Euphytica (2013) 189:111–122
DOI 10.1007/s10681-012-0731-z
Generation mean analysis of leaf chlorophyll concentration
from mid-silking to physiological maturity in some tropical
maize (Zea mays L.) genotypes under low and high nitrogen
dosages
A. A. Mushongi • J. Derera • P. Tongoona
N. G. Lyimo
•
Received: 20 August 2011 / Accepted: 29 May 2012 / Published online: 27 June 2012
Ó Springer Science+Business Media B.V. 2012
Abstract Genetic control of the leaf chlorophyll
concentration (LCC) in tropical maize (Zea mays L.)
genotypes has not been established, especially at
different reproduction growth stages under low and
high nitrogen (N). A generation mean analysis study
was conducted to identify the genetic effects that
govern the LCC from mid-silking to physiological maturity under high (120 kg N ha-1) and low
(60 kg N ha-1) (N) regimes for two seasons in a
randomized complete block design with two replications during main and offseason of year 2009 at Inyala
Agricultural Training Institute in Mbeya, Tanzania.
This study revealed that mid-parent heterosis for the
LCC increased with growth stages under both N
dosages but it was more pronounced under low N
dosage. Generally, genetic effects for LCC were more
easily estimable under high than under low N dosage.
Additive gene effects decreased with growth stage,
whereas dominance effects increased, irrespective of
the genotype and N regime. All genetic effects except
A. A. Mushongi (&)
Agricultural Research Institute (ARI)-Uyole, Mbeya,
P.O. Box 400, Tanzania
e-mail: [email protected]
A. A. Mushongi J. Derera P. Tongoona
African Centre for Crop Improvement, University of
KwaZulu-Natal, Pietermaritzburg, South Africa
N. G. Lyimo
Highlands Seed Growers Ltd, Mbeya, Tanzania
dominance 9 dominance interaction were significant
at some stages only in cross T20 9 NG8. The ratio
of fixable (additive plus additive 9 additive) to the
non-fixable (dominance plus additive 9 dominance,
and dominance 9 dominance) genetic effects was
74–26 % under high N dosage, and 35–65 % under the
low N dosage for the T20 9 C58, while for the
T20 9 NG8, the ratio was 37–63 % under high N
dosage, and 20–80 % under the low N dosage. The
trend observed suggest that fixable effects are preponderant under the high N dosage in one cross, while
non-fixable effects prevailed under the low N dosage
in the other cross.
Keywords Genetic effects Growth stages Leaf
chlorophyll concentration Nitrogen Tropical maize
Introduction
The genetic control of post-silking leaf chlorophyll
concentration (LCC) and its retention to physiological
maturity has not been resolved, especially under
contrasting soil nitrogen (N) conditions in tropical
maize (Zea mays L.) plant. The LCC relates directly to
the amounts of plant and soil N status (Hawkins et al.
2007). Thus genotypes which retain functional LCC
until physiological maturity (SG) hold the high N
dosage and water in their leaves (Thomas and Smart
1993), high N content in kernels and they yield higher
(Robson et al. 2001) than senescent genotypes
123
112
particularly in stressful environments. Such genotypes
have strong root systems that allow prolonged N
uptake during grain-filling (Rajcan and Tollenaar
1999), they minimise foliar diseases (Robinson
1996) and heat and drought stresses (Earl and
Tollenaar 1997; Bänziger et al. 2000; Joshi et al.
2007), when compared to senescent genotypes.
Because the SG is an adaptive trait, non senescent
genotypes would therefore mitigate stresses that affect
maize around the grain-filling stages. However, Joshi
et al. (2007) could not come up with a conclusive
answer whether maize genotypes that stay-green for a
longer period require additional N or not. Contentiously, Worku et al. (2007) reported that small-scale
maize farmers in sub-Saharan Africa apply less than
20 kg N ha-1. It is therefore crucial to comprehensively study the inheritance of the LCC character in
tropical maize, particularly under low N conditions,
because farmers apply less N to maize, soil N is
volatile, the soils are inherently deficient of the N, and
prices for the N fertilisers are higher than farmers can
afford.
In tropical areas maize is produced under harsh
conditions, the seasons are getting shorter and the rains
are becoming unpredictable thus failure to realise
investments from fertiliser applications. In these areas
the maize genotypes are source-limited such that
incorporating the SG trait in these genotypes would
improve grain yields by accelerating grain-filling
duration (Khanna-Choppra and Maheswari 1998;
Bänziger et al. 2000; Subedi and Ma 2005). Cavalieri
and Smith (1985) and Willman et al. (1987) reported
on the switching role of photosynthetic organs other
than leaves to supply the sink with the assimilates
when defoliation occurs in maize, unfortunately this
may not recover grain yield once the LCC has been
reduced before grain-filling has been completed.
Genotypic variation under low N conditions and the
fact that N metabolism plays a major role in low N
than it does in optimum N conditions is the essence of
G 9 N interaction (Bertin and Gallais 2000). Care
must therefore be taken when comparing genotypes
that widely differ in flowering dates and plant stature,
especially under stress due to differential accumulation of dry matter.
