Genetic Control of the Leaf Angle and Leaf Orientation Value as Revealed by Ultra-High Density Maps in Three Connected Maize Populations
Genetic Control of the Leaf Angle and Leaf Orientation Value as Revealed by Ultra-High Density Maps in Three Connected Maize Populations
Chunhui Li 0 1
Yongxiang Li 0 1
Yunsu Shi 0 1
Yanchun Song 0 1
Dengfeng Zhang 0 1
Edward S. Buckler 0 1
Zhiwu Zhang 0 1
Tianyu Wang 0 1
Yu Li 0 1
0 1 Institute of Crop Science, Chinese Academy of Agricultural Sciences , Beijing, 100081, China , 2 Institute for Genomic Diversity, Cornell University , Ithaca, New York , United States of America
1 Academic Editor: Rongling Wu, Pennsylvania State University , UNITED STATES
Plant architecture is a key factor for high productivity maize because ideal plant architecture with an erect leaf angle and optimum leaf orientation value allow for more efficient light capture during photosynthesis and better wind circulation under dense planting conditions. To extend our understanding of the genetic mechanisms involved in leaf-related traits, three connected recombination inbred line (RIL) populations including 538 RILs were genotyped by genotyping-by-sequencing (GBS) method and phenotyped for the leaf angle and related traits in six environments. We conducted single population quantitative trait locus (QTL) mapping and joint linkage analysis based on high-density recombination bin maps constructed from GBS genotype data. A total of 45 QTLs with phenotypic effects ranging from 1.2% to 29.2% were detected for four leaf architecture traits by using joint linkage mapping across the three populations. All the QTLs identified for each trait could explain approximately 60% of the phenotypic variance. Four QTLs were located on small genomic regions where candidate genes were found. Genomic predictions from a genomic best linear unbiased prediction (GBLUP) model explained 459% to 688% of the variation in the remaining RILs for the four traits. These results extend our understanding of the genetics of leaf traits and can be used in genomic prediction to accelerate plant architecture improvement.
Funding: This study was supported by the Ministry of
Science and Technology (2011CB100105,
2011DFA30450), National Natural Science
Foundation (91335206), US-NSF(0820619,
1238014), USDA-ARS. The funders had no role in
study design, data collection and analysis, decision to
publish, or preparation of the manuscript.
Over the past few decades, plant architecture improvement has greatly increased maize grain
yields . The leaves of maize hybrids in particular have become more upright. Erect leaves
can effectively contribute to the maize grain yield by enhancing light capture for
photosynthesis, serving as nitrogen reservoirs for grain filling and enabling denser planting with a higher
leaf area index . Therefore, understanding the genetic mechanisms of plant leaf
architecture will not only address a fundamental issue in plant science but also facilitate the genetic
Competing Interests: The authors have declared
that no competing interests exist.
improvement of maize breeding . The leaf angle, leaf length, and leaf width are important
components of maize leaf architecture. The leaf orientation value accounts for the ability of
leaves to maintain the same orientation for their entire length, and it is a good selection index
for plant leaf orientation .
Natural variations in leaf architecture have been identified by using quantitative trait loci
(QTL) mapping in only a few different bi-parental maize mapping populations .
Mickelson et al.  found nine QTLs for the leaf angle by using a B73 x Mo17 population with 180
recombination inbred lines (RILs) and 192 genetic markers. Pelleschi et al.  detected five
QTLs for leaf length and seven QTLs for leaf width under different water environments by
using 120 RILs and 153 markers. Yu et al.  located a total of nine QTLs for the leaf angle in
two different populations (120 F2:3 families with 102 SSR markers and 114 F2:3 families with 90
SSR markers). Lu et al.  located six QTLs for the leaf angle and eight QTLs for the leaf
orientation value by using 397 F2:4 families and 137 SSR markers. Ku et al.  detected three
QTLs for the leaf angle, three QTLs for the leaf length, four QTLs for the leaf width, and five
QTLs for the leaf orientation value by using 229 F2:3 families and 222 SSR markers. The QTLs
obtained through these studies usually had large confidence intervals, and thus it was difficult
to identify the narrow recombination bins and underlying causal genes. Moreover, inconsistent
results in term of the QTL location as found in the above studies had to be further validated.
