Upload
claudio-lopes
View
220
Download
3
Embed Size (px)
Citation preview
Genetic analysis of kernel oil content in tropical maizewith design III and QTL mapping
Gustavo Vitti Moro • Mateus Figueiredo Santos •
Dyeme Antonio Vieira Bento • Aurelio
Mendes Aguiar • Claudio Lopes de Souza Jr.
Received: 12 July 2011 / Accepted: 17 December 2011 / Published online: 1 January 2012
� Springer Science+Business Media B.V. 2011
Abstract Oil content and grain yield in maize are
negatively correlated, and so far the development of
high-oil high-yielding hybrids has not been accom-
plished. Then a fully understand of the inheritance of
the kernel oil content is necessary to implement a
breeding program to improve both traits simulta-
neously. Conventional and molecular marker analyses
of the design III were carried out from a reference
population developed from two tropical inbred lines
divergent for kernel oil content. The results showed
that additive variance was quite larger than the
dominance variance, and the heritability coefficient
was very high. Sixteen QTL were mapped, they were
not evenly distributed along the chromosomes, and
accounted for 30.91% of the genetic variance. The
average level of dominance computed from both
conventional and QTL analysis was partial dominance.
The overall results indicated that the additive effects
were more important than the dominance effects, the
latter were not unidirectional and then heterosis could
not be exploited in crosses. Most of the favorable
alleles of the QTL were in the high-oil parental inbred,
which could be transferred to other inbreds via marker-
assisted backcross selection. Our results coupled with
reported information indicated that the development of
high-oil hybrids with acceptable yields could be
accomplished by using marker-assisted selection
involving oil content, grain yield and its components.
Finally, to exploit the xenia effect to increase even
more the oil content, these hybrids should be used in the
Top CrossTM
procedure.
Keywords Tropical maize � Oil � SSR markers �QTL � Design III � Genetic variances
G. V. Moro � M. F. Santos � D. A. V. Bento �A. M. Aguiar � C. L. de Souza Jr. (&)
Department of Genetics, University of Sao Paulo,
Agriculture College ‘‘Luiz de Queiroz’’, P.O. Box 83,
13.400-970 Piracicaba, Sao Paulo, Brazil
e-mail: [email protected]
G. V. Moro
e-mail: [email protected]
M. F. Santos
e-mail: [email protected]
D. A. V. Bento
e-mail: [email protected]
A. M. Aguiar
e-mail: [email protected]
Present Address:
M. F. Santos
Embrapa/Vegetables, P.O. Box 218, 70.359-970 Gama,
DF, Brazil
Present Address:
D. A. V. Bento
Nunhes do Brasil Comercio de Sementes Ltda,
P.O. Box 2081, 59.611-970 Mossoro, RN, Brazil
Present Address:
A. M. Aguiar
Fibria Celulose, Rodovia Gen. Euryale de Jesus Zerbini,
Km 84, 12.340-010 Jacareı, SP, Brazil
123
Euphytica (2012) 185:419–428
DOI 10.1007/s10681-011-0604-x
Introduction
The primary use of maize grain in several countries as
Brazil and USA is for feeding livestock as poultry and
swine (Goldman et al. 1994; Mangolin et al. 2004).
Several reports have shown that feed prepared with
high oil maize kernels increased the daily weight gain
in both species above cited, which was attributable to
the greater caloric content of oil than that of starch on a
weight basis (Lambert et al. 2004). Current maize
hybrids have about 4% of oil content in the kernels
(Laurie et al. 2004) and, although high-oil ([6%)
maize cultivars is in demand, the negative genetic
correlation between grain yield and oil content has
hampered the development of high-oil high-yielding
cultivars since the goal of the breeding programs is to
increase grain yield (Val et al. 2009).
Inheritance studies have shown that maize kernel oil
content is a quantitatively inherited trait that presents
high heritability, additive effects are more important
than non-additive effects (dominance and epistasis),
low magnitude or absence of genotype by environment
interaction and, also, presents xenia effect (Goldman
et al. 1994; Berke and Rocheford 1995; Lambert et al.
2004; Laurie et al. 2004; Mittelman et al. 2003).
Several studies have been conducted to map QTL for
kernel oil content in temperate maize germplasm, and
as summarized by Val et al. (2009) the number of QTL
mapped and reported from them varied from 7 to 70.
However, to the author’s knowledge only one study has
been conducted to map QTL for kernel oil content in
tropical maize germplasm (Mangolin et al. 2004),
which mapped 13 QTL. The number, position, mag-
nitudes of the QTL effects, and the magnitudes of the
phenotypic variances accounted for by the QTL in
these reports varied greatly, which could be due to the
differences in the genetic background of the popula-
tion, the map saturation, the design and the model used
to map QTL (Lima et al. 2006).
The design III was outlined to estimate the com-
ponents of the genetic variance as well as the average
level of dominance (Comstock and Robinson 1952),
and it has been used to study the inheritance of several
traits in maize (Hallauer and Miranda Filho 1988).
Recently, Lu et al. (2003) presented a procedure to use
the design III to map QTL, and coupled with the
classical analysis this procedure is a powerful frame-
work for inheritance studies. Despite the importance
of kernel oil content there is limited information on its
inheritance in tropical maize germplasm, which is
needed to design an effective breeding program to
develop high-oil genotypes with acceptable yields.
Thus, this research was aimed to study the inheritance
of the kernel oil content in a tropical maize population
using the design III coupled with molecular markers.