The LCC stabilises more from the periods of silking
to physiological maturity than during the vegetative
stages (Martinez and Guiamet 2004; Subedi and Ma
2005), suggesting that breeders should select for LCC
123
Euphytica (2013) 189:111–122
during this period. The Chlorophyll Meter (Model
SPAD-502 Camera Minolta Co. Ltd, Japan) readings
correlate highly with other methods used to determine
the status of N in plants and soils, indicating that the
meter may estimate the LCC precisely. The meter is
non-destructive so it allows for the collection of other
data on the same leaves.
Little literature that exists supports the polygenic
inheritance of the LCC character, such that the
estimation of individual genes is not applicable. Thus,
methods which detect and estimate the amount and
type of genetic effects rather than the individual genes
become relevant. Generation mean analyses (GMA)
has been used to detect the genetic effects of
quantitatively inherited traits in maize and other crops
(Azizi et al. 2006; Checa et al. 2006; Smith et al. 2009;
Shashkumar et al. 2010). GMA was therefore conducted to determine gene action and inheritance of the
LCC character from mid-silking to physiological
maturity in maize genotypes under high and low N
dosages.
Materials and methods
Plant materials
Three maize (T20, C58 and NG8) inbred lines
contrasting for post mid-silking LCC character were
selected during screening in season 2007/08. Inbred
line T20 was a common female parent to both F1
crosses T20 9 C58 and T20 9 NG8 (Table 1). The F1
was advanced to F2, and then backcrossed to the
respective parents to generate BCP1 and BCP2 to
constitute six generations of each cross for the study.
Inbred parent T20 is a locally adapted, high yielding
commercial inbred line with high combining ability
for early maturity. All the three inbred parents were
free of foliar diseases which would affect the SPAD
readings if these lines succumbed to such diseases.
Design and management of experiments
The trials were conducted at the Inyala Agricultural
Training Institute (S08o51.0110 and E033o38.2270 and
above sea level is 1520 m) in Mbeya, Tanzania. Two
harvesting crops were obtained with supplemented
irrigation in dry and wet seasons in 2009. The same
field plots were maintained throughout the study. The
Euphytica (2013) 189:111–122
113
Table 1 Maize inbred lines used for generation mean analyses
Inbred line
Pedigree of inbred line
Source of inbred line
Status of LCC
(SPAD values)
$
#
T20
UYL 15-11-1-8-5
Tanzania
Low
H
C58
MAS[MSR/312]-117-2-2-1-B 9 5/MAS[202/312]86-1-3-1-B 9 4
CIMMYT-Harare
High
H
NG8
TZE-Y Pop Co S6 Inb 62-3-3
IITA-Nigeria
High
H
LCC leaf chlorophyll concentration, $ female inbred parent, # male inbred parent
two environments were 120 and 60 kg N ha-1, which
correspond to the recommended fertiliser rates (or
high N dosage) and reduced N fertiliser rates, respectively. Phosphorus was applied at the recommended
rate of 30 kg P ha-1 for the study area in the form of
P2O5. In each season six generations of each cross
were field evaluated in a randomized complete block
design in two replications, 5.1 m plot length at 75 cm
between and 30 cm within rows under two N dosages
in two-row plots for non-segregating generations (P1,
P2, and F1) to make 36 plants, and seven-row plots for
segregating generations (F2, BCP1, and BCP2) to make
126 plants. Only two replications were used because
the number of individual plants in each evaluated
generations was adequate to allow estimation of the
genetic effects. Allard and Bradshaw (1964) reported
that the gradients related to low soil fertility are at least
systematic, as compared with random stresses such as
drought, which justifies further use of two replications
in two seasons in the present study. According to Kang
(1994) a few individuals are required for non-segregating generations, whereas many individuals were
required for segregating generations because these
generations undergo segregation and recombination.
Recently harvested seeds in the same season for all
generations of the two crosses were planted each
season to minimise variation in genotypes that may
lead to uneven germination and low plant vigour if
seeds from different seasons were used thus affecting
plant stand and plant stature. Prior to planting, the soils
and irrigation water were sampled and subjected to full
standard analysis. All standard agronomic practices
were followed.
Measuring the traits
The data were collected for the LCC character at four
intervals of 14 days. These intervals were considered
as separate traits since the LCC character is affected
by time. The mid-silking days stage was set as a
reference point to collect the LCC data in SPAD
values using the SPAD meter. Silking was considered
when the silk had extruded by 0.5–1.0 cm. The data
were recorded on individual plants from the second
leaf below the flag leaf after each generation had
reached that stage. Such a leaf was assumed to be
relatively free from mechanical injury and shading
that would affect SPAD meter values. The data were
taken between 09:30 and 11:30 a.m. The leaf was
cleared of dust and water before the recording of
SPAD meter data. The data were recorded at three
points: top, middle and at the base of the leaf, and
averaged to get a single SPAD meter value per plant
following the six reproductive stages in maize denoted
as R with each stage regarded at a seven-day interval
(Lafitte 1994; Hawkins et al. 2007), which meant after
every 2 weeks with regards to this study. The SPAD
meter values at the mid-silking stage reflected R1 and
were recorded as (LCC1), R2 = blister stage, R3 =
milk stage (LCC2), R4 = dough stage, R5 = dent
stage (LCC3) and R6 = physiological maturity, thus
(LCC4) was recorded in the 7th week from the midsilking stage––the stages are as defined according to
Lafitte (1994) and Hawkins et al. (2007).