Tian et al.  used a nested association mapping (NAM) population in maize to conduct joint
linkage mapping for the leaf architecture, resulting in 30 QTLs for the leaf angle, 36 QTLs for
the leaf length, and 34 QTLs for the leaf width. Although 100 QTLs for leaf-related traits were
obtained by Tian et al. , the parents of the NAM population only include a portion of the
global maize diversity. Over approximately 500 years of selection, the Chinese maize
germplasm has become well-adapted to the numerous ecological regions of China and is
substantially different from US and Latin American germplasms . Therefore, further research into the
genetic mechanism underlying the leaf architecture could provide more favorable alleles for
maize genetic improvement. In addition, when this NAM panel was studied, the RILs were
only genotyped with 1106 SNP markers, which limited the high-resolution linkage
High-throughput genotyping based on whole-genome sequencing data provides
informative genome-wide and high density markers for mapping a population . High-density
markers can greatly improve the QTL mapping resolution and facilitate the identification of
additional recombination events and exact recombination breakpoints. In rice, two different
research groups used high-density genetic maps of two different RIL populations that were
genotyped by sequencing to identify the QTLs, resulting in two QTLs for the grain length and
width in regions of less than 200 kb that contained the GS3 and GW5 genes, respectively [17,
18]. In maize, an ultra-high density map of a large F2 population genotyped through GBS
identified one QTL in a 700 kb region containing an r1 gene that controlled the color of silk .
Therefore, the high-density map constructed by sequencing will be promising for candidate
gene identification. Additionally, high-density genome-wide markers will lead to better
genomic prediction in plant breeding. A genomic best linear unbiased prediction (GBLUP) that
employs genomic relations to estimate the genetic merit of an individual has been shown to obtain
accurate breeding values in breeding programs [20, 21]. This model allows for the prediction of
breeding values in genotyped individuals before evaluating the phenotypic values of complex
traits. This method will help breeders to select lines with superior potential performance from
a larger germplasm pool and to accelerate genetic gains.
In this study, HUANGZAOSI was crossed as a common parent with three elite Chinese
maize inbred lines to construct three connected RIL populations. These populations were used
to conduct high-resolution QTL mapping for four leaf architecture traits within a single
population and across all populations through GBS. A joint population analysis coupled with a
high density genetic map provides high resolution for many QTL positions, permitting the
robust evaluation of underlying candidate genes. Based on the genomic identity-by-state (IBS)
matrices constructed through the GBS genotyping of the three populations, the cross-validated
GBLUP model was applied to assess the accuracy of predicting each lines mean trait value
across the three populations.
Three recombination inbred line (RIL) populations were obtained by crossing the common
parent HUANGZAOSI (HZS) with each of other three inbred lines, namely HUOBAI,
WEIFENG322, and LV28. From the F2 progeny of each cross, a single seed descent was applied to
produce RILs at the F7 generation. The HUOBAI, WEIFENG322, and LV28 populations
included 183, 172, and 183 RILs, respectively. The parents of these populations were chosen on
the basis of their different leaf architecture and maize germplasm groups. The common parent
(HZS) is an important elite foundation inbred line with compact leaf architecture derived from
Chinese Tangsipintou germplasm, a heterotic group used broadly in China. HZS was
frequently used in Chinese maize breeding. In using HZSs as parental lines, more than 70 descended
inbred lines and 80 important hybrids were released , with the total planting area of these
hybrids exceeding more than 10 million ha 17 years ago . HUOBAI and WEIFENG322
were two foundation inbred lines with semi-compact and expanded leaf architecture,
respectively. LV28 is an elite foundation inbred line with expanded leaf architecture derived from
Chinese Luda Red Cob germplasm, a heterotic group used broadly in China.
Fields environments and trait evaluations
The phenotypes were measured in six field environments and performed over two years (2009
and 2010) at three different locations (Xinxiang of Henan province, Beijing, and Urumqi of
Xinjiang province), where the institute of crop science belonging to the Chinese Academy of
Agricultural Sciences has set up experimental field bases. The institute of crop science was
approved for field experiments, and the field studies did not involve endangered or protected
species. For each population, all lines were randomly grown within each replication with
singlerow plots of 11 plants. Two replications of each population were planted adjacent to one
another. Each plot was 3 m in length and 0.6 m apart. Three representative plants from the middle of
each plot were chosen to measure the four-leaf traits as follows at the 10th day after anthesis.
The leaf angle (LA) was scored as the angle of each leaf from a plane defined by the stalk below
the node subtending the leaf. The leaf length (LL) was measured as the length from the base of
the ligula to tip of the leaf. The leaf width (LW) was measured by taking the width of the widest
section of the leaf. The leaf orientation value (LOV) was calculated as follows: LOV = 1/n (90
)(Lf / LL) where is the leaf angle, Lf is the distance from the base of the ligule to the
flagging point of the measured leaf, LL is the leaf length, and n is the number of measured leaves
. Three consecutive leaves including the first leaf above the primary ear, the primary ear
leaf and the first leaf below the primary ear were measured for each of three plants. The trait
value for each RIL was averaged for the three measured plants in each replication.
