Content uploaded by Zuxin Zhang
Author content
All content in this area was uploaded by Zuxin Zhang on Mar 29, 2016
Content may be subject to copyright.
1 3
Theor Appl Genet (2014) 127:1019–1037
DOI 10.1007/s00122-014-2276-0
ORIGINAL PAPER
Genetic analysis and major QTL detection for maize kernel size
and weight in multi‑environments
Ying Liu · Liwei Wang · Chuanlong Sun · Zuxin Zhang ·
Yonglian Zheng · Fazhan Qiu
Received: 15 October 2013 / Accepted: 26 January 2014 / Published online: 20 February 2014
© Springer-Verlag Berlin Heidelberg 2014
of significant QTL play an important role in controlling
the development of maize kernel. These putative QTL may
have great promising for further fine-mapping with more
markers, and genetic improvement of maize kernel size and
weight through marker-assisted breeding.
Abbreviations
KL 20-Kernel length
KW 20-Kernel width
KT 20-Kernel thickness
HKW 100-Kernel weight
CIM Composite interval mapping
MCIM Mixed linear model-based composite interval
mapping
QTL Quantitative trait loci
SSR Single sequence repeat
MAS Marker-assisted selection
QEI QTL × environment interaction
Introduction
Maize (Zea mays L.) is one of the most important cereal
crops—it is widely consumed and plays a crucial role in
sustaining food security. In addition, forage production
and industrial energy require maize as a raw material. The
wide range of demand makes grain yield a major target of
maize breeding. Grain yield is a quantitative feature with
a complex genetic basis and various regulatory quantita-
tive trait loci (QTL)/genes affected by environmental fac-
tors (Austin and Lee 1996; Beavis et al. 1994; Messmer
et al. 2009). Compared with grain yield, yield components
have higher heritability and better stability across environ-
ments (Messmer et al. 2009; Peng et al. 2011). Dissecting
a complex quantitative trait into several related components
Abstract
Key Message Twelve major QTL in five optimal clus‑
ters and several epistatic QTL are identified for maize
kernel size and weight, some with pleiotropic will be
promising for fine‑mapping and yield improvement.
Abstract Kernel size and weight are important target traits
in maize (Zea mays L.) breeding programs. Here, we report
a set of quantitative trait loci (QTL) scattered through the
genome and significantly controlled the performance of
four kernel traits including length, width, thickness and
weight. From the cross V671 (large kernel) × Mc (small
kernel), 270 derived F2:3 families were used to identify
QTL of maize kernel-size traits and kernel weight in five
environments, using composite interval mapping (CIM)
for single-environment analysis along with mixed linear
model-based CIM for joint analysis. These two mapping
strategies identified 55 and 28 QTL, respectively. Among
them, 6 of 23 coincident were detected as interacting with
environment. Single-environment analysis showed that
8 genetic regions on chromosomes 1, 2, 4, 5 and 9 clus-
tered more than 60 % of the identified QTL. Twelve stable
major QTLs accounting for over 10 % of phenotypic vari-
ation were included in five optimal clusters on the genetic
region of bins 1.02–1.03, 1.04–1.06, 2.05–2.07, 4.07–4.08
and 9.03–9.04; the addition and partial dominance effects
Communicated by Natalia de Leon.
Electronic supplementary material The online version of this
article (doi:10.1007/s00122-014-2276-0) contains supplementary
material, which is available to authorized users.
Y. Liu · L. Wang · C. Sun · Z. Zhang · Y. Zheng · F. Qiu (*)
National Key Laboratory of Crop Genetic Improvement,
Huazhong Agricultural University, Wuhan 430070, China
e-mail: qiufazhan@gmail.com
1020 Theor Appl Genet (2014) 127:1019–1037
1 3
will be aided and so increase their genetic effect by identi-
fying more QTL associated with such causal traits (Yang
et al. 2012) as yield components, physiological processes
during grain filling and kernel internal components (Aus-
tin and Lee 1996, 1998; Goldman et al. 1993; Gupta et al.
2006; Li et al. 2007, 2009, 2012, 2013; Liu et al. 2011; Lu
et al. 2011; Messmer et al. 2009; Veldboom and Lee 1996;
Wassom et al. 2008). During domestication, kernel size and
weight are recognized as important yield components for
improving grain yield (Doebley et al. 2006). Kernel size,
referring to the space bounded by the husks and measured
by kernel length, width and thickness, serves as a compo-
nent of grain yield that determines kernel weight (Borrás
and Otegui 2001; Li et al. 2009; Xing and Zhang 2010).
Grain yield has been demonstrated to significantly correlate
with kernel size, especially kernel length (Li et al. 2009,
2013). Meanwhile kernel size is also a positive factor influ-
encing the end-use quality of maize (Gupta et al. 2006),
grain filling (Liu et al. 2011) as well as seedling vigor in
early growing maize in cool humid regions (Revilla et al.
1999). Therefore, improving kernel size and weight is
a prime breeding target to facilitate the improvement of
maize yield.
Great progress has been made in identifying major QTL
and isolating underlying genes for kernel size and weight in
grain crops, such as rice (Ishimaru 2003; Li et al. 2011; Qiu
et al. 2012; Song et al. 2007; Wan et al. 2006, 2008), soy-
bean (Han et al. 2012; Xu et al. 2011), wheat (Breseghe-
llo and Sorrells 2007; Ramya et al. 2010; Sun et al. 2009)
and barley (Ayoub et al. 2002; Backes et al. 1995). Espe-
cially for rice, several genes, GS3 (Fan et al. 2006), qGL3
(Zhang et al. 2012) and GW2 (Song et al. 2007), GS5 (Li
et al. 2011), GW8 (Wang et al. 2012b) and qSW5/GW5
(Shomura et al. 2008; Wan et al. 2008), which are associ-
ated with seed size and grain yield have been identified and
cloned through map-based cloning. Results of previous
studies revealed that grain yield is significantly determined
by kernel-size traits.
Compared with related research in rice, the molecular
cloning of genes associated with kernel size and weight has
lagged behind in maize (Arumuganathan and Earle 1991).
Kernel size and weight, as important agronomic traits and
yield components, can be used to facilitate maize yield and
have been increasingly attractive in molecular genetics in
recent years (Austin and Lee 1996; Gupta et al. 2006; Li
et al. 2009, 2012; Peng et al. 2011; Ribaut et al. 1997).
Mutant analysis was used to demonstrate the first gene
gln1-4 (glutamine synthetase) known to influence maize
kernel size (Martin et al. 2006); in addition, ZmGS3 and
ZmGW2 in maize, consistent with previous relevant QTL
analyses, were identified to be highly homologous with
rice GS3 and GW2 through an orthologous cloning method
(Li et al. 2010a, b). However, the effects on kernel size of
these two maize genes were not as remarkable as that of
their orthologs in rice (Li et al. 2010a, b). Therefore, more
attention should be paid to ‘mining’ favorable QTL/genes
to enhance the understanding of the genetic basis of maize
kernel-related traits, and applying them to marker-assisted
selection (MAS).
The lack of consistent QTL across environments is usu-
ally the major impediment to applying the achievements
generated from a handful of studies on QTL mapping and
genetic analysis for maize yield, particularly for kernel-
related traits. Recently, Peng et al. (2011) reported that
QTL for kernel-related traits were clearly more stable than
that for grain yield across diverse environments, indicating
that more efficient selection would be performed if robust
QTL for kernel-related traits were fine mapped.
In the present study, an F2:3 segregating population
derived from Mc × V671 was used to (a) identify QTL
for kernel size and weight in multiple agri-ecological envi-
ronments; (b) detect the QTL × environment interactions
(QEIs) to find crucial stable QTL and characterize the
epistatic QTL for kernel-related traits; and (c) investigate
the genetic basis and correlation between kernel size and
weight. This study aims to improve the understanding of
the intricate genetic basis of kernel size and weight and to
contribute favorable kernel-related QTL for fine-mapping
to aid yield improvement in maize breeding.
Materials and methods
Plant materials
An F2 population derived from a cross between two maize
elite inbred lines, Mc and V671, which have significantly
different kernel size and were created for QTL analysis.
Mc has small kernels while V671 has much larger kernels
(Fig. 1). The F2 population was planted in Hainan, China
during the winter of 2010; and 270 F2 plants were success-
fully self-pollinated. The seeds of the 270 F2:3 families with
less missing phenotypic data according to the subsequent
phenotypic analysis were harvested from the 270 F2 selfed-
plants, respectively, and used for validating the phenotype
in multi-environments.
Field trials
The trials were performed at three experimental stations
located in Wuhan (WH), Huanggang (HG) and Enshi (ES),
during 2011 and 2012, respectively. Each location and year
combination was considered as an experimental environ-
ment. Abbreviations were used to identify the different
environments, i.e. WH11, HG11, ES11, HG12 and ES12
indicated environments of Wuhan in 2011, Huanggang
1021Theor Appl Genet (2014) 127:1019–1037
1 3
in 2011, Enshi in 2011, Huanggang in 2012 and Enshi in
2012, respectively.
All trials were laid out as randomized complete block
designs with two replications, except for ES11 which was
not replicated. Each plot consisted of two rows with spaced
0.30 m apart on a raised bed, 3 m in length, 0.50 m in width
and with spacing between plots of 0.25 m. Each genotype
was grown in a single-row. Twelve open-pollinated individ-
ual plants were harvested, and all shelled from the middle
part of ears at maturity, for each genotype in the five trial
circumstances, respectively. Then kernels were bulked for
each genotype and used to measure the kernel-related traits.
Kernel-related traits were measured for each genotype as
follows:
100-kernel weight (HKW, g) was the average weight
of three repeated measurements of 100 kernels randomly
sampled from the bulked kernels and weighed by electronic
balance; Kernel length (KL, mm), width (KW, mm) and
thickness (KT, mm) were estimated by the average of three
replicated measurements of 20 kernels randomly chosen
from the bulked kernels using electronic digital calipers.
Phenotypic data analysis
The phenotype performance of kernel-related traits in sin-
gle environment was determined by the average of each
family from two replications. SPSS17.0 software (http://
www.spss.com) was used to calculate the variance com-
ponents including genotype, environment, replication and
interaction between genotype and environment of each trait
by general linear model (GLM) program. Broad-sense her-
itability (H2) for each trait was estimated as described by
Hallauer and Miranda (1998).
H2=
σ2
g
σ2
g+σ2
ge/n+σ2
ε/rn
Here, σg
2 is the genetic variance, σ2
ge is the interaction of
genotype with environments, σε
2 is the residual error, while
n is the number of environments with replications, and r is
the number of replications per environment. σ2
g, σ2
ge and σε
2
were obtained from variance components of GLM analysis
by SPSS17.0 software as well as by regression analysis.
Phenotypic correlation coefficients (r) between ker-
nel-related traits in each environment were estimated by
SPSS17.0 software (http://www.spss.com). Coefficients of
genotypic correlations (rg) between two traits were con-
ducted with PLABSTAT software (Utz 1997).