Data analysis
Genetic assumptions for generation mean analyses
(GMA) were observed as defined by Wright (1968),
Mather and Jinks (1977), and Lande (1981).The SPAD
meter data were analysed in SAS software using the
PROC GLM and PROC REG procedures and modules. First, the overall model was submitted to analysis
to check the significance of main and interaction
effects: Y = replication ? generation ? nitrogen ?
season ? generation 9 nitrogen ? generation 9
season ? generation 9 nitrogen 9 season ? error,
where Y = overall mean for trait. The replications and
123
123
**, * Data statistically significant at p B 0.0001, 0.05, respectively
11.40
59.7
SD silking date, SV source of variation, DF degrees of freedom, Gen generation, R coefficient of determination, CV coefficient of variation, Pr probability level, F F distribution
2
21.47
16.58
12.75
13.50
12.54
24.78
17.86
12.06
–
CV ( %)
13.57
16.90
37.50
84.7
30.5
37.82
26.59
91.2
43.8
26.42
19.07
73.5
32.1
43.85
40.02
67.1
69.8
22.09
–
R2 ( %)
76.6
22
Error mean square
26.59
0.769
0.597
0.847
0.896
0.845
0.305
0.912
0.588
0.158
0.438
0.735
0.0295*
0.0825
0.321
0.671
0.0838
0.0339*
0.698
0.766
5
6
Gen 9 season
Gen 9 season 9 N
0.0532
0.345
0.109
0.344
0.242
0.375
0.155
0.343
0.311
0.665
0.662
5
Gen 9 N
0.306
\0.0001**
\0.0001**
0.305
0.777
0.0017*
0.0001*
0.0138*
0.142
0.0051*
0.0005*
0.607
0.649
0.0167*
0.0016*
0.763
0.351
0.003*
0.0115*
5
1
Generation
Season
1
0.389
0.0005*
0.0025*
0.0095*
0.0056*
0.0417*
0.0004*
0.0017*
0.0029*
0.0018*
Maturity
Dent
Milk
Nitrogen
0.052
Average
Pr [ F
Maturity
Dent
Milk
Mid-SD
Mid-SD
Average
Pr [ F
T20 9 NG
The nitrogen (except at mid-silking stage in the
T20 9 C58) and generation main effects were statistically significant in both crosses (Table 2). The R2
values were less than 50 % at maturity stage in the
T20 9 C58 and at mid-silking and dent stages in the
T20 9 NG8. Both parents (P1 and P2) were not
statistically significantly different (LSD0.05) at all
grain-filling stages under the low N dosage (LN) for
both crosses, hence the data were not submitted for
GMA, as the criteria for contrasting parents was not
met under the LN dosage conditions. The High N
dosage (HN) trials had statistically significant parents
for all growth stages, except at the physiological
maturity stage in the T20 9 C58 (Table 3a). The
SPAD meter values peaked at the milk stage and began
to drop at the dent stage and worsened at maturity,
regardless of generations and N regimes. Generally,
the mid-parent heterosis (MPH) built up with growth
stages, increasing sharply at the dent stage in both
crosses and N conditions, although this was more
pronounced under the LN dosage than the HN dosage
as indicated at the average of growth stages (Tables 3
a, b).
T20 9 C58
Main- and interaction effects, and mean separation
DF
Results
SV
seasons were considered random, while generations
and nitrogen levels were fixed effects. The GMA were
performed as described by Kang (1994), and the
following model was used: Y = m ? aa ? bd ? a2aa ? abad ? b2dd, where Y = generation
mean, m = mean of the F2 generation as base
population and intercept values, and other parameters
in the formulae for main and interaction genetic
effects and associated coefficients are as defined by
Gamble (1962) and Kang (1994). Separation of means
was performed using the LSD procedure for pair-wise
mean comparisons (p B 0.05) (Steel and Torrie 1980)
in the software of SAS. The narrow sense heritability
(h2) estimates were calculated following Warner
(1952) as: h2 = 100 9 [(2r2F2 - (r2BCP1 ? r2BCP2)/
r2F2], where r2F2, r2BCP1, and r2BCP2 = variances of F2,
BCP1 and BCP2, respectively. The numerator part in
the formula of the h2 estimate represents additive
genetic variance, whereas the denominator represents
the phenotypic variance.
