true causative loci, reinforcing the importance to validate the SNPs. The LD may be allied for the use of the SNP in marker assisted selection if the linkage is so strong that it prevents the recombination between the SNP and the causal gene, i.e. the presence of the SNP will, most of the time, be associated to a gene or genes that confer the favorable phenotype. Approaches for the reduction of the set of SNP to be used in molecular breeding have been already explored in cattle (Biffani et al. 2015) and sugar beet (Biscarini et al. 2015).

Two out of the eight SNP markers were able to identify a group of rice genotypes with higher productivity when evaluated by drought experiments in the field (Porangatu) and greenhouse (Sitis Platform). These results are encouraging for deriving markers for the routine analysis of marker assisted selection. The conversion of the GWAS analysis results into useful breeding information is challenging because the marker-assisted selection must be based in a genotyping platform that produces faster results and has low cost per sample. The SNP alleles associated to the drought tolerant phenotype then could be used to select rice accessions from genebanks, from segregating populations or to monitor the allele present in a backcross introgression scheme (inbred line conversion), without the need to conduct field trials to evaluate drought tolerance of these materials. Shamsudin et al. (2016) described the pyramiding of three drought yield QTLs (qDTY 2.2 , qDTY 3.1 , and qDTY 12.1 ) in the Malaysian rice cultivar MR219 using SSRs as flanking markers.

The SNPs C3_288777, C6_249141 and C6_250764 were associated to yield in both experiments. A hypothesis to be considered is that since they are part of linkage blocks, with 18, 23 and 5 genes, respectively, the possibility to include the genes that confer plasticity for the accessions that have the best allele combinations in these sets reinforces the importance to use these SNPs in a marker assisted selection program, this could be useful to select nucleotides that positively affect the yield independent of water availability. From these three SNP markers, only C3_288777 was present in the SNP set developed from the drought experiment.

The use of GWAS in the identification of QTLs for morphological traits in European temperate rice has been described by Biscarini et al. (2016). In our study, we used the same methodology to identify QTLs for drought tolerance in tropical japonica rice. This trait is of great importance for the sustainability of rice cultivation, both for small as for large plantations, and the identification of these QTLs paves the way for the use of assisted selection in developing rice cultivars more tolerant to this stress.

Drought GWAS: potential impact for molecular geneticists

The SNP C2_49592 was found in the gene LOC_Os02g09650, related to the gene Apetala2/Ethylene Responsive Factor (AP2/ERF), and previously associated to drought (Licausi et al. 2013). AP2/ERF is a transcription factor that binds to the promoter regions of stress-responsive genes. ERF genes are induced by biotic and abiotic stress, including pathogen infection, salt stress, osmotic stress, wounding, drought, hypoxia, jasmonic acid and ABA. The plant specific AP2/ERF class of transcription factors are a large family with ~163 members in rice that regulate important functions of responses to environmental stimuli or plant growth and development depending on the presence of one or two highly conserved 60 amino acid AP2 domains in the protein (Jisha et al. 2015).

The SNP C8_82588 was located on LOC_Os08g13840, which encodes the WRKY transcription factor, previously related to biotic and abiotic stresses by Chen et al. (2012). WRKY proteins are a family of transcription factors that are unique to plants, showing enhanced expression and/or DNA-binding activity following induction by a range of pathogens, defense signals, and wounding (Perez-Clemente et al. 2013).

Most of the SNP markers identified by GWAS analysis (57 %) were positioned in linkage blocks, preventing the identification of the causal gene associated to drought tolerance. In a linkage block the probability of recombination is low, especially considering the narrow-based group of elite parents of breeding programs, and it is possible, therefore, to correlate the desirable phenotype with the nucleotide with a positive effect in the trait, no matter which gene was involved in the trait expression. To find evidence about the causative gene, a search for the putative gene function in databanks is the starting point. However, to have a better idea about which gene or genes are directly involved with the phenotype, it would be necessary to knock out all the genes on the region by means of the T-DNA approach, or knocking down the genes by interference RNA, and observe the resultant phenotype (Alonso and Ecker 2006).