Phenotypic data analysis
For each line of a single RIL population, the best linear unbiased prediction (BLUP) for all
traits across environments were obtained by PROC MIXED in SAS 9.2. In the model, the
environment, line, replication environment and line environment were considered as random
effects. The broad-sense heritability (H2) for each trait across the environments was calculated
on a plot basis by using the ANOVA tool in QTL IciMapping Version 3.3 . The correlation
coefficients among four traits were obtained on the basis of the BLUP with the cor function in
the R software package .
Genetic map construction and QTL mapping
The three RIL populations have been genotyped by using genotyping-by-sequencing (GBS)
technology [26, 27]. The GBS data are available from S1 File. Recombination bin maps were
constructed for each of the three populations, resulting in 1595, 1981, and 2091 bins for the
HUOBAI, WEIFENG322, and LV28 populations, respectively. The composite recombination
bin map including 4932 bins was also built across all the populations (unpublished). The bins
were treated as genetic markers to construct a linkage map. The high-density maps had a
powerful resolution for QTL mapping. The QTL analysis for the individual RIL population was
conducted with inclusive composite interval mapping in QTL IciMapping software Version 3.3
. The LOD threshold was determined by a 1000 permutation test. The P values for entering
a variable (PIN) and removing a variable (POUT) were set at 0.001 and 0.002, and the scanning
step was set to 1.0 cM. Joint linkage mapping across the three populations was conducted by
PROC GLMSelect in SAS 9.2. The detailed mapping procedure was previously described .
The phenotypic variation explained (PVE) by each QTL was counted in a previously described
study . A 2 LOD-drop in the confidence interval was used for each QTL.
Cross-validated genomic prediction across three RIL populations
The GBLUP of the rrBLUP package in R v3.0.2 was applied for genomic prediction . We
used the van Raden method  to construct the identity-by-state (IBS) genomic relationship
matrix based on the GBS data of the three populations. All the RILs of the three populations
were randomly divided into five disjointed subsets for cross-validation, where the line values
from combinations of one to four subsets were used to calibrate models and predict the line
values of remaining subsets . This process was repeated 20 times. The prediction accuracy
was counted as the coefficient of determination obtained by BLUP line means against the
predicted line means obtained by GBLUP averaged over all the cross-validation runs.
Phenotypic variations were identified for LA, LL, LW, and LOV within three RIL populations
(Table 1). The WEIFENG322 population had the greatest phenotypic variation for the four
traits. The variation ranges for the four traits were similar between the HUOBAI and LV28
populations. The broad-sense heritability for LA, LL, LW, and LOV reached 0.68, 0.75, 0.63,
and 0.64, respectively. Approximately 30.2% and 29.8% for LA and LOV variations across the
three populations were attributed to environmental variations (Fig. 1). These variations were
greater than that observed for LL or LW. Nonetheless, the manual phenotyping method for LA
may confound measures of environmental variation.
The phenotypic correlations among the four traits are shown in Fig. 2. The largest
correlation was estimated between LA and LOV (r = 0.81, p<0.0001). LA showed the most
significant negative correlation to LL (r = 0.18, p<0.0001) and LW (r = 0.23, p<0.0001). The most
significant positive correlation was observed between LL and LW (r = 0.51, p<0.0001).
Linkage maps of recombination bins
Linkage maps for the HUOBAI, WEIFENG322, and LV28 populations were separately
constructed by using 1595, 1981, and 2091 recombination bins, which were obtained through the
genotyping-by-sequencing of the three RIL populations. The total lengths of the genetic linkage
maps were 1746.61, 1960.57, and 1761.19 cM long with averages of 1.10, 0.99, and 0.84 cM
between two neighboring bins for the HUOBAI, WEIFENG322, and LV28 populations,
respectively. The composite map constructed from 4932 bins across the three populations was
1700.44 cM long with an average of 0.34 cM between two adjacent bins.
Single population and combined multiple population QTL mapping
A QTL analysis of each population was separately conducted for each trait, resulting in a total
of 13, 6, 8, and 13 QTLs for the LA, LL, LW, and LOV, respectively (Fig. 3). We found that
Fig 1. Partitioning variations in LA, LL, LW, and LOV across three populations.