Materials and methods
Genetic material
Parental inbred lines L-08-05F (orange flint kernels)
and L-14-04B (yellow dent kernels) with 71.33 and
47.00 g kg-1 of kernel oil content, respectively, were
used to develop a reference F2 population. Both inbreds
were developed by the Maize Breeding Program at the
Department of Genetics, Agriculture College ‘‘Luiz de
Queiroz’’, University of Sao Paulo, Brazil. Inbred
L-08-05F was derived from IG-1, an early-flowering
with orange flint kernels population, developed from
crosses of Brazilian and Thai populations; and inbred
L-14-04B was derived from BR-106, an early-flower-
ing with yellow dent kernels population, developed
from crosses of Brazilian and Mexican populations
(Silva et al. 2004). These populations and their
respective inbreds are in different heterotic groups
and they are genetically divergent for several traits
besides kernel oil content, and both populations were
derived from tropical germplasm (Sabadin et al. 2008;
Moreira et al. 2009). The parental inbreds were crossed
and three F1 plants, previously tested against the
parental inbreds to check their genetical identity with
microsatellite markers, were selfed to develop the F2
population. Two-hundred and fifty plants were ran-
domly taken and selfed to develop F2:3 progenies,
which were then backcrossed to the parental inbreds in
isolation blocks. The F2:3 progenies were used as
females and the parental inbreds as males. The males
were sown at each set of four progenies and each plot
consisted of 30 plants. Each plot was harvested
separately and their seeds bulked. Thus, from each F2
plant two backcrossed progeny was developed, for a
total of 500 backcrossed progenies.
Experimental evaluation
The 500 backcrossed progenies and their parental
inbreds were evaluated at the Experimental Station
420 Euphytica (2012) 185:419–428
123
Department of Genetics in the 2004/2005 growing
season, near the city of Piracicaba, the state of Sao
Paulo, Brazil. The progenies were allocated in five
experiments following the 10 9 10 lattice design,
with 50 progenies backcrossed to each parental inbred
allocated in each experiment (lattice), with three
replications. Plots were one row 4.0 m long spaced
0.8 m between rows, and 0.2 m between plants within
rows; plots were overplanted and thinned to 20 plants
per plot (62,500 plants ha-1). Ten plants per plot were
selfed to prevent against the xenia effect, and after
harvest 10 kernels from the mid part of each ear of
each backcrossed progeny from each replication were
randomly taken and bulked. The three random samples
of the kernels from each backcrossed progeny were
used to record the kernel oil content in a nuclear
magnetic resonance device (NMR). The trait was
recorded in only one environment with three replica-
tions since kernel oil content presents higher narrow-
or broad-sense heritability coefficient ([70%)
(Hallauer and Miranda Filho 1988), absence (Berke
and Rocheford 1995; Mittelman et al. 2003) or presence
of non-crossover genotype by environment interaction
(Goldman et al. 1994), and even when genotype by
environment interactions were significant their magni-
tudes ranged from about 12 (Laurie et al. 2004) to 43
(Wasson et al. 2008b) times inferior than the genotype
variation indicating that likely they were non-crossover
interactions. Furthermore, the long-term breeding pro-
grams to increase kernel oil content have been
conducted with one cycle of phenotypic selection per
year in only one experimental station, and their
successful results (Dudley and Lambert 2004; Song
and Chen 2004) could be only explained because of its
high heritability and the absence or presence of non-
crossover genotype by environment interaction.
Analysis of variance
Analyses of variance were computed for each experi-
ment following the lattice design, and these analyses
were then pooled in a single analysis of variance
summing up their respective degrees of freedom and
sums of squares. All analyses were computed using
PROC GLM from SAS software (SAS Institute and Inc.
2001). From the pooled analysis of variance, estimates
of additive ðr2AÞ and dominance ðr2
DÞ variances, and
the average level of dominance ½ALD ¼ ð2r2D=r
2AÞ
0:5�
were computed following the design III procedure
(Comstock and Robinson 1952). Also, the genetic
variance ðr2G ¼ r2
A þ r2DÞ, the phenotypic variance on
a progeny mean-basis ðr2Ph�x ¼ r2
G þ r2=RÞ and on a
plant-basis ðr2Ph ¼ r2
G þ r2Þ, and the broad-sense on a
progeny mean-basis ðh2�x ¼ r2
G=r2Ph�xÞ and the narrow-
sense ðh2r ¼ r2
A=r2PhÞ heritability coefficients were
computed; r2and R stand for the error variance and
number of replications, respectively. Confidence
intervals at the 95% probability level for the estimates
of the genetic and phenotypic variances and of the
heritability coefficient were computed following the
Burdick and Graybill (1992) procedures.
Genetic map
The genetic map used, and the procedures used to
develop it, was previously described by Sibov et al.
(2003). Briefly, the F2 plants that gave rise to the F2:3
progenies were genotyped with microsatellite markers.
The genetic map was developed using the MAP-
MAKER/EXP version 3.0b (Lincoln et al. 1992) with
a LOD threshold of 3.0 and a maximum distance
between adjacent markers of 50 cM; i.e., 0.38 as the
maximum recombination frequency, to establish the
linkage groups, and the Kosambi’s (1944) mapping
function was used to convert recombination frequencies
into map distances. Sixty new microsatellite markers
were added to the map of Sibov et al. (2003), for a total
of 177 markers distributed along the 10 linkage groups.
The genetic map spanned 2,052 cM in length with an
average interval of 11.60 cM between adjacent markers.
QTL mapping
The composite interval mapping method (CIM)
extended to perform a joint analysis of multiple
environments (mCIM) was used to map QTL, since it
is more powerful than the CIM procedure (Jiang and
Zeng 1995). The least square means of kernel oil content
from each replication were used to perform the analysis,
using the sample of each replication as an environment.
