Genome-wide association mapping of iron homeostasis in the maize association population
© Benke et al.; licensee BioMed Central. 2015
Received: 9 June 2014
Accepted: 25 September 2014
Published: 30 January 2015
Iron (Fe) deficiency in plants is the result of low Fe soil availability affecting 30% of cultivated soils worldwide. To improve our understanding on Fe-efficiency this study aimed to (i) evaluate the influence of two different Fe regimes on morphological and physiological trait formation, (ii) identify polymorphisms statistically associated with morphological and physiological traits, and (iii) dissect the correlation between morphological and physiological traits using an association mapping population.
The fine-mapping analyses on quantitative trait loci (QTL) confidence intervals of the intermated B73 × Mo17 (IBM) population provided a total of 13 and 2 single nucleotide polymorphisms (SNPs) under limited and adequate Fe regimes, respectively, which were significantly (FDR = 0.05) associated with cytochrome P450 94A1, invertase beta-fructofuranosidase insoluble isoenzyme 6, and a low-temperature-induced 65 kDa protein. The genome-wide association (GWA) analyses under limited and adequate Fe regimes provided in total 18 and 17 significant SNPs, respectively.
Significantly associated SNPs on a genome-wide level under both Fe regimes for the traits leaf necrosis (NEC), root weight (RW), shoot dry weight (SDW), water (H 2O), and SPAD value of leaf 3 (SP3) were located in genes or recognition sites of transcriptional regulators, which indicates a direct impact on the phenotype. SNPs which were significantly associated on a genome-wide level under both Fe regimes with the traits NEC, RW, SDW, H 2O, and SP3 might be attractive targets for marker assisted selection as well as interesting objects for future functional analyses.
Iron (Fe) deficiency in plants is the result of a low Fe availability which might be induced by lime-chlorosis that affects 30% of cultivated soils worldwide . As an adaptation to the sparingly available Fe, plants evolved two different strategies to mobilize and uptake Fe . Dicotyledonous and non graminaceous plant species acquire Fe by the so-called strategy I mechanism . The characteristic of this strategy is the release of protons into the rhizosphere that facilitate the mobilization and subsequent reduction of Fe(III) to Fe(II) via a plasma membrane bound Fe(III) chelate reductase . The soluble Fe(II) is finally taken up by the iron regulated transporter 1 (IRT1) .
For the crop plants which are graminaceous plant species such as barley, rice, and maize, Fe is acquired using the so-called strategy II . Characteristic for this strategy is the release of non proteinogenic compounds named phytosiderophores. These compounds chelate the Fe(III) in the rhizosphere. Phyto-siderophore-Fe(III) complexes are transported by the specific transporter yellow stripe 1 (YS1) into the plant . It was shown by  that the amount of exudated phytosiderophores is crucial for a chlorosis tolerance and therefore, Fe-efficient plant. However, for an Fe-efficient genotype, the balance of Fe dependent systems like Fe mobilization and uptake into the plant and the homeostasis related mechanisms like translocation and regulation of the Fe level in the cell to avoid shortage or toxicity [8,9] is essential.
To improve our understanding of the mechanisms which are responsible for Fe-efficiency in maize, two different methods have been applied so far. The RNA-Sequencing approach used by  focused on genes which were differentially expressed between the Fe-efficient and inefficient inbred lines under sufficient and deficient Fe regimes. This study provided a tremendous amount of putative candidate genes for Fe-efficiency. The same inbred lines were used for the establishment of the intermated B73 × Mo17 (IBM) segregating population . Benke et al., 2014  observed a considerable phenotypic variation for Fe-efficiency in this population which was used to map quantitative trait loci (QTL). An alternative to linkage mapping is association mapping which has the potential to provide a higher mapping resolution as well as allows the evaluation of a higher number of alleles at a time. To our knowledge, no genome-wide association study has been conducted to dissect Fe-efficiency in maize.
The objectives of our study were to (i) evaluate the influence of different Fe regimes on morphological and physiological trait formation, (ii) identify polymorphisms statistically associated with morphological and physiological traits, and (iii) dissect the correlation between morphological and physiological traits using an association mapping population.