Genotyping and the construction of genetic linkage map
Total genomic DNA was extracted and purified with modi-
fied CTAB method (Saghai-Maroof et al. 1984) from the
fresh leaf tissue of 270 individual F2 plants whose kernel-
related traits were estimated based on their F2:3 family prog-
eny test. In accordance with bin location among genomes, a
total of 1102 single sequence repeat (SSR) molecular mark-
ers chosen from the maize genome database (http://www.m
aizegdb.org/) were used to detect polymorphisms between
the two parental lines, using the protocol available at http://
www.maizegdb.org/documentation/maizemap/ssr_protocol,
with slight modification. The 270 F2 individuals were even-
tually genotyped by 256 distinct co-dominant SSR markers.
PCR products were separated on 6 % denaturing polyacryla-
mide gels with a 19:1 ratio of acrylamide:bisacrylamide and
then silver stained as described by Santos et al. (1993).
A molecular linkage map (Fig. 2) of total length
1,351.7 cM across maize genome with an average interval
between adjacent markers of 5.28 cM, was constructed by
Mapmaker/EXP V3.0 software (Lander et al. 1987; Lincoln
et al. 1992) with ‘error detection on’ at logarithm of odds
(LOD) threshold >3.72. The Kosambi mapping function
(Kosambi 1943) was used to calculate genetic distance.
The linear order of most markers in the linkage map was
Fig. 1 Kernel phenotypes of the two parental inbred lines used for
QTL mapping in this study. a 20-kernel length, b 20-kernel width,
c 20-kernel thickness. Scale bars 10 mm for a, b and c. Kernels in
upper lines belong to the large-kernel parent V671 and those in the
lower lines from the small-kernel parent Mc
1022 Theor Appl Genet (2014) 127:1019–1037
1 3
Fig. 2 Distribution of identified quantitative trait loci (QTL) for
kernel size and kernel weight on genetic linkage maps in this study.
The marks denoted peak positions of QTL. WH11, HG11 and ES11
represent Wuhan, Huanggang and Enshi in 2011, respectively; HG12
and ES12 represent Huanggang and Enshi in 2012, respectively. KL
(20-kernel length), KW (20-kernel width) and KT (20-kernel thick-
ness) are measured in the unit of millimeter (mm); and the unit of
HKW (100-kernel weight) is gram (g). J only represented the QTLs
that were detected through joint mapping only. Numbers on the left
side are the genetic distances between two flanking markers with the
unit of centiMorgan (cM). The eight important QTL clusters’ regions
overlapping with that derived from previous studies were designed as
orange color box on chromosome bars (color figure online)
1023Theor Appl Genet (2014) 127:1019–1037
1 3
in agreement with their order of physical positions (Sup-
plementary Fig. S2).
QTL mapping
QTL analysis was performed by composite interval map-
ping (CIM), presented in the Windows QTL Cartographer
software 2.5 (Wang et al. 2007), at 1 cM walk speed and
with 10 cM window size to determine whether the two
adjacent test-statistic peaks represented two QTLs. Model
6 in CIM was employed to identify QTL for each trait
in each environment, with the values greater than LOD
threshold considering 1,000 permutations (P = 0.05) to
determine whether the presence of a QTL at a certain chro-
mosomal region was significantly associated with target
trait, as suggested by Lander and Kruglyak (1995). QTLs
which were stably identified from different environments
for a target trait with clearly similar positions (overlapping
1-LOD confidence intervals provided by software) were
assumed to be the same. QTL, which could be identified
in multiple environments and explain more than 10 % of
phenotypic variation, was considered as major QTL. QTL
detected for different traits with overlapped confidence
intervals and common marker(s), or couples of overlapped
QTL with distance less than 2 cM was defined as a QTL
cluster in which at least one stable major/large effect (phe-
notypic variation explained >10 %) QTL was included.
The phenotypic data used for QTL analysis of each trait
was based on the means of two replications in a single
environment.
Analysis of joint QTL, binary epistatic interaction in a
single environment and QEIs based on the datasets of all
experimental environments was performed by mixed lin-
ear model-based composite interval mapping (MCIM)
with best linear unbiased predictors (BLUP) for random
effect prediction of QTLNetwork software version 2.0
(Yang et al. 2007). Window size, working speed and filtra-
tion window were set at 10, 2 and 10 cM, respectively. The
F-test using Henderson method III was employed to deter-
mine significance, and the critical F-value was estimated
by 1,000 permutation tests (Doerge and Churchill 1996).
QTL designations were defined adopting the nomencla-
ture of McCouch et al. (1997). The designation for a QTL
starts with ‘q’, followed by an abbreviation of the trait
name, then the number of the chromosome on which the
QTL was located, and finally, the serial number assigned
to the related trait of QTL on a specific chromosome. The
last number was omitted in QTL nomenclature under the
situation that there is only one QTL detected on the specific
chromosome for a trait. In addition, if the QTL was identi-
fied only by joint analysis among all environments but not
a single-environment QTL detection, then ‘J’ was placed
after the numbers representing the chromosome of the sig-
nificant QTL.
Results
Trait performance
The two parents, Mc and V671, showed highly significant
differences (P < 0.001) in all examined kernel-related traits
(Fig. 1; Table 1) with higher values generally for V671.
Among the F2:3 families, most traits were approximately
normally distributed, and there were wide variations in the
performance measurements at the three locations during
the 2 years (Table 1 and Supplementary Fig. S1)—notably,
the phenotypic values of all four traits exhibited obvious
bi-directional transgressive segregation in all environments,
indicating polygenic quantitative genetic control. Broad-
sense heritability (H2) of the four kernel-related traits ranged
from 0.881 (KL) to 0.944 (KW), suggesting that genetic fac-
tors played an important role in the formation of these traits.
The highly significant difference (P < 0.001) was found in
genotype and environments for all traits, and the interactions
G × E were significant for KW, KT and HKW. The vari-
ances of the replications for all traits were non-significant
(P < 0.05) except for KW (Table 2), which is the reason that
the mean of two replications in one location for each geno-
type was used for the subsequent QTL mapping.
The phenotypic and genotypic correlation coefficients
between kernel-related traits across environments revealed
highly significance in F2:3 families (Table 3). Only in HG11
and HG12 were there no significant phenotypic correlations
between KL and KW, with significant (P < 0.01) positive
phenotypic and genotypic correlations in the other three envi-
ronments and among all environments, respectively, suggest-
ing that differences between experimental locations affected
kernel development. It is noteworthy that a significant nega-
tive phenotypic and genotypic correlation only occurred
between KT and KL across all environments (P < 0.01). Out-
standing phenotypic and genotypic correlations were found
between KL and KW, KW and KT, and between HKW and
these kernel-size traits, implying the important role of ker-
nel size in determining HKW and potential for simultaneous
improvement. Interestingly, similar results were reported in
previous study using F2:3 families (Li et al. 2009; Peng et al.
2011). Simultaneously, regression analysis revealed that, the
largest contributor to HKW was KT, with KL and KW fol-
lowed (Supplementary Table S1). Overall, the significant cor-
relation of the majority of character pairs indicated closely
genetic association among kernel size and weight, and that
the population was deserved for further studies of QTL map-
ping for these kernel-related traits.
1024 Theor Appl Genet (2014) 127:1019–1037
1 3
QTL analysis
The results of QTL analysis for the four kernel-related
traits in F2:3 families are shown in Fig. 2 and the analyses
of putative QTL are summarized in Table 4 and Supple-
mentary Table S2. A total of fifty-five QTLs were identified
for four traits through single-environment QTL analysis
and spread over all ten chromosomes (Table 4; Fig. 2). The
phenotypic variation explained by individual QTL ranged
from 0.46 (qHKW7) to 20.56 % (qKW1-2). Over 49.09 %
of the identified QTL had positive additive effect, indicat-
ing that alleles from the large-kernel parent V671 contrib-
uted on increasing phenotype. Results concerning the QTL
detected in the study are presented below.
20-Kernel length
Six QTLs for KL were identified by single-environment
mapping and individually accounted for 1.18–12.92 % of
the phenotypic variation while they explained 8.84–40.03 %
together of KL variation in each environment (Table 4;
Fig. 2). The major QTL qKL9-1 accounted for 2.39–
11.98 % of the phenotypic variation with LOD value 2.77–
7.38. Another major QTL, qKL9-2, explained up to 12.92 %
of phenotypic variation with higher LOD of 3.89–8.27. The
positive additive effects of all QTL on chromosome 9 indi-
cated that their alleles were derived from large-kernel parent
V671. In contrast, the negative additive effect of the rest two
environment-specific QTL on chromosome 2 indicated that
alleles from small-kernel parent Mc at these loci were bene-
ficial for increasing KL. The four KL QTLs on chromosome
9 mainly characterized by A or PD effects, while the other
two on chromosome 2 showed OD and D effects.
20-Kernel width
KW was governed by 16 QTLs dispersed on chromo-
somes 1, 2, 3, 4, 5 and 9. Each explained 1.7–20.51 % of
Table 1 Phenotypic performance of the four maize kernel-related traits in F2:3 families under five environments
V671: parent inbred line with large kernel; Mc: parent inbred line with small kernel
KL (20-kernel length), KW (20-kernel width) and KT (20-kernel thickness) are measured in the unit of millimeter (mm); and the unit of HKW
(100-kernel weight) is gram (g)
Env., represents environment; WH11, HG11 and ES11 represent Wuhan, Huanggang and Enshi in 2011, respectively; HG12 and ES12 represent
Huanggang and Enshi in 2012, respectively
%TS, represents the transgressive segregation which refers to the percentage of F2:3 families with phenotype beyond the range of two parents
P value, results from the Shapiro–Wilk test for normalized detection
SD standard deviation
Trait Env. Mc V671 F2:3 families
Mean SD Min Max % TS Skew Kurt P value
KL WH11 186.2 205.1 193.8 12.3 160.6 224.8 43.6 −0.05 −0.08 0.76
HG11 189.9 201.4 214.6 11.6 187.5 249.9 87.4 0.27 0.02 0.18
ES11 205.1 220.6 237.9 12.9 195.9 281.4 91.7 −0.11 0.32 0.51
HG12 180.4 208.0 208.1 13.7 163 249.6 57.7 −0.42 0.67 0.01
ES12 200.3 217.8 224.1 13.1 182.8 270.8 72.3 0.07 0.83 0.23
KW WH11 150.5 169.0 154.0 8.0 134.5 180.4 36.5 0.14 −0.02 0.74
HG11 152.1 168.8 159.3 8.0 132.7 178.3 32.0 −0.07 −0.12 0.52
ES11 156.9 168.5 163.0 9.2 139.4 193.9 52.7 −0.02 0.15 0.53
HG12 156.6 173.8 165.4 8.5 133.1 189.5 31.3 −0.21 0.37 0.39
ES12 164.5 175.2 163.9 8.4 143.6 185.8 60.6 0.05 −0.36 0.37
KT WH11 83.2 104.9 98.8 9.4 77.1 123.5 26.3 0.18 −0.28 0.13
HG11 92.5 117.8 99.1 7.6 73.2 119.7 18.9 0.01 0.02 0.90
ES11 84.2 95.6 86.7 6.0 71.5 105 42.0 0.13 −0.01 0.79
HG12 90.5 103.7 92.4 7.2 75.5 118 46.9 0.29 0.26 0.27
ES12 83.9 86.7 85.9 5.4 72.5 100.9 78.8 0.14 −0.20 0.38
HKW WH11 22.5 26.3 22.5 2.1 17.4 29.4 56.0 0.14 0.03 0.52
HG11 22.1 26.6 24.6 2.3 19.3 31.7 32.6 0.12 −0.22 0.65
ES11 23.4 28.3 28.8 3.4 20.3 37.7 64.8 −0.04 −0.37 0.24
HG12 20.0 26.7 24.6 2.3 17.3 30.1 24.4 −0.41 0.02 0.02
ES12 22.9 28.4 25.1 3.7 12.9 33.6 43.8 −0.42 0.23 0.03
1025Theor Appl Genet (2014) 127:1019–1037
1 3
phenotypic variation with qKW1-2 contributing the highest
percentage in most environments. Among them, half QTLs
were repeatedly detected in multiple environments. Nota-
bly, qKW1-2 was the only KW QTL found in all five envi-
ronments with LOD varied among 6.86–12.38. Besides,
another three major QTLs (qKW2-2, qKW1-2 and qKW4-
2) and four environment-specific QTLs were identified in
correspondence with joint analysis. Of which, qKW1-2,
qKW2-3 and qKW5 were co-located with QTL for HKW,
qHKW1-4, qHKW2-3 and qHKW5 on chromosomes 1, 2
and 5, respectively. Out of the 16 QTLs associated with
KW, the positive additive effects of six QTLs on chromo-
somes 1, 4 and 9 indicated that their positive alleles (alleles
which increased the trait) were consistently contributed
by the large-kernel parent V671, while the positive alleles
of the other ten QTLs on chromosomes 2, 3, 5 and 6 were
contributed by small-kernel parent Mc. All of the QTL
associated with KW showed A or PD effects, except for
qKW3.