The underlying mixture model for QTL mapping is:
Yjk ¼ b0k þ a�kx�j þ d�k z�j þXt
l
ðalkxjl þ dlkzjlÞ þ ejk
where Yjk is the phenotypic mean of the jth progeny at
the kth replication; b0k is the mean effect of the model
Euphytica (2012) 185:419–428 421
123
for replication k; a�k is the additive effect of the
putative QTL at replication k; x�j counts the number of
alleles from parental inbred L-08-05F at the putative
QTL taking values of 2, 1, and 0 for genotypes QQ,
Qq, and qq, respectively, with probabilities depending
on genotypes of the markers flanking the putative
QTL and the recombination frequencies between
the QTL and the markers; d�k is the dominance effect
of the putative QTL in replication k; z�j is an indicator
variable taking values of 1 for genotype Qq and 0 for
genotypes QQ and qq of the putative QTL, with
probabilities depending on genotypes of the markers
flanking the putative QTL and the recombination
frequencies between the QTL and the markers; xjl and
zjl are corresponding variables for marker l, with
t markers selected as cofactors for controlling residual
variation; alk and dlk are the partial regression
coefficients of Yjk on xjl and on zjl, respectively; and
ejk is the residual of the model. The cofactors were
selected using the stepwise regression procedure
(P B 0.05) for each replication and combined for the
joint QTL analysis.
The QTL mapping was performed for each set of
the progeny backcrossed to one of the parental inbreds,
and then two analyses were performed: one for the
progenies backcrossed to the parental inbred L-08-05F
(BC1) and the other one to the parental inbred L-14-
04B (BC2). The LR threshold used to declare a
presence of QTL was 13.83, which corresponds to a
LOD score of 3.0. This value was used since the LOD
score values used for other authors have ranged from
2.5 to 3.8 (Mangolin et al. 2004; Song et al. 2004;
Willmot et al. 2006; Zhang et al. 2008; Wasson et al.
2008a; Wasson et al. 2008b). Only the mapped QTL
stable across replications were considered for this
study. The QTL mapping was performed with the
software Windows QTL Cartographer, version 2.5,
JZmap procedure (Wang et al. 2005), using a window
size of 10 cM and walking speed of 1 cM.
Since the analyses were performed for each of the
backcrossed progenies the correct values of the additive
(a) and dominance (d) effects for the QTL mapped were
not provided by the mCIM analyses, which was
developed for progenies as F2:3, F3:4, etc. Lu et al.
(2003) showed that they can be computed from the
pseudo-additive values (‘‘a’’) computed from each
backcross analysis as follows: the ‘‘a’’ value for a
QTL mapped for the BC1 progenies is actually
‘‘a1’’ = (a - d)/2, and for a QTL mapped for the BC2
progenies ‘‘a2’’ = (a ? d)/2, and then the additive and
dominance effects for each mapped QTL could be
computed as: a = (‘‘a1’’ ? ‘‘a2’’) and d = (‘‘a2’’-
‘‘a1’’), respectively. In addition, Melchinger et al. (2008)
reported that the genetic effects of the ith QTL are H1 ¼ai � di þ ð1=2Þ
Pij aaij � ð1=2Þ
Pij daij and H2 ¼
ai þ di � ð1=2ÞP
ij aaij � ð1=2ÞP
ij daij for the BC1
and BC2 progenies, respectively. In these expressions
aaij and daij stand for the additive by additive and
dominance by additive epistatic effects, respectively.
Notice that ‘‘a1’’ = H1/2 and ‘‘a2’’ = H2/2 , and then
the additive effect is actually a�i ¼ ½ðH1 þ H2Þ=2� ¼ai � ð1=2Þ
Pij daij and the dominance effect is actu-
ally d�i ¼ ½ðH2 � H1Þ=2� ¼ di � ð1=2ÞP
ij aaij. The
a�i and d�i effects were termed as the augmented
additive and augmented dominance effects, which
contains the additive ðaiÞ and dominance ðdiÞ effects
of the ith QTL and half of the epistatic effects daij and
aaij of the QTL with other QTL underlying the trait;
i.e., with the genetic background of the population,
and they are related to the parental inbreds difference
PD ¼P
2a�i and to the mid-parent heterosis MPH ¼Pd�i (Melchinger et al. 2008). Thus, the additive and
dominance effects of the QTL reported by Lu et al.
(2003), and those ones estimated in this paper, actually
estimate the a�i and d�i effects, which contains the
additive ðaiÞ and dominance ðdiÞ effects of the ith QTL
confounded with their epistatic interactions daij and
aaij, respectively, with the genetic background of the
population.
The phenotypic variances explained by each QTL
and by the set of all mapped QTL ðR2PhÞ were also
computed using the Windows QTL Cartographer. All
QTL previously mapped using the mCIM procedure
were fitted in a single model with multiple QTL, and
then, the software uses the EM algorithm to evaluate
the likelihood of this mixture model with multiple
QTL. Once the model was fitted, the coefficients of
phenotypic determination of each QTL and of all QTL
were automatically computed by the software. The
genotypic coefficient of determination was computed
dividing the R2Ph by the coefficient of heritability; i.e.,
ðR2G ¼ R2
Ph=h2Þ.Parental inbred L-08-05F has higher oil content in
the kernels than L-14-04B and, therefore, the positive
additive effects of the mapped QTL indicated that the
422 Euphytica (2012) 185:419–428
123
favorable alleles (alleles that increase oil content)
were from the L-08-05F; whereas the negative signs
indicated that the favorable alleles were from L-14-
04B. Levels of dominance were computed as LD ¼d�j j= a�j j ratio for each QTL, and the average level of
dominance (ALD) was computed weighing each LD
ratio by its respective R2G value. Following several
reports (Sibov et al. 2003; Mangolin et al. 2004; Lima
et al. 2006; Moreira et al. 2009), gene action was
characterized as: additive (A) 0.00 B LD B 0.20;
partial dominance (PD) 0.21 B LD B 0.80; domi-
nance (D) 0.81 B LD B 1.20; and overdominance
LD C 1.21.
