THE PLANT GENOME Â„ JULY 2017 Â„ VOL. 10 NO. 2 t 1 OF 12 QTLs Associated with Crown Root Angle, Stomatal Conductance, and Maturity in SorghumJose R. Lopez, John E. Erickson,* Patricio Munoz, Ana Saballos, Terry J. Felderhoff, Wilfred VermerrisAbstractThree factors that directly affect the water inputs in cropping systems are root architecture, length of the growing season, and stomatal conductance to water vapor ( g s ). Deeper-rooted cultivars will perform better under water-limited conditions because they can access water stored deeper in the soil prole. Reduced g s limits transpiration rate ( E ) and thus throughout the vegetative phase conserves water that may be used during grain lling in water-limited environments. Additionally, growing early-maturing varieties in regions that rely on soil-stored water is a key water management strategy. To further our understanding of the ge netic basis underlying root depth, growing season length, and g s we conducted a quantitative trait locus (QTL) study. A QTL for crown root angle (a proxy for root depth) new to sorghum was identied in chromosome 3. For g s a QTL in chromosome seven was identied. In a follow-up eld study it was determined that the QTL for g s was associated with reduced E but not with net carbon assimilation rate ( A ) or shoot biomass. No differences in guard-cell length or stomatal density were observed among the lines, leading to the conclusion that the observed differences in g s must be explained by partial stomatal closure. The well-studied maturity gene Ma1 was identied in the QTL for maturity. The transgressive segregation of the population was explained by the possible interaction of Ma1 with other loci. Finally, the most prob able position of the genes underlying the QTLs and candidate genes were proposed.T HE competing demand for water resources is expected to intensify due to the increasing use of water for power generation, the current rate of population growth, and climate change (Roy et al., 2012). Considering that crop production accounts for 92% of the global water footprint of humanity (Hoekstra and Mekonnen, 2012), investigating dierent strategies to reduce agricultural water consumption should be a research priority. Even in industrialized countries like the USA, where the major ity of the water withdrawal goes to thermoelectric energy production, irrigation accounts for 37% of the total fresh water use (Kenny et al., 2009). In theory, crop irrigation requirements can be reduced by breeding for cultivars Published in Plant Genome Volume 10. doi: 10.3835/plantgenome2016.04.0038 Crop Science Society of America 5585 Guilford Rd., Madison, WI 53711 USA This is an open access article distributed under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).J.R. Lopez, J.E. Erickson, P. Munoz, and A. Saballos, Agronomy Dep., Univ. of Florida, PO Box 110500, Gainesville, FL, 32611; A. Saballos, Chromatin Inc., Alachua, FL, 32615; A. Saballos, T.J. Felderhoff, W. Vermerris, UF Genetics Institute, Univ. of Florida, PO Box 103610, Gainesville, FL, 32610; W. Vermerris, Dep. of Micro biology and Cell ScienceÂ–IFAS, Univ. of Florida, PO Box 110700, Gainesville, FL, 32611. Received 14 Apr. 2016. Accepted 4 Jan. 2017. *Corresponding author (email@example.com). Abbreviations: DAP, days after planting; GBS, genotyping by se quencing; NCBI, National Center for Biotechnology Information; QTL, quantitative trait locus; SNP, single-nucleotide polymorphism; WUE, water use efciency. Core Ideas t QTLs for crown root angle, stomatal conductance, and maturity were identied in two eld studies through the construction of a high-density bin map. t e QTL for stomatal conductance was associated with reduced leaf transpiration but not reduced net assimilation rate. t Candidate genes are proposed based on the physical location of the QTLs and the function of known genes in those locations. Published online July 13, 2017
2 OF 12 t THE PLANT GENOME Â„ JULY 2017 Â„ VOL. 10 NO. 2 that transpire less and intercept more rainfall water. For this purpose, a better understanding of the genetic basis of crop traits associated with plant rainfall water uptake and transpiration is needed. Sorghum [ Sorghum bicolor (L.) Moench] makes an excellent model plant to study the genetic basis of plant water uptake and utilization. e diploid nature of sor ghum, use of inbred lines, and the ability to cross dier ent genotypes enables the genetic mapping of traits of interest. Indeed, dierent drought adaptation traits and QTLs associated with plant water uptake have been iden tied in this widely grown crop (Xu et al., 2000; Borrell et al., 2014). Additionally, the sorghum genome has been fully sequenced (Paterson et al., 2009). Furthermore, because sorghum is closely related to corn ( Zea mays L.) and rice ( Oryza sativa L.), the function of sorghum genes can be reasonably inferred on the basis of the known function of the studied genes of these and other species based on sequence similarity, synteny, and homology. Recent advances in genotyping technologies allow whole-genome high-throughput genotyping of hundreds of lines with thousands of single-nucleotide polymor phisms (SNPs) markers (Elshire et al., 2011). While a cer tain number of genotyping errors are expected with this high-throughput technology, through the construction of high-density maps researchers can lter genotyping errors in biparental populations and identify the position of genes causing phenotypic variation with high accuracy (Huang et al., 2009). ese methods, coupled with novel statistical approaches that increase the power of detection in genetic studies (Kang et al., 2008), provide a powerful tool for the identication of candidate genes in eld studies. ree factors that potentially aect the water inputs of cropping systems are root depth, duration of the crop growth cycle, and stomatal conductance to water vapor ( g s ). Deeper-rooted cultivars take up more rainfall water because they can access water stored deeper in the soil prole (Uga et al., 2013) and therefore may require less irrigation. e duration of the crop growth cycle is also associated with the amount of water used by the crop and can be used by plant breeders to optimize rainfall water use (Blum, 1970). Reduced g s limits transpiration rate ( E ), potentially reduces crop water requirements (Allen et al., 1998) and can conserve water, through the vegetative phase, that may be used during grain lling in water-limited environments (Richards and Passioura, 1989; Choudhary et al., 2014). Root depth is associated with crown root angle in sorghum and other members of the Poaceae family. To study the root system of sorghum, one needs to dieren tiate among the seminal roots, which will be important for seedling establishment, and the shoot-borne roots, which will grow larger and deeper than the crown roots by the sixth-leaf stage (Singh et al., 2010) and will be responsible for most of the water and nutrient take-up by the plant thenceforth. e study of the angle of the crown rootsÂ—that is, the rst whorl of shoot-borne roots to appear aer germination (Hochholdinger, 2009)Â—has been proposed as a high-throughput strategy to evaluate root architecture in maize (Trachsel et al., 2011), whose root system morphology and development are similar to those of sorghum (Singh et al., 2010). In rice, another member of the Poaceae family, a gene that controls root angle has been shown to improve yield under drought conditions (Uga et al., 2013). Reductions in g s can limit E by increasing the resis tance to water-vapor movement from the leaf intercel lular spaces to the atmosphere. However, it also limits the rate of CO2 diusion from the atmosphere into the leaf intercellular spaces and potentially net CO2 assimilation rate ( A ). erefore, whenever a genetically controlled physiological or anatomical trait limits g s it is important to investigate how this limitation aects CO2 diusion as well. Fortunately, reductions in g s oen aect E more than A (Egea et al., 2011). For example, Balota et al. (2008) tested sorghum plants with dierent genetic back grounds and found some lines with dierent leaf E but similar A and biomass production. is could imply that water use eciency (WUE; i.e., A divided by E ) can be optimized without reducing yield. Previous research has led to the identication of genomic regions associated with root angle and g s in sorghum (Table 1; Mace et al., 2012; Kapanigowda et al., 2014). However, the studies evaluating these traits relied on low-density linkage maps or were conducted under controlled conditions only. Because of genotype q environment interactions, it is not uncommon to identify dierent QTLs for the same trait in dierent locations even within the same study (Zou et al., 2012). is will be particularly true for root architecture (de Dorlodot et al., 2007) that may be aected, for example, by soil proper ties. e objective of this project was to identify candi date sorghum genes that aect crown root angle and g s erefore, a QTL mapping study was conducted under eld conditions in two dierent locations using a highdensity SNP map. Additionally, we tested the hypothesis that small changes in g s across individuals associated with a single QTL aect E but not A or shoot biomass. Table 1. Summary of QTLs associated with nodal root angle and transpiration rate reported in the literature. Trait Chromo some Peak LOD position Physical position LOD score R 2 cM % Nodal root angleÂ† 5 51.8 13,414Â–45,780 3.7 10.0 Nodal root angleÂ† 5 34.0 55,170Â–55,691 5.0 29.7 Nodal root angleÂ† 8 25.4 8068Â–41,532 2.7 6.7 Nodal root angleÂ† 10 204.4 57,495Â–58,574 2.3 11.7 Transpiration rateÂ‡ 1 104.3 Not available 9.7 12.0 Transpiration rateÂ‡ 7 107.0 Not available 13.7 13.0 Transpiration rateÂ‡ 1 48.0 Not available 24.6 15.0Â† Reported by Mace et al. (2012). Â‡ Reported by Kapanigowda et al. (2014).