In the drought experiment, linkage blocks detected by SNPs ranged from five to 23 genes, and it is reasonable to infer that these genes are inherited together in most of the rice accessions. From the overall 46 genes from linkage blocks, 18 had no predicted function. In the linkage block 7065, where the SNP marker C3_288777 is located, nine genes with unknown function and nine annotated rice genes were identified. The gene where the SNP was identified (LOC_Os03g50570) did not have its function determined in rice, and also did not show hit with the aminoacid sequence of Arabidopsis. From the nine annotated genes, the mitogen-activated protein kinase (MAPK) OsMKK10-2 was the only one previously related to abiotic stress responses, according to Kumar et al. (2008). The MAPK pathways in eukaryotes play a pivotal role in basic cellular processes, development, hormone biosynthesis/signaling, senescence, plant immunity as well as in producing responses to several stress conditions (Wankhede et al. 2013). Mitogen-activated protein kinase cascade is an evolutionarily conserved signal transduction module involved in transducing extracellular signals to the nucleus for appropriate cellular adjustment. This cascade essentially consists of three components, a MAPK kinase kinase (MAPKKK), a MAPK kinase (MAPKK) and a MAPK connected to each other by the event of phosphorylation. These kinases play various roles in intra- and extra-cellular signaling in plants by transferring the information from sensors to responses (Sinha et al. 2011). Since the SNP marker C3_288777 was related to yield in both the drought and control experiments, and considering that the genes of the linkage block co-segregate, it is not possible to infer that the same gene was associated to yield in both experiments.

The interaction between the genes WRKY TF and MAPKs (both identified in this work as being associated to yield in the drought experiment) has already been described by Shen et al. (2012). These authors pointed out that OsWRKY30 could be a substrate of MAPKs—the phosphorylation of OsWRKY30 activated its transcriptional activity and allowed it to perform its function as a transcription factor. From the result of overexpression transgenic lines, it was proven that the activation of OsWRKY30 by phosphorylation improved drought tolerance in rice (Shen et al. 2012).

The linkage block 12816 had the SNP markers C6_249141, C6_249167 and C6_249165, the first two present in the gene LGC1 (low glutelin content). This block had 23 genes, nine of them with unknown functions, and from the 14 annotated genes, four were previously related to drought, of which two related to the d-mannose binding lectin family, one F-box domain containing protein, one receptor-like protein kinase, and one VQ motif family protein. Lectins are a class of carbohydrate-binding proteins that exhibit differential expression abundance under various abiotic stresses including temperature shock, drought and high salinity stresses (Jiang et al. 2010). F-box proteins constitute a large family in eukaryotes and are characterized by a conserved F-box (approximately 40 aminoacids), these are classified into 10 subfamilies, and at least 43 F-box protein-encoding genes have been found to be differentially expressed in rice seedlings subjected to different abiotic stress conditions (Jain et al. 2007). The receptor-like kinases (RLKs) are signaling proteins that feature an extracellular domain connected via a transmembrane domain to a cytoplasmic kinase, perceiving external signals, transducing them into the cell, with abiotic stress roles. RLKs comprise the largest gene family of receptors in plants, with more than 1100 RLKs in rice (Morillo and Tax 2006; Todaka et al. 2015). The VQ motif-containing proteins is a class of plant specific transcriptional regulators related to diverse developmental processes including responses to biotic and abiotic stresses, seed development and photomorphogenesis. There is evidence that these VQ proteins interact with WRKY transcription factors, detected in this study and above mentioned (Jing and Lin 2015).

Considering the genes with unknown function in this linkage block, after a search for homologous aminoacid sequences in Arabidopsis, two genes were identified, the CBL-interacting protein kinase 6 (CIPK6, AT4G30960, E-value 5e-05), related to salt and drought stresses, and largely accumulated in abscisic acid treated seedlings (Chen et al. 2013), and another receptor-like protein kinase (RLK1, AT5G60900, 7e-07) associated to drought by Kilian et al. (2007) and Skirycz et al. (2011).

The SNP marker C6_250764 was located at the linkage block 12824, which had five genes, two of them with two copies with high identity of sequences: LOC_Os06g41820 and LOC_Os06g41830 (strictosidine synthase) showed 69.9 % similarity on nucleotide and 62.07 % on aminoacid sequences, while LOC_Os06g41810 and LOC_Os06g41840 (Cinnamoyl CoA Reductase, or CCR) showed 93.3 % of nucleotide and 93.88 % of aminoacid sequences. CCR is a key gene for lignin biosynthesis, which has been shown to be over-expressed under stress conditions (Srivastava et al. 2015). Another gene previously related to drought in this linkage block was LOC_Os06g41800 (dihydroflavonol-4-reductase), significantly up-regulated by drought and cold stresses (Wu et al. 2014).