Results
Means and components of variance
Highly significant differences (P B 0.01) were detected
in the pooled analysis of variance of the Design III for
progenies and for progenies x parents, indicating that
both additive and dominance effects were present in the
variation for kernel oil content. The coefficient of
experimental variation was 11.45%, value similar to
other reports (Goldman et al. 1994; Berke and Roche-
ford 1995; Zhang et al. 2008) indicating that the
experimental data presented good precision. The overall
means of the parental inbreds differed significantly
(P \ 0.05), with values of 71.33 g kg-1 for the inbred
L-08-05F and 47.00 g kg-1 for the inbred L-14-04B.
The average of the 500 progenies (49.10 g kg-1)
differed significantly from the mean of the parental
inbreds (59.17 g kg-1), and the average of the progenies
backcrossed to L-08-05F (58.18 g kg-1) was signifi-
cantly higher than the average of those backcrossed to
L-14-04B (40.04 g kg-1). The means of the progenies
assessed ranged from 28.30 to 75.27 g kg-1, showing
the presence of transgressive progenies for both lower
and higher kernel oil content (Table 1).
The estimate of the genetic variance ðr2GÞ, as well as
its components additive ðr2AÞ and dominance variances
ðr2DÞ, the broad-sense on a progeny mean-basis ðh2
�xÞand the narrow-sense ðh2
r Þ heritability coefficients, and
the average level of dominance differed significantly
from zero. Also, additive variance was quite larger
than the dominance variance, and the level of dom-
inance was 0.58 (Table 2).
QTL mapping
Sixteen QTL were mapped for kernel oil content on
chromosomes 1–8; i.e.; QTL were not mapped only on
chromosomes 9 and 10. The number of QTL mapped
on each chromosome varied greatly, with one QTL
mapped on each one of the chromosomes 2, 3, 6, 7 and
8, three QTL mapped on each one of the chromosomes
1 and 5, and five QTL mapped on chromosome 4. The
augmented additive effects ða�i Þ of the QTL ranged
from -3.39 to 2.57 g kg-1, and the augmented domi-
nance effects ðd�i Þ ranged from -1.66 to 1.48 g kg-1.
The mapped QTL presented all types of gene action
Table 1 Mean values and confidence intervals (CI) of the
parental inbreds (L-08-05F and L-14-04D), of the backcrossed
progenies to the parental inbreds (BC1 and BC2) and the
average of the backcrosses, for kernel oil content in a tropical
maize population
Entriesa Means (g kg-1) CIb
L-08-05F 71.33 [62.32; 80.35]
L-14-04B 47.00 [37.99; 56.01]
BC1 (L-08-05F) 58.18 [49.17; 67.20]
BC2 (L-14-04B) 40.04 [31.03; 49.05]
Average (BC1 ? BC2)/2 49.10 [40.09; 58.12]
a L-08-05F: high oil parental inbred, L-14-04B: normal oil
parental inbred, BC1: backcross to parental inbred L-08-05F,
BC2: backcross to parental inbred L-14-04Bb CI Confidence intervals at the 95% probability level
Table 2 Estimates of the genetic variances, heritability coef-
ficients, average level of dominance and their confidence
intervals for kernel oil content in a tropical maize population
Parametersa Estimates C.I.b
r2A
75.05 [60.93; 96.10]
r2D
12.41 [9.30; 17.87]
r2G
87.46 [70.22; 113.96]
h2�x
0.89 [0.80; 0.99]
h2r
0.77 [0.68; 0.87]
d^ 0.58 [0.66; 1.47]
a r2A additive genetic variance, r2
D dominance genetic
variance, r2G ¼ r2
A þ r2D, r2
�Phphenotypic variance on a
progeny mean-basis,h2�x broad-sense heritability coefficient on
a mean-basis, h2r narrow-sense heritability coefficient, d
^
average level of dominanceb CI Confidence intervals at the 95% probability level
Euphytica (2012) 185:419–428 423
123
since seven QTL displayed additive gene action, six
partial dominance, one complete dominance, and two
QTL displayed overdominance; and the average level
of dominance (ALD) was 0.44 characterizing partial
dominance. From the 16 QTL mapped, 11 (68.75%)
with favorable alleles (alleles increasing oil content)
were from the parental inbred L-08-05F, and five
(31.25%) from the parental inbred L-14-04B. The
percentage of the phenotypic (genotypic) variance
accounted for by the QTL ranged from 0.17 (0.19) to
6.58% (7.37%), and collectively the set of the QTL
mapped accounted for by 27.57 and 30.91% of the pheno-
typic and genotypic variance, respectively (Table 3).