Fig 2. Trait correlations between LA, LL, LW, and LOV variations based on BLUP values across three
populations. Right-top represents the correlation coefficients among four traits. Diagonal represents the
frequency distribution for each of four traits. Left-bottom represents the scatter distribution among four traits.
most of the QTLs for each trait were unique to a single mapping population. In general, the
joint population analysis had much higher accuracy and power than single population
mapping for detecting common QTLs . Furthermore, joint linkage mapping across the three
populations was performed to estimate the QTL position accurately for the four traits (Fig. 3,
S1 Table). The joint analysis identified 17 QTLs for LA, 9 for LL, 6 for LW, and 13 for LOV,
which collectively explained 58%, 59%, 51%, and 55% of the total phenotypic variations,
respectively. The phenotypic variation explained (PVE) by individual QTLs across all
populations ranged from 1.2% to 29.2% (S1 Table). A QTL for LL on chromosome 3 explained the
largest proportion of phenotypic variance. The 2-LOD support interval of all QTLs mapped
with the joint linkage analysis ranged from 0.602 Mb to 33.168 Mb, with a median of 2.838 Mb.
The high resolution of the QTL position in the joint analysis greatly narrowed the genome
distance in which causal genes could be searched through a forward genetic method.
The separate effect at each QTL for all three populations can be estimated by joint linkage
analysis. A total of 34 alleles for LA, 14 for LL, 12 for LW, and 27 for LOV were significant at
P<0.05 (S1 Table). All QTL additive allelic effects were small relative to the amount of variance
observed among parental differences, with the largest allelic effects for LA, LL, LW, and LOV
being 2.49, 2.42 cm, 0.32 cm, and 2.67, respectively. The allelic series with both positive and
negative additive allelic effects were identified in 28% to 61% of the QTL, depending on
Fig 3. The leaf angle (LA), leaf length (LL), leaf width (LW), and leaf orientation value (LOV) QTLs obtained from single and joint-linkage QTL
analyses across all the 10 chromosomes. Other parents were crossed with the HZS common parent to derive each bi-parental population as shown on the
right vertical axis. The result of joint linkage mapping across the three populations is indicated as JL. The physical distance for each chromosome is
represented in Mb units on the horizontal axis. Chromosomes were separated by solid gray lines. A solid circle size represents different LOD values ranging
from 4.0 to 16.0. Different circle colors represent different traits. Arrows represent the physical location of the known maize leaf architecture genes
overlapping with or adjacent to joint QTL.
With a relatively high mapping resolution, the small genomic regions of some QTLs can be
used to search for candidate genes related to traits. For the leaf angle QTL, qLA2a was located
in the region that ranged from 3.09 to 4.30 Mb on chromosome 2. The strong candidate gene
lg1 (liguleless1) that controls the leaf angle was previously cloned by mutagenesis in maize 
and was located in this region. The lg1 mutant has no ligule or auricle, leading to considerably
more upright leaves than their normal counterparts. For the leaf length, qLL2b was mapped to
the region that ranged from 233.55 to 236.61 Mb on chromosome 2, and the gene gn (gnarley1)
that regulates leaf initiation and was cloned in maize  was found in that interval. qLL4 is
located in the 236.29237.51 Mb region and was approximately 90 kb apart from the knox7
(knotted related homeobox 7) gene, which regulated leaf initiation in maize . For the leaf
width, qLW1 was overlapped with the position of the homolog for Arabidopsis PHB, which
acts in the establishment of Arabidopsis leaf polarity . For the leaf orientation value, the
qLOV4 spaced from 18.37 to 21.66 Mb on chromosome 4 was approximately 400 kb apart
from the position of the ZmCLA4 gene, which controlled the leaf angle and was cloned in
maize . qLOV5 overlapped with the location of YABBY15, which was expressed on the
adaxial side of the incipient and developing leaf primordia in maize .
The genomic prediction of four traits
A genomic prediction by GBLUP was conducted to determine the ability of all genotyped
diversity to explain variations in the BLUP line means for LA, LL, LW, and LOV across the three
populations (Fig. 4). For the cross-validation of LA across all lines, 20% of the lines (106 RILs)
that were randomly chosen from all the lines to calibrate a model could explain 235% of the
Fig 4. Genomic predictions of the leaf angle (LA), leaf length (LL), leaf width (LW), and leaf orientation
value (LOV) in all RILs by GBLUP. A random sample of 20%, 40%, 60%, and 80% of the RILs in all
populations for the calibrated GBLUP model to predict the BLUP line means variation in (A) LA, (C) LL, (E)
LW, and (G) LOV in the remaining lines. All RILs in all populations were used to calibrate the GBLUP model
to explain variations in (B) LA, (D) LL, (F) LW, and (H) LOV.