Euphytica (2013) 189:111–122
Table 2 ANOVA summary of main- and interaction effects on the statistical significance (Pr [ F) of leaf chlorophyll concentration during grain-filling stages in the T20 9 C58
and T20 9 NG8 crosses of maize over N dosages in two seasons
114
(a) Cross T20 9 C58
40.7
39.5
39.2
38.9
32.5
0.3
BCP2
F1
F2
BCP1
P1
MPH ( %)
38.2
36.6
35.6
35.6
31.3
2.4
P2
BCP1
F2
F1
P1
MPH ( %)
ABC
42.9
37.6
36.0
32.3
10.1
BCP1
F2
BCP2
P1
MPH ( %)
40.8
40.4
F1
BCP2
Low N
AB
43.1
A
A
C
BC
AB
46.0
F1
A
B
AB
AB
AB
AB
A
B
AB
AB
AB
AB
P2
High N
(b) Cross T20 9 NG8
41.3
BCP2
Low N
46.2
P2
High N
A
F1
BCP2
P1
F2
BCP1
BCP2
F1
P2
P1
BCP1
F2
F1
P2
BCP2
P1
BCP1
F1
F2
BCP2
P2
Generation
Generation
Mean
Milk stage
Mid-silking stage
43.7
46.6
17.4
33.3
39.4
42.9
45.9
47.3
47.3
8.1
30.9
32.5
34.2
37.7
38.8
45.4
32.8
5.6
40.9
41.5
42.6
44.7
45.8
Mean
AB
A
C
B
AB
A
A
A
B
B
B
AB
AB
A
B
A
A
A
A
A
BCP2
F1
P1
BCP2
BCP1
F2
P2
F1
P1
P2
BCP1
F2
F1
BCP2
P1
BCP1
P2
F2
F1
BCP2
Generation
Dent stage
40.7
43.4
31.8
29.5
39.0
40.5
40.8
41.2
46.6
61.2
24.1
26.4
32.5
33.0
37.4
40.7
30.0
27.1
36.3
39.2
40.2
41.1
44.0
Mean
AB
A
B
A
A
A
A
A
C
BC
ABC
ABC
AB
A
B
AB
A
A
A
A
BCP2
F1
P1
P2
F2
BCP1
BCP2
F1
F2
P2
P1
BCP1
BCP2
F1
P1
P2
BCP1
F1
F2
BCP2
Generation
Maturity stage
27.3
38.1
78.0
18.8
27.7
7
34.30.6
35.9
41.4
57.6
18.5
18.6
19.4
24.1
29.2
29.9
20.4
26.4
29.9
31.7
31.8
32.8
34.2
Mean
B
A
C
B
B
AB
AB
A
B
B
AB
AB
AB
A
C
AB
A
A
A
A
BCP2
F1
P1
F2
BCP2
P2
BCP1
F1
P1
F2
P2
BCP1
F1
BCP2
P1
BCP1
F1
F2
P2
BCP2
Generation
38.7
41.5
28.9
28.7
37.1
39.2
40.6
40.6
44.6
24.7
26.4
30.3
30.5
31.4
35.5
39.2
28.9
11.2
36.9
38.5
38.8
40.4
40.9
Mean
Average of stages
A
A
C
B
B
AB
AB
A
C
BC
BC
BC
AB
A
B
A
A
A
A
A
Table 3 Least significant differences in pair-wise means comparison tests and heterosis under the HN and LN dosages across the grain-filling stages of six generations in the
maize crosses T20 9 C58 and T20 9 NG8
Euphytica (2013) 189:111–122
115
123
MPH mid-parent heterosis = [(F1-MP)/(MP)] 9 100 (Hill et al. 1998). P1 inbred parent 1, P2 inbred parent 2, F1 single cross between P1 and P2, F2 selfing F1, BCP1 P1 crossed
back to F1, BCP2 P2 crossed back to F1, LCC leaf chlorophyll concentration (SPAD values), HN high nitrogen fertiliser application rate (120 kg N ha-1), LN low nitrogen
fertiliser application rate (60 kg N ha-1)
B
Means with the same letter in the column of each cross and N dosage are not statistically significantly different. a: 0.05, 17 error degrees of freedom for t distribution
B
40.4
27.8
P1
C
19.0
94.6
45.0
P2
C
29.4
P1
C
30.7
17.9
MPH ( %)
P1
31.1
P1
28.0
B
30.3
BCP1
BC
AB
AB
BCP1
33.8
BCP1
34.4
BC
BCP1
P1
C
30.4
20.2
31.3
32.3
F2
P2
BC
BC
25.5
C
BC
P2
BCP1
F2
BC
30.5
37.5
ABC
123
F2
P2
34.6
F2
AB
38.1
P2
Mean
Generation
Mean
Generation
35.9
F2
33.0
Mean
Generation
Generation
Mean
22.7
Mean
Generation
Average of stages
Maturity stage
Dent stage
Milk stage
Mid-silking stage
Table 3 continued
B
Euphytica (2013) 189:111–122
B
116
Genetic effects
Genetic effects were only estimated when the data
showed a clear statistically significant difference of P1
from P2 means based on the LSD0.05 values as required
by GMA model assumptions. Thus an estimation of
genetic effects was only made under high N (HN)
dosage (Table 4). Only positive additive effects were
statistically significant for all traits under all growth
stages in the T20 9 C58. The R2 values were
marginally greater than 50 % at all stages except at
mid-silking and dent stages. The R2 values were C70 %
at the milk stage and mean LCC for the T20 9 NG8.
All the growth stages had statistically significant
positive additive genetic effects. The dominance
genetic effects were statistically significant and
positive at the milk stage. The epistatic genetic effects
of negative additive 9 dominance type were statistically significant at mid-silking, dent and average of
growth stages, while the positive epistatic genetic
effects of additive 9 additive nature were significant
at milk stage.