Traits recorded in the current study for two deficient and sufficient iron (Fe) regimes, where H 2 is the repeatability on an entry means basis for the association mapping population
SPAD value at leaf 3
SPAD value at leaf 4
SPAD value at leaf 5
SPAD value at leaf 6
Shoot dry weight
Shoot water content
Ratio of dry shoot weight
compared to shoot length
Branching at the terminal 5 cm
score 1 - 9
Lateral root formation
score 1 - 9
score 1 - 9
In the ASMP, the population structure explained on average 2.02% of the phenotypic variation with a minimum of 0.08% (SL) and a maximum of 5.32% (RL) under the Fe-deficient regime (Additional file 1: Table S1). Under the Fe-sufficient regime, the population structure accounted on average for 2.42% of the phenotypic variation ranging from 0.35% (SDW) to 5.09% (RL).
Single nucleotide polymorphism (SNP) markers significantly (FDR = 0.05) associated in the association mapping population which were located within confidence intervals of QTL detected for the same trait in the IBM population [ 12 ]
% r 2
205.0 - 208.5
Pentatricopeptide repeat-containing protein At5g47360
205.0 - 208.5
Pentatricopeptide repeat-containing protein At5g47360
205.0 - 208.5
Cytochrome P450 94A1
205.0 - 208.5
Cytochrome P450 94A1
73.3 - 74.4
73.3 - 74.4
CLE family OsCLE305 protein
410.8 - 413.6
410.8 - 413.6
Uncharacterized membrane protein At3g27390
464.0 - 466.5
220.7 - 223.9
Sec14p-like phosphatidylinositol transfer family protein
220.7 - 223.9
220.7 - 223.9
Late embryogenesis abundant protein 4-5
825.8 - 833.0
833.0 - 839.3
833.0 - 839.3
S-ribonuclease binding protein
Under the Fe-sufficient regime, the QTL FM analyses revealed in total two significant (FDR = 0.05) SNPs for SP4 QTL1 (Table 2, Figure 3). The maximum proportion of phenotypic variance of SNPs was 6.32%. The phenotypic proportion was 10.31% for both SNPs in a simultaneous fit.
Single nucleotide polymorphism (SNP) markers significantly (FDR = 0.05) associated with traits evaluated under Fe-deficient and the Fe-sufficient iron regime
% r 2
Beta-fructofuranosidase, insoluble isoenzyme 6
Polyadenylate-binding protein RBP47C
Low-temperature-induced 65 kDa protein
E3 ubiquitin-protein ligase EL5
CASP-like protein Os11g0549625
SAUR-like auxin-responsive protein
Transcription factor HBP-1b(c38)
RING-finger, DEAD-like helicase, PHD and SNF2 domain
Protein CLP1 homolog
Putative zinc finger protein SHI
Mitogen-activated protein kinase 8
Sex determination protein tasselseed-2
The GWA analyses under the Fe-sufficient regime revealed in total 17 significant (FDR = 0.05) SNPs where H 2O (9) included the highest number (Table 3, Figure 3, Additional file 4: Figure S2;A, Additional file 5: Figure S4; A). The proportion of the explained phenotypic variance was highest for H 2O (21.21%). In a simultaneous fit of all significant (FDR = 0.05) SNPs, proportion of the phenotypic variance maximally explained was 57.47% (H 2O) and the minimum was 10.99% (SP3).
Under consideration of the global extent of LD, 18 and 9 unique genes were linked to the significantly (FDR = 0.05) associated SNPs under the Fe-deficient and Fe-sufficient regime, respectively (Tables 2 and 3). None of the Sanger-sequenced genes evaluated in Additional file 2: Figure S1 included SNPs that were significantly (FDR = 0.05) associated with the morphological and physiological traits.
Environmental factors such as pH variation in the soil, temperature, water stress, and mineral concentration effects have a strong influence on Fe availability for plants . To reveal genotypic effects that contribute to Fe-efficiency and avoid an overlap with other mineral nutrients, hydroponic culture has been proven to be the method of choice providing standard environmental conditions . Such a culture has been used in our study to examine the Fe-efficiency in a broad germplasm set of maize.
Dissection of phenotypic diversity and relation between the examined traits
We observed for all traits moderate to high repeatabilities under both Fe regimes (Table 1). This finding indicated that the genetic contribution to variation was minimally covered by experimental variation of hydroponics which in turn increases the power of the genetic dissection of Fe-efficiency by association mapping methods.