Three of linkage blocks have been associated with productivity in the two experiments. The reasons why it may have occurred in each linkage block may be due to distinct genes acting in each irrigation condition or genes having epistatic action for both irrigation conditions.

There were two linkage blocks that occurred exclusively in the control experiment, and only the block 12820 showed a gene previously related to drought tolerance, the expansin gene (LOC_Os06g41700). The expansin proteins are prime candidates for cell wall–loosening factors that mediate the growth of plant cells. Transgenic expression of expansin genes in tobacco and Arabidopsis lead to various changes in growth and enhanced resistance to biotic and abiotic stresses, while in maize, changes in expansin expression facilitate the root growth as an adaptation to increase the tolerance to drought (Cosgrove 2015). In rice, expansin overexpression increased the root development (Ma et al. 2013; Wang et al. 2014), but no relation to drought tolerance has been described to date.

Conclusions

This study identified yield-related genes in rice cultivated under drought and normal irrigation conditions. From the drought experiment, including the genes inherited in linkage blocks, 50 genes were identified, from which 30 were annotated, and 10 were previously related to drought and/or abiotic stress tolerance, such as the transcription factors WRKY and Apetala2, MAP and receptor-like kinases. The eight SNP markers related to yield under drought were converted to TaqMan assays, and there is a good potential to use at least the SNP2 and SNP7 into a routine PCR analysis to identify drought tolerant rice accessions by means of marker assisted selection.

Notes

Acknowledgments

National Council for Scientific and Technological Development (CNPq) for the grants to CB and RPV; the Coordination for the Improvement of Higher Education Personnel/Ministry of Education (CAPES/MEC) for the grants to GFP; and the Brazilian Agricultural Research Corporation (EMBRAPA) for financial support for this research.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

Supplementary material

10709_2016_9932_MOESM1_ESM.docx (32 kb)
List of rice accessions drought-evaluated in field (Porangatu, F) and greenhouse (Sitis platform, S) experiments (DOCX 31 kb)
10709_2016_9932_MOESM2_ESM.docx (22 kb)
Putative annotation of rice transcripts identified by SNP markers related to yield in drought and control experiments (DOCX 22 kb)
10709_2016_9932_MOESM3_ESM.docx (21 kb)
Arabidopsis, Brachypodium, maize and sorghum transcripts homologous of rice transcripts identified by SNP markers related to yield in drought and control experiments (DOCX 21 kb)