LOPEZ ET AL.: CROWN ROOT ANGLE, STOMATAL CONDUCTANCE & MATURITY QTLS IN SORGHUMt 3 OF 12 Materials and MethodsPlant Material and Experimental Design for PhenotypingA biparental mapping population developed from the cross between cultivars Early Hegari-Sart (EH) and Bk7 was evaluated. e development of this population was described by Felderho et al. (2016). Early HegariSart is an experimental forage sorghum line (Pedersen et al., 1982; Oliver et al., 2005). Bk7 is a grain sorghum cultivar derived from the GPP5BR germplasm popula tion (Duncan et al., 1991), following nine generations of selection by Dr. Daniel Gorbet (University of Florida, retired) for resistance to several diseases prevalent in Florida. In a preliminary screening of adult plants under water-limited eld conditions, Bk7 was identied as drought-sensitive on the basis of the observation that its leaves were wilted and partially desiccated, whereas the EH plants were unaected (Fig. 1). e owering times of these lines were similar under well-watered conditions, but diered in as much as 3 wk under water-limited con ditions. In this preliminary study, the root systems were excavated and shown to have quite dierent architec tures, with EH plants having steeper root angles (Fig. 1). A mapping population of 28 F3:4 and 107 F4:5 individuals was evaluated in two locations: Live Oak, FL (3018 a 29 N, 8254 a 1 W) and Citra, FL (2924 a 38 N, 828 a 30 W). Both locations have sandy soils with no restrictions to root growth. e experimental design was a randomized complete block design (RCBD) with two blocks in Live Oak and four blocks in Citra. e planting dates were 11 June 2013 for Citra and 12 June 2013 for Live Oak. e row-by-plant distance was 0.76 0.10 m in Live Oak and 0.76 0.25 m in Citra.DNA Extraction and GenotypingDNA was extracted from leaves of three young plants per line using a DNeasy Plant Mini Kit (Qiagen). e DNA of the three plants was bulked and sent to the Cornell Uni versity Institute for Genomic Diversity for genotyping. e plants were genotyped following the genotyping-bysequencing protocol (GBS; Elshire et al., 2011), and SNP calls anchored to the sorghum reference genome (Pater son et al., 2009) were identied using the TASSEL-GBS analysis pipeline, version 3.0.166 (Glaubitz et al., 2014). An initial lter of the SNPs was performed to remove markers present in less than 100 individuals, more than 30% heterozygous (because these lines are expected to be ~90% homozygous), with a minor allele frequency equal to or below 3%, or with the same genotype call for both parents. Genotypes with distorted segregation ratios were ltered using the library qtl (Broman et al., 2003) in R (R Core Team, 2015). Tight double recombinations (double recombinations occurring within 2 kb) were set to missing as described by Truong et al. (2014).High-Density Bin Map Constructione relative distance in centimorgans between the remaining SNPs was calculated using the qtl package in R (Broman et al., 2003) with the ext.map function. Four misplaced markers in chromosome 6 were removed on the basis of their recombination distance with the rest of the markers. Tight double crossovers occurring within 2 cM were set to missing (Truong et al., 2014). Note that tight double cross overs are ltered out twice: rst, on the basis of physical distance and second, on the basis of relative distance. Missing markers were imputed on the basis of the anking marker alleles using the R/qtl ll. geno function. When the anking markers on opposite sides of the missing marker had dierent alleles, the missing marker was not imputed. Adjacent markers with the same alleles for all individuals were considered to describe the same recombination block and clustered into one recombination bin (Huang et al., 2009). e nal recombination map distance was calculated with the Haldane function using the es.rf.exHet function (Truong et al., 2014) in R/qtl.PhenotypingAt each location, plants were monitored for growth stage approximately weekly aer owering. At physiologi cal maturity following hard-dough stage and readily Figure 1. Parental lines. (a) Â‘Early Hegari-SartÂ’ (EH; left ) and Â‘Bk7Â’ ( right ) have contrasting wilting responses to mild drought. (b) Root angle of the parental lines.
4 OF 12 t THE PLANT GENOME Â„ JULY 2017 Â„ VOL. 10 NO. 2 distinguishable grain black-layer formation, one rep resentative plant per plot was harvested. is practice resulted in three separate harvest dates. Days to maturity for the QTL analysis was then determined from the dif ference of the harvest date and the planting date. e number of leaves per plant of two representative plants per plot, in two blocks per location, was counted 107 d aer planting (DAP) in Citra, and 111 DAP in Live Oak. A modication of the method described by Trachsel et al. (2011) was used for root excavation at physiological maturity. In short, the roots were removed from the eld along with a cylinder of soil of 0.40 m in diameter and 0.25 m deep. e soil was gently washed away from the roots with a standard garden hose. e angles in relation to the ground of two roots of the crown root whorl were mea sured with a digital protractor Model 1702 (General Tools & Instruments). e reported plot-based root angle was the average of six measurements (three plants per plot and two roots per plant). e root angle of mature plants was evalu ated in three blocks at Citra and two blocks at Live Oak. Stomatal conductance to water vapor ( g s ) was mea sured between 45 and 48 DAP in Citra and between 57 and 59 DAP in Live Oak with a LI-COR LI-6400XT Por table Photosynthesis System (LI-COR Biosciences). One representative plant per plot was measured in four blocks in Citra and two in Live Oak between 11:00 AM and 2:00 PM on sunny days in well-watered elds.Statistical AnalysisBroad-sense heritability and marker-based estimates of narrow-sense heritability were obtained using the heri tability package (Kruijer et al., 2015) in R (R Core Team, 2015). Least square means from the genotypes were cal culated by tting a linear mixed model using ASReml v4.1.979 (Gilmour et al., 2009). In the model, location and block nested within location were considered as xed eects, while genotype [~ N (0, T 2 g G )] and genotype q environment interaction were treated as random eects. e matrix G is the genomic relationship matrix esti mated from the markers (Endelman and Jannink, 2012). e QTL analysis was implemented in R using the library rrBLUP (Endelman, 2011). A mixed model (Yu et al., 2005) with block and location as xed eects was solved using the ecient mixed-model association method (Kang et al., 2008). e additive eect and the phenotypic variance explained by the identied QTL ( R 2 ) were calculated by tting a simple linear-regression model with the phenotypic value as the response vari able and the content of alleles contributed by EH as the explanatory variable (Lynch and Walsh, 1998). We used the false-discovery-rate criteria for QTL detection (Storey and Tibshirani, 2003).Identication of Candidate GenesOnce the QTLs