Ratio of fixable to non-fixable genetic effects
All genetic effects under both N dosages were shown
because of small sums of squares they contributed to
the total sums of squares of the models. The ratio of
fixable (a, aa) to non-fixable (d, ad, dd) genetic effects
of a particular trait within N dosage adds to 100 %
(Table 5). In the T20 9 C58 under the HN dosage, the
fixable effects predominated at all grain-filling stages,
however, the fixable and non-fixable effects were equal
at the physiological maturity stage. In the LN dosage,
both effects prevailed at all stages, except that the
fixable effects were negligible at the dent stage. The
mid-silking and milk stages had equal proportions of
fixable and non-fixable effects. Overall, the ratio of
fixable to non-fixable effects was 73.90:26.10 under
HN dosage and 35.10:64.67 for LN dosage. Whereas
for the T20 9 NG8 under HN dosage, the fixable
effects were only high at the milk stage, at about 71 %;
the rest i.e. about 30 % of the non-fixable effects
predominated at other growth stages. In the LN dosage,
apart from at the mid-silking stage, other generations
had larger non-fixable effects than the fixable genetic
effects. Generally, the ratio of fixable to non-fixable
effects was 37.43:62.57 for the HN dosage and
19.97:80.11 under the LN dosage. The general trends
Euphytica (2013) 189:111–122
indicate that the additive genetic effects decreased as
growth stage increased, while the dominance effects
increased at later growth stages, irrespective of genotypes and N regimes (data not shown).
Narrow sense heritability
The narrow sense heritability (h2) estimates were
highest at the dent (54 %) and milk (62 %) stages for
the high and low N, respectively for the T20 9 C58.
The mid-silking stage had more or less equal estimates
of h2 of about 40 % each under both N regimes. In the
T20 9 NG8 under high N, only the milk and physiological maturity stages had positive h2 but the later
stage had the highest positive h2 of about 46 %. The
dent stage had the highest magnitude of negative h2 of
about 100 %. Only the negative h2 of less than 10 %
was observed under the low N dosage. Clearly the
negative h2 was associated with data where P1 = P2.
The h2 estimates increased from milk stage and
reached 105 % at the physiological maturity stage in
the low N dosage. In both crosses and N conditions,
the h2 estimates for the mean LCC were comparable in
sign and magnitudes. The h2 was positive at 44 %
under high N dosage in the T20 9 C58 and it was
about 54 % also of positive sign in the T20 9 NG8
under low N but only a 20 % negative h2 estimate was
observed for the both crosses and N regimes.
Discussion
Mean separation of generations
The main effects of N regimes and generations except
at mid-silking date in the T20 9 C58 were statistically
significant (Table 2) which allowed to proceeding for
the GMA. The requirement for GMA of contrasting
parents for the LCC character (P1 and P2), based on the
mean separation test (LSD0.05) was only met under
high N for all grain-filling stages in both crosses
(Tables 3a, b). This may suggest that genotypic
differences for the LCC character could be clearer
under the HN than under the LN dosages. The results
revealing LCC peaking at milk stage, irrespective of
cross and N conditions, would suggest that leaf N is
needed the most in maize at this growth stage. This
may be called the linear graining-filling stage; any
stress at this stage would result in irreparable effects,
117
since dry matter accumulation would be reduced and
result in a low final grain yield. These findings
corroborate Duncan et al. (1965), Beauchamp et al.
(1976), and Weiland and Ta (1992) that leaf N must be
maintained at optimum levels if high grain yield is
desired in maize as no any other photosynthetic parts
could compensate for loss of leaf N if stress comes at
the milk stage. Genotypes that maintain higher levels
of the LCC at and past the milk stage under HN and
LN would therefore translate into a higher grain yield
than that of genotypes which quickly lose LCC at
similar conditions. The change in position of generations with genes that are of interest in replenishing
declining LCC with grain-filling age and their consistency with N regimes could demonstrate both adaptive
and performance phenomena of the LCC character.
This was the case between the superior parents, backcrossed progenies to superior parents for the trait, and
F1 generations. And this could be associated with the
phenomenon of heterosis observed in this study which
Kang (1994) referred to as a measure of adaptability of
a genotype. The changing positions of the generations
for LCC could be used as markers or indications that
grain-filling is taking place, further demonstrating the
response of genotypes to N conditions and/or the
status of soil N. Furthermore, the increase of heterosis
with grain-filling stages and poor environments (i.e. at
low N dosages) would prove the predominance of
dominance genetic effects at physiological maturity
(envisaged in next section of this paper) and as the
environment got impoverished. Mather and Jinks
(1982) reported that heterosis levelled off linearly as
the production environment improved, thus partly
confirming the findings of the present study. Regarding the shift of heterosis with age from low at the early
grain-filling stages (implying negligible non-allelic
interactions), to high heterosis at later growth stages,
which suggest a prevalence of non-allelic genetic
effects, no literature is available to strongly support
this finding. The SPAD meter values were consistent
with generations, crosses, and N regimes, which
suggest that the SPAD-502 Chlorophyll Meter was
the appropriate and precise technique to quantify the
trends of LCC character from kernel set and maintenance until physiological maturity. It further demonstrates that the N regimes used for evaluation of the
LCC character were ideal, since HN dosage had
consistently higher SPAD meter mean values than LN
dosage.