20-Kernel thickness
A total of 18 QTLs influencing the KT were identified on
chromosomes 1, 2, 4, 5, 8, 9 and 10 in the present study,
individually explaining 0.84–17.98 % of phenotypic vari-
ation and totally accounting for 39.23–63.6 % of KT vari-
ation in each environment. Among them, nine QTLs were
significant in multiple environments and the additional nine
were environment-specific. The V671 alleles had a posi-
tive effect on increasing KT for nine QTLs distributed on
chromosomes 1, 4, 5 and 8, including two location-specific
major QTL (qKT1-1 and qKT1-2), one major (QTL qKT1-
4) detected across all five environments with 5.07–17.98 %
of the phenotypic variation and another major QTL (qKT1-
3) explaining 3.09–14.93 % of the phenotypic variation in
four environments. Five of the 18 QTLs detected for KT
were located on the same map position with the QTL for
KW and HKW. Two QTLs, qKT9-1 and qKT9-2 on chro-
mosome 9 were co-located with QTL for KL and the major
QTL qKT1-4 on chromosome 1 was corresponding with
one major QTL for HKW. Thirteen QTLs for KT showed
A or PD effects and the rest QTLs were basically character-
ized by dominance effects (D or OD).
100-Kernel weight
Fifteen QTLs influencing the HKW were detected
(Table 4) with seven on chromosome 1, five on chromo-
some 2 and one each on chromosomes 4, 5, and 7. Six of
the 15 QTLs were identified across 2 environments and
nine were environment-specific QTL. All positive alleles
of QTL on chromosomes 1 and 4 were derived from large-
kernel parent V671. The phenotypic variation explained by
Table 2 Analysis of variance (ANOVA) for kernel-related traits of
F2:3 families in four environments
H2 the broad-sense heritability
*, ** and *** indicate significant level at P < 0.05, P < 0.01 and
P < 0.001, respectively
Trait Source of variation F H2
KL Environment (E) 476.675*** 0.881
Genotype (G) 2.971***
Replication 2.004
G × E 1.111
KW Environment (E) 267.217*** 0.944
Genotype (G) 6.921***
Replication 6.090*
G × E 1.145*
KT Environment (E) 393.669*** 0.920
Genotype (G) 4.880***
Replication 1.771
G × E 1.186**
HKW Environment (E) 88.687*** 0.884
Genotype (G) 3.231***
Replication 0.576
G × E 1.201**
Table 3 Phenotypic (r) and genotypic (rg) correlation coefficients
between kernel-related traits across five environments
rg, genotypic correlation coefficients of two kernel-related traits
among four environments with replications
Env. represents environments; WH11, HG11 and ES11 represent
Wuhan, Huanggang and Enshi in 2011, respectively; HG12 and ES12
represent Huanggang and Enshi in 2012, respectively
** and ns indicate significance at P < 0.01 and non-significant effect,
respectively
Trait Env. KL KW KT
KW WH11 0.23**
HG11 0.06ns
ES11 0.19**
HG12 0.09ns
ES12 0.13**
rg0.17*
KT WH11 −0.54** 0.16**
HG11 −0.39** 0.40**
ES11 −0.17** 0.56**
HG12 −0.47** 0.24**
ES12 −0.27** 0.54**
rg−0.43** 0.51**
HKW WH11 0.32** 0.63** 0.30**
HG11 0.26** 0.62** 0.44**
ES11 0.50** 0.66** 0.56**
HG12 0.32** 0.58** 0.21**
ES12 0.44** 0.45** 0.32**
rg0.29** 0.92** 0.57**
1026 Theor Appl Genet (2014) 127:1019–1037
1 3
Table 4 Putative QTL for maize kernel size and weight in F2:3 families through single-environment QTL mapping
Trait QTLaMarker interval Range (cM) Size
(Mb)bEnv. cPeak position
(cM)
BindAeDfGene
actiongPVEh
(%)
LOD
KL qKL2-1*bnlg1175-umc1285 75.5–79.7 9.60 ES11 78.71 2.04 2.09 2.86 OD 1.18 4.45
qKL2-2umc2023-umc1890 82.3–94.2 22.18 ES11 90.91 2.07 2.84 2.8 D 2.28 3.85
qKL9-1umc1893-umc1634 48.3–59.5 6.48 ES12 49.61 9.02 −5.17 3 PD 7.67 3.12
WH11 55.11 9.03 −2.75 −1.96 PD 2.4 2.77
ES11 56.11 9.03 −6.38 −0.57 A 11.99 7.38
HG12 56.11 9.03 −5.89 3.96 PD 8.84 3.35
qKL9-2*bnlg1209-umc1771 60.3–65.7 11.46 ES11 61.41 9.04 −6.36 −0.85 A 11.96 8.27
ES12 61.71 9.04 −5.39 1.71 PD 8.3 3.89
HG11 62.71 9.04 −6.09 2.91 PD 12.92 5.56
WH11 65.51 9.04 −4.99 0.54 A 7.99 4.45
qKL9-3umc1519-umc1231 65.7–68.1 6.36 ES11 67.11 9.05 −6.59 −0.71 A 12.62 8.03
qKL9-4umc1494-umc2346 70.1–85.6 10.61 WH11 74.81 9.05 −4.51 −0.68 A 6.44 3.78
KW qKW1-1umc2225-bnlg1007 58.5–63.6 12.55 ES11 62.01 1.02 −5.73 2.21 PD 16.06 9.62
qKW1-2*bnlg1007-bnlg439 63.62–77.88 16.67 ES12 67.61 1.02 −3.92 −0.24 A 9.55 7.93
HG11 69.51 1.03 −4.66 1.61 PD 15.63 9.00
ES11 69.51 1.03 −5.72 1.66 PD 17.75 11.31
HG12 69.51 1.03 −5.76 1.06 A 20.51 12.38
WH11 72.51 1.03 −4.45 2.35 PD 14.17 6.86
qKW2-1umc2245-umc1227 19.6–23.6 1.66 WH11 21.71 2.01 4.02 −1.87 PD 10.76 5.51
qKW2-2*bnlg1831-bnlg1138 79.7–82.3 20.13 ES11 81.01 2.05 4.17 0.43 A 8.88 11.38
ES12 81.02 2.05 4.55 −1.03 PD 12.83 10.60
qKW2-3umc2023-umc1890 83.8–91 22.18 ES12 84.91 2.07 4.25 −0.49 A 11.86 10.66
ES11 86.91 2.07 4.44 0.67 A 10.78 12.07
qKW2-4umc1946-umc2625 93.9–97 9.37 HG11 94.81 2.07 2.58 0.34 A 5.15 5.65
ES12 95.81 2.07 2.49 1.13 PD 4.36 6.58
ES11 96.81 2.07 3.13 1.65 PD 5.72 9.6
qKW2-5*bnlg1662-bnlg1316 102.7–111 6.51 HG11 107.71 2.08 3.04 0.49 A 7.38 7.77
qKW2-6bnlg1316-umc1464 112–115.3 1.46 HG12 112.02 2.08 2.83 0.76 PD 5.58 6.97
qKW2-7umc1526-umc1230 117.1–130.4 15.20 HG12 122.31 2.08 3.34 0.76 PD 7.77 7.84
HG11 123.31 2.08 3.4 −0.49 A 9.07 6.85
qKW3 umc1320-umc1273 87–102.15 0.23 HG12 88.91 3.08 2.13 2.03 D 2.63 3.5
HG11 98.21 3.08 1.56 1.89 OD 1.7 3.1
qKW4-1umc1667-umc2041 81.1–88.2 6.15 HG11 84.31 4.08 −3.69 2.48 PD 9.37 5.12
ES12 87.51 4.08 −3.92 0.41 A 9.36 7.45
WH11 87.51 4.08 −3.26 0.18 A 7.1 5.02
qKW4-2*umc1051-bnlg292b 91.4–99.8 22.39 HG11 93.41 4.08 −3.43 1.42 PD 8.04 4.52
1027Theor Appl Genet (2014) 127:1019–1037
1 3
Table 4 continued
Trait QTLaMarker interval Range (cM) Size
(Mb)bEnv. cPeak position
(cM)
BindAeDfGene
actiongPVEh
(%)
LOD
ES11 96.41 4.08 −4.73 0.12 A 11.85 9.91
WH11 95.39 4.08 −3.28 0.17 A 7.41 5
qKW5*umc2294-umc2161 62–68.2 47.10 ES11 67.01 5.03 2.6 0.05 A 3.91 3.62
qKW6*phi070-umc2165 106.68–115.94 2.18 ES12 110.81 6.07 2.18 0.29 A 3.19 2.92
qKW9-1*umc2084-bnlg1583 10.9–20.2 5.08 HG12 17.91 9.01 −2.15 −1.48 PD 2.74 3.71
qKW9-2umc2346-umc1714 94.3–114.1 HG11 108.01 9.07 −3.23 0.68 PD 7.26 4.14
KT qKT1-1*umc1568-umc1403 64.8–70.8 17.20 ES11 67.61 1.02 −3.15 0.29 A 12 9.59
ES12 67.61 1.02 −3.07 −0.78 PD 10.79 14.2
qKT1-2* umc1403-umc2171 76.1–79 18.53 ES11 77.91 1.03 −4.47 0.94 PD 11 5.95
ES12 77.91 1.03 −2.61 −0.37 A 6.02 6.72
qKT1-3*umc1144-bnlg1811 85.6–89.6 6.48 WH11 85.81 1.04 −3.13 −0.68 PD 5.47 4.48
ES11 85.81 1.04 −2.68 0.45 A 5.64 3.65
ES12 87.81 1.04 −1.68 −0.58 PD 3.09 4.53
HG11 88.81 1.04 −4.13 0.14 A 14.93 10.01
qKT1-4*bnlg2086-umc1323 94.8–98.7 86.83 ES11 95.71 1.04 −2.27 0.17 A 5.07 3.73
HG12 96.01 1.05 −3.5 −0.2 A 11.36 7.81
ES12 96.01 1.05 −2.18 −0.39 A 5.85 6.45
WH11 97.01 1.05 −4.09 −1 PD 8.81 6.44
HG11 98.01 1.05 −4.7 −0.4 A 17.98 13.87
qKT1-5umc2234-umc1335 101.9–109 10.00 ES12 103.21 1.06 −1.6 −0.7 PD 3.23 5.23
qKT2 mmc0111-umc1422 42.2–61.3 5.57 ES12 49.41 2.02 2.07 −1.26 PD 6.66 5.03
qKT4-1*umc1791-umc2027 65–66.7 58.22 HG12 66.51 4.06 −1.71 −1.98 D 2.83 5.94
qKT4-2umc1899-umc2135 87.3–97.8 17.59 HG11 87.01 4.08 −2.15 −0.85 PD 3.42 4.9