were detected, the most probable loca tion of the genes of interest was approximated using the 1-LOD support interval approach (Lynch and Walsh, 1998; Dupuis and Siegmund, 1999). Since the output of rrBLUP is the negative logarithm of the p-value, DaleÂ’s conversion was used to calculate the LOD score for each bin (Nyholt, 2000). e annotated genes contained within the 1-LOD support interval were downloaded from the Phytozome website (Paterson et al., 2009; Goodstein et al., 2012) using the BioMart tool (Durinck et al., 2005) and considered posi tional candidate genes. ese candidate genes were further evaluated using the sequence cluster feature within the UniGene tool available at the National Center for Biotech nology Information (NCBI) website (Pontius et al., 2003) in two ways. e rst approach was to use genes from related species known to be involved in pathways associated with variation in the trait of interest to identify orthologs within the sorghum QTL. e second approach was to evaluate all of the positional candidates for sequence similarities with genes studied in other plant species and associated with the traits evaluated. UniGene sequence clusters were obtained by evaluating expressed sequence tag databases from dier ent species and clustering them based on dierent sources of information (Pontius et al., 2003). ese alignments can suggest the function of the members of the cluster.Stomatal Density and Guard-Cell LengthTwenty lines with contrasting g s and homozygous for either Bk7 or EH alleles at qSC7 were selected to test the hypoth esis that stomatal density and/or guard cell length were the physiological explanation for the QTL eect on g s ese lines were planted in Live Oak with a row-by-plant spacing of 0.76 0.10 m in two blocks in 2015. Leaf samples were collected from two plants per block from the last fully expanded leaf. A stomatal imprint was collected using nail polish as a viscous emulsion as described by Horanic and Gardner (1967) from each side (abaxial and adaxial) of the upper-third of the leaf and photographed under a microscope using an innity one camera (Lumera Corporation) under a 20 q objective lens. ree pictures were taken at dierent parts of the imprint. e pixel-size was calculated using ImageJ soware (Schnei der et al., 2012) by taking a picture of a Petro-Housser counting chamber. It was determined that 1 mm was equiva lent to 2631 pixels and that each picture covered an area of 0.45 mm2 e total number of stomatal pores of each picture was recorded as well as the length of four guard cells per picture. An analysis of variance was conducted in R to test whether the genotype at qSC7 was associated with stomatal density or guard-cell length.Results and DiscussionHigh-Density Bin MapAer ltering, a total of 6128 SNPs were retained from the initial 282,267 SNP calls that described 2833 recombina tion bins (Table 2). e length of the map was 1559.9 cM, which is consistent with previously reported high-density linkage maps of Sorghum bicolor that have lengths span ning from 1059.2 cM to 1713 cM (Menz et al., 2002; Bow ers et al., 2003; Mace et al., 2009; Burow et al., 2011; Zou et al., 2012; Zhang et al., 2013; Truong et al., 2014). e highdensity bin map is available in Supplemental File S1.
LOPEZ ET AL.: CROWN ROOT ANGLE, STOMATAL CONDUCTANCE & MATURITY QTLS IN SORGHUMt 5 OF 12 Phenotypic Evaluatione distribution of crown root angles within the mapping population was approximately normal (Fig. 2), which is consistent with a quantitative trait. Since H 2 (broad-sense heritability) was more than twofold the h 2 (narrowsense heritability) for root angle (Table 3), and since h 2 represents the proportion of the genetic variance that is explained by the additive genetic eects alone (Poehlman and Sleper, 1995; Kruijer et al., 2015), we inferred that non additive genetic eects combined surpassed the genetic variance explained by additive genetic eects. While it is unclear from these results whether dominance or epistasis are responsible for the remainder of the H 2 for root angle, de Dorlodot et al. (2007) suggested that epistasis explains an important part of the variation in root traits. e distribution of g s was also normal, but h 2 and H 2 for g s (Table 3) and the range of the genotype LS means (Fig. 2) was low, most probably as a result of the known high eect of the environment on g s (Jarvis, 1976; Radin and Eidenbock, 1984), which decreased the genetically controlled variability for g s within this population. How ever, similar values of H 2 for this trait have been reported for at least one cotton population (Percy et al., 1996). e phenotypic values of the parents for days to matu rity and number of leaves at maturity were very similar, but the progeny segregated for these traits (Fig. 2). e narrowsense heritability across locations for days to maturity and number of leaves at maturity were high, 49.52 and 81.96%, respectively (Table 3). Interestingly, the observed segrega tion for these two traits was transgressive, with the parents having similar phenotypes but the progeny exhibiting a wide range of phenotypes for these quantitative traits.QTL AnalysisWe identied a total of seven QTLs when the data across locations was combined (Table 3). A QTL in chromo some three ( qRA3 ; Fig. 3) explained only 2.6% of the variation in root angle. In contrast, overlapping QTLs in chromosome 6 ( qDM6 and qLN6 ) explained 25.69 and 57.74% of the observed phenotypic variation in days to maturity and leaf number, respectively. Two additional QTLs for days to maturity, which explained 0.4 and 3.3% of the phenotypic variation, were identied in chro mosomes 8 and 10, respectively. For stomatal conduc tance, two QTLs were identied, one in chromosome 7, explaining 4.32% of the phenotypic variation, and one in chromosome 10, explaining 1.25% of the phenotypic variation. While these QTLs only explain a small portion of the overall phenotypic variation, they explain most of the additive genetic variation, that is, h 2 which is 5.73%. Our initial analysis identied only one QTL for days to maturity, qDM6 (Table 4), which was puzzling consid ering that the parents had contrasting alleles at this locus yet similar phenotypic values. e similar maturity of the parents implies that they also contain dierent alleles at other loci that mask the eect of qDM6 As a conse quence, the mapping population segregated for dierent alleles at dierent loci possibly interacting with Ma1 resulting in the observed transgressive segregation. To identify other genes contributing to the observed transgressive segregation for days to maturity in the prog eny, two follow-up QTL analyses were conducted. First, a QTL analysis was performed using only the lines homo zygous for EH alleles at the peak bin for qDM6, and then only the lines homozygous for Bk7 alleles at the peak bin for qDM6 were analyzed. e analysis using lines homozygous for EH alleles revealed two new QTLs in the combined data set, qDM8 and qDM10 (Table 4). While qDM10 was detected in the two eld