123
123
35.66 ± 12.89**
Mean stages
4.13 ± 1.97*
20.57 ± 10.15**
38.93 ± 17.43**
-4.54 ± 16.96
22.89 ± 8.88**
Milk
Dent
Maturity
Mean stages
34.32 ± 20.75
89.05 ± 39.78**
-19.23 ± 40.10
48.52 ± 23.84*
-1.32 ± 35.75
16.97 ± 29.96
34.26 ± 46.63
20.28 ± 37.26
21.22 ± 31.69
5.08 ± 35.53
d
11.27 ± 8.49
27.34 ± 16.11
-10.48 ± 16.65
19.55 ± 9.71**
2.47 ± 14.69
2.57 ± 12.34
2.29 ± 19.26
2.91 ± 15.35
5.45 ± 13.02
1.91 ± 14.62
aa
Genetic effects
-14.97 ± 5.54**
-5.33 ± 10.70
-20.62 ± 10.41*
-8.87 ± 6.45
-32.33 ± 9.55**
-4.19 ± 7.94
-3.12 ± 12.11
5.65 ± 9.84
-5.68 ± 8.45
-11.96 ± 9.48
ad
-12.83 ± 12.94
-42.12 ± 25.11
20.43 ± 24.82
-21.70 ± 14.89
7.82 ± 22.21
-10.26 ± 18.53
-23.05 ± 28.80
-10.23 ± 23.08
-13.52 ± 19.61
-2.52 ± 21.97
dd
76.85
69.34
54.10
71.49
52.15
53.70
54.34
46.06
51.85
38.65
R2 (%)
1.50
4.20
2.85
1.49
2.56
2.25
5.17
2.71
2.05
2.49
CV (%)
***, **, * indicates significant at p B 0.001, 0.01, 0.05, respectively
R2 coefficient of determination, CV coefficient of variation, LCC leaf chlorophyll concentration (SPAD values), m mean of the F2 generation and intercept, a pooled additive
effects (homozygote loci), d pooled dominance effects (heterozygote loci), aa additive 9 additive (homozygote 9 homozygote) interactions, ad additive 9 dominance
(homozygote 9 heterozygote) interactions, dd dominance 9 dominance (heterozygote 9 heterozygote) interactions
5.85 ± 1.17**
7.05 ± 1.39***
5.79 ± 2.25**
40.20 ± 15.29**
6.77 ± 2.15**
5.81 ± 1.70**
Mid-silking
T20 9 NG8
4.65 ± 2.32*
4.72 ± 2.07**
34.32 ± 16.04**
28.26 ± 19.99
Dent
Maturity
6.75 ± 2.11***
6.56 ± 1.83**
36.54 ± 15.23**
37.92 ± 13.68**
a
Milk
m
T20 9 C58
Mid-silking
Growth stage
Table 4 Estimated gene effects of LCC under high nitrogen dosage in the T20 9 C58 and T20 9 NG8 crosses
118
Euphytica (2013) 189:111–122
Euphytica (2013) 189:111–122
Genetic effects
With the T20 9 C58, only the additive genetic effects
conditioned the LCC at all the grain-filling stages
under high N regime. Predominance of additive effects
implied that recurrent (RS) or any form of cyclic
selection could be effective for the LCC at all growth
stages. This supports the idea that the genetic potential
of a genotype is well exploited under ideal production
conditions. It may suggest further that only inbred
lines could be generated at all grain-filling stages from
segregating generations for the LCC under high N.
However, significant differences for only additive
genetic effects for the LCC under HN dosage contradicts Elings et al. (1997), Bänziger and Lafitte (1997),
Bänziger et al. (2000) and Worku et al. (2007), who
reported a preponderance of additive genetic effects in
secondary traits, especially under low N. The LCC has
been associated with the physiological maturity of
genotypes in that the greater the time it takes to reach
maturity, the more the LCC becomes relevant. On the
other hand, shorter maturing genotypes may lose the
LCC earlier and faster under similar conditions. In
short-season areas, therefore, the high LCC would
translate to high yield (Tollenaar and Daynard 1978a,
b), as opposed to long season or varieties with
extended growth. This suggests that early-maturing
or short-season cultivars are source-limited, such that
the extended LCC may increase dry matter accumulation and so result in improved grain yield.
Both fixable (a, and aa) and non-fixable (d, ad, and
dd) genetic effects for the T20 9 NG8 were preponderant under HN dosage. This may suggest that
reciprocal recurrent selection (RRS) could be
employed for the LCC. Epistasis was observed for
the T20 9 NG8, and it could be defined as any
interaction between genes at non-homologous loci
(Sprague et al. 1962). A greater preponderance of
epistatic interactions in the T20 9 NG8 than in the
T20 9 C58 supported the hypothesis that epistasis is
real in maize, although it is specific for this cross, trait
and environment, as opposed to many researchers who
ignore such phenomenon. This further suggests that
positive epistasis could have contributed significantly
to heterosis, which was common in the T20 9 NG8.