LA variation in the remaining RILs (Fig. 4A). When 80% of the lines (432 RILs) were used to
calibrate the model, the variation in LA that was explained increased to 459%. This finding
suggested that the prediction accuracy of LA might be greatly improved as the calibration
subset size increases. For LL variation, the prediction accuracy was strong in different calibration
subset sizes (Fig. 4C). The calibration subsets of 106 RILs (20% of lines) explained 605% of LL
variation in the remaining RILs. As the calibration subset size increased to 432 RILs (80% of
lines), the prediction accuracy (688%) could not be substantially improved further. The LW
cross-validation revealed that 20% of the RILs could explain only 3710% of the LW variation
in the remaining RILs (Fig. 4E). This percentage was much lower than the prediction accuracy
of LL in the same calibration subset size. The prediction accuracy for LOV was a little more
than that of the LA in different calibration subset sizes (Fig. 4A and 4G).
We performed a literature review on the leaf architecture QTL as reported in maize in linkage
mapping studies [914, 39] and compared the published QTL with those identified in this
study. In the present study, fourteen of the 45 QTLs for leaf architecture could be validated by
using the published QTL. For example, the known classic lg1 gene controlling the leaf angle
 was found to be located in an approximately 1-Mb region of qLA2a on chromosome 2.
Tian et al.  also detected one significant QTL in the 2-Mb region at the nearby lg1 on
chromosome 2. Ku et al.  used a meta-QTL analysis to find an mQTL in the region between
umc1165 and bnlg1297, which overlapped with the location of lg1. These QTLs that were
detected in different genetic backgrounds and environments shared a high congruence, which
strongly supported the candidacy of lg1 for qLA2a. One QTL for LOV called qLOV5 was
identified on chromosome 5 in the region between 181.888 and 204.347 Mb. Mickelson et al. 
and Lu et al.  also detected two QTLs for LA on chromosome 5, one QTL nearest to marker
bnlg5.02 and the other one between umc1822 and phi048. Ku et al.  also found one
important QTL for LA on chromosome 5 between bnlg1287 and mmc0282 and predicted YABBY15
to be the candidate gene in this region. These previous results consistently suggested that this
region contains a key gene for LA. Because the highest correlation occurred between LA and
LOV, this region might include a gene that commonly controls the two traits. There was no
such strong evidence for the remaining QTL in this study. However, we found that some
candidate genes, such as ZmCLA4 and knox7, perhaps support some of the detected QTLs.
In this research, we found thirty-one QTLs that were not reported in previous
investigations. Several reasons may explain why these QTLs could be found in our current study. First,
these QTLs were polymorphic in at least one population because QTL detection is generally
based on natural allelic differences between parents [40, 41]. Second, the joint QTL mapping of
multiple connected populations provides an improved power to detect QTLs . For
example, a joint linkage analysis of the three connected RIL populations could detect 17 QTLs for
LA, 9 for LL, 6 for LW, and 13 for LOV; however, single RIL population mapping could only
identify averages of 4, 2, 3, and 4 QTLs for LA, LL, LW, and LOV, respectively. Finally,
highdensity markers could increase the power of QTL mapping, especially for QTLs with small
genetic effects . Seven QTLs with PVE values >10%, namely qLA9a, qLL1a, qLL3, qLW5,
qLW7, qLOV3b and qLOV7, were found in the thirty-one QTLs. qLL1a, qLL3, qLOV3b and
qLOV7 were still detected by individual population mapping. Although these major QTLs were
not overlapped with the reported maize leaf architecture genes, they could be important target
regions of fine mapping to identify candidate genes for leaf architecture, especially for the four
major QTLs detected by joint linkage mapping and single population mapping. These new
reported QTLs not only provide new genomic regions for the further identification and
characterization of genes responsible for maize leaf architecture, but they also facilitate
marker-assisted selection (MAS) for maize plant architecture improvement to develop hybrids that are
better suited to high-density planting.
A total of six co-localized QTL groups were detected among the 45 QTLs obtained by the joint
linkage mapping for the four traits. Among these, the four groups located on chromosomes 1,
6, 9, and 10 were involved with the traits of both LA and LOV. To test the pleiotropy between
LA and LOV, the correlation of allelic effect estimates across the three no-common parents at
co-localized QTLs was found to be high (r = 0.61). Therefore, the same sets of genetic variants
likely control natural variation at these co-localized loci between LA and LOV. The other two
groups were identified as co-localizations between LW and LOV on chromosome 5, and
among the LA, LL, and LW on chromosome 2. Few genetic overlaps were found among LA,
LL, and LW in this study, which was consistent with the result obtained by Tian et al. .