Discussion
Means and variances
The parental inbred line L-08-05F could be considered
as high oil genotype since it presented above 60 g kg-1
kernel oil content, but the other parental inbred L-14-
04B could not be considered as low oil genotype since
its kernel oil content is similar to the cultivars selected
for grain yield. Nevertheless, the difference between
their means was quite high 24.33 g kg-1 (51.76%),
which allowed the development of a population with
high genetic variability, which displayed transgressive
progenies for lower (28.30 g kg-1) and higher
(75.27 g kg-1) kernel oil content. The presence of a
few transgressive genotypes for higher kernel oil
content indicates that the inbred L-14-04B has a small
set of loci with favorable alleles for increasing kernel
oil content, and the presence of several transgressive
progenies for lower oil content indicates that the inbred
L-14-04B has most of the loci underlying kernel oil
content with alleles decreasing this trait. Then, as
expected the progenies backcrossed to the inbred L-08-
05F presented higher kernel oil content than the mean
of the progenies backcrossed to the inbred L-14-04B;
and the difference between their means was quite high;
i.e., 18.14 g kg-1 (45.30%). Mangolin et al. (2004) did
Table 3 Genomic position, genetic effects, levels of dominance, direction of the favorable alleles, and coefficients of determination
for the mapped QTL
QTLa Bin Marker interval P (cM)b LOD a* d* LDc DdR2
Ph R2G
Qoil1a 1.01 umc1177–bnlg1014 4 3.89 1.66 -1.46 0.88 (D) L-08-05 1.95 2.19
Qoil1b 1.01 bnlg1014–umc1106 14 4.01 2.06 -0.28 0.14 (A) L-08-05 2.18 2.45
Qoil1c 1.01–1.02 umc1106–bnlg1627 22 4.18 2.02 -0.53 0.27 (PD) L-08-05 2.15 2.41
Qoil2 2.08 umc1464–umc1633 201 3.66 0.39 -0.94 2.42 (OD) L-08-05 0.30 0.34
Qoil3 3.04 bnlg1452–bnlg602 52 3.07 0.57 0.08 0.13 (A) L-08-05 0.17 0.19
Qoil4a 4.06–4.07 bnlg2291–dupssr34 94 3.24 1.05 1.48 1.42 (OD) L-08-05 1.12 1.25
Qoil4b 4.07–4.08 dupssr34–bnlg2244 100 3.07 -0.74 0.10 0.14 (A) L-14-04 0.28 0.31
Qoil4c 4.08 bnlg2244–bnlg2162 114 3.11 -3.39 -1.66 0.49 (PD) L-14-04 6.58 7.37
Qoil4d 4.08 bnlg2162–umc1086 116 3.19 -2.49 -0.56 0.23 (PD) L-14-04 3.24 3.63
Qoil4e 4.08–4.09 umc1051–umc1989 126 3.36 -1.21 0.27 0.23 (PD) L-14-04 0.77 0.86
Qoil5a 5.04 bnlg1892–dupssr10 106 3.47 -2.36 -0.21 0.09 (A) L-14-04 2.86 3.20
Qoil5b 5.07–5.08 phi0128–bnlg1711 238 3.49 2.57 -0.42 0.16 (A) L-08-05 3.40 3.81
Qoil5c 5.08–6.00 bnlg1711–umc1023 256 3.26 1.85 -0.23 0.12 (A) L-08-05 1.75 1.96
Qoil6 6.00–6.01 phi0126–bnlg1371 17 3.01 1.87 0.99 0.53 (PD) L-08-05 2.04 2.29
Qoil7 7.05–8.01 umc1407–umc1139 194 3.27 1.74 -0.83 0.47 (PD) L-08-05 1.72 1.93
Qoil8 8.03 bnlg1863–phi0115 82 3.50 2.12 -0.20 0.10 (A) L-08-05 2.31 2.59
– – – – ALD 0.44 (PD) Total 27.57 30.91
a QTL are indicated as Qoil followed by a chromosome number and by a word for more than one QTL in the same chromosomeb P position of QTL refers to the distance in centimorgans (cM) from the first marker of the chromosome to the mapped QTLc LD level of dominance A additive, PD partial dominance, D dominance, OD overdominance, ALD average level of dominanced D direction of the favorable allele: negative and positive a* values indicates that favorable alleles were from L-14-04B and L-08-
05F, respectively
424 Euphytica (2012) 185:419–428
123
not report transgressive progenies in tropical germ-
plasm, may be because their parental inbreds had
greater difference in oil content (29.24 vs. 79.25
g kg-1) than the parental inbred lines used in this
study. Yet the range of their entry means (44.93–
72.63 g kg-1) was lower than the range of the inbred
parental means even using a number of genotypes
greater (408 F2 plants) than we used.
The magnitude of the additive variance was about
six times greater than the dominance variance and the
level of dominance was partial dominance. As is well-
known estimates of additive and dominance variances
are biased in populations with high levels of linkage
disequilibrium, with the dominance variance always
biased upward regardless the phase of linkage dis-
equilibrium and the additive variance biased upward
in association-phase and downward in repulsion-phase
linkage disequilibrium, respectively (Comstock and
Robinson 1952). Thus, the estimate of the level of
dominance is probably biased upward because the
population used was in high level of linkage disequi-
librium, evidencing that the additive effects are more
important than the dominance effects in the loci
underlying kernel oil content in maize (Val et al.
2009). The narrow- and broad-sense heritability
coefficients were quite high (0.77 and 0.89), which
agreed with the reported results (Hallauer and Miranda
Filho 1988; Mangolin et al. 2004; Wasson et al. 2008b;
Yang et al. 2010).
QTL mapping
The 16 mapped QTL were not evenly distributed in the
chromosomes; instead 11 QTL out of 16 were on only
three chromosomes (five on chromosome 4, and three
on chromosomes 1 and 3). The set of the mapped QTL
accounted for only 30.91% of the genetic variance,
with Qoil4c with the greatest contribution (7.37%) to
the genetic variance. The low magnitude of the genetic
variance accounted for by the mapped QTL has also
been reported for almost all studies for this trait
(Mangolin et al. 2004; Wasson et al. 2008b; Yang et al.
2010). These results indicate that several QTL were not
mapped which could be due to (1) failure of the model
to map QTL with very small effects, (2) the model
allows to map only one QTL per interval, and then sets
of QTL in the same interval will be mapped as only one
QTL, and (3) furthermore if the genetic effects of sets
of QTL within the same interval are different they
could have the total effect canceled or highly reduced
and not being mapped as a unique QTL with small
effect (Lima et al. 2006). From the QTL mapped, 11
displayed positive and five displayed negative aug-
mented additive effects ða�i Þ. Thus, there were favor-
able alleles, i.e., alleles that increase kernel oil content,
in both parental inbred lines. Although most of these
QTL were in the parental inbred L-08-05F (68.75%),
the presence of QTL with favorable alleles in the
parental inbred L-14-04B allowed the occurrence of
transgressive progenies as above mentioned. Aug-
mented dominance effects ðd�i Þ of the QTL displayed
positive and negative signs showing that this effect
was not unidirectional, and the average level of
dominance was partial dominance (0.44); notice that
this result was similar to that computed from the
estimates of variance components (0.58). Two QTL
Qoil2 and Qoil4a presented overdominance, but since
true overdominance is not expected, this might be due
to repulsion-phase linkages of sets of QTL with
dominance and/or partially dominance effects mim-
icking pseudo-overdominances (Lu et al. 2003). Most
of the reported results have shown that the average
level of dominance was in the range of additive to
partial dominance and that the dominance effects was
not unidirectional (Berke and Rocheford 1995;
Mangolin et al. 2004; Song et al. 2004; Zhang et al.