ES11 90.21 4.08 −1.58 −0.66 PD 2.93 4.36
ES12 92.31 4.08 −1.32 −0.3 PD 2.61 3.92
qKT5-1umc1496-umc1761 32.3–56.4 12.58 WH11 42.91 5.01 1.98 −6.14 OD 2.14 4
qKT5-2umc1761-umc2294 55.2–65.7 19.56 HG11 62.21 5.03 −1.49 −1.26 D 1.87 3.58
qKT5-3*umc1784-umc1747 66.9–71.4 83.29 WH11 68.21 5.03 7.42 −3.24 PD 13.68 4.05
qKT8-1umc1075-umc1974 19–44.2 9.99 ES12 37.41 8.02 1.8 −0.42 PD 5.46 5.47
qKT8-2*umc2052-umc1638 111–120.1 3.18 HG11 119.11 8.08 −1.69 −1.11 PD 2.47 4.86
qKT9-1umc1893-umc2370 52.3–59.3 11.46 HG12 54.11 9.03 2.61 −0.03 A 6.12 3.94
ES12 54.11 9.03 2.45 −1.11 PD 9.68 6.78
ES11 56.11 9.03 2.45 −0.21 A 7.87 6.2
qKT9-2*bnlg1159-umc1494 61.4–69.4 25.13 ES12 61.71 9.04 2.03 −0.88 PD 6.82 5.09
ES11 67.41 9.05 2.55 −0.2 A 8.26 6.38
qKT10-1phi059-umc1863 30.1–34.8 6.96 ES11 30.71 10.02 1.02 1.02 D 1.42 4.47
1028 Theor Appl Genet (2014) 127:1019–1037
1 3
Table 4 continued
Trait QTLaMarker interval Range (cM) Size
(Mb)bEnv. cPeak position
(cM)
BindAeDfGene
actiongPVEh
(%)
LOD
ES12 30.71 10.02 0.7 1.29 OD 0.84 5.6
qKT10-2*umc1785-bnlg1655 38–43 60.69 ES12 40.31 10.03 0.82 0.78 D 0.98 3.68
HG12 41.71 10.03 2.28 0.67 PD 4.22 5.09
HG11 42.71 10.03 1.26 0.85 PD 1.12 3.2
qKT10-3umc1911-umc1930 47–52.6 12.47 HG12 47.31 10.04 2.25 1.04 PD 4 5.38
HKW qHKW1-1bnlg1014-umc1222 34.4–45.2 4.50 HG12 39.31 1.01 −0.93 0.28 PD 7.33 4.11
qHKW1-2umc1727-umc2224 41.33–52.6 3.61 ES12 48.21 1.01 −1.45 0.09 A 5.82 3.43
HG12 49.21 1.01 −0.87 0.21 PD 5.73 3.1
qHKW1-3umc2224-umc1568 56.4–58.5 3.64 ES12 58.31 1.02 −1.82 0.4 PD 7.41 3.41
qHKW1-4*bnlg1007-bnlg439 63.5–77.88 16.67 ES11 68.51 1.03 −1.79 0.55 PD 12.23 6.82
WH11 73.51 1.03 −0.97 0.66 PD 10.03 4.69
qHKW1-5*umc1144-bnlg2086 81.2–90.7 19.37 HG12 85.81 1.04 −1.16 0.69 PD 12.02 5.62
HG11 88.81 1.04 −1.13 0.15 A 12.2 6.68
qHKW1-6umc1323-umc2234 98.9–101.5 22.03 HG11 100.51 1.06 −1.24 0.11 A 12.8 7.52
qHKW1-7umc2151-umc1356 105.9–109 13.91 HG11 107.91 1.06 −1.08 0.36 PD 10.21 5.2
qHKW2-1umc1165-umc1265 19.5–25.1 1.37 HG11 21.81 2.02 0.51 0.2 PD 2.09 2.63
ES11 21.81 2.02 0.67 0.67 D 1.64 4.19
qHKW2-2bnlg1831-bnlg1138 80.3–83.3 20.13 WH11 81.01 2.05 0.79 −0.12 A 6.43 5.13
qHKW2-3*umc2023-umc1890 86.3–95.7 22.18 ES11 90.91 2.07 1.31 0.91 PD 6.72 10.57
ES12 90.91 2.07 1.42 1.2 D 6.83 8.98
qHKW2-4umc1049-bnlg1316 97.5–112.2 8.02 ES12 100.11 2.08 1.15 0.66 PD 4.84 6.98
HG11 107.71 2.08 0.4 0.7 OD 1.6 5.64
qHKW2-5umc1526-umc1230 115.3–123.5 15.20 HG11 120.31 2.08 0.4 0.47 D 1.55 3.2
qHKW4*umc1847-umc1667 78.7–87.03 7.48 ES11 83.11 4.07 −1 0.09 A 3.84 2.97
qHKW5*umc2060-umc1784 63.5–69.4 47.59 ES11 66.51 5.03 0.78 0.54 PD 2.43 3.88
qHKW7*bnlg2259-umc2197 93.3–117.3 7.40 ES12 102.31 7.05 0.04 1.45 OD 0.46 3.68
a The names followed by * were QTL simultaneously detected by joint QTL analysis among all environments
b Physical size of marker interval
c Environment. WH11, HG11 and ES11 represent Wuhan, Huanggang and Enshi in 2011, respectively; HG12 and ES12 represent Huanggang and Enshi in 2012, respectively
d The specific genetic region included the peak position of QTL in differential environments. Bins in maize are designated with the chromosome number followed by a two-digit decimal (e.g.,
1.00, 1.01, 1.02, etc.) (http://www.maizegdb.org/cgi-bin/bin_viewer.cgi)
e The additive effect of the QTL, with negative effect was contributed by V671 and positive effect was contributed by Mc; the unit of KL, KW and KT was millimeter (mm) and HKW was
measured in gram (g)
f The dominant effect of the QTL
g A, D, PD, and OD represent additive, dominance, partial dominance, over-dominance effect, respectively, based on Stuber et al. (1987)
h The percentage of phenotypic variation explained by corresponding QTL
1029Theor Appl Genet (2014) 127:1019–1037
1 3
QTL together in single environment ranged from 16.46 to
40.45 %. Two of the six QTLs identified across two envi-
ronments, qHKW1-4 and qHKW1-5, had major effect on
HKW with phenotypic variation explained >10 %. All
QTL for HKW co-located with QTL for kernel-size traits
except for qHKW7, indicating the close genetic correlation
between kernel size and kernel weight which may result
from pleiotropy or ‘multifactorial linkage’. Ten QTLs for
HKW were characterized by A or PD effects, whereas the
rest five QTLs with minor phenotypic variation explained
showed dominance effects (D or OD).
QEIs
There were 28 putative QTLs associated with 4 kernel-
related traits that were detected by joint analysis with
MCIM method; of these, 23 QTLs exhibited significant
coincidence with the results of single-environment map-
ping (Table 4 and Supplementary Table S2). For KL,
KW, KT and HKW, all significant QTL identified through
MCIM method could explain 4.32, 31.12, 31 and 36.61 %
of phenotype variance, respectively. Most of these QTLs
were detected with additive main effect, while six of them
were involved in significant QTL × environment interac-
tion (QEI) (Table 5) through joint analysis. Of which, two
QTLs for KT have significant dominant × environment
interactions (P < 0.05), while one of them, qKT1-3 has
additive × environment interaction (P < 0.05) simultane-
ously. This suggests that QTL for KT may be modified by
environmental conditions. Other four QTLs were detected
just with significant additive × environment interactions
(P < 0.05). The additive × environment interactions for
the target traits were responsible for 0.22–1.75 % of the
heritability, while dominant × environment interactions
contributed less, with 0.18–0.43 %. In addition, the three
environment-specific QTLs with QEI, qKW5, qKT5-3 and
qHKW5 were identified in overlapped genomic regions
on chromosome 5, indicating that the interactions of this
region with environments might involve in the influence on
kernel development.
Epistatic interaction
A total of five significant epistatic interaction loci
(P < 0.05) for KW and KT were identified with additive-
by-additive interaction, additive-by-dominance interac-
tion, dominance-by-additive interaction or dominance-
by-dominance interaction effects (Table 6). Four epistatic
interactions occurred within the genetic regions of signifi-
cant QTL, while the other one was distinct between signifi-
cant QTL and a non-significant locus on chromosome 8.
Although none of these epistatic interactions were consist-
ent in different environments, four referred loci, umc1403–
bnlg439, phi448880–umc1714, bnlg1662–umc2005 and
umc2135–bnlg292b, were repeatedly detected as involved
in two epistatic interactions, even for different traits like
umc1403–bnlg439 which was responsible for two major
QTL, qKW1-2 and qKT1-1, in different interaction pairs—
indicating that epistasis at this genomic region may partici-
pate in regulating maize kernel size. The phenotypic varia-
tion of target traits explained by epistatic interactions was
less than main effects of relevant QTL, indicating that main
effect of QTL may play an essential role in determining
maize kernel size and weight.