Therefore, there were few pleiotropy locus functions among these traits.
Predicting the leaf angle and related traits
Predicting the phenotypic value before trait evaluation could help breeders to optimize the
allocation of evaluation resource to lines with the most predicted promise . The GBLUP
prediction model based on identity-by-state (IBS) has attracted interest in plant breeding [20, 21].
With the rapid progress of next-generation DNA sequencing technologies, millions of
molecular markers could be used to estimate IBS genomic relations accurately . To assess the
prediction accuracies of our genetic models for LA and other traits, we evaluated the GBLUP
performance across the three RIL populations. In cross-validations of the four traits, although
the prediction accuracies of a given calibration set size were informative, variations in the
prediction accuracies were present at all calibration set sizes. Variations in accuracy were relatively
higher for LA and LOV than for LL and LW (Fig. 4). This finding may be attributed to the
higher environmental variation for LA and LOV (Fig. 1). The GBLUP revealed significant and
strong prediction accuracies for LL in different calibration set sizes (Fig. 4C). This trend
suggested that new prediction models are needed to improve the prediction accuracy of LL. For
LA, LW, and LOV, the prediction accuracies will largely depend on the calibration subset size.
When all RILs were included in the calibration subset, variations in LA, LL, LW, and LOV
explained by the IBS genomic relationship matrix were greater than 85% (Fig. 4). This finding
revealed that substantial variations in the BLUP line means for LA, LL, LW, and LOV were
predictable when using the GBLUP model. In conclusion, predicting the genetic value for leaf
architecture can be an effective strategy for selecting lines whose phenotypes have yet to be
observed in maize breeding.
The authors are grateful to Bo Peng, Qingchao Li and Zhaozhao Yang for collecting the
phenotype data. We thank Dr. Huihui Li for calculating the broad-sense heritability. We also thank
Dr. Rongling Wu and two anonymous reviewers for valuable suggestions and
Conceived and designed the experiments: TW Yu Li. Performed the experiments: CL
Yongxiang Li Yunsu Shi Yanchun Song DZ ESB ZZ. Analyzed the data: CL. Wrote the paper: CL.
1. Duvick DN . The contribution of breeding to yield advances in maize . In: Duvick DN, editor. San Diego: Academic Press ; 2005 . pp. 83 - 145 .
2. Li Y , Ma XL , Wang TY , Li YX , Liu C , Liu Z , et al. Increasing maize productivity in China by planting hybrids with germplasm that responds favorably to higher planting densities . Crop Sci . 2011 ; 51 : 2391 - 2400 .
3. Wang TY , Ma XL , Li Y , Bai DP , Liu C , Liu Z , et al. Changes in yield and yield components of singlecross maize hybrids released in China between 1964 and 2001 . Crop Sci . 2011 ; 51 : 512 - 525 .
4. Duvick DN . Genetic progress in yield of United States maize ( Zea mays L.). Maydica . 2005 ; 50 : 193 - 202 .
5. Stewart DW , Costa C , Dwyer LM , Smith DL , Hamilton RI , Ma BL. Canopy structure, light interception, and photosynthesis in maize . Agron J . 2003 ; 95 : 1465 - 1474 .
6. Vazin F , Hassanzadeh M , Madani A , Nassiri-Mahallati M , Nasri M. Modeling light interception and distribution in mixed canopy of common cocklebur (Xanthium stramarium) in competition with corn . Planta Daninha . 2010 ; 28 : 455 - 469 . doi: 10.1177/0734242X09335694 PMID: 19723829
7. Wang Y , Li J. Molecular basis of plant architecture . Annu Rev Plant Biol . 2008 ; 59 : 253 - 279 . doi: 10. 1146/annurev. arplant.59.032607.092902 PMID: 18444901
8. Pepper GE , Pearce RB , Mock JJ . Leaf orientation and yield of maize . Crop Sci . 1977 ; 17 : 883 - 886 .
9. Mickelson SM , Stuber CS , Senior L , Kaeppler SM . Quantitative trait loci controlling leaf and tassel traits in a B73 x Mo17 population of maize . Crop Sci . 2002 ; 42 : 1902 - 1909 .
10. Pelleschi S , Leonardi A , Rocher JP , Cornic G , de Vienne D , Thevenot C , et al. Analysis of the relationships between growth, photosynthesis and carbohydrate metabolism using quantitative trait loci (QTLs) in young maize plants subjected to water deprivation . Mol Breed . 2006 ; 17 : 21 - 39 .