References

  1. Abadie T, Cordeiro CMT, Fonseca JR, Alves RBN, Burle ML, Brondani C, Rangel PHN, Castro EM, Silva HT, Freire MS, Zimmermann FJP, Magalhães JR (2005) Construção de uma coleção nuclear de arroz para o Brasil. Pesqui Agropecu Bras 40:129–136CrossRefGoogle Scholar
  2. Alonso JM, Ecker JR (2006) Moving forward in reverse: genetic technologies to enable genome-wide phenomic screens in Arabidopsis. Nat Rev Genet 7:524–536CrossRefPubMedGoogle Scholar
  3. Barrett JC, Fry B, Maller J, Daly MJ (2005) Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21:263–265CrossRefPubMedGoogle Scholar
  4. Bernier J, Kumar A, Venuprasad R, Spaner D, Atlin GN (2007) A large-effect QTL for grain yield under reproductive-stage drought stress in upland rice. Crop Sci 47:507–516CrossRefGoogle Scholar
  5. Biffani S, Dimauro C, Macciotta N, Rossoni A, Stella A, Biscarini F (2015) Predicting haplotype carriers from SNP genotypes in Bos taurus through linear discriminant analysis. Genet Sel Evol 47:4CrossRefPubMedPubMedCentralGoogle Scholar
  6. Biscarini F, Marini S, Stevanato P, Broccanello C, Bellazzi R, Nazzicari N (2015) Developing a parsimonius predictor for binary traits in sugar beet (Beta vulgaris). Mol Breed 35:10CrossRefGoogle Scholar
  7. Biscarini F, Cozzi P, Casella L, Riccardi P, Vattari A, Orasen G, Perrini R, Tacconi G, Tondelli A, Biselli C, Cattivelli L, Spindel J, McCouch S, Abbruscato P, Valé G, Piffanelli P, Greco R (2016) Genome-Wide Association Study for traits related to plant and grain morphology, and root architecture in temperate rice accessions. PLoS One 11:e0155425CrossRefPubMedPubMedCentralGoogle Scholar
  8. Bradbury PJ, Zhang Z, Kroon DE, Casstevens TM, Ramdoss Y, Buckler ES (2007) TASSEL: software for association mapping of complex traits in diverse samples. Bioinformatics 19:2633–2635CrossRefGoogle Scholar
  9. Chen L, Song Y, Li S, Zhang L, Zou C, Yu D (2012) The role of WRKY transcription factors in plant abiotic stresses. Biochim Biophys Acta 1819:120–128CrossRefPubMedGoogle Scholar
  10. Chen L, Wang QQ, Zhou L, Ren F, Li DD, Li XB (2013) Arabidopsis CBL-interacting protein kinase (CIPK6) is involved in plant response to salt/osmotic stress and ABA. Mol Biol Rep 40:4759–4767CrossRefPubMedGoogle Scholar
  11. Cosgrove DJ (2015) Plant expansins: diversity and interactions with plant cell walls. Curr Opin Plant Biol 25:162–172CrossRefPubMedPubMedCentralGoogle Scholar
  12. Courtois B, Audebert A, Dardou A, Roques S, Ghneim-Herrera T, Droc G, Frouin J, Rouan L, Goz E, Kilian A, Ahmadi N, Dingkuhn M (2013) Genome-wide association mapping of root traits in a japonica rice panel. PLoS One. doi: 10.1371/journal.pone.0078037 Google Scholar
  13. Earl DA, Vonholdt BM (2011) STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conserv Genet Resour 4:359–361CrossRefGoogle Scholar
  14. Elshire RJ, Glaubitz JC, Sun Q, Poland JA, Kawamoto K, Buckler ES, Mitchell SEA (2011) Robust, simple genotyping-by-sequencing (GBS) approach for high diversity species. PLoS One. doi: 10.1371/journal.pone.0019379 PubMedPubMedCentralGoogle Scholar
  15. Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel B, Higgins J, Defelice M, Lochner A, Faggart M, Liu-Cordero SN, Rotimi C, Adeyemo A, Cooper R, Ward R, Lander ES, Daly MJ, Altshuler D (2002) The structure of haplotype blocks in the human genome. Science 296:2225–2229CrossRefPubMedGoogle Scholar
  16. He J, Zhao X, Laroche A, Lu ZX, Liu H, Li Z (2014) Genotyping-by-Sequencing (GBS), an ultimate marker-assisted selection (MAS) tool to accelerate plant breeding. Front Plant Sci 5:484CrossRefPubMedPubMedCentralGoogle Scholar
  17. Henry R (2014) Genomics strategies for germplasm characterization and the development of climate resilient crops. Front Plant Sci 5:68. doi: 10.3389/fpls.2014.00068 CrossRefPubMedPubMedCentralGoogle Scholar
  18. Jain M, Aashima N, Arora R, Agarwal P, Ray S, Sharma P, Kapoor S, Tyagi AK, Khurana P (2007) F-box proteins in rice. Genome-wide analysis, classification, temporal and spatial gene expression during panicle and seed development, and regulation by light and abiotic stress. Plant Physiol 143:1467–1483CrossRefPubMedPubMedCentralGoogle Scholar