Generally, the R2 values were slightly above 50 % and
about 70 % for the T20 9 C58 and T20 9 NG8,
respectively suggesting that the genetic effects for the
LCC character would be more easily detected in and
119
estimated for the latter, rather than the former cross. In
both crosses and at all growth stages for high N regime
the sign of the genetic effects refer to the relative
position of the parents to the mid-parent for the case of
dominance effects plus associated epistatic effects. In
short, this refers to the heterosis (Shashkumar et al.
2010). With regards to the additive genetic effects and
related epistatic effects, the signs imply which parent
was chosen as superior or inferior for the studied traits
(Azizi et al. 2006). The positive sign observed for
additive genetic effects would therefore imply that the
choice of parents was appropriate during mating since
the individual growth stages for the LCC were
considered as separate traits. Viana (2006) reported
the additive 9 additive and additive 9 dominance to
be inestimable, such that their relative importance is
difficult to assess. This assertion could further confirm
the unknowns on the inheritance of the LCC in maize.
The results of the present study do not completely
agree with other reports because no systematic work
has been done to quantify the LCC across grain-filling
stages under high and low dosageses of the N in
tropical maize. Thus, the inheritance of the LCC is not
fully understood but it may be deduced that the LCC is
only estimable under the ideal N conditions.
Ratio of fixable to non-fixable genetic effects
The ratio of fixable to non-fixable genetic effects were
estimated under both N regimes since the genetic
component sums of squares contributed little to total
sums of square for the models. The separation of total
genetic effects into individual genetic effects (i.e.
fixable and non-fixable) for the LCC during the grainfilling stages in maize under high and low N conditions
seem not to be documented in literature. In T20 9
C58, the ratio of fixable to non-fixable effects of
73.90–26.10 % under HN and 35.10–64.67 % for LN
suggested that fixable genetic effects govern the LCC
under high N, although dominance genetic effects
prevailed under low N. There was a ratio of fixable to
non-fixable genetic effects of 37.43–62.57 % for HN
and 19.97–80.11 % for LN in T20 9 NG8. These ratios
did not agree across the two crosses and N dosages
(Table 5). In T20 9 C58, fixable genetic effects predominated at HN dosage whereas non-fixable genetic
effects were large under LN dosage while in
T20 9 NG8, non-fixable effects prevailed under both
regimes of N but with the magnitude of non-fixable
123
120
Euphytica (2013) 189:111–122
Table 5 Ratio of fixable to non-fixable gene effects computed from their relative contribution to the models’ total sums of squares
over growth stages in two crosses under high and low nitrogen dosages
Growth stage
T20 9 NG8
T20 9 C58
Fix.
HN
Non-Fix.
Fix.
LN
Non-Fix.
Fix.
HN
Non-Fix.
Fix.
LN
Non-Fix.
36.87
Mid-silking
84.31
15.68
62.73
37.28
25.76
74.23
62.12
Milk
88.74
11.26
62.75
37.26
70.98
29.03
18.31
81.69
Dent
64.34
35.67
4.06
95.93
28.08
71.92
3.75
96.25
Maturity
51.06
48.94
22.05
77.94
19.34
80.66
0.11
99.89
Mean grain fill stages
81.03
18.97
24.42
75.58
42.97
57.03
15.54
85.86
-1
Fix. fixable, Non-fix. non-fixable, HN high nitrogen fertiliser application rate (120 kg N ha ), LN low nitrogen fertiliser application
rate (60 kg N ha-1)
effects increasing at LN relative to HN. This may
further indicate the difficulty in breeding for the LCC
as Subedi and Ma (2005) and Hawkins et al. (2007)
suggested. The general trend for both crosses and N
dosages suggested that additive and dominance genetic
effects for the LCC decreased and increased respectively across the grain-filling stages.
Narrow sense heritability
Only a few estimates of narrow sense heritability (h2)
were consistent with fixable genetic components in
Table 5. The consistency of h2 with genetic effects
was more evident in T20 9 C58 than in the
T20 9 NG8. Effective breeding strategies could be
expected for the LCC at dent stage under HN, and milk
stage in LN dosage, as the h2 estimates of 54 and 62 %
respectively were observed in the T20 9 C58. Equal
h2 of about 40 % at the mid-silking stage, under both N
conditions, would suggest gain from selection in both
N environments for the trait. In addition, effective
breeding strategy is expected only under the HN
dosage in T20 9 NG8 due to h2 of about 46 % that was
observed at physiological maturity (i.e. SG trait).
Under the low N dosage, the dent, physiological
maturity and average of growth stages would benefit
from selection. The h2 for the mean LCC was 44 % for
T20 9 C58 under HN and 54 % for T20 9 NG8 under
the LN dosage, which further suggests a lack of
consistency for the LCC, particularly under the N
dosages. The inconsistency of h2 estimates evident in
this study could be associated with the polygenic
nature or uncertainty of the inheritance of the LCC.
Furthermore, negative h2 estimates observed in the
123
present study could partly be attributed to cases where
P1 equalled P2, especially under low N conditions
where GMA assumptions could not hold. Unusual
estimates of h2 have been reported in literature.
Robinson et al. (1955) reported that negative values
of h2 estimates can be assumed to be zero, whereas
Dudley and Moll (1969) and Robinson et al. (1951)
cited by Amand and Wehner (2001) proposed that
such estimates they be presented for reference. Other
researchers reported that these negative estimates
could be due to the effects of small sample size,
environment, and used techniques.