We observed, under the Fe-deficient regime, variation for the trait BTR (Figure 1). Long et al. 2010  revealed an Fe sensing gene named POPEYE in Arabidopsis roots during Fe-deficiency. Their finding indicated that Fe deficiency sensing mechanisms regulate terminal root branching. However, in contrast to Arabidopsis , in maize the mechanism of root branching under Fe-deficiency is not yet understood.
The whole set of traits evaluated in one Fe regime showed mostly moderate to high pairwise correlations (Figure 2). This finding suggests that for each of the Fe-sufficient and Fe-deficient regimes most of the examined traits have a joint regulation. One of the few exception was the correlation between leaf necrosis and water content, which was only observed in the Fe-sufficient regime. This positive correlation might be caused by a nutrient distortion, also known as concentration effect .
Marker-phenotype associations for QTL confidence intervals and on genome-wide scale
Using the ASMP we were able to validate 13% and 3% of detected QTLs from our former study  for Fe-deficient and Fe-sufficient regimes, respectively. Among the SNPs that were located within QTL confindence intervals , we identified a SNP (S1_28765627) in the cytochrome P450 94A1 (CYP94A1) (GRMZM2G036257) gene that was significantly associated with NEC (Table 2). CYP94A1 is responsible for modifying lipophilic compounds like fatty acids . Its involvement in plant development, repair, and defense  might indicate the contribution of stress response mechanisms during Fe-deficiency. Furthermore, cytochrome P450 family proteins might also play a role in Fe sensing  as Fe is incorporated into a heme group of the cytochrome P450 proteins .
We did not observe a clear clustering of genotypes with high NEC values in the individual subgroups. Furthermore, when examing the subgroups individually (Additional file 1: Table S1), we detected no significant associations neither for NEC nor for RW under both Fe regimes (data not shown). Additionally, excluding genotypes with a higher NEC susceptibility from the association analysis changed the results only marginally compared to the analyses with all genotypes. These results suggested that the concentration effect does not influence the conclusions of our study.
Despite the variation observed for BTR under the Fe-deficient regime, no significant associations have been detected. Therefore, further research is required on the genetics of BTR. In that context, the genes identified in our companion study  using an RNA sequencing approach can be promising starting points.
In our study, genes, known being mechanistically involved in strategy II related processes for Fe mobilization, uptake and storage, were resequenced (Additional file 6: Table S2). For polymorphisms in these genes, no significant associations were detected for both Fe regimes. This finding could be explained by a correlation of allele frequency of the mechanistically involved genes and population structure as was observed previously for flowering time and Dwarf8 [22,23]. As we did not observe a strong correlation between population structure and phenotypic variation of the studied traits this explanation is not likely to be true (Additional file 1: Table S1). The reason could be that these mechanistically involved genes have been identified by mutant screening only and that natural genetic variation at these genes leads to evolutionary disadvantages. Therefore, only neutral polymorphisms with respect to the phenotype are observed in the maize ASMP. This might reflect purified selection of these adaptive genes that does not contribute to phenotypic variation of quantitative trait .
An overlap between associated SNPs of traits were not observed putatively due to minor effect associations and a stringent significance thresholds applied in our study. Nevertheless, significant association of SNPs and their corresponding genes as described above provide an insight in the genetic architecture of biological processes characteristic for each trait that is in a direct relation to Fe-homeostasis. However, association mapping analyses provide only an indirect statistical evidence for a contribution of the considered allele to phenotypic variation  a direct functional validation is indispensable. Furthermore, additional traits like protein and transcriptome expression profiling could be performed on the association mapping population to further dissect Fe-homeostasis.
The QTL confidence intervals of the traits NEC, RW, SDW/SL, SP3, SP4, and SP6, from a previous study contained hundreds of genes and millions of base pairs. A dissection of these QTL confidence intervals using association mapping methods allowed a confirmation of the previously detected QTLs as well as the fine-mapping. In addition, our study described SNPs which were significantly associated on a genome-wide level under both Fe regimes with the traits NEC, RW, SDW, H 2O, and SP3. Several of these SNPs were located in genes (coding) or recognition sites (non-coding) of transcriptional regulators, which indicates a direct impact on the phenotype. Beside being attractive targets for marker assisted selection, these loci are interesting objects for future functional analyses.
A set of 302 maize inbred lines representing world-wide maize diversity  was used as association mapping population (ASMP) in the current study. Due to the unavailability of sufficient amounts of seeds for 35 inbred lines, a final set of 267 inbred lines was evaluated in the frame of this study (Additional file 7: Table S4).