experiments independently and when the data across locations was combined, qDM8 was detected only in the combined data set (Table 4). Subsequently, we calculated the expected phenotypes for each homozygous allele combination of qDM6 and qDM10 following FisherÂ’s decomposition of the geno typic value (Fisher, 1941; Lynch and Walsh, 1998) to dis sect the role of each parental allele in the transgressive segregation for days to maturity (Table 5). On average, the individuals with the same genotype as parent Bk7, namely qDM6BB qDM10BB matured 120 DAP. e indi viduals with the same genotypes as parent EH, namely qDM6EE qDM10EE matured on average 130 DAP. ese values are close to the mean phenotype of the population Table 2. Summary of the high-density bin map. Bins represent recombination events captured by at least one single nucleotide polymorphism (SNP). Chromosome No. of bins No. of SNPs Length Mean bin size Mean bin size Max bin size Max bin size Â—Â—Â—Â—Â—Â—Â—Â— cM Â—Â—Â—Â—Â—Â—Â—Â— Mb cM Mb 1 451 960 228.84 0.51 0.16 3.39 8.00 2 347 825 175.87 0.51 0.22 3.20 20.81 3 334 727 177.46 0.53 0.22 2.89 28.65 4 293 670 147.10 0.50 0.23 3.31 8.07 5 265 545 160.92 0.61 0.23 3.03 19.66 6 255 604 131.81 0.52 0.24 2.34 6.30 7 222 477 131.70 0.59 0.29 9.52 12.86 8 187 386 106.88 0.57 0.30 3.34 14.75 9 231 509 128.01 0.55 0.26 3.60 9.21 10 248 425 171.27 0.69 0.24 10.81 7.66 Total 2833 6128 1559.85 0.56 0.24 10.81 28.65
6 OF 12 t THE PLANT GENOME Â„ JULY 2017 Â„ VOL. 10 NO. 2 and are also similar to values of the phenotypes of the parents. In contrast, individuals with alleles from con trasting sources at each of the studied loci matured either signicantly earlier or signicantly later than the par ents. Individuals with the genotype qDM6BB qDM10EE matured 114 DAP, and individuals with the genotype qDM6EE qDM10BB matured 150 DAP (Table 5). ere fore, we conclude that the maturity of the parental line EH is delayed by its allele at qDM6 but is partially accel erated by its allele at qDM10 whereas the maturity of line Bk7 is accelerated by its allele at qDM6 but partially delayed by its allele at qDM10 Figure 2. Histograms of the genotype least square means across locations. (a) Citra, Florida. (b) Live Oak, Florida. (c) Across locations. Gray and black triangles mark least square means of the parent genotypes Â‘BK7Â’ and Â‘Early Hegari-SartÂ’ (EH) respectively. Table 3. Description of the identied quantitative trait loci (QTLs) for the traits of interest across locations and the position of the candidate genes. Trait h 2 Â† H 2 Â† QTLÂ‡ R 2 QTL Candidate gene Gene start Gene end Â—Â—Â—Â—Â—Â—Â— % Â—Â—Â—Â—Â—Â—Â— % Â—Â—Â—Â—Â—Â—Â— bp no. Â—Â—Â—Â—Â—Â—Â— Root angle 27.84 68.07 qRA3 2.60 Sobic.003G052700 4757527 4762604 Days to maturity 49.52 77.93 qDM6 25.69 Sobic.006G057900 40266956 40277108 qDM8 Â§ 0.4 qDM10 Â§ 3.3 Leaf number 81.96 92.44 qLN6 52.74 Stomatal conductance 5.73 15.47 qSC7 4.32 qSC10 1.25Â† h 2 narrow-sense heritability; H 2 broad-sense heritability. Â‡ QTL nomenclature follows the nomenclature system for rice (McCouch and CGSNL, 2008), with the difference that the letters C or L come after the QTL name for the QTLs identied using the 2013 Citra or Live Oak data alone. Â§ QTLs for days to maturity identied when analyzing only the lines homozygous for Â‘Early Hegari-SartÂ’ alleles at qDM6 R 2 was calculated using the whole data set.
LOPEZ ET AL.: CROWN ROOT ANGLE, STOMATAL CONDUCTANCE & MATURITY QTLS IN SORGHUMt 7 OF 12 A QTL in the same physical position as qDM10 asso ciated with owering time was identied by Chantereau et al. (2001) using RFLP (restriction fragment length polymorphism) markers and was anchored to the sor ghum reference genome (Paterson et al., 2009) by Mace and Jordan (2011). Interestingly, Chantereau et al. (2001) identied this QTL in only one out of six experiments conducted at dierent planting dates: in the experiment with the longest day length. Considering that qDM10 was identied in the present study only when the alleles at qDM6 that delay maturity were present, and that long days also delay maturity in sorghum, the eect of qDM10 on maturity may be enhanced when maturity is delayed. e QTLs for days to maturity and leaf number were overlapping, which raises the question whether these two traits are being controlled by the same or dierent loci. Two recombination bins, 1763 and 1764, are included in the QTL for days to maturity, while only one is located in the QTL for leaf number, 1764. A close evaluation of loci 1763 and 1764 reveals that bin 1763 is supported by only one marker while bin 1764 contains three markers (Sup plementary File S1). Since these adjacent loci dier in only two calls, it is possible that bins 1763 and 1764 are in fact describing only one recombination bin. Other authors have observed that there is a strong correlation between days to maturity and leaf number in sorghum and maize as well (Sieglinger, 1936; Chase and Nanda, 1967). In this study, no QTL for g s was identied when ana lyzing the Live Oak data alone, but a QTL for this trait in chromosome 7 was detected in Citra (Table 4). A QTL in a similar position was also identied when the data from Live Oak and Citra were combined. Combining the phenotypic data collected in both locations increased the power of QTL detection for the traits with low h 2 namely root angle and g s (Table 3). is nding was con sistent with previous studies that show that increasing the number of replicates within lines increases the power of QTL detection, especially in traits with low heritability, because repeated measurements provide more accurate estimates of the phenotype associated with the genotype (Belknap, 1998). Furthermore, the expression of the g s QTL, qSC7 in both locations was supported by the followup eld study that conrmed that qSC7 is associated with g s but does not aect A or biomass production negatively. To further improve our understanding of the physi ological processes under study, that is, g s root angle development, and maturity, and to provide a framework for future research, some of the QTLs discovered were studied in more detail. Since combining all data provides a higher power of QTL detection, the rest of the discus sion will focus only on the QTLs detected on the basis of the combined data set. Additionally, QTLs discovered when relaxing the false-discovery-rate threshold to 0.1, and those discovered using only a subsample of the pop ulation (Table 4) were also excluded from this section. e variance explained by the QTLs identied for root angle and g s were low (Table 4). is was not surprising considering that the plants were grown in the eld and these Figure 3. Manhattan plots for the quantitative trait locus analysis for various traits using the combined dataset. (a) Root angle. (b) Stoma tal conductance to water vapor. (c) Days to maturity. (d) Number of leaves at maturity. Black and gray dashed lines correspond to false discovery rates of 0.05 and 0.1, respectively. Black and gray points represent loci located in odd and even chromosomes, respectively.