  19. Jiang SY, Ma Z, Ramachandran R (2010) Evolutionary history and stress regulation of the lectin superfamily in higher plants. BMC Evol Biol 10:79. doi: 10.1186/1471-2148-10-79 CrossRefPubMedPubMedCentralGoogle Scholar
  20. Jing Y, Lin R (2015) The VQ motif-containing protein family of plant-specific transcriptional regulators. Plant Physiol 169:371–378CrossRefPubMedPubMedCentralGoogle Scholar
  21. Jisha V, Dampanaboina L, Vadassery J, Mithöfer A, Kappara S, Ramanan R (2015) Overexpression of an AP2/ERF type transcription factor OsEREBP1 confers biotic and abiotic stress tolerance in rice. PLoS One. doi: 10.1371/journal.pone.0127831 PubMedPubMedCentralGoogle Scholar
  22. Kang Y, Sakiroglu M, Krom N, Stanton-Geddes J, Wang M, Lee YC, Young ND, Udvardi M (2015) Genome-wide association of drought-related and biomass traits with HapMap SNPs in Medicago truncatula. Plant, Cell Environ 38:1997–2011. doi: 10.1111/pce.12520 CrossRefGoogle Scholar
  23. Kawahara Y, Bastide MDL, Hamilton JP, Kanamori H, Mccombie WR, Ouyang S, Schwartz DC, Tanaka T, Wu J, Zhou S, Childs KL, Davidson RM, Lin H, Quesada-Ocampo L, Vaillancourt B, Sakai H, Lee SS, Kim J, Numa H, Itoh T, Buell CR, Matsumoto T (2013) Improvement of the Oryza sativa Nipponbare reference genome using next generation sequence and optical map data. Rice 6:1–10CrossRefGoogle Scholar
  24. Kilian J, Whitehead K, Horak J, Wanke D, Weinl S, Batistic O, D’Angelo C, Bauer EB, Kudla J, Harter K (2007) The AtGenExpress global stress expression data set: protocols, evaluation and model data analysis of UV-B light, drought and cold stress responses. Plant J 50:347–363CrossRefPubMedGoogle Scholar
  25. Kole C, Muthamilarasan M, Henry R, Edwards D et al (2015) Application of genomics-assisted breeding for generation of climate resilient crops: progress and prospects. Front Plant Sci 6:563. doi: 10.3389/fpls.2015.00563 CrossRefPubMedPubMedCentralGoogle Scholar
  26. Kumar K, Rao KP, Sharma P, Sinha AK (2008) Differential regulation of rice mitogen activated protein kinase kinase (MKK) by abiotic stress. Plant Physiol Biochem 46:891–897CrossRefPubMedGoogle Scholar
  27. Kumar A, Dixit S, Ram T, Yadaw RB, Mishra KK, Mandal NP (2014) Breeding high-yielding drought-tolerant rice: genetic variations and conventional and molecular approaches. J Exp Bot 65:6265–6278CrossRefPubMedPubMedCentralGoogle Scholar
  28. Licausi F, Ohme-Takagi M, Perata P (2013) Apetala2/ethylene responsive factor (AP2/ERF) transcription factors: mediators of stress responses and developmental programs. New Phytol 199:639–649CrossRefPubMedGoogle Scholar
  29. Ma N, Wang Y, Qiu S, Kang Z, Che S, Wang G, Huang J (2013) Overexpression of OsEXPA8, a root-specific gene, improves rice growth and root system architecture by facilitating cell extension. PLoS One. doi: 10.1371/journal.pone.0075997 Google Scholar
  30. Morillo SA, Tax RE (2006) Functional analysis of receptor-like kinases in monocots and dicots. Curr Opinion Plant Biol 9:460–469CrossRefGoogle Scholar
  31. Perez-Clemente RM, Vives V, Zandalinas SI, Lopez-Climent MF, Munoz V, Gomez-Cadenas A (2013) Biotechnological approaches to study plant responses to stress. Biomed Res Int 2013:654120. doi: 10.1155/2013/654120 CrossRefPubMedGoogle Scholar
  32. Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genetics 155:945–959PubMedPubMedCentralGoogle Scholar
  33. Rebolledo MC, Dingkuhn M, Courtois B, Gibon Y, Clément-Vidal A, Cruz DF, Duitama J, Lorieux M, Luquet D (2015) Phenotypic and genetic dissection of component traits for early vigour in rice using plant growth modelling, sugar content analyses and association mapping. J Exp Bot 66:5555–5566CrossRefPubMedPubMedCentralGoogle Scholar
  34. Scheet P, Stephens M (2006) A fast and flexible statistical model for large-scale population genotype data: applications to inferring missing genotypes and haplotypic phase. Am J Hum Genet 78:629–644CrossRefPubMedPubMedCentralGoogle Scholar
  35. Shamsudin NAZ, Swamy BPM, Ratnam W, Sta Cruz MT, Raman A, Kumar A (2016) Marker assisted pyramiding of drought yield QTLs into a popular Malaysian rice cultivar, MR219. BMC Genet 17:30. doi: 10.1186/s12863-016-0334-0 CrossRefPubMedPubMedCentralGoogle Scholar