Culture conditions and evaluated traits
Maize seeds were sterilized with 60°C hot water for 30 minutes. Afterwards, seeds were placed between two filter paper sheets moistened with saturated CaSO 4 solution for germination in the dark at room temperature. After 6 days, the germinated seeds were transplanted to a continuously aerated nutrient solution with nutrient concentrations as described by . The plants were supplied with 100 μM Fe(III)-EDTA for 7 days. From day 14 to 28, plants were cultured at 10 (Fe-deficient) and 300 (Fe-sufficient) μM iron regimes. The nutrient solution was exchanged every third day. Plants were cultivated from day 7 to day 28 in a growth chamber at a relative humidity of 60%, light intensity of 170 μmol m −2 s −1 in the leaf canopy, and a day-night temperature regime of 16 h/24°C and 8 h/22°C, respectively.
Each genotype was grown in one shaded pot of 600 milliliter volume. All pots of one Fe regime were arranged in an alpha lattice design with 13 incomplete blocks. The entire experiment was replicated b= 3 times for the Fe-deficient and sufficient regime, respectively.
Under both Fe regimes, the following traits were evaluated: the relative chlorophyll content of the 3rd, 4th, 5th, and 6th leaf (SP) measured with a SPAD meter (Minolta SPAD 502). Branching at the terminal 5 cm of the root (BTR) was evaluated with 1 for strong presence and 9 for absence of terminal root branching. Leaf necrosis (NEC) was recorded as a visual score on a scale from 1 for high trait expression and 9 for low trait expression. The lateral root formation (LAT) was recorded on a scale from 1 for absence to 9 for high trait expression. Furthermore, root length (RL), root weight (RW), shoot length (SL), shoot dry weight (SDW), water content (H 2O) as well as the ratio between SDW and SL (SDW/SL) was according to .
In our study, the data collected in this way for both Fe regimes were not directly combined to calculate a response variable for each trait in order to avoid problems related to error propagation. Instead, we followed examples from the literature and analysed data from the regimes individually but compared the results afterwards.
SNP marker data
A data set with 437,650 SNP markers for the ASMP is publicly available from http://www.panzea.org. If for one SNP more than 20% of the marker information across all inbreds was unknown or denoted as missing data, this mSNP was skipped from the following analyses. Furthermore, SNPs with minor allele frequency lower than 2.5% were excluded from the following analyses.
A set of 16 candidate genes for mobilization, uptake, storage, and transport of Fe as well as regulatory function on these processes was selected for sequence analyses to detect additional polymorphisms compared to the above mentioned SNP data set (Additional file 2: Figure S1). Primers for candidate genes were designed using software Primer3  (Additional file 8: Table S3). Each region of the candidate gene sequence was PCR amplified for the ASMP. PCR products were sequenced by the DNA core facility of the Max-Planck-Institute for Plant Breeding Research on Applied Biosystems (Weiterstadt, Germany) Abi 3730XL sequencers using BigDye-terminator v3.1 chemistry. Premixed reagents were from Applied Biosystems. The gene sequences were aligned with the software ClustalW2 (http://download.famouswhy.com/clustalw2/) and edited with BioLign (http://en.bio-soft.net/dna/BioLign.html) manually. The SNPs were filtered as described above and the remaining 562 SNPs were added to the above mentioned set of genome-wide distributed SNPs.
The residuals for each trait under both Fe regimes were tested with a Kolmogorov-Smirnov test  for their normal distribution. Pairwise correlation coefficients were assessed between all pairs of traits for the ASMP. Student’s t-tests were calculated for each trait to examine the significance of the difference between the Fe-deficient and sufficient regimes.
Physical map positions of QTL confidence intervals detected in the linkage mapping study of  were used for fine-mapping.
For each SNP of the marker set, the information about the physical map position was available. The extent of linkage disequilibrium in the maize ASMP which was estimated by  was used to determine the genes which are linked to the detected SNP in the association analysis: up and downstream of a significant association the genes included in the region 2,000 base pairs were extracted from the filtered gene set of the maize genome sequence version 5b.
If not stated differently, all analyses were performed using statistical software R .
We would like to thank the North Central Regional Plant Introduction Station (NCRIPS) for providing seeds of the association mapping population. We also thank Nicole Kliche-Kamphaus, Andrea Lossow, Nele Kaul, and Isabel Scheibert for the excellent technical support. This work was supported by research grants from the Deutsche Forschungsgemeinschaft (STI596/4-1 and WI1728/16-1) and the Max Planck Society.