Clusters with co-located QTL for kernel-related traits
In the overview of the identified QTL in this study, 8 QTL
clusters comprising 34 QTLs were scattered on chromo-
somes 1, 2, 4, 5 and 9 (Tables 4, 7; Fig. 2). Half of the
clusters with co-located QTL for three traits spread on
chromosomes 1, 2, 4 and 5; while the other four clusters
on chromosomes 1, 2 and 9 encompassed QTL for two
kernel-related traits. Among the aforementioned clusters,
QTLs for HKW were always detected together with QTL
for kernel size apart from QTL cluster VIII where only
Table 5 QTL × environment interactions influencing kernel-related traits in different environments
a AE is the additive by designated environment interaction effect
b DE is the dominance by designated environment interaction effect
c H2(ae) is heritability of the additive by designated environment interaction effect
d H2(de) is heritability of the dominance by designated environment interaction effect
Trait QTL AE2aAE3aAE4aAE5aDE1bH2(ae)cH2(de)d
KL qKL2-1−2.46** 0.0035
KW qKW5 1.27* 0.0069
KT qKT1-30.96* −1.46* 0.0022 0.0018
KT qKT5-3−2.14*** 0.0043
HKW qHKW2-30.80*** −0.82*** 0.79*** 0.0175
HKW qHKW5 0.60** −0.62** 0.55** 0.0128
1030 Theor Appl Genet (2014) 127:1019–1037
1 3
some QTLs for KL and KT were co-located. The opposite
parents conferring positive alleles of co-located QTL for
KL and KT may account for the no significance for HKW
in QTL cluster VIII and the negative correlation between
them, while the positive alleles of co-located QTL in other
seven QTL clusters were consistently contributed by the
same parent. The series of QTL clusters in the present
study indicated that the underlying genetic correlation and
pleiotropic effect may influence maize kernel-related traits.
Furthermore, multiple kernel-related traits are likely to be
simultaneously improved.
Discussion
Complex genetic basis of QTL detected for kernel-related
traits
In this study, there was a wide variation of kernel-related
traits in the F2:3 families derived from two parents with con-
trasting kernel-related traits: Mc (small kernel) and V671
(large kernel). The close correlation among kernel-related
traits and the relatively high broad-sense heritability indi-
cated stable phenotypic and genetic association between
kernel size and kernel weight. In term of QTL mapping
results, the disparate numbers of significant QTL per trait
which ranged from 6 for KL to 18 for KT as well as an
asymmetric and clustered distribution among genomic
regions revealed the complex nature of the kernel-related
traits. Besides, the phenotypic variation explained by sta-
ble QTL also differed in magnitude among different envi-
ronments, as a result of interaction with environments (Xu
2010). At the same time, some environment-specific QTL
for target trait also accounted for relatively large propor-
tions of phenotypic variation. This complex phenomenon
may be due to context-dependent effects and regulation
of minor polygenes (Mackay et al. 2009). According to
Malosetti et al. (2008) and Messmer et al. (2009), the posi-
tions and stability of QTL as well as the direction and mag-
nitude of genetic effects on target traits were not exactly
predicted by QTL analysis in single environment. How-
ever, QTL that has obvious genetic effect could be prefer-
entially identified in varying environments. In this study,
twelve significant QTLs contributed over 10 % of pheno-
typic variation and were mapped in multiple environments;
ten of them were found by both single-environment QTL
analysis and joint analysis. For example, two robust QTLs
with best repeatability, qKW1-2 and qKT1-4, were detected
in all five environments. Moreover, more than half of all
detected loci were environment-specific QTL, indicating
that a complex genetic constitution with major and minor
effects controlled maize kernel-related traits. The complex
genetic basis of kernel-size traits and kernel weight was
Table 6 Digenetic epistatic QTL for maize kernel-related traits identified in F2:3 families under different environments
-/chr.8, a non-significant quantitative loci on chromosome 8, but involved in epistatic interaction
*, ** and *** indicate significant level at P < 0.05, P < 0.01 and P < 0.001, respectively
a AA, AD, DA and DD are the additive-by-additive, additive-by-dominant, dominant-by-additive and dominant-by-dominant epistatic interaction effects, respectively
b H2(aa), H2(ad), H2(da) and H2(dd) are the heritability of the additive-by-additive, additive-by-dominant, dominant-by-additive and dominant-by-dominant epistatic interaction effects, respec-
tively
Env Trait QTL_i interval_i QTL_j interval_j AAaADaDAaDDaH2(aa)bH2(ad)bH2(da)bH2(dd)b
HG11 KW qKW1-2umc1403-bnlg439 qKW9-2phi448880-umc1714 −2.97* 3.38* −4.71* 0.0106 0.0071 0.0106
KW qKW2-5bnlg1662-umc2005 qKW9-2phi448880-umc1714 3.48** −4.45* 0.001 0.0096
ES11 KW qKW2-5bnlg1662-umc2005 qKW4-2umc2135-bnlg292b −2.90* −4.30* 0.0034 0.0087
KW qKW4-2umc2135-bnlg292b -/chr.8 umc1069-umc1638 3.65** 0.0002
KT qKT1-1umc1403-bnlg439 qKT1-3umc2171-umc1144 17.04*** −19.98*** 17.63*** −17.96*** 0.0007 0.0013 0.005 0.0154
1031Theor Appl Genet (2014) 127:1019–1037
1 3
also reflected on the gene action of QTL: A, PD, D and
OD were all shown. Most QTL (42/55) expressed A or PD
effect across different environments, especially the major
QTL. These results demonstrated that additive effects and
partial dominance effects may play important roles in con-
trolling the development of maize kernel size and weight.
The large-kernel parent V671 contributed to increasing
effects for 27 QTL (49.09 %) including 10 of the 12 sig-
nificant major QTL mentioned above, meaning that V671
is a good donor for improving maize kernel-related traits.
Whereas, the small-kernel parent Mc contributed to the
other 28 QTLs (50.91 %), embracing the last two major
QTL and other QTL which were found in several environ-
ments explained >5 % of phenotypic variation. Obviously,
the QTL alleles from parent with low values on favored
traits also played an important role in increasing pheno-
typic value. Therefore, alteration in direction of parental
increasing alleles would be a critical component of QTL
assessment and valuable in illuminating the genetic basis
of kernel yield.
QTL clusters associated with multiple kernel-related traits
One of the central concepts in genetical genomics is
the existence of QTL clusters, in which widespread
downstream changes in expression of genes result from
a distant single polymorphism that located in the same
genomic regions (Schadt et al. 2003). Associative traits
are prone to share regions with significant QTL (Austin
and Lee 1996, 1998; Li et al. 2007). Domestication has
increased the size of maize kernels compared to its pro-
genitor teosinte. In rice, QTL of domestication-related
traits tends to form clusters that coincide with the regions
harboring favorable genes (Cai and Morishima 2002).
In an overview of QTL distribution in the present study
34 QTLs were clustered in 8 genetic regions on chromo-
somes 1, 2, 4, 5 and 9 (Tables 4 and 5) including all 12
robust major QTLs.
As demonstrated through multi-environmental trials in
previous studies (Li et al. 2013; Peng et al. 2011; Veldboom
and Lee 1996), we also found that KL was controlled by
several genetic loci on chromosome 9, which have also
been reported to significantly influence kernel yield (Li
et al. 2009, 2013). KL in our study was primarily regulated
by the major QTL qKL9-1 (bin 9.02–9.03) and qKL9-2 (bin
9.04) in cluster VIII, and co-localized with QTL qKT9-
1 (bin 9.03) and qKT9-2 (bin 9.03) for KT, respectively.
QTLs in this cluster were only for KL and KT with no
significant effect on HKW, suggesting a tension or trade-
off between the two kernel developmental dimensions that
would both benefit from assimilates. The genomic region
of QTL cluster VIII also harbored co-localized QTL for
yield and its component traits in other studies.Austin and
Table 7 Characterization of QTL clusters for four maize kernel-related traits identified in F2:3 families across multiple environments
Chr. represents chromosome
a V671: parent with large kernel; Mc: parent with small kernel; according to the additive effect of QTL in the same cluster, each parent responsible for increasing phenotype was listed as the
order of ‘trait included’
QTL cluster no. Chr. Trait included Marker interval Physical position (bp) Bin Positive allelesaRang of explained
phenotypic varia-
tion (%)
I 1 KW + KT + HKW bnlg1007-bnlg439 26648076–43864136 1.02–1.03 V671 + V671 + V671 6.02–20.51
II 1 KT + HKW umc1144-umc1335 64259586–196926685 1.04–1.06 V671 + V671 3.09–17.98
III 2 KW + HKW umc1165-bnlg1227 3945514–4688370 2.01–2.02 Mc + Mc 1.64–10.76
IV 2 KL + KW + HKW umc1884-umc1890 107384428–190855347 2.05–2.07 Mc + Mc + Mc 2.12–11.86
V 2 KW + HKW umc2005-umc1464 206227032–213768587 2.08 Mc + Mc 1.54–9.07
VI 4 KW + KT + HKW umc1194-umc2135 177331242–201133111 4.07–4.08 V671 + V671 + V671 2.36–11.86
VII 5 KW + KT + HKW umc2294-umc2161 33097126–74217977 5.03 Mc + Mc + Mc 2.4–13.68
VIII 9 KL + KT umc1634-umc1519 23407901–123304894 9.03–9.04 V671 + Mc 1.26–12.62
1032 Theor Appl Genet (2014) 127:1019–1037
1 3
Lee (1998) detected a QTL at bin 9.03, stably influencing
300-kernel weight in stress and non-stress environments. In
three different recombinant inbred lines (RILs) among six
environments Li et al. (2013) identified co-located QTL for
kernel length, width and yield on bin 9.03. In the genetic
region of bin 9.04, Peng et al. (2011) detected a stable
major QTL for grain yield at bin 9.04 in an F2:3 population
across multi-environments. This region was also frequently
reported to be involved in the QTL for kernel weight
in other studies (Austin and Lee 1998; Lu et al. 2006).
Therefore, bins 9.03 and 9.04 are noteworthy for genetic
improvement of maize kernel size and yield.
Another QTL for KL, qKL2-2 clustered with qHKW2-
2, qHKW2-3, qKW2-2, qKW2-3 and qKW2-4 in bin 2.05–
2.07 on chromosome 2 (cluster IV), which simultaneously
influenced HKW, KW and KL. The small-kernel parent Mc
contributed increasing alleles for all these significant loci.
Around this genetic region, Li et al. (2013) found a QTL
for KW on bin 2.07 clustered with three QTLs each for
KL, KT and grain yield, identified by several RILs among
multi-environments. QTL for kernel weight under water-
stressed conditions (Lu et al. 2006) and kernel volume
(Peng et al. 2011) were also identified on bin 2.07. These
QTLs in bin 2.05–2.06 seem to indicate a novel genomic
region for kernel-related traits.
QTLs for KW and HKW were co-located in another two
clusters (III and V) on chromosome 2 (bins 2.01–2.02 and
2.08). Both single-environment and joint QTL analyses
identified qKW2-2 and qHKW2-3 (bin 2.08) with moderate
phenotypic contributions. A QTL for kernel width on bin
2.08 was also revealed with moderate contribution (<10 %)
in specific environments (Peng et al. 2011), as were QTL
in cluster III. No QTL for KW on bin 2.01–2.02 has been
detected with any confidence in the past, but several yield
components, like cob diameter, restricting kernel devel-
opment, were reported previously (Austin and Lee 1996),
ear length (Lu et al. 2011), semi-diameter for cob and ear
(Li et al. 2009), seemed to be controlled by efficient loci
around this genomic region.