8 OF 12 t THE PLANT GENOME Â„ JULY 2017 Â„ VOL. 10 NO. 2 are complex traits probably controlled by many small-eect genes and the environment (de Dorlodot et al., 2007; Nilson and Assmann, 2007). Our results diered from previous studies that reported QTLs explaining on average 14.6 and 14.8% of the variability in root angle and g s respectively (Mace et al., 2012; Kapanigowda et al., 2014). However, these dierences can be easily explained. Mace et al. (2012) minimized the environmental eect on root angle by grow ing the plants in temperature-controlled glasshouses and in custom-made pots 3 mm thick. Kapanigowda (2014) evaluated a biparental recombinant inbred line population of 70 individuals from parents with contrasting g s whereas the parents of our mapping population did not dier widely in g s (Fig. 2) since the parents were selected based on con trasting root architecture and wilting responses, not g s Additionally, the linear model used to calculate the R 2 for all the QTLs assumed an additive genetic eect. However, the additive control ( h 2 ) for root angle was less than half of the total genetic control ( H 2 ) for the same trait ( h 2 = 28% and H 2 = 68%; Table 3). Other authors have also identied QTLs for root traits with even lower R 2 (Gowda et al., 2011) than the 2.6% found in this study.Water Use Efciency, gs and qSC7Selecting for plants with reduced g s has the potential to reduce irrigation water requirements because plants with reduced g s may transpire less. However, reduced g s could also be associated with reductions in the net CO2 assimila tion rate ( A ; Farquhar and Sharkey, 1982), since partially closed stomatal pores may also limit CO2 diusion from the air into the site of carboxylation. Interestingly, while Table 4. Quantitative trait locus (QTL) peaks, number of single nucleotide polymorphisms (SNPs), and recombina tion bins within the one logarithm of the odds (1-LOD) support interval for evaluated traits sorted by their location in the sorghum genome. QTLÂ† Chromosome Peak binÂ‡ No. of bins No. of SNPs LOD support interval LOD support interval LOD score cM kb Combined data qRA3 3 844 7 14 21.23Â–23.08 4631Â–5166 4.49**b qDM6 6 1763 2 4 32.87Â–33.72 40,260Â–40,489 14.42** qLN6 6 1764 1 3 33.15Â–33.72 40,399Â–40,489 20.82** qSC7 7 2101 4 10 100.13Â–101.59 58,192Â–58,594 6.92** qDM8 Â§ 8 2171 5 5 4.18Â–6.87 1584Â–2334 2.75** qSC10 10 2586 3 4 0Â–4.71 0Â–781 2.66* qDM10c 10 2618 1 1 23.15Â–23.41 2519Â–2556 9.33** Citra qDM6C 6 1764 2 4 32.87Â–33.72 40,260Â–40,489 17.76** qLN6C 6 1764 1 4 33.15Â–33.72 40,399Â–40,489 23.84** qSC7C 7 2103 5 14 100.13Â–102.16 58,192Â–58,742 4.31** qDM10C Â§ 10 2618 11 16 23.15Â–40.14 2519Â–6669 5.20** Live Oak qLN1L 1 450 3 3 226.32Â–228.84 72,821Â–73,728 2.28* qDM6L 6 1763 3 4 31.71Â–33.72 39,708Â–40,489 6.46** qLN6L 6 1764 4 4 33.15Â–33.72 40,399Â–40,489 20.15** qDM10L1 10 2608 2 6 17.03Â–18.94 2258Â–2328 3.02** qDM10L2 Â§ 10 2618 2 5 23.15Â–23.94 2519Â–2575 5.35***Signicant at B = 0.10. **Signicant at B = 0.05. Â‡ Peak bin refers to the bin with the maximum probability of containing the underlying gene based on the LOD value. Â† QTL nomenclature follows the nomenclature system for rice (McCouch and CGSNL, 2008), with the difference that the letters C or L come after the QTL name for the QTLs identied using the Citra or Live Oak data alone. Trait abbreviations: RA, crown root angle; DM, days to maturity; LN, leaf number; SC, stomatal conductance. Â§ QTLs for days to maturity identied when analyzing only the lines homozygous for EH alleles at qDM6 Table 5. Expected phenotypes (mean) and standard errors for days to maturity for all possible homozy gous combinations of alleles at qDM6 and qDM10 and homozygous effect coefcient ( a ), derived from the 2013 experiments across locations. The contrasting a of Â‘Early Hegari-SartÂ’ (EH) alleles explain the observed transgressive segregation. qDM10BB Â† qDM10EE Â‡ a Â§ qDM6BB 120.3 o 1.5 113.7 o 1.9 3.3 qDM6EE # 150.1 o 1.2 129.6 o 3.3 10.3 a 14.9 8.0Â† Homozygous for Â‘Bk7Â’ alleles at locus qDM10 Â‡ Homozygous for EH alleles at locus qDM10 Â§ Homozygous effect coefcient for EH alleles, dened as half the difference between the expected homozygous phenotypes at that locus. Homozygous for Bk7 alleles at locus qDM6 # Homozygous for EH alleles at locus qDM6
LOPEZ ET AL.: CROWN ROOT ANGLE, STOMATAL CONDUCTANCE & MATURITY QTLS IN SORGHUMt 9 OF 12 plants with homozygous EH alleles at qSC7 had lower transpiration rates ( E ), they had similar A and overall bio mass as lines homozygous for Bk7 alleles at qSC7 (Fig. 4). To conrm that these results are replicable and to further study the underlying physiology leading to this reduction in transpiration rate, in 2015 we evaluated 20 lines with contrasting homozygous alleles at qSC7 in the eld for leaf gas exchange and leaf anatomy. Once again, we observed that the lines with contrasting homozygous alleles at qSC7 had dierent E values but similar A values (Fig. 5). erefore, qSC7 was associated with increased WUE but not with reductions in net assimilation rate across years. However, neither stomatal density nor guard cell length explained the dierences in E and g s erefore, the dierences in g s E and WUE explained by qSC7 may be associated with partial stomatal closure.Candidate GenesQTL mapping has been proven useful for the identication of candidate genes that ultimately lead to the characteriza tion of functional genes aecting quantitative traits of inter est (Pieger et al., 2001; Price, 2006). Furthermore, as shown by Price (2006), genetic maps can successfully identify the underlying functional genes accurately, even when the genes have a minimal eect on the trait. To narrow down the QTL region to the most probable location of the underlying gene, previous experience has shown that the 1-LOD sup port interval can be a useful strategy in high-density maps (Dupuis and Siegmund, 1999; Van Eerdewegh et al., 2002; Matsuda et al., 2012). e QTL for days to maturity detected with the combined data set, qDM6 contains a total of eight anno tated genes within the 1-LOD support interval (Table 4; Supplementary File S2) based on the Phytozome website (Goodstein et al., 2012). e high resolution and accu racy of our genetic map can be further supported by the identication of the well-studied Ma1 gene within these eight genes (Table 3). Ma1 has been mapped to the gene annotation Sb06g014570 in the sorghum genome (Mur phy et al., 2011), which corresponds to the gene annota tion Sorbic.006G057900 in version 2.1 (Goodstein et al., 2012). Ma1 encodes the pseudoresponse regulator protein 37 (PRR37), which in long days activates the expression of the oral inhibitor CONSTANS and represses various oral activators (Murphy et al., 2011). It is hypothesized that Ma1 is also responsible for the variation in leaf number at maturity. Support for this hypothesis can be found in the work of Chase and Nanda (1967), who observed very strong correlations between number of leaves and days to maturity in maize, and a highly signicant positive correlation between days to maturity and leaf number in this study ( P = 2.2 10 16 ). However, the strongest evidence for this hypothesis is the narrow overlapping QTLs for leaf number and days to maturity observed in both locations (Table 2) e angle of the root system has been associated with gravitropism (Uga et al., 2013), that is, the downward growth of the root system in response to gravity. In the gravitropic response, an asymmetric accumulation of auxins inhibits cell elongation in the lower side of the root (Abas et al., 2006). Figure 4. Follow-up study of plants with contrasting genotypes at the quantitative trait locus for stomatal conductance ( qSC7 ). (a) Sto matal conductance ( g s ). (b) Transpiration rate ( E ). (c) Net assimilation rate ( A ). (d) Shoot dry weight. Different letters indicate statistically signicant differences.