2008), which are consistent with the results from this
research. Therefore, the additive effects being more
important than the dominance effects, and the latter
being not unidirectional, likely the magnitude of
inbreeding depression and of heterosis for kernel oil
content will be of low magnitude.
The number of QTL mapped has varied greatly
across several studies; i.e., from 7 to 70. There are
several reasons for these differences as the populations,
genetic designs, number of markers, methods used for
mapping etc., but those that reported a high number of
QTL mapped were in studies involving populations
that have been subjected to a long-term divergent
selection for kernel oil content (Val et al. 2009). All
QTL mapped in this study have already been mapped in
the same bins in other studies (Goldman et al. 1994;
Berke and Rocheford 1995; Laurie et al. 2004;
Mangolin et al. 2004; Song et al. 2004; Willmot et al.
2006; Dudley et al. 2007; Wasson et al. 2008b; Yang
et al. 2010; Zhang et al. 2008). Although in the same
Euphytica (2012) 185:419–428 425
123
bins, they could not be considered as the same QTL.
However, the QTL Qoil3, Qoil5a, and Qoil8 were
mapped in at least five studies, and therefore, they were
stable across populations, environments and methods
of mapping.
Implications for selection
The results of this research and those already reported
in tropical (Mangolin et al. 2004) as well as in
temperate germplasm (Goldman et al. 1994; Berke and
Rocheford 1995; Song et al. 2004; Willmot et al. 2006;
Dudley et al. 2007; Zhang et al. 2008; Wasson et al.
2008a; Wasson et al. 2008b) showed that the kernel oil
content in maize presents high heritability, additive
effects are more important than non-additive effects,
and the dominance effects are not unidirectional.
Thus, the magnitude of heterosis is expected to be low
and could not be exploited in crosses. Also, the
magnitude of inbreeding depression should be low and
coupled with the high magnitude of the coefficient of
heritability simple selection procedures as individual
phenotypic selection could be used to develop inbred
lines and/or populations with high kernel oil content.
In fact, reported results of long-term phenotypic
recurrent selection have shown steady increases in
the kernel oil content (Dudley and Lambert 2004;
Song and Chen 2004). Notice, also, that a transgres-
sive progeny had 75.27 g kg-1 of kernel oil content,
indicating that we can develop inbred lines from this
genotype with high levels of kernel oil content.
We mapped 16 QTL underlying kernel oil content,
and three of them were already reported in other studies
with temperate germplasm, and then these genomic
regions should be considered for fine mapping and/or
cloning; also these mapped QTL could be used for
marker-assisted selection (MAS). Although MAS will
not be as effective as phenotypic selection (PS) on a per
cycle basis, PS could be coupled with MAS to design
more effective breeding programs since MAS could be
conducted under both growing and off-growing sea-
sons speeding-up breeding programs aimed to increase
kernel oil content in maize. For instance, the favorable
alleles of the set of QTL Qoil1b, Qoil1c, Qoil5b, and
Qoil8, whose additive effects are 2.06, 2.02, 2.57, and
2.12 g kg-1 of oil content are in the parental inbred
L-08-05F and could be transferred by marker-assisted
backcross selection to pyramid them in other inbreds,
as has been successfully reported for maize and other
species for several traits (Stuber and Sisco 1992;
Neereja et al. 2007; Garzon et al. 2008).
However, the major constraint to develop high oil
maize hybrids is the high negative genetic correlation
of kernel oil content and grain yield (Lambert 2001).
To overcome this drawback the Top CrossTM
procedure
has been proposed to exploit the xenia effect to allow
the production of high-oil maize while maintaining the
productivity (Berquist et al. 1998). This procedure
consists in sowing a mixture of seeds of a normal
kernel oil content high-yielding male-sterile single-
cross (&90%) and a high-oil (but low-yielding) male
pollinator (&10%). Lambert et al. (1998) reported that
hybrids pollinated with high oil and normal oil content
strains had comparable yield, but those hybrids
pollinated with high oil strains yielded 19 g kg-1
more oil than those pollinated with the normal oil
content strains. Also, Thomison et al. (2003) reported
that with the Top CrossTM
procedure the kernel oil
content was quite higher (72.12 g kg-1) than in the
normal single-crosses crop (41.25 g kg-1), but a
significant reduction on grain yield was reported
(Thomison et al. 2002). Although there were divergent
results on the use of the Top CrossTM
procedure, the
reduction on grain yield is expected since the yields of
the pollinator genotypes are too low and they occupy
about 10% of the field, and also because oil requires
2.25 times more energy than starch to be produced
(Val et al. 2009).
Thus, further researches with molecular markers
could be useful to design new approaches to develop
high oil genotypes without significant reduction on
grain yield. Sibov et al. (2003) evaluated F2:3 prog-
enies of the same population used in this study and
mapped QTL for grain yield, but none of them were
mapped in the same genomic regions of the QTL for
kernel oil content mapped in this study. Also, Wasson
et al. (2008b) mapped QTL for grain yield and kernel
oil content in different genomic regions. These results
indicate that the introgression of specific genomic
regions associated with kernel oil but not with yield
into inbred lines could allow the development of
cultivars with above-average oil content while main-
taining yield. Furthermore, Berke and Rocheford
(1995) mapped QTL for kernel weight and for kernel
oil content in different genomic regions. They empha-
sized that ‘‘three major QTL for oil concentration on
chromosomes 2, 6, and 9 were not associated with
changes in kernel weight, indicating that selection for
426 Euphytica (2012) 185:419–428
123
increased oil concentration without changing kernel
weight should be possible in this population’’. Kernel
weight is one of the grain yield components, along
with the number of ears per plant (prolificacy), kernels
per row and kernel-row number. Thus, it is necessary
to investigate the genetic architecture of these traits
jointly; i.e., to estimate the genetic correlations
between kernel oil content and grain yield compo-
nents, to map QTL for kernel oil content and for yield
components, and to investigate whether there are QTL
underlying kernel oil content and grain yield compo-
nents in different genomic regions.