- Mori S: Iron acquisition by plants. Curr Opin Plant Biol1999, 2:250–253.View ArticlePubMedGoogle Scholar
- Marschner H: Mineral Nutrition of Higher Plants (Second Edition), UK: Elsevier; 1995.Google Scholar
- Curie C, Briat JF: Iron transport and signaling in plants. Annu Rev Plant Biol2003, 54:183–206.View ArticlePubMedGoogle Scholar
- Guerinot M: It’s elementary: Enhancing Fe 3+ reduction improves rice yields. Proc Nat Acad Sci USA2007, 104:7311–7312.View ArticlePubMed CentralPubMedGoogle Scholar
- Vert G, Grotz N, Dédaldéchamp F, Gaymard F, Guerinot M, Briat JF, Curie C: IRT1, an Arabidopsis transporter essential for iron uptake from the soil and for plant growth. Plant Cell2002, 14:1223–1233.View ArticlePubMed CentralPubMedGoogle Scholar
- Römheld V: Existence of two different strategies for the acquisition of iron in higher plants. In Iron Transport in Microbes, Plants and Animals. Edited by Winkelmann G, van der Helm D, Wiley-VCH. Federal Republic of Germany; 1987:353–374.Google Scholar
- Curie C, Panaviene Z, Loulergue C, Dellaporta S, Briat JF, Walker E: Maize yellow stripe1 encodes a membrane protein directly involved in Fe III uptake. Nature2001, 409:346–349.View ArticlePubMedGoogle Scholar
- Kobayashi T, Nishizawa N: Iron uptake, translocation, and regulation in higher plants. Annu Rev Plant Biol2012, 63:131–152.View ArticlePubMedGoogle Scholar
- Lee S, Ryoo N, Jeon JS, Guerinot M, An G: Activation of rice Yellow stripe1-like 16 (OsYSL16) enhances iron efficiency. Mol Cells2012, 33:117–126.View ArticlePubMed CentralPubMedGoogle Scholar
- Urbany C, Benke A, Marsian J, Huettel B, Reinhardt R, Stich B: Ups and downs of a transcriptional landscape shape iron deficiency associated chlorosis of the maize inbreds B73 and Mo17. BMC Plant Biol2013, 13:213.View ArticlePubMed CentralPubMedGoogle Scholar
- Lee M, Sharopova N, Beavis W, Grant D, Katt M, Blair D, Hallauer A: Expanding the genetic map of maize with the intermated B73 x Mo17 (IBM) population. Plant Mol Biol2002, 48:453–461.View ArticlePubMedGoogle Scholar
- Benke A, Urbany C, Marsian J, Shi R, von Wirén N, Stich B: The genetic basis of natural variation for iron homeostasis in the maize IBM population. BMC Plant Biol2014, 14:12.View ArticlePubMed CentralPubMedGoogle Scholar
- Nguyen V, Ribot S, Dolstra O, Niks R, Visser R, van der Linden C: Identification of quantitative trait loci for ion homeostasis and salt tolerance in barley (Hordeum vulgare L.) Mol Breeding2013, 31:137–152.View ArticleGoogle Scholar
- Long T, Tsukagoshi H, Busch W, Lahner B, Salt D, Benfey P: The bHLH transcription factor POPEYE regulates response to iron deficiency in arabidopsis roots. Plant Cell2010, 22:2219–2236.View ArticlePubMed CentralPubMedGoogle Scholar
- Tijet N, Helvig C, Pinot F, Le Bouquin R, Lesot A, Durst F, Salaün JP, Benveniste I: Functional expression in yeast and characterization of a clofibrate-inducible plant cytochrome P-450 (CYP94A1) involved in cutin monomers synthesis. Biochem J1998, 332:583–589.PubMed CentralPubMedGoogle Scholar
- Colangelo E, Guerinot M: The essential basic helix-loop-helix protein FIT1 is required for the iron deficiency response. Plant Cell2004, 16:3400–3412.View ArticlePubMed CentralPubMedGoogle Scholar
- Mizutani M, Ward E, Ohta D: Cytochrome p450 superfamily in Arabidopsis thaliana: Isolation of cDNAs, differential expression, and RFLP mapping of multiple cytochromes P450. Plant Mol Biol1998, 37:39–52.View ArticlePubMedGoogle Scholar