Clusters VI and VII, both with multiple QTL, simulta-
neously facilitated KT, KW and HKW—and were mapped
on chromosomes 4 (bin 4.07–4.08) and 5 (bin 5.03),
respectively. Major QTLs qKW4-1, qKT4-2 and qKW4-
2 co-located on bin 4.08 were stably detected, where Li
et al. (2013) also found clustered QTL for KW and KT
in several RIL populations. A series of previous studies
reported numerous QTL on bin 4.08 for kernel weight
(Lu et al. 2006; Veldboom and Lee 1996) and yield com-
ponents influencing kernel shape: such as kernel row
number (Austin and Lee 1996; Lu et al. 2011), semi-
diameter of ear (Li et al. 2009) and ear length (Lu et al.
2006). ZmGW2-CHR4 and ZmGW2-CHR5 are two maize
homologs of GW2, which controls grain width and weight
in rice (Li et al. 2010a). ZmGW2-CHR4 was demonstrated
significantly influencing kernel weight and located just
in the genetic region of qKW4-2. A QTL focus on grain
yield was identified on bin 5.03–5.04 with 256 F2:3 fami-
lies in five environments (Lima et al. 2006). In the present
study, all the three QTLs in cluster VII (bin 5.03) were
environment-specific with relative less stable influence on
kernel size and weight.
Two more notable QTL clusters, I (bin 1.02–1.03) and
II (bin 1.04–1.06) on chromosome 1, possessed several
significant QTL for kernel-related traits. QTL cluster I
consisted of a range of QTL affecting KW, KT and HKW,
while QTL in cluster II regulated KT and HKW. Two major
QTLs, qKW1-2 (bin 1.03) and qKT1-4 (bin 1.05), were sol-
idly identified in all environments with the corresponding
highest effects of up to 20.51 and 17.98 %, respectively,
strongly implying their determining effect on phenotype
and the presence of kernel trait-related genes. The majority
of other QTL in these two clusters, including major QTL
qKT1-4 (bin 1.04), was also detected as stably influencing
maize kernel development across different agro-ecological
environments. Veldboom and Lee (1996) published a suite
of QTL for grain yield, including such yield components
as kernel length, around the genetic regions of cluster I
in an F3 population. The importance of bin 1.04 on maize
kernel weight and other yield components was verified by
Austin and Lee (1996, 1998) in a RIL population derived
from Mo17 and H99. In recent years, QTLs on bins 1.03
and 1.04–1.06 with obvious contributions to grain yield
and kernel size and weight were discovered in F2:3 (Li et al.
2007; Peng et al. 2011; Ribaut et al. 1997), BC2F2 (Li et al.
2007) and RIL populations (Li et al. 2012; Messmer et al.
2009) under different experimental environments. Within
the abundant candidate genes, ZmGS3 as a putative GS3
ortholog located between qKT1-3 and qKT1-4 and involved
in maize kernel development was successfully cloned and
characterized by Li et al. (2010b). ZmGS3 with influence
on KW and HKW but not KT and with different functional
polymorphisms from rice GS3 imply that other relevant
genes regulating maize kernel size with distinctive mecha-
nisms must remain concealed. Collectively, the adjacent
genetic regions of bins 1.03 and 1.04–1.06 should have
great potential for improving maize kernel-related traits
and kernel yield.
The genetic regions aforementioned with clustered/
co-located robust QTL are worth of further investigation
due to the importance of genetic control of kernel devel-
opment as well as grain yield. Peng et al. (2011) reported
seven major QTLs on chromosomal regions responsible
for maize kernel-related traits and yield components in two
F2:3 populations. Although studies were conducted in the
same generation, none of the major QTL was coincident
with the results of the present study, possibly due to the
1033Theor Appl Genet (2014) 127:1019–1037
1 3
different genetic background. According to Li et al. (2013),
one of the seven important QTL clusters located on bin
4.08, for kernel-related traits and yield components in RIL
populations was in agreement with the result of this study.
Through QTL detection and meta-analysis in three popula-
tions (F2:3, BC2F2 and RIL), three main genetic regions of
bins 7.02–7.03, 1.03–1.04 and 10.05–10.06 were identified
for maize kernel development (Li et al. 2012). Most of the
QTL clusters identified in the present study would be novel
loci regulating kernel-related traits in maize, especially
the five optimal QTL clusters in the genetic regions of
bins 1.02–1.03, 1.04–1.06, 2.05–2.07, 4.08 and 9.03–9.04.
Remarkably, the physical distance of the most QTL clus-
ters were too large to isolate candidate genes, and it was
more expansive of the QTL clusters near or across chro-
mosomal centromere, which makes it necessary to conduct
fine-mapping of the major QTL in cluster regions with
enriched markers. However, these crucial clusters provide
more opportunities to identify important agriculturally ben-
eficial genes underlying these genetic regions. The linked
or co-located QTLs are inferred to benefit from the asso-
ciation of adaptive phenotypes during domestication and
will lead to a cumulative increase in kernel yield due to the
integrative positive effect (Marathi et al. 2012). Pleiotropic
regulator(s) or ‘multifactorial linkage’ of multiple traits
offers selective advantages and provides a rational expla-
nation for the numerous clustered QTL across the genome.
Stability of QTL for kernel-related traits
among environmental trials
The stability of significant genetic regions influencing ker-
nel-related traits was displayed well by comparing the joint
QTL mapping with single-environment QTL analysis, as
recommended by Messmer et al. (2009) and Malosetti et al.
(2008). Of the 23 consistent QTLs, twelve were stably iden-
tified with the same direction of increasing alleles among
multiple environments: one for KL (qKL9-2), three for
KW (qKW1-2, qKW2-2 and qKW4-2), five for KT (qKT1-
1, qKT1-3, qKT1-4, qKT9-2 and qKT9-2) and three for
HKW (qHKW1-4, qHKW1-5 and qHKW2-3). However, the
other consistent QTLs were basically significant loci with
relatively lower phenotypic contributions, suggesting that
kernel-related traits in maize appeared to be controlled by
some major QTL and a large number of minor-effect QTL
identified in specific environments or locations (Li et al.
2013). QEI may epitomize the existence of environment-
or location-specific QTL which lacked stability or main
effects among varied environments/locations (Cho et al.
2007; Hittalmani et al. 2003; Messmer et al. 2009). Mean-
while, QTL with major effect does not mean lack of QEI.
This is supported by the fact that QEI was also detected
for some repeatedly identified QTL (Hosseini et al. 2012),
resembling qKT1-3, a major KT QTL, as well as one of the
nine location-specific QTL, qHKW2-3; and by the fact that
QTL effects estimated in multiple environments could be
very different. For target traits, those QEIs were detected
accounting for much less phenotypic variation than that
explained by QTL through joint analysis, as a consequence,
they did not appreciably alter the main effects of QTL either
by magnitude or direction. Peng et al. (2011) reported that
QTLs for kernel-related traits were more consistent across
environments and genetic backcrosses than QTL of grain
yield influenced by QEI. Similar results were found in
eleven RILs derived from one common parent by Li et al.
(2013). Grain yield in maize might be regulated by a large
number of minor-effect QTLs that are sensitive to environ-
ment (Beavis et al. 1994). In addition, the majority of QTL
for grain yield presented instability and less co-localization
across diverse water regimes, locations, years or cropping
seasons in several studies (Austin and Lee 1998; Li et al.
2013; Lima et al. 2006; Lu et al. 2006; Messmer et al. 2009;
Ribaut et al. 1997). Yield component traits including ker-
nel size and weight displayed more advantages for genetic
improvement. Improving kernel size and weight is such an
efficient strategy for increasing grain yield, which has been
demonstrated by related research in rice.
As members of the grass family (Poaceae), maize and
rice share good synteny of genomes and most gene fami-
lies are the same (Schnable et al. 2009), and important
agronomic and domestication-related QTLs were revealed
in orthologous regions (Yan et al. 2004). Do the maize
orthologs of isolated rice grain genes coincide with the
stable major QTL in our study that also have an impor-
tant influence on maize kernel size and yield? There are
limited results available on this question. One of the two
orthologs of the SPL14 (Miura et al. 2010), which con-
trols panicle branching and grain production of rice, lays
in the prominent genetic region on chromosome 4 where
qKW4-2 and qKT4-2 were co-located, and both were sta-
bly expressed among multi-environments. Two co-located
QTLs for KL, qKL9-2 and qKT9-2, on chromosome 9 con-
tain one of the two orthologs of another rice yield-related
gene, APO1 (Terao et al. 2010). However, the influence
on maize kernel size and yield of the above two candi-
date orthologs requires further investigation. Although
the cloned homologs ZmGS3 and ZmGS2 were the only
two maize orthologs confirmed as involved in maize ker-
nel development, they were marginally associated with
maize kernel size and just adjacent to our major genetic
regions. Orthologs of cloned rice yield-related genes seem
to change or reduce their function in maize, indicating the
underlying of the major QTL in the present study, maize
kernel size and weight were controlled by distinct genetic
mechanisms or other members of the same gene families
found in rice grain yield and grain development.
1034 Theor Appl Genet (2014) 127:1019–1037
1 3
These stably identified QTL could contribute to effective
selection of genotypes with broad adaptation across diver-
sity agro-ecological conditions. Meanwhile, the abundant
environment-specific QTL, particularly the consistent ones
among two mapping strategies, would allow more oppor-
tunities to understand the importance and genetic basis
of QEI. As interaction with environment is the natural of
creatures, QEI significantly associating with plant perfor-
mance cannot be ignored during breeding for stability and
adaption, especially for resource-limited environments. To
increase crop productivity, there were two strategies deal-
ing with QEI in the breeding program. First, identifica-
tion of some stable robust QTL or QTL with minor QEI
should be highly desirable. Second, development of widely
adapted cultivars by pyramiding some stable QTL and/or
cultivars with specific adaption by pyramiding stable major
QTL and reliable environment-/location-specific QTL
together might be very useful for optimizing MAS of ker-
nel yield.
Epistatic interactions between QTL
For complex quantitative traits, the interaction effects
between loci/genes, or epistasis have been considered as
essential to understanding genetic regulation (Carlborg
and Haley 2004; Phillips 2008). The reduced genetic het-
erogeneity after crosses and the missing proportion of phe-
notypic variation explained by the identified QTL, com-
pared to the heritability of a certain trait, could be partly
due to epistasis (Doebley et al. 2006; Mackay et al. 2009;
Miedaner et al. 2011; Phillips 2008; Reif et al. 2011; Xu
and Jia 2007). Although the effect of epistasis may not be
significant in different crops and traits, these interactions
could be selected and retained to impact on the pheno-
type of target traits (Würschum 2012), and will obviously
influence the efficiency and accuracy of marker-assisted
breeding (Carlborg and Haley 2004; Steinhoff et al. 2012).