One can speculate that if the findings of Berke and
Rocheford (1995) for kernel weight could be verified
for the others grain yield components we could be able
to increase kernel oil content without decreasing
significantly grain yield. Thus, selection to increase
grain yield and kernel oil content simultaneously could
be conducted by manipulating the QTL underlying
grain yield and its components and kernel oil content
that are in different genomic regions using a selection
index in a marker-assisted selection or in a genomic-
wide selection framework (Lorenzana and Bernardo
2009). This procedure could allow the development of
high-yielding above-average kernel oil content (say
60 g kg-1) elite single-crosses, and single-crosses
with high kernel oil content (say 90 g kg-1) with a
lower reduction in grain yield. Then, the high-yielding
above-average kernel oil content single-cross should
be used as female (male-sterile plants) and the high
kernel oil content single-cross as pollinator in the Top
CrossTM
procedure to allow the production of grains
with high oil content (say 80 g kg-1) with a minimum
loss in productivity.
Concluding, although this research did not include
grain yield components, it contributed to increase the
knowledge on the inheritance of the kernel oil content
in maize, mainly in tropical maize, and how to use this
information to implement an effective maize breeding
program aimed to increase this trait. Also, coupled with
reported information, we proposed further research
aimed to develop hybrids to be used in the Top CrossTM
procedure to exploit the xenia effect to increase even
more the oil content in the grains. Finally, the inbred
line L-08-05F is an important source of favorable
alleles (QTL) for oil content and these QTL could be
transferred to other inbreds to develop high kernel oil
content inbreds or populations.
Acknowledgements This research was supported by Conselho
Nacional de Desenvolvimento Cientıfico e Tecnologico (CNPq-
308499/2006-9), and by Department of Genetics at the
Agriculture College ‘‘Luiz de Queiroz’’/University of Sao
Paulo. C. L. Souza Jr. is recipient of a research fellowship
from CNPq. Authors are grateful to Dr. Anete Pereira de Souza,
from the University of Campinas, for the genetic mapping of the
population; to Dr. Luis Alberto Colnago, from Embrapa/
Instrumentacao, for the analysis of the kernel oil content; and
to A. J. Desiderio, A. S. Oliveira, A. O. Gil, C. R. Segatelli, and A.
Silva for their assistance with the field experiments. We are also
grateful to three anonymous reviewers for their constructive
suggestions.
References
Berke TG, Rocheford TR (1995) Quantitative trait loci for
flowering, plant and ear height, and kernel traits in maize.
Crop Sci 35:1542–1549
Berquist RR, Nubel DS, Thompson DL (1998) Production
method for corn with enhanced quality grain traits. Patent-
USA-5706603
Burdick RK, Graybill FA (1992) Confidence intervals on vari-
ance components. M. Dekker Inc., New York
Comstock RE, Robinson HF (1952) Estimation of average
dominance genes. In: Gowen JW (ed) Heterosis. Iowa State
College Press, Ames, pp 494–516
Dudley JW, Lambert RJ (2004) 100 generations of selection for
oil and protein in corn. In: Janick J (ed) Plant breeding
reviews: long-term selection-Maize, vol 24 (1). Wiley,
Oxford, pp 79–110
Dudley JW, Clark D, Rocheford TR, LeDeaux JR (2007)
Genetic analysis of corn kernel chemical composition in
the random mated 7 generation of the cross of generations
70 of IHP 9 ILP. Crop Sci 47:45–57
Garzon LN, Ligareto GA, Blair MW (2008) Molecular marker-
assisted backcrossing of anthracnose resistance into
Andean climbing beans. Crop Sci 48:562–570
Goldman IL, Rocheford TR, Dudley JW (1994) Molecular
marker associated with maize kernel oil concentration in an
Illinois high protein 9 Illinois low protein cross. Crop Sci
34:908–915
Hallauer AR, Miranda Filho JB (1988) Quantitative genetics in
maize breeding, 2nd edn. Iowa State University Press, Ames
Jiang C, Zeng ZB (1995) Multiple trait analysis of genetic map-
ping for quantitative trait loci. Genetics 140:1111–1127
Kosambi DD (1944) The estimation of map distances from
recombination values. Ann Eugenics 12:172–175
Lambert RJ (2001) High-oil corn hybrids. In: Hallauer AR (ed)
Specialty corns, 2nd edn. CRC Press, Boca Raton, pp 123–145
Lambert RJ, Alexander DE, Han ZJ (1998) A high oil pollinator
enhancement of kernel oil and effects on grain yields of
maize hybrids. Agron J 90:211–215
Lambert RJ, Alexander DE, Mejaya IJ (2004) Single kernel
selection for increased grain oil in maize synthetics and
high-oil hybrid development. In: Janick J (ed) Plant
breeding reviews: long-term selection-Maize, vol 24 (1).
Wiley, Oxford, pp 153–175
Euphytica (2012) 185:419–428 427
123
Laurie CC, Chasalow SD, LeDeaux JR, McCarroll R, Bush D,
Hauge B, Lai CQ, Clark D, Rocheford TR, Dudley JW
(2004) The genetic architecture of response to long-term
artificial selection for oil concentration in the maize kernel.