11. Yu Y , Zhang J , Shi Y , Song Y , Wang T , Li Y. QTL analysis for plant height and leaf angle by using different populations of maize . J Maize Sci . 2006 ; 14 : 88 - 92 .
12. Lu M , Zhou F , Xie C , Li M , Xu Y , Marilyn W , et al. Construction of a SSR linkage map and mapping of quantitative trait loci (QTL) for leaf angle and leaf orientation with an elite maize hybrid . Yi Chuan . 2007 ; 29 : 1131 - 1138 . PMID: 17855265
13. Ku L , Zhao W , Zhang J , Wu L , Wang C , Wang P , et al. Quantitative trait loci mapping of leaf angle and leaf orientation value in maize ( Zea mays L.). Theor Appl Genet . 2010 ; 121 : 951 - 959 . doi: 10.1007/ s00122- 010 - 1364 -z PMID : 20526576
14. Tian F , Bradbury PJ , Brown PJ , Hung H , Sun Q , Flint-Garcia S , et al. Genome-wide association study of leaf architecture in the maize nested association mapping population . Nat Genet . 2011 ; 43 : 159 - 162 . doi: 10.1038/ng.746 PMID: 21217756
15. Jiao Y , Zhao H , Ren L , Song W , Zeng B , Guo J , et al. Genome-wide genetic changes during modern breeding of maize . Nat Genet . 2012 ; 44 : 812 - 815 . doi: 10.1038/ng.2312 PMID: 22660547
16. Huang X , Feng Q , Qian Q , Zhao Q , Wang L , Wang A , et al. High-throughput genotyping by whole-genome resequencing . Genome Res . 2009 ; 19 : 1068 - 1076 . doi: 10.1101/gr. 089516.108 PMID: 19420380
17. Yu H , Xie W , Wang J , Xing Y , Xu C , Li X , et al. Gains in QTL detection using an ultra-high density SNP map based on population sequencing relative to traditional RFLP/SSR markers . PLoS One . 2011 ; 6 : 617595 .
18. Wang L , Wang A , Huang X , Zhao Q , Dong G , Qian Q , et al. Mapping 49 quantitative trait loci at high resolution through sequencing-based genotyping of rice recombinant inbred lines . Theor Appl Genet . 2011 ; 122 : 327 - 340 . doi: 10.1007/s00122- 010 - 1449 - 8 PMID: 20878143
19. Chen Z , Wang B , Dong X , Liu H , Ren L , Chen J , et al. An ultra-high density bin-map for rapid QTL mapping for tassel and ear architecture in a large F2 maize population . BMC Genomics . 2014 ; 15 : 433. doi: 10.1186/ 1471 - 2164 - 15 -433 PMID: 24898122
20. Technow F , Schrag TA , Schipprack W , Bauer E , Simianer H , Melchinger AE . Genome Properties and Prospects of Genomic Prediction of Hybrid Performance in a Breeding Program of Maize . Genetics. 2014 ; 197 : 1343 - 1355 . doi: 10.1534/genetics. 114.165860 PMID: 24850820
21. Daetwyler HD , Bansal UK , Bariana HS , Hayden MJ , Hayes BJ . Genomic prediction for rust resistance in diverse wheat landraces . Theor Appl Genet . 2014 ; 127 : 1795 - 1803 . doi: 10.1007/s00122- 014 - 2341 - 8 PMID: 24965887
22. Li Y , Wang TY . Germplasm base of maize breeding in China and formation of foundation parents . J Maize Sci . 2010 ; 18 : 1 - 8 .
23. Li S. The development and application of maize inbred line "Huangzaosi . Beijing Agric Sci. 1997 ; 15 : 19 - 21 (in Chinese).
24. Wang J , Li H , Zhang L , Li C , Meng L. QTL iciMapping software . Isbreeding , Beijing, China. 2012 . Available: http://www.isbreeding.net.