Particularly, epistatic interaction as an important regulator
for maize grain yield and its components has been often
detected between non-significant major loci (Ma et al.
2007; Peng et al. 2011). In the present study, five pairs of
epistatic interactions regulating KW and KT were detected,
mostly among significant QTL, indicating that these sig-
nificant loci could influence each other’s genetic back-
ground while controlling kernel size. Properly pyramid-
ing this epistasis QTL may improve kernel yield (Wang
et al. 2012a). Recently, Wang et al. (2012b) showed that
the genetic interaction between two rice seed-size genes,
OsSPL16 and GS3, led to improved grain yield and quality
through simultaneously targeting GS3 and OsSPL16 within
a marker-assisted strategy. Zhao et al. (2011) found that the
pyramiding effect could be obvious among QTL for the
rice grain length, with epistasis occurring only when the
direction of epistatic effects was the same as the additive
effect of target QTL/genes. In addition, there was an epi-
static effect even between closely linked QTL (Kroymann
and Mitchell-Olds 2005; Mackay et al. 2009). Among the
five pairs of epistatic interactions observed in the present
study, four significant QTLs, qKW9-2, qKW2-5, qKW4-2
and qKT1-1/qKW1-2 were involved in two pairs of inter-
actions, that is, they co-operated with two different loci
either for different traits or in different environments, sug-
gesting that these loci may be regarded as epistatic regula-
tors prone to taking part in multiple binary interactions—
similar results were previously observed (Reif et al. 2011;
Würschum 2012).
As a focal point for the unification of many traditionally
research areas that used to be disparate, epistasis of dispa-
rate loci is important for elucidating the functional nature
of complex genetic system and the long-term change of
biological evolution (Phillips 2008). To explore and clarify
the potential epistatic network and the genetic regulation
of complex quantitative traits, large population sizes, high-
throughput screens from DNA level to phenotype and rig-
orous analysis of gene interaction with good bioinformatic
tools will be required in the future (Carlborg and Haley
2004; Phillips 2008). These tools will enable the pursuit
of epistatic QTL and confirmation of the epistatic effect on
maize kernel size and yield in efficient pyramiding through
MAS.
Conclusion
In the present study, the estimates of phenotype and gen-
otype of F2:3 families in multiple agro-ecological circum-
stances reveal the significant association between kernel
size of KL, KW and KT and kernel weight. Numbers of
QTL dramatically influencing kernel size and weight with
additive and partial dominance effects were co-localized
in eight genomic regions (bins 1.03, 1.04–1.06, 2.01–2.02,
2.05–2.07, 2.08, 4.08, 5.03 and 9.03–9.04). The integrative
positive effects of the clustered QTL will lead to a cumu-
lative increasing kernel yield due to the improved kernel
size and kernel weight. Meanwhile, the pleiotropic effect of
enriched QTL and genetic epistasis between two genomic
regions may play an important role in mutual interactions
of these kernel-related traits. The information generated
in this study could well aid in understanding the genetic
basis of maize kernel-related traits and fine-mapping genes
underlying the robust major QTL in the optimal clusters
(bins 1.03, 1.04–1.06, 2.05–2.07, 4.08 and 9.03–9.04).
However, challenging research questions remains such
as how to accurately estimate the ‘real’ breeding value of
a significant QTL for kernel-related traits without genetic
background interference and how to best clone and transfer
1035Theor Appl Genet (2014) 127:1019–1037
1 3
the robust QTL to other populations. And due to the limited
resolution in the present study, there could be hundreds or
thousands of genes underlying the major QTL. Therefore,
further QTL validation will be proceeding with advanced
backcross population (QTL-NIL or chromosomal seg-
ment substitution lines). Besides, support from other omics
researches, like transcriptomics, should also be considered
to promote genetic analysis of kernel development and the
isolation of favorable alleles for molecular breeding of
high-yield maize based on this study.
Acknowledgments This research was supported by the National
Basic Research Program of China (973 Program) (2014CB138203)
and the National High Technology Research and Development Pro-
gram of China (863 Program) (2012AA101104), the National Nature
Science Foundation of China (No. 91335205) and the Fundamental
Research Funds for the Central University (No. 2013PY027).
Conflict of interest The authors declare that they have no conflict
of interest.
Ethical standards The experiments comply with the current laws
of the country in which they were performed.
References
Arumuganathan K, Earle E (1991) Nuclear DNA content of some
important plant species. Plant Mol Biol Rep 9:208–218
Austin D, Lee M (1996) Comparative mapping in F2:3 and F6:7 gen-
erations of quantitative trait loci for grain yield and yield compo-
nents in maize. Theor Appl Genet 92:817–826
Austin DF, Lee M (1998) Detection of quantitative trait loci for grain
yield and yield components in maize across generations in stress
and nonstress environments. Crop Sci 38:1296–1308
Ayoub M, Symons S, Edney M, Mather D (2002) QTLs affecting
kernel size and shape in a two-rowed by six-rowed barley cross.
Theor Appl Genet 105:237–247
Backes G, Graner A, Foroughi-Wehr B, Fischbeck G, Wenzel G,
Jahoor A (1995) Localization of quantitative trait loci (QTL) for
agronomic important characters by the use of a RFLP map in bar-
ley (Hordeum vulgare L.). Theor Appl Genet 90:294–302
Beavis W, Smith O, Grant D, Fincher R (1994) Identification of quan-
titative trait loci using a small sample of topcrossed and F4 prog-
eny from maize. Crop Sci 34:882–896
Borrás L, Otegui ME (2001) Maize kernel weight response to post-
flowering source–sink ratio. Crop Sci 41:1816–1822
Breseghello F, Sorrells ME (2007) QTL analysis of kernel size and
shape in two hexaploid wheat mapping populations. Field Crop
Res 101:172–179
Cai H, Morishima H (2002) QTL clusters reflect character associations
in wild and cultivated rice. Theor Appl Genet 104:1217–1228
Carlborg Ö, Haley CS (2004) Epistasis: too often neglected in com-
plex trait studies? Nat Rev Genet 5:618–625
Cho YG, Kang HJ, Lee JS, Lee YT, Lim SJ, Gauch H, Eun MY,
McCouch SR (2007) Identification of quantitative trait loci in rice
for yield, yield components, and agronomic traits across years
and locations. Crop Sci 47:2403–2417
Doebley JF, Gaut BS, Smith BD (2006) The molecular genetics of
crop domestication. Cell 127:1309–1321
Doerge RW, Churchill GA (1996) Permutation tests for multiple loci
affecting a quantitative character. Genetics 142:285–294
Fan C, Xing Y, Mao H, Lu T, Han B, Xu C, Li X, Zhang Q (2006)
GS3, a major QTL for grain length and weight and minor QTL
for grain width and thickness in rice, encodes a putative trans-
membrane protein. Theor Appl Genet 112:1164–1171
Goldman I, Rocheford T, Dudley J (1993) Quantitative trait
loci influencing protein and starch concentration in the Illi-
nois long term selection maize strains. Theor Appl Genet
87:217–224
Gupta PK, Rustgi S, Kumar N (2006) Genetic and molecular basis of
grain size and grain number and its relevance to grain productiv-
ity in higher plants. Genome 49:565–571
Hallauer AR, Miranda JB (1998) Quantitative genetics in maize
breeding, 2nd edn. Iowa State Univ Press, Ames
Han Y, Li D, Zhu D, Li H, Li X, Teng W, Li W (2012) QTL analy-
sis of soybean seed weight across multi-genetic backgrounds and
environments. Theor Appl Genet 125:671–683
Hittalmani S, Huang N, Courtois B, Venuprasad R, Shashidhar H,
Zhuang J, Zheng K, Liu G, Wang G, Sidhu J (2003) Identification
of QTL for growth-and grain yield-related traits in rice across
nine locations of Asia. Theor Appl Genet 107:679–690
Hosseini M, Houshmand S, Mohamadi S, Tarang A, Khodambashi M,
Rahimsoroush H (2012) Detection of QTLs with main, epistatic
and QTL × environment interaction effects for rice grain appear-
ance quality traits using two populations of backcross inbred
lines (BILs). Field Crop Res 135:97–106
Ishimaru K (2003) Identification of a locus increasing rice yield
and physiological analysis of its function. Plant Physiol 133:
1083–1090
Kosambi D (1943) The estimation of map distances from recombina-
tion values. Ann Eugen 12:172–175
Kroymann J, Mitchell-Olds T (2005) Epistasis and balanced polymor-
phism influencing complex trait variation. Nature 435:95–98
Lander E, Kruglyak L (1995) Genetic dissection of complex traits
guidelines for interpreting and reporting linkage results. Nat
Genet 11:241–247
Lander ES, Green P, Abrahamson J, Barlow A, Daly MJ, Lincoln SE,
Newburg L (1987) MAPMAKER: an interactive computer pack-
age for constructing primary genetic linkage maps of experimen-
tal and natural populations. Genomics 1:174–181
Li Y, Niu S, Dong Y, Cui D, Wang Y, Liu Y, Wei M (2007) Identifica-
tion of trait-improving quantitative trait loci for grain yield com-
ponents from a dent corn inbred line in an advanced backcross
BC2F2 population and comparison with its F2:3 population in pop-
corn. Theor Appl Genet 115:129–140
Li Y, Wang Y, Shi Y, Song Y, Wand T, Li Y (2009) Correlation analysis
and QTL mapping for traits of kernel structure and yield compo-
nents in maize. Sci Agric Sin 42:408–418
Li Q, Li L, Yang X, Warburton ML, Bai G, Dai J, Li J, Yan J (2010a)
Relationship, evolutionary fate and function of two maize co-
orthologs of rice GW2 associated with kernel size and weight.
BMC Plant Biol 10:143
Li Q, Yang X, Bai G, Warburton ML, Mahuku G, Gore M, Dai J, Li
J, Yan J (2010b) Cloning and characterization of a putative GS3
ortholog involved in maize kernel development. Theor Appl
Genet 120:753–763
Li Y, Fan C, Xing Y, Jiang Y, Luo L, Sun L, Shao D, Xu C, Li X, Xiao
J (2011) Natural variation in GS5 plays an important role in regu-
lating grain size and yield in rice. Nat Genet 43:1266–1269
Li Y, Yang M, Dong Y, Wang Q, Zhou Y, Zhou Q, Shen B, Zhang F,
Liang X (2012) Three main genetic regions for grain develop-
ment revealed through QTL detection and meta-analysis in
maize. Mol Breed 30:195–211
Li C, Li Y, Sun B, Peng B, Liu C, Liu Z, Yang Z, Li Q, Tan W, Zhang
Y (2013) Quantitative trait loci mapping for yield components
and kernel-related traits in multiple connected RIL populations in
maize. Euphytica 193:303–316
1036 Theor Appl Genet (2014) 127:1019–1037
1 3
Lima MDA, de Souza CL, Bento DAV, de Souza AP, Carlini-Garcia
LA (2006) Mapping QTL for grain yield and plant traits in a trop-
ical maize population. Mol Breed 17:227–239
Lincoln S, Daly M, Lander E (1992) Constructing genetic maps with
MAPMAKER/EXP 3.0. Whitehead Institute Technical Report
Liu Z, Ji H, Cui Z, Wu X, Duan L, Feng X, Tang J (2011) QTL
detected for grain-filling rate in maize using a RIL population.