Genetics 168:2141–2155
Lima MLA, Souza CL Jr, Bento DAV, Souza AP, Garcia LC
(2006) Mapping QTL for grain yield and plant traits in a
tropical maize population. Mol Breed 17:227–239
Lincoln SE, Daly MJ, Lander ES (1992) Constructing genetic
maps with Mapmaker/Exp. 3.0, 3rd edn. Whitehead Insti-
tute for Biometrical Research, MAS
Lorenzana RE, Bernardo R (2009) Accuracy of genotypic value
predictions for marker-based selection in biparental plant
populations. Theor Appl Genet 120:151–161
Lu H, Romero-Severson J, Bernardo R (2003) Genetic basis of
heterosis explored by simple sequence repeat markers in a
random-mated maize population. Theor Appl Genet 107:
494–502
Mangolin CA, Souza CL Jr, Garcia AAF, Garcia AF, Sibov ST,
Souza AP (2004) Mapping QTLs for kernel oil content in a
tropical maize population. Euphytica 137:251–259
Melchinger AE, Utz HF, Schon CC (2008) Genetic expectations
of quantitative trait loci main and interaction effects
obtained with the triple testcross design and their relevance
for the analysis of heterosis. Genetics 178:2265–2274
Mittelman A, Miranda Filho JB, Lima GJMM, Hara-Klein C,
Tanaka RT (2003) Potential of the ESA23B maize popu-
lation for protein and oil content improvement. Scientia
Agricola 60:319–327
Moreira JUV, Bento DAV, Souza AP, Souza CL Jr (2009) QTL
mapping for reaction to Phaeosphaeria leaf spot in a trop-
ical maize population. Theor Appl Genet 119:1361–1369
Neereja CN, Maghirang-Rodriguez R, Pamplona A, Heuer S,
Collard BCY, Sptiningsih EM, Vergara G, Sanchez D, Xu
K, Ismail AM, Mackill DJ (2007) A marker-assisted
backcross approach for developing submergence-tolerant
rice cultivars. Theor Appl Genet 115:767–776
Sabadin PK, Souza CL Jr, Souza AP, Garcia AAF (2008) QTL
mapping for yield components in a tropical maize popula-
tion using microsatellite markers. Hereditas 145:194–203
SAS Institute Inc (2001) SAS/STAT User’s guide, v.6.03. SAS
Institute, Cary
Sibov ST, Souza CL Jr, Garcia AAF, Silva AR, Mangolin CA,
Benchimol LL, Souza AP (2003) Molecular mapping in
tropical maize using microsatellite markers. 2. Quantitative
trait loci (QTL) for grain yield, ear height, and grain
moisture. Hereditas 139:107–115
Silva AR, Souza CL Jr, Aguiar AM, Souza AP (2004) Estimates
of genetic variance and level of dominance in a tropical
maize population.I. Grain yield and plant traits. Maydica
49:65–71
Song TM, Chen SJ (2004) Long term selection for oil concen-
tration in five maize populations. Maydica 49:9–14
Song XF, Song TM, Daı JR, Rocheford TR, Li JS (2004) QTL
mapping of kernel oil concentration with high-oil maize by
SSR markers. Maydica 49:41–48
Stuber CW, Sisco P (1992) Marker-facilitated transfer of QTL
alleles between inbred lines and responses in hybrids. In:
Proceedings of 46th Ann Corn Sorghum Res. Conference.
ASTA, Washington pp 104–113
Thomison PR, Geyer AB, Lotz LD, Siegrist HJ, Dobbels TL
(2002) TopCross high-oil corn production: agronomic
performance. Agron J 94:290–299
Thomison PR, Geyer AB, Lotz LD, Siegrist HJ, Dobbels TL
(2003) TopCross high oil corn production: select grain
quality attributes. Agron J 95:147–154
Val DL, Schwartz SH, Kerns MR, Deikman J (2009) Devel-
opment of a high oil trait for maize. In: Kriz AL, Larkins
BA (eds) Biotechnology in agriculture and forestry, vol 63:
molecular genetic approaches to maize improvement.
Springer-Verlag Press, Berlin, pp 303–323
Wang S, Basten CJ, Zeng ZB (2005) Windows QTL Cartogra-
pher 2.5. Department of Statistics, North Caroline State
University, Raleigh. http://statgen.ncsu.edu/qtlcart/WQTL
Cart.htm. Accessed March 17, 2010
Wasson JJ, Mikkelineni V, Bohn MO, Rocheford TR (2008a)
QTL for fatty acid composition of maize kernel oil in
Illinois High Oil 9 B73 backcross-derived lines. Crop Sci
48:69–78
Wasson JJ, Wong JC, Martinez E, King JJ, DeBaene J, Hotch-
kiss JR, Mikkilineni V, Bohn MO, Rocheford TR (2008b)
QTL associated with maize kernel oil, protein, and starch
concentrations; kernel mass; and grain yield in Illinois high
oil 9 B73 backcross-derived lines. Crop Sci 48:243–252
Willmot DB, Dudley JW, Rocheford TR, Bari AL (2006) Effect
of random mating on marker QTL associations for grain
quality traits in the cross of Illinois high oil 9 Illinois low
oil. Maydica 51:187–199
Yang XH, Guo YQ, Yan JB, Zhang J, Song TM, Rocheford TR,
Li JS (2010) Major and minor QTL and epistasis contribute
to fatty acid compositions and oil concentration in high-oil
maize. Theor Appl Genet 20:665–678
Zhang J, Lu XQ, Song XF, Yan JB, Song TM, Dai JR, Roche-
ford TR, Li JS (2008) Mapping quantitative trait loci for
oil, starch, and protein concentrations in grain with high-oil
maize by SSR markers. Euphytica 162:335–344
428 Euphytica (2012) 185:419–428
123