25. R Development Core Team . R: a language and environment for statistical computing . 2010 .
26. Elshire RJ , Glaubitz JC , Sun Q , Poland JA , Kawamoto K , Buckler ES , et al. A robust, simple genotyping-by-sequencing (GBS) approach for high diversity species . PLoS One . 2011 ; 6 : e19379. doi: 10. 1371/journal. pone.0019379 PMID: 21573248
27. Glaubitz JC , Casstevens TM , Lu F , Harriman J , Elshire RJ , Sun Q , et al. TASSEL- GBS: A High Capacity Genotyping by Sequencing Analysis Pipeline . PLoS One . 2014 ; 9 : e90346. doi: 10.1371/journal. pone.0090346 PMID: 24587335
28. Buckler ES , Holland JB , Bradbury PJ , Acharya CB , Brown PJ , Browne C , et al. The genetic architecture of maize flowering time . Science . 2009 ; 325 : 714 - 718 . doi: 10.1126/science.1174276 PMID: 19661422
29. Li H , Bradbury PJ , Ersoz E , Buckler ES , Wang J. Joint QTL linkage mapping for multiple-cross mating design sharing one common parent . PLoS One . 2011 ; 6 : e17573. doi: 10.1371/journal. pone.0017573 PMID: 21423655
30. Endelman JB . Ridge Regression and Other Kernels for Genomic Selection with R Package rrBLUP . Plant Genome . 2011 ; 4 : 250 - 255 .
31. Van Raden PM . Efficient methods to compute genomic predictions . J Dairy Sci . 2008 ; 91 : 4414 - 4423 . doi: 10.3168/jds. 2007-0980 PMID: 18946147
32. Peiffer JA , Flint-Garcia SA , De Leon N , McMullen MD , Kaeppler SM , Buckler ES . The genetic architecture of maize stalk strength . PLoS One . 2013 ; 8 : e67066. doi: 10.1371/journal. pone.0067066 PMID: 23840585
33. Moreno MA , Harper LC , Krueger RW , Dellaporta SL , Freeling M. liguleless1 encodes a nuclear-localized protein required for induction of ligules and auricles during maize leaf organogenesis . Genes Dev . 1997 ; 11 : 616 - 628 . PMID: 9119226
34. Bolduc N , Hake S. The maize transcription factor KNOTTED1 directly regulates the gibberellin catabolism gene ga2ox1 . Plant Cell . 2009 ; 21 : 1647 - 1658 . doi: 10.1105/tpc. 109.068221 PMID: 19567707
35. Kerstetter R , Vollbrecht E , Lowe B , Veit B , Yamaguchi J , Hake S. Sequence analysis and expression patterns divide the maize knotted1-like homeobox genes into two classes . Plant Cell . 1994 ; 6 : 1877 - 1887 . PMID: 7866030
36. McConnell JR , Emery J , Eshed Y , Bao N , Bowman J , Barton MK . Role of PHABULOSA and PHAVOLUTA in determining radial patterning in shoots . Nature . 2001 ; 411 : 709 - 713 . PMID: 11395776
37. Zhang J , Ku L , Han Z , Guo S , Liu H , Zhang Z , et al. The ZmCLA4 gene in the qLA4-1 QTL controls leaf angle in maize ( Zea mays L.). J Exp Bot . 2014 ; 65 : 5063 - 5076 . doi: 10.1093/jxb/eru271 PMID: 24987012
38. Juarez MT , Twigg RW , Timmermans MC . Specification of adaxial cell fate during maize leaf development . Development . 2004 ; 131 : 4533 - 4544 . PMID: 15342478
39. Ku L , Zhang J , Guo S , Liu H , Zhao R , Chen Y. Integrated multiple population analysis of leaf architecture traits in maize ( Zea mays L.). J Exp Bot . 2012 ; 63 : 261 - 274 . doi: 10.1093/jxb/err277 PMID: 21984652
40. Li YL , Li XH , Li JZ , Fu JF , Wang YZ , Wei MG . Dent corn genetic background influences QTL detection for grain yield and yield components in high-oil maize . Euphytica . 2009 ; 169 : 273 - 284 .
41. Blanc G , Charcosset A , Mangin B , Gallais A , Moreau L. Connected populations for detecting quantitative trait loci and testing for epistasis: an application in maize . Theor Appl Genet . 2006 ; 113 : 206 - 224 . PMID: 16791688
42. Verhoeven KJF , Jannink JL , McIntyre LM . Using mating designs to uncover QTL and the genetic architecture of complex traits . Heredity . 2006 ; 96 : 139 - 149 . PMID: 16304603
43. Yu J , Holland JB , McMullen MD , Buckler ES . Genetic design and statistical power of nested association mapping in maize . Genetics . 2008 ; 178 : 539 - 551 . doi: 10.1534/genetics. 107.074245 PMID: 18202393
44. Crossa J , de los Campos G , Perez P , Gianola D , Burgueno J , Araus JL , et al. Prediction of genetic values of quantitative traits in plant breeding using pedigree and molecular markers . Genetics . 2010 ; 186 : 713 - 724 . doi: 10.1534/genetics. 110.118521 PMID: 20813882