10 OF 12 t THE PLANT GENOME Â„ JULY 2017 Â„ VOL. 10 NO. 2 ere were a total of 39 annotated genes in the 1-LOD support interval for qSC7 (Supplementary File S2). Our ini tial hypothesis was that gene annotation Sobic.010G007600 was the strongest candidate gene to explain the observed variation in this trait (Table 4). e reasoning behind this hypothesis was the nding that this locus is homologous to CYCLIN D4;1, which in A. thaliana regulates stomatal development (Kono et al., 2007). Previous studies have asso ciated stomatal density and guard-cell length to drought tolerance (Karande and Lad, 2015) and stomatal conduc tance (Muchow and Sinclair, 1989) in sorghum. However, no dierences were observed in a follow-up study in guard cell length or stomatal density (Fig. 5) and this hypothesis was rejected. erefore, the improved WUE conferred by this QTL might be associated with partial stomatal closure rather than with the size or density of the pores. In summary, through the construction of a high-reso lution genetic map, QTLs associated with days to maturity, crown root angle, and leaf stomatal conductance in sorghum were identied. Results from the present study further our understanding of the genetic basis of these morphologi cal and physiological processes that are important for crop e proposed indole-3-pyruvic acid (IPA) auxin biosynthesis pathway in maize and Arabidopsis thaliana L. (Phillips et al., 2011) requires tryptophan aminotransferases (Phillips et al., 2011). According to Phillips et al., genes encoding tryp tophan aminotransferases are highly conserved among land plants. erefore, gene annotation Sobic.003G052700, which encodes a putative tryptophan aminotransferase (Pontius et al., 2003), was considered a strong candidate gene for this trait among the 85 genes contained in the 1-LOD support interval for qRA3 (Table 3; Supplementary File S2). According to the Unigene cluster, Sobic.003G052700 is homologous to the rice FISH BONE ( FIB ) gene (LOC_Os01g07500 or LOC_ Os01g0169800). Knock out mutants of the rice FIB gene showed a lack of gravitropism. Sobic.003G052700 was placed in the same transcript cluster as the rice FIB gene using the NCBI UniGene tool (Pontius et al., 2003). Furthermore, homology between Sobic.003G052700 the rice FIB gene, and other genes encoding tryptophan aminotransferases in other species was supported by a Bayesian phylogenetic analysis of 82 land plants (Phillips et al., 2011). In the sorghum genome v2.1, gene annotation Sobic.003G052700 corresponds to Sb06g014570 (Goodstein et al., 2012).Figure 5. Results from second eld experiment to conrm the increase in water use efciency associated with qSC7 and investigate its association with leaf anatomy. (a) The effect of qSC7 on transpiration rate ( E ) was conrmed. (b) qSC7 was not associated with changes in net assimilation rate ( A ). (c) Stomatal density is not associated with qSC7 (d) Guard-cell length is not associated with qSC7 Bars represent standard errors of the least square means. Different letters indicate statistically signicant differences.
LOPEZ ET AL.: CROWN ROOT ANGLE, STOMATAL CONDUCTANCE & MATURITY QTLS IN SORGHUMt 11 OF 12 improvement in water-limited environments. However, improvement for quantitative physiological traits with rela tively low heritability that are probably controlled by several genes with traditional breeding techniques would be very slow. Molecular techniques that identify plants with several desirable alleles for a specic environment would be needed. Further understanding of the interactions of these QTL with other loci and the environment will also be required. Never theless, the identication of these QTLs and possible underly ing genes represent a rst step toward developing sorghum genotypes that use less water and/or rely on less irrigation. us, future studies should focus on the validation of the pro posed candidate genes and understanding their interactions with other loci and the environment.Supplemental Information AvailableSupplemental File S1. High-density bin map constructed from parents EH and BK7. Supplemental File S2. Gene annotations in QTLs.Acknowledgmentse authors gratefully acknowledge funding from USDA-NIFA Biomass Research and Development Initiative competitive grant No. 2011-1000630358 (JEE; WV) and the Southeastern Sungrant Center and USDA-NIFA award No. 2010-38502-21854 (WV). Publication of this article was funded in part by the University of Florida Open Access Publishing Fund.ReferencesAbas, L., R. Benjamins, N. Malenica, T. Paciorek, J. WiÂšniewska, J.C. MoulinierAnzola, T. Sieberer, J. Friml, and C. Luschnig. 2006. Intracellular tracking and proteolysis of the Arabidopsis auxin-eux facilitator PIN2 are involved in root gravitropism. Nat. Cell Biol. 8:249Â–256. doi:10.1038/ncb1369 Allen R.G. L.S. Pereira D. Raes and M. Smith 1998 Crop evapotranspiration:. Guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper 56. FAO, Rome. Balota M. W.A. Payne W. Rooney and D. Rosenow 2008 Gas exchange and transpiration ratio in sorghum Crop Sci. 48 : 2361 Â– 2371 doi:10.2135/ cropsci2008.01.0051 Belknap J.K. 1998 Eect of within-strain sample size on QTL detection and mapping using recombinant inbred mouse strains Behav. Genet. 28 : 29 Â– 38 doi:10.1023/A:1021404714631 Blum A. 1970 Eect of plant density and growth duration on grain sorghum yield under limited water supply Agron. J. 62 : 333 Â– 336 doi:10.2134/agronj 1970.00021962006200030007x Borrell A.K. E.J. Oosterom J.E. Mullet B. George-Jaeggli D.R. Jordan P.E. Klein and G.L. Hammer 2014 Stay-green alleles individually enhance grain yield in sorghum under drought by modifying canopy development and water uptake patterns New Phytol. 