- Cho JI, Lee SK, Ko S, Kim HK, Jun SH, Lee YH, Seong H, Lee KW, An G, Hahn TR, Jeon JS: Molecular cloning and expression analysis of the cell-wall invertase gene family in rice (Oryza sativa L.) Plant Cell Rep2005, 24:225–236.View ArticlePubMedGoogle Scholar
- Nordin K, Vahala T, Palva E: Differential expression of two related, low-temperature-induced genes in Arabidopsis thaliana (L.) Heynh. Plant Mol Biol1993, 21:641–653.View ArticlePubMedGoogle Scholar
- Hundertmark M, Hincha D: LEA (Late Embryogenesis Abundant) proteins and their encoding genes in Arabidopsis thaliana. BMC Genomics2008:9–118. 9.Google Scholar
- O’Rourke J, Charlson D, Gonzalez D, Vodkin L, Graham M, Cianzio S, Grusak M, Shoemaker R: Microarray analysis of iron deficiency chlorosis in near-isogenic soybean lines. BMC Genomics2007:8–476. 8.Google Scholar
- Van Inghelandt D, Melchinger A, Martinant JP, Stich B: Genome-wide association mapping of flowering time and northern corn leaf blight (Setosphaeria turcica) resistance in a vast commercial maize germplasm set. BMC Plant Biol2012:12–56. 12.Google Scholar
- Larsson S, Lipka A, Buckler E: Lessons from Dwarf8 on the strengths and weaknesses of structured association mapping. PLoS Genet2013, 9:e1003246.View ArticlePubMed CentralPubMedGoogle Scholar
- Benke A, Stich B: An analysis of selection on candidate genes for regulation, mobilization, uptake, and transport of iron in maize. Genome2011, 54:674–683.View ArticlePubMedGoogle Scholar
- Andersen J, Lübberstedt T: Functional markers in plants. Trends Plant Sci2003, 8:554–560.View ArticlePubMedGoogle Scholar
- Flint-Garcia S, Thuillet AC, Yu J, Pressoir G, Romero S, Mitchell S, Doebley J, Kresovich S, Goodman M, Buckler E: Maize association population: A high-resolution platform for quantitative trait locus dissection. Plant J2005, 44:1054–1064.View ArticlePubMedGoogle Scholar
- von Wirén N, Marschner H, Römheld V: Roots of iron-efficient maize also absorb phytosiderophore-chelated zinc. Plant Physiol1996, 111:1119–1125.PubMed CentralPubMedGoogle Scholar
- Rozen S, Skaletsky H: Primer3 on the WWW for general users and for biologist programmers. Methods Mol Biol (Clifton, N.J.)2000, 132:365–386.Google Scholar
- Chakravarti I, Laha R, Roy J: Handbook of Methods of Applied Statistics, Volume I, New York: John Wiley and Sons; 1967.Google Scholar
- Stich B, Möhring J, Piepho HP, Heckenberger M, Buckler E, Melchinger A: Comparison of mixed-model approaches for association mapping. Genetics2008, 178:1745–1754.View ArticlePubMed CentralPubMedGoogle Scholar
- Kang H, Zaitlen N, Wade C, Kirby A, Heckerman D, Daly M, Eskin E: Efficient control of population structure in model organism association mapping. Genetics2008, 178:1709–1723.View ArticlePubMed CentralPubMedGoogle Scholar
- Benjamini Y, Hochberg Y: Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J R Stat Soc. Ser B (Methodological)1995, 57:289–300.Google Scholar
- Sun G, Zhu C, Kramer M, Yang SS, Song W, Piepho HP, Yu J: Variation explained in mixed-model association mapping. Heredity2010, 105:333–340.View ArticlePubMedGoogle Scholar
- Remington D, Thornsberry J, Matsuoka Y, Wilson L, Whitt S, Doebley J, Kresovich S, Goodman M, Buckler E, IV: Structure of linkage disequilibrium and phenotypic associations in the maize genome. Proc Nat Acad Sci USA2001, 98:11479–11484.View ArticlePubMed CentralPubMedGoogle Scholar
- R Core Team: R: A Language and Environment for Statistical Computing, Vienna, Austria: R Foundation for Statistical Computing; 2012.Google Scholar
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.