  36. Shen H, Liu C, Zhang Y, Meng X, Zhou X, Chu C, Wang X (2012) OsWRKY30 is activated by MAP kinases to confer drought tolerance in rice. Plant Mol Biol 80:241–253CrossRefPubMedGoogle Scholar
  37. Sinha AK, Jaggi M, Raghuram B, Tuteja N (2011) Mitogen-activated protein kinase signaling in plants under abiotic stress. Plant Signal Behav 6:196–203CrossRefPubMedPubMedCentralGoogle Scholar
  38. Skirycz A, Claeys H, Bodt S, Oikawa A, Shinoda S, Andriankaja M, Maleux K, Eloy NB, Coppens F, Yoo SD, Saito K, Inzé D (2011) Pause-and-stop: the effects of osmotic stress on cell proliferation during early leaf development in Arabidopsis and a role for ethylene signaling in cell cycle arrest. Plant Cell 23:1876–1888CrossRefPubMedPubMedCentralGoogle Scholar
  39. Srivastava S, Vishwakarma RK, Arafat YA, Gupta SK, Khan BM (2015) Abiotic stress induces change in Cinnamoyl CoA Reductase (CCR) protein abundance and lignin deposition in developing seedlings of Leucaena leucocephala. Physiol Mol Biol Plants 21:197–205CrossRefPubMedPubMedCentralGoogle Scholar
  40. The R Foundation for statistical computing (2016) R: a language and environment for statistical computing. R Core Team, Vienna. http://www.R-project.org. Accessed 30 March 2016
  41. Todaka D, Shinozaki K, Yamaguchi-Shinozaki K (2015) Recent advances in the dissection of drought-stress regulatory networks and strategies for development of drought-tolerant transgenic rice plants. Front Plant Sci 6:84. doi: 10.3389/fpls.2015.00084 CrossRefPubMedPubMedCentralGoogle Scholar
  42. Vikram P, Swamy MB, Dixit S, Ahmed UH, Sta Cruz MT, Singh AK, Kumar A (2011) 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. doi: 10.1186/1471-2156-12-89 CrossRefPubMedPubMedCentralGoogle Scholar
  43. Vikram P, Swamy BP, Dixit S, Singh R, Singh BP, Miro B, Kohli A, Henry A, Singh NK, Kumar A (2015) Drought susceptibility of modern rice varieties: an effect of linkage of drought tolerance with undesirable traits. Sci Rep 5:14799. doi: 10.1038/srep14799 CrossRefPubMedPubMedCentralGoogle Scholar
  44. Wang Y, Ma N, Qiu S, Zou H, Zang G, Kang Z, Wang G, Huang J (2014) Regulation of the alpha-expansin gene OsEXPA8 expression affects root system architecture in transgenic rice plants. Mol Breed 34:47–57CrossRefGoogle Scholar
  45. Wankhede DP, Misra M, Singh P, Sinha AK (2013) Rice mitogen activated protein kinase kinase and mitogen activated protein kinase interaction network revealed by in silico docking and yeast two-hybrid approaches. PLoS One 8:e65011. doi: 10.1371/journal.pone.0065011 CrossRefPubMedPubMedCentralGoogle Scholar
  46. Wu Y, Wei W, Pang X, Wang X, Zhang H, Dong B, Xing Y, Li X, Wang M (2014) Comparative transcriptome profiling of adesert evergreen shrub, Ammopiptanthus mongolicus, in response to drought and cold stresses. BMC Genom 15:671. doi: 10.1186/1471-2164-15-671 CrossRefGoogle Scholar
  47. Yang J, Lee SH, Goddard ME, Visscher PM (2011) GCTA: a tool for genome-wide complex trait analysis. Am J Hum Genet 88:76–82CrossRefPubMedPubMedCentralGoogle Scholar
  48. Zhang Z, Ersoz E, Lai CQ, Todhunter RJ, Tiwari HK, Gore MA, Bradbury PJ, Yu J, Arnett DK, Ordovas JM, Buckler ES (2010) Mixed linear model approach adapted for genome-wide association studies. Nat Genet 42:355–360CrossRefPubMedPubMedCentralGoogle Scholar
  49. Zhang Z, Ober U, Erbe M, Zhang H, Gao N, He J, Li J, Simianer H (2014) Improving the accuracy of whole genome prediction for complex traits using the results of genome wide association studies. PLoS One 9:e93017CrossRefPubMedPubMedCentralGoogle Scholar

Copyright information

© Springer International Publishing Switzerland 2016

Authors and Affiliations

  • Gabriel Feresin Pantalião
    • 1
  • Marcelo Narciso
    • 2
  • Cléber Guimarães
    • 2
  • Adriano Castro
    • 2
  • José Manoel Colombari
    • 2
  • Flavio Breseghello
    • 2
  • Luana Rodrigues
    • 2
  • Rosana Pereira Vianello
    • 2
  • Tereza Oliveira Borba
    • 2
  • Claudio Brondani
    • 2
  1. 1.Escola de AgronomiaUniversidade Federal de GoiásGoiâniaBrazil
  2. 2.Embrapa Arroz e FeijãoGoiâniaBrazil

Personalised recommendations