203 : 817 Â– 830 doi:10.1111/nph.12869 Bowers J.E. C. Abbey S. Anderson C. Chang X. Draye A.H. Hoppe R. Jessup C. Lemke J. Lennington and Z. Li 2003. A high-density genetic recombination map of sequence-tagged sites for sorghum, as a framework for comparative structural and evolutionary genomics of tropical grains and grasses. Genetics 165 : 367 Â– 386 Broman K.W. H. Wu Sen, and G.A. Churchill 2003. R/qtl: QTL mapping in experimental crosses Bioinformatics 19 : 889 Â– 890 doi:10.1093/bioinfor matics/btg112 Burow G.B. R.R. Klein C.D. Franks P.E. Klein K.F. Schertz G.A. Pederson Z. Xin and J.J. Burke 2011 Registration of the BTx623/IS3620C recom binant inbred mapping population of sorghum J. Plant Reg. 5 : 141 Â– 145 doi:10.3198/jpr2010.04.0219crmp Chantereau J. G. Trouche J.F. Rami M. Deu C. Barro and L. Grivet 2001 RFLP mapping of QTLs for photoperiod response in tropical sorghum Euphytica 120 : 183 Â– 194 doi:10.1023/A:1017513608309 Chase S.S. and D.K. Nanda 1967 Number of leaves and maturity classication in Zea mays L Crop Sci. 7 : 431 Â– 432 doi:10.2135/cropsci1967.0011183 X000700050007x Choudhary S. T.R. Sinclair C.D. Messina and M. Cooper 2014 Hydraulic conductance of maize hybrids diering in transpiration response to vapor pressure decit Crop Sci. 54 : 1147 Â– 1152 doi:10.2135/cropsci2013.05.0303 de Dorlodot S. B. Forster L. Pags A. Price R. Tuberosa and X. Draye 2007 Root system architecture: Opportunities and constraints for genetic improvement of crops Trends Plant Sci. 12 : 474 Â– 481 doi:10.1016/j. tplants.2007.08.012 Duncan R.R. P.R. Hepperly S. Torres-Cardona and A. Sotomayor-Rios 1991 Registration of GPP5BR (M/H/F) C3 anthracnose-resistant sor ghum germplasm population Crop Sci. 31 : 241 Â– 242 doi:10.2135/cropsci19 91.0011183X003100010075x Dupuis J. and D. Siegmund 1999 Statistical methods for mapping quantitative trait loci from a dense set of markers Genetics 151 : 373 Â– 386 Durinck S. Y. Moreau A. Kasprzyk S. Davis B. De Moor A. Brazma and W. Huber 2005 BioMart and Bioconductor: A powerful link between biolog ical databases and microarray data analysis Bioinformatics 21 : 3439 Â– 3440 doi:10.1093/bioinformatics/bti525 Egea G. A. Verhoef and P.L. Vidale 2011 Towards an improved and more exible representation of water stress in coupled photosynthesisÂ–stomatal conductance models Agric. For. Meteorol. 151 : 1370 Â– 1384 doi:10.1016/j. agrformet.2011.05.019 Elshire R.J. J.C. Glaubitz Q. Sun J.A. Poland K. Kawamoto E.S. Buckler and S.E. Mitchell 2011 A robust, simple genotyping-by-sequencing (GBS) approach for high diversity species PLoS One 6 : E19379 doi:10.1371/jour nal.pone.0019379 Endelman J.B. 2011 Ridge regression and other kernels for genomic selection with R package rrBLUP Plant Genome 4 : 250 Â– 255 doi:10.3835/plantgen ome2011.08.0024 Endelman J.B. and J. Jannink 2012 Shrinkage estimation of the realized relationship matrix. G3-Genes Genom Genet. 2 : 1405 Â– 1413 Farquhar G.D. and T.D. Sharkey 1982 Stomatal conductance and photo synthesis Annu. Rev. Plant Physiol. 33 : 317 Â– 345 doi:10.1146/annurev. pp.33.060182.001533 Felderho T.J. L.M. McIntyre A. Saballos and W. Vermerris 2016 Using genotyping by sequencing to map two novel anthracose reistance loci in Sorghum bicolor G3: Genes Genomes Genet. 6 : 1935 Â– 1946 Fisher R.A. 1941 Average excess and average eect of a gene substitution Ann. Eugen. 11 : 53 Â– 63 doi:10.1111/j.1469-1809.1941.tb02272.x Gilmour A.R. B.J. Gogel B.R. Cullis R. ompson and D. Butler 2009 ASReml user guide. Release 3.0. VSN Int. Hemel Hempstead, UK Glaubitz J.C. T.M. Casstevens F. Lu J. Harriman R.J. Elshire Q. Sun and E.S. Buckler 2014 TASSEL-GBS: A high capacity genotyping by sequencing analysis pipeline PLoS One 9 : E90346 doi:10.1371/journal.pone.0090346 Goodstein D.M. S. Shu R. Howson R. Neupane R.D. Hayes J. Fazo T. Mitros W. Dirks U. Hellsten and N. Putnam 2012 Phytozome: A comparative platform for green plant genomics. Nucleic Acids Res. 40(D1): D1178Â–D1186. Gowda, V.R.P., A. Henry, A. Yamauchi, H.E. Shashidhar, and R. Serraj. 2011. Root biology and genetic improvement for drought avoidance in rice Field Crops Res. 122 : 1 Â– 13 doi:10.1016/j.fcr.2011.03.001 Hochholdinger F. 2009 e maize root system: Morphology, anatomy, and genetics In: J.L. Bennetzen and S.C. Hake, editors, Handbook of maize: Its biology Springer New York. p. 145 Â– 160 doi:10.1007/978-0-387-79418-1_8 Hoekstra A.Y. and M.M. Mekonnen 2012 e water footprint of humanity Proc. Natl. Acad. Sci. USA 109 : 3232 Â– 3237 doi:10.1073/pnas.1109936109 Horanic G.E. and F.E. Gardner 1967 An improved method of making epidermal imprints Bot. Gaz. 128 : 144 Â– 150 doi:10.1086/336390 Huang X. Q. Feng Q. Qian Q. Zhao L. Wang A. Wang J. Guan D. Fan Q. Weng and T. Huang 2009 High-throughput genotyping by whole-genome resequencing Genome Res. 19 : 1068 Â– 1076 doi:10.1101/gr.089516.108 Jarvis P.G. 1976 e interpretation of the variations in leaf water potential and stomatal conductance found in canopies in the eld Philos. Trans. R. Soc. Lond. B Biol. Sci. 273 : 593 Â– 610 doi:10.1098/rstb.1976.0035 Kang H.M. N.A. Zaitlen C.M. Wade A. Kirby D. Heckerman M.J. Daly and E. Eskin 2008 Ecient control of population structure in model organism association mapping Genetics 178 : 1709 Â– 1723 doi:10.1534/ genetics.107.080101 Kapanigowda M.H. W.A. Payne W.L. Rooney J.E. Mullet and M. Balota 2014 Quantitative trait locus mapping of the transpiration ratio related to preowering drought tolerance in sorghum ( Sorghum bicolor ) Funct. Plant Biol. 41 : 1049 Â– 1065 doi:10.1071/FP13363
12 OF 12 t THE PLANT GENOME Â„ JULY 2017 Â„ VOL. 10 NO. 2 Karande S.P. and D.B. Lad 2015 Genetics of traits imparting drought toler ant in post rainy season sorghum ( Sorghum bicolor L. Moench) Int. J. Pure App. Biosci 3 : 93 Â– 98 doi:10.18782/2320-7051.2137 Kenny J.F. N.L. Barber S.S. Hutson K.S. Linsey J.K. Lovelace and M.A. Maupin 2009 Estimated use of water in the United States in 2005. US Geological Survey Reston, VA Kono A. C. Umeda-Hara S. Adachi N. Nagata M. Konomi T. Nakagawa H. Uchimiya and M. Umeda 2007 e Arabidopsis D-type cyclin CYCD4 controls cell division in the stomatal lineage of the hypocotyl epidermis Plant Cell 19 : 1265 Â– 1277 doi:10.1105/tpc.106.046763 Kruijer W. M.P. Boer M. Malosetti P.J. Flood B. Engel R. Kooke J.J.B. Keurentjes and F.A. van Eeuwijk 2015 Marker-based estimation of heritability in immortal populations Genetics 199 : 379 Â– 398 doi:10.1534/ genetics.114.167916 Lynch M. and B. Walsh 1998 Genetics and analysis of quantitative traits. Sinauer Associates Sunderland, MA Mace E.S. and D.R. Jordan 2011 Integrating sorghum whole genome sequence information with a compendium of sorghum QTL studies reveals uneven distribution of QTL and of gene-rich regions with signi cant implications for crop improvement eor. Appl. Genet. 123 : 169 Â– 191 doi:10.1007/s00122-011-1575-y Mace E.S. J.-F. Rami S. Bouchet P.E. Klein R.R. Klein A. Kilian P. Wenzl L. Xia K. Halloran and D.R. Jordan 2009 A consensus genetic map of sorghum that integrates multiple component maps and high-throughput Diversity Array Technology (DArT) markers BMC Plant Biol. 9 : 13 doi:10.1186/1471-2229-9-13 Mace E.S. V. Singh E.J. Van Oosterom G.L. Hammer C.H. Hunt and D.R. Jordan 2012 QTL for nodal root angle in sorghum ( Sorghum bicolor L. Moench) co-locate with QTL for traits associated with drought adaptation eor. Appl. Genet. 