Mol Breed 27:25–36
Lu GH, Tang JH, Yan JB, Ma XQ, Li JS, Chen SJ, Ma JC, Liu ZX,
Zhang YR, Dai JR (2006) Quantitative trait loci mapping of
maize yield and its components under different water treatments
at flowering time. J Integr Plant Biol 48:1233–1243
Lu M, Xie CX, Li XH, Hao ZF, Li MS, Weng JF, Zhang DG, Bai
L, Zhang SH (2011) Mapping of quantitative trait loci for ker-
nel row number in maize across seven environments. Mol Breed
28:143–152
Ma X, Tang J, Teng W, Yan J, Meng Y, Li J (2007) Epistatic interac-
tion is an important genetic basis of grain yield and its compo-
nents in maize. Mol Breed 20:41–51
Mackay TF, Stone EA, Ayroles JF (2009) The genetics of quantitative
traits: challenges and prospects. Nat Rev Genet 10:565–577
Malosetti M, Ribaut JM, Vargas M, Crossa J, van Eeuwijk FA (2008)
A multi-trait multi-environment QTL mixed model with an appli-
cation to drought and nitrogen stress trials in maize (Zea mays
L.). Euphytica 161:241–257
Marathi B, Guleria S, Mohapatra T, Parsad R, Mariappan N, Kurun-
gara VK, Atwal SS, Prabhu KV, Singh NK, Singh AK (2012)
QTL analysis of novel genomic regions associated with yield and
yield related traits in new plant type based recombinant inbred
lines of rice (Oryza sativa L.). BMC Plant Biol 12:137
Martin A, Lee J, Kichey T, Gerentes D, Zivy M, Tatout C, Dubois F,
Balliau T, Valot B, Davanture M (2006) Two cytosolic glutamine
synthetase isoforms of maize are specifically involved in the con-
trol of grain production. Plant Cell 18:3252–3274
McCouch SR, Cho YG, Yano M, Paul E, Blinstrub M, Morishima H,
Kinoshita T (1997) Report on QTL nomenclature. Rice Genet
Newlett 14:11–13
Messmer R, Fracheboud Y, Bänziger M, Vargas M, Stamp P, Ribaut
J-M (2009) Drought stress and tropical maize: QTL-by-envi-
ronment interactions and stability of QTLs across environments
for yield components and secondary traits. Theorl Appl Genet
119:913–930
Miedaner T, Würschum T, Maurer HP, Korzun V, Ebmeyer E, Reif JC
(2011) Association mapping for Fusarium head blight resistance
in European soft winter wheat. Mol Breed 28:647–655
Miura K, Ikeda M, Matsubara A, Song X-J, Ito M, Asano K, Mat-
suoka M, Kitano H, Ashikari M (2010) OsSPL14 promotes pan-
icle branching and higher grain productivity in rice. Nat Genet
42:545–549
Peng B, Li Y, Wang Y, Liu C, Liu Z, Tan W, Zhang Y, Wang D, Shi
Y, Sun B (2011) QTL analysis for yield components and kernel-
related traits in maize across multi-environments. Theor Appl
Genet 122:1305–1320
Phillips PC (2008) Epistasis—the essential role of gene interactions
in the structure and evolution of genetic systems. Nat Rev Genet
9:855–867
Qiu X, Gong R, Tan Y, Yu S (2012) Mapping and characterization of
the major quantitative trait locus qSS7 associated with increased
length and decreased width of rice seeds. Theor Appl Genet
125:1717–1726
Ramya P, Chaubal A, Kulkarni K, Gupta L, Kadoo N, Dhaliwal H,
Chhuneja P, Lagu M, Gupt V (2010) QTL mapping of 1000-ker-
nel weight, kernel length, and kernel width in bread wheat (Triti-
cum aestivum L.). J Appl Genet 51:421–429
Reif JC, Maurer HP, Korzun V, Ebmeyer E, Miedaner T, Würschum T
(2011) Mapping QTLs with main and epistatic effects underlying
grain yield and heading time in soft winter wheat. Theor Appl
Genet 123:283–292
Revilla P, Butrón A, Malvar R, Ordás R (1999) Relationship among ker-
nel weight, early vigor, and growth in maize. Crop Sci 39:654–658
Ribaut J-M, Jiang C, Gonzalez-de-Leon D, Edmeades G, Hoisington
D (1997) Identification of quantitative trait loci under drought
conditions in tropical maize. 2. Yield components and marker-
assisted selection strategies. Theor Appl Genet 94:887–896
Saghai-Maroof MA, Soliman KM, Jorgensen RA, Allard RW (1984)
Ribosomal DNA spacer-length polymorphisms in barley: mende-
lian inheritance, chromosomal location, and population dynam-
ics. Proc Natl Acad Sci 81:8014–8018
Santos FR, Pena SD, Epplen JT (1993) Genetic and population study
of a Y-linked tetranucleotide repeat DNA polymorphism with a
simple non-isotopic technique. Hum Genet 90:655–656
Schadt EE, Monks SA, Drake TA, Lusis AJ, Che N, Colinayo V, Ruff
TG, Milligan SB, Lamb JR, Cavet G (2003) Genetics of gene
expression surveyed in maize, mouse and man. Nature 422:297–302
Schnable PS, Ware D, Fulton RS, Stein JC, Wei F, Pasternak S, Liang
C, Zhang J, Fulton L, Graves TA (2009) The B73 maize genome:
complexity, diversity, and dynamics. Science 326:1112–1115
Shomura A, Izawa T, Ebana K, Ebitani T, Kanegae H, Konishi S, Yano
M (2008) Deletion in a gene associated with grain size increased
yields during rice domestication. Nat Genet 40:1023–1028
Song XJ, Huang W, Shi M, Zhu MZ, Lin HX (2007) A QTL for rice
grain width and weight encodes a previously unknown RING-
type E3 ubiquitin ligase. Nat Genet 39:623–630
Steinhoff J, Liu W, Reif JC, Della Porta G, Ranc N, Würschum T
(2012) Detection of QTL for flowering time in multiple families
of elite maize. Theor Appl Genet 125:1539–1551
Stuber CW, Edwards MD, Wendel JF (1987) Molecular marker-facil-
itated investigations of quantitative trait loci in maize. II. Factors
influencing yield and its component traits. Crop Sci 27:639–648
Sun XY, Wu K, Zhao Y, Kong FM, Han GZ, Jiang HM, Huang XJ,
Li RJ, Wang HG, Li SS (2009) QTL analysis of kernel shape
and weight using recombinant inbred lines in wheat. Euphytica
165:615–624
Terao T, Nagata K, Morino K, Hirose T (2010) A gene controlling
the number of primary rachis branches also controls the vascular
bundle formation and hence is responsible to increase the harvest
index and grain yield in rice. Theor Appl Genet 120:875–893
Utz HF (1997) PLABSTAT: a computer program for statistical anal-
ysis of plant breeding experiments. Institute of Plant Breeding,
Seed Science, and Population Genetics, University of Hohenheim,
Stuttgart, Germany. http://www.uni-hohenheim.de/~ipspwww/
soft.html (version 3 Bwin of Feb 2010)
Veldboom LR, Lee M (1996) Genetic mapping of quantitative trait
loci in maize in stress and nonstress environments: I. Grain yield
and yield components. Crop Sci 36:1310–1319
Wan XY, Wan JM, Jiang L, Wang JK, Zhai HQ, Weng JF, Wang HL,
Lei CL, Wang JL, Zhang X (2006) QTL analysis for rice grain
length and fine mapping of an identified QTL with stable and
major effects. Theor Appl Genet 112:1258–1270
Wan X, Weng J, Zhai H, Wang J, Lei C, Liu X, Guo T, Jiang L, Su
N, Wan J (2008) Quantitative trait loci (QTL) analysis for rice
grain width and fine mapping of an identified QTL allele gw-5
in a recombination hotspot region on chromosome 5. Genetics
179:2239–2252
Wang S, Basten C, Zeng Z (2007) Windows QTL cartographer
2.5. Department of Statistics, North Carolina State University,
Raleigh, NC
Wang P, Xing Y, Li Z, Yu S (2012a) Improving rice yield and quality
by QTL pyramiding. Mol Breed 29:903–913
Wang S, Wu K, Yuan Q, Liu X, Liu Z, Lin X, Zeng R, Zhu H, Dong
G, Qian Q (2012b) Control of grain size, shape and quality by
OsSPL16 in rice. Nat Genet 44:950–954
1037Theor Appl Genet (2014) 127:1019–1037
1 3
Wassom JJ, Wong JC, Martinez E, King JJ, DeBaene J, Hotchkiss
JR, Mikkilineni V, Bohn MO, Rocheford TR (2008) QTL asso-
ciated with maize kernel oil, protein, and starch concentrations;
kernel mass; and grain yield in Illinois high oil × B73 backcross-
derived lines. Crop Sci 48:243–252
Würschum T (2012) Mapping QTL for agronomic traits in breeding
populations. Theor Appl Genet 125:201–210
Xing Y, Zhang Q (2010) Genetic and molecular bases of rice yield.
Annu Rev Plant Biol 61:421–442
Xu Y (2010) Molecular plant breeding. CAB International,
Wallingford
Xu S, Jia Z (2007) Genomewide analysis of epistatic effects for quan-
titative traits in barley. Genetics 175:1955–1963
Xu Y, Li HN, Li GJ, Wang X, Cheng LG, Zhang YM (2011) Mapping
quantitative trait loci for seed size traits in soybean (Glycine max
L. Merr.). Theor Appl Genet 122:581–594
Yan JB, Tang H, Huang YQ, Zheng YL, Li JS (2004) Comparative
analyses of QTL for important agronomic traits between maize
and rice. Acta Genet Sin 31:1401–1407
Yang J, Zhu J, Williams RW (2007) Mapping the genetic architecture
of complex traits in experimental populations. Bioinformatics
23:1527–1536
Yang X, Ma H, Zhang P, Yan J, Guo Y, Song T, Li J (2012) Char-
acterization of QTL for oil content in maize kernel. Theor Appl
Genet 125:1169–1179
Zhang X, Wang J, Huang J, Lan H, Wang C, Yin C, Wu Y, Tang H,
Qian Q, Li J (2012) Rare allele of OsPPKL1 associated with
grain length causes extra-large grain and a significant yield
increase in rice. Proc Natl Acad Sci 109:21534–21539
Zhao F, Zhang G, Zeng R, Yang Z, Ling Y, Sang X, He G (2011)
Analysis of epistatic and additive effects of QTLs for grain shape
using single segment substitution lines in rice (Oryza sativa L.).
Acta Agron Sin 37:469–476