124 : 97 Â– 109 doi:10.1007/s00122-011-1690-9 Matsuda F. Y. Okazaki A. Oikawa M. Kusano R. Nakabayashi J. Kikuchi, J. Yonemaru K. Ebana M. Yano and K. Saito 2012 Dissection of genotypeÂ– phenotype associations in rice grains using metabolome quantitative trait loci analysis Plant J. 70 : 624 Â– 636 doi:10.1111/j.1365-313X.2012.04903.x McCouch, S.R., and CGSNL (Committee on Gene Symbolization, Nomenclature and Linkage, Rice Genetics Cooperative). 2008. Gene nomenclature system for rice. Rice 1: 72Â–84. doi:10.1007/s12284-008-9004-9 Menz M.A. R.R. Klein J.E. Mullet J.A. Obert N.C. Unruh and P.E. Klein 2002 A high-density genetic map of Sorghum bicolor (L.) Moench based on 2926 AFLP, RFLP and SSR markers Plant Mol. Biol. 48 : 483 Â– 499 doi:10.1023/A:1014831302392 Muchow R.C. and T.R. Sinclair 1989 Epidermal conductance, stomatal den sity and stomatal size among genotypes of Sorghum bicolor (L.) Moench. Plant. Cell Environ. 12 : 425 Â– 431 doi:10.1111/j.1365-3040.1989.tb01958.x Murphy R.L. R.R. Klein D.T. Morishige J.A. Brady W.L. Rooney F.R. Miller D.V. Dugas P.E. Klein and J.E. Mullet 2011 Coincident light and clock regulation of pseudoresponse regulator protein 37 (PRR37) controls photoperiodic owering in sorghum Proc. Natl. Acad. Sci. USA 108 : 16469 Â– 16474 doi:10.1073/pnas.1106212108 Nilson S.E. and S.M. Assmann 2007 e control of transpiration. Insights from Arabidopsis Plant Physiol. 143 : 19 Â– 27 doi:10.1104/pp.106.093161 Nyholt D.R. 2000 All LODs are not created equal Am. J. Hum. Genet. 67 : 282 Â– 288 doi:10.1086/303029 Oliver A.L. J.F. Pedersen, R.J. Grant and T.J. Klopfenstein 2005 Comparative Eects of the Sorghum-6 and -12 Genes: I. Forage Sorghum Yield and Quality Crop Sci. 45 : 2234 Â– 2239 doi:10.2135/cropsci2004.0644 Paterson A.H. J.E. Bowers R. Bruggmann I. Dubchak J. Grimwood H. Gundlach G. Haberer U. Hellsten T. Mitros and A. Poliakov 2009 e Sorghum bicolor genome and the diversication of grasses Nature 457 : 551Â– 556 doi:10.1038/nature07723 Pedersen, J.F. H.J. Gorz F.A. Haskins and W.M. Ross 1982 Variability for quality and agronomic traits in forage sorghum hybrids Crop Sci. 22 ( 4 ): 853 Â– 856 doi:10.2135/cropsci1982.0011183X002200040037x Percy R.G. Z. Lu J.W. Radin E.L. Turcotte and E. Zeiger 1996 Inheritance of stomatal conductance in cotton ( Gossypium barbadense ) Physiol. Plant. 96 : 389 Â– 394 doi:10.1111/j.1399-3054.1996.tb00449.x Pieger S. V. Lefebvre and M. Causse 2001 e candidate gene approach in plant genetics: A review Mol. Breed. 7 : 275 Â– 291 doi:10.1023/A:1011605013259 Phillips K.A. A.L. Skirpan X. Liu A. Christensen, T.L. Slewinski C. Hudson S. Barazesh J.D. Cohen S. Malcomber and P. McSteen 2011 vanishing tassel2 encodes a grass-specic tryptophan aminotransferase required for vegetative and reproductive development in maize Plant Cell 23 : 550 Â– 566 doi:10.1105/tpc.110.075267 Poehlman J.M. and D.A. Sleper 1995 Breeding eld crops. 4th ed. Iowa State Univ. Press Ames Pontius J.U. L. Wagner and G.D. Schuler 2003. UniGene: A unied view of the transcriptome. In: NCBI Handbook. Natl. Center for Biotechnol. Information Bethesda, MD. Chapter 21. Price A.H. 2006 Believe it or not, QTLs are accurate! Trends Plant Sci. 11 : 213 Â– 216 doi:10.1016/j.tplants.2006.03.006 Radin J.W. and M.P. Eidenbock 1984 Hydraulic conductance as a factor limiting leaf expansion of phosphorus-decient cotton plants Plant Physiol. 75 : 372 Â– 377 doi:10.1104/pp.75.2.372 R Core Team 2015 R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.rproject.org/ (accessed 20 Feb. 2017). Richards R.A. and J.B. Passioura 1989 A breeding program to reduce the diameter of the major xylem vessel in the seminal roots of wheat and its eect on grain yield in rain-fed environments Crop Pasture Sci. 40 : 943 Â– 950 doi:10.1071/AR9890943 Roy S.B. L. Chen E.H. Girvetz E.P. Maurer W.B. Mills and T.M. Grieb 2012 Projecting water withdrawal and supply for future decades in the US under climate change scenarios Environ. Sci. Technol. 46 : 2545 Â– 2556 doi:10.1021/es2030774 Schneider C.A. W.S. Rasband and K.W. Eliceiri 2012 NIH Image to ImageJ: 25 years of image analysis Nat. Methods 9 : 671 Â– 675 doi:10.1038/nmeth.2089 Sieglinger J.B. 1936 Leaf number of sorghum stalks J. Am. Soc. Agron. 28 : 636 doi:10.2134/agronj1936.00021962002800080005x Singh V. E.J. van Oosterom D.R. Jordan C.D. Messina M. Cooper and G.L. Hammer 2010 Morphological and architectural development of root systems in sorghum and maize Plant Soil 333 : 287 Â– 299 doi:10.1007/ s11104-010-0343-0 Storey J.D. and R. Tibshirani 2003. Statistical signicance for genome wide studies Proc. Natl. Acad. Sci. USA 100 : 9440 Â– 9445 doi:10.1073/ pnas.1530509100 Trachsel S. S.M. Kaeppler K.M. Brown and J.P. Lynch 2011 Shovelomics: High throughput phenotyping of maize ( Zea mays L.) root architecture in the eld Plant Soil 341 : 75 Â– 87 doi:10.1007/s11104-010-0623-8 Truong S.K. R.F. McCormick D.T. Morishige and J.E. Mullet 2014 Resolution of genetic map expansion caused by excess heterozygosity in plant recombinant inbred populations. G3: Genes Genom es Genet. 4 : 1963 Â– 1969 Uga Y. K. Sugimoto, S. Ogawa J. Rane M. Ishitani N. Hara Y. Kitomi Y. Inukai K. Ono and N. Kanno 2013 Control of root system architecture by DEEPER ROOTING 1 increases rice yield under drought conditions Nat. Genet. 45 : 1097 Â– 1102 doi:10.1038/ng.2725 Van Eerdewegh P. R.D. Little J. Dupuis R.G. Del Mastro K. Falls J. Simon D. Torrey S. Pandit J. McKenny and K. Braunschweiger 2002 Associa tion of the ADAM33 gene with asthma and bronchial hyperresponsive ness. Nature 418 : 426 Â– 430 doi:10.1038/nature00878 Xu W. P.K. Subudhi O.R. Crasta D.T. Rosenow J.E. Mullet and H.T. Nguyen 2000 Molecular mapping of QTLs conferring stay-green in grain sorghum ( Sorghum bicolor L. Moench) Genome 43 : 461 Â– 469 doi:10.1139/ gen-43-3-461 Yu J. G. Pressoir W.H. Briggs I.V. Bi M. Yamasaki J.F. Doebley M.D. McMullen B.S. Gaut D.M. Nielsen and J.B. Holland 2005 A unied mixed-model method for association mapping that accounts for multiple levels of relatedness Nat. Genet. 38 : 203 Â– 208 doi:10.1038/ng1702 Zhang Z. J. Sang L. Ma G. Wu H. Wu D. Huang D. Zou S. Liu A. Li and L. Hao 2013 RiceWiki: A wiki-based database for community curation of rice genes Nucleic Acids Res. 42 : D1222 Â– D1228 Zou G. G. Zhai Q. Feng S. Yan A. Wang Q. Zhao J. Shao Z. Zhang J. Zou and B. Han 2012 Identication of QTLs for eight agronomically impor tant traits using an ultra-high-density map based on SNPs generated from high-throughput sequencing in sorghum under contrasting photoperiods J. Exp. Bot. 63 : 5451Â– 5462 doi:10.1093/jxb/ers205