Conclusion
Generally, it was established that different gene effects
govern the LCC at different grain-filling stages and
this could be a function of the examined genotypes, N
dosage, time of maturity and the interaction of these
factors. However, this study failed many of the
assumptions of requirements of the GMA. Some of
the assumptions of GMA are lack of dominance and
epistasis, which prevailed in this study. The additive
and non-additive genetic effects increased and
decreased consistently with mid-parent heterosis
across age and N, regardless of the genotype. And
this was expressed better under low N than under high
N, pointing to the possibility of taking advantage of the
heterosis for LCC under low N condition leaving
development of inbred lines at the early grain-filling
stages. Inconsistency of heritability estimates for the
LCC with other genetic parameters confirmed that h2 is
specific to the genotype, environment (i.e. N) and trait.
Euphytica (2013) 189:111–122
Acknowledgments The first author acknowledges the
Rockefeller Foundation through an Alliance for a Green
Revolution in Africa for funding; the University of KwaZuluNatal for the scholarship and the Government of the United
Republic of Tanzania for supportive environment during the study.
References
Allard RW, Bradshaw AD (1964) Implications of genotypeenvironment interactions in applied plant breeding. Crop
Sci 4:503–508
Amand PC, Wehner TC (2001) Generation mean analysis of leaf
and stem resistance to gummy stem blight in cucumber.
J Am Soc Hort Sci 126(1):95–99
Azizi F, Rezai AM, Saeidi G (2006) Generation means analysis
to estimate genetic parameters for different traits in two
crosses of corn inbred lines at three planting densities.
J Agric Sci Technol 8:153–189
Bänziger M, Lafitte HR (1997) Efficiency of secondary traits for
improving maize for low-nitrogen target environments.
Crop Sci 37:1110–1117
Bänziger M, Edmeades GO, Beck D, Bellon M (2000) Breeding
for drought and nitrogen stress tolerance in maize: from
theory to practice. CIMMYT, Mexico, p 68
Beauchamp EG, Kannenberg LW, Hunter RB (1976) Nitrogen
accumulation and translocation in corn genotypes following silking. Agron J 68:418–422
Bertin P, Gallais A (2000) Genetic variation for nitrogen use
efficiency in a set of recombinant maize inbred lines I.
Agrophysiological results. Maydica 45:53–66
Cavalieri AJ, Smith OS (1985) Grain filling and field drying of
maize hybrids released from 1930–1982. Crop Sci 25:
856–860
Checa O, Ceballos H, Blair MR (2006) Generation means
analysis of climbing ability in common beans (Phaseolus
vulgaris L.). J Hered 94:456–465
Dudley JW, Moll RH (1969) Interpretation and use of estimates
of heritability and genetic variance in plant breeding. Crop
Sci 9:257–262
Duncan WG, Hatfield AL, Rangland JL (1965) The growth and
yield of corn. Agron J 57:221–223
Earl HJ, Tollenaar M (1997) Maize leaf absorptance of photosynthetically active radiation and its estimation using
chlorophyll meter. Crop Sci 37:436–440
Elings A, White JW, Edmeades GO (1997) Options for breeding
for greater maize yields in the tropics. Eur J Agron
7:119–132
Gamble EE (1962) Gene effects in corn (Zea mays L.). I. Separation and relative importance of gene effects for yield. Can
J Plant Sci 42:339–348
Hawkins JA, Sawyer JL, Barker DW, Lundvall JP (2007) Using
relative chlorophyll meter values to determine N application rates for corn. Agron J 99:1034–1040
Hill J, Becker HC, Tigerstedt PMA (1998) Quantitative and
ecological aspects of plant breeding. Chapman & Hall,
London, pp 50–51
Joshi AK, Kumari M, Singh VP, Reddy CM, Kumar S, Rane J,
Chand R (2007) Stay green trait: variation, inheritance and
121
its association with spot blotch resistance in spring wheat
(Triticum aestivum L.). Euphytica 153:59–71
Kang MS (1994) Applied quantitative genetics. Department of
Agronomy, Louisiana State University, p 157
Khanna-Choppra R, Maheswari M (1998) Effect of altering
source availability on expression of sink capacity in a
maize hybrid and its parents. Eur J Agron 9:101–107
Lafitte HR (1994) Identifying production problems in tropical
maize: a field guide. CIMMYT, Mexico, p 122
Lande R (1981) The minimum number of genes contributing to
quantitative variation between and within populations.
Genetics 99:541–553
Martinez DE, Guiamet JJ (2004) Distortion of the SPAD 502
chlorophyll meter readings by changes in irradiance and
leaf water status. Agronomie 24:41–46
Mather K, Jinks JL (1977) Introduction to biometrical genetics.
Cornell University Press, Ithaca, p 291
Mather K, Jinks JL (1982) Biometrical genetics. Chapman and
Hall, London, p 396
Rajcan I, Tollenaar M (1999) Source: sink ratio and leaf
senescence in maize: II. Nitrogen metabolism during grain
filling. Field Crops Res 60:255–265
Robinson RA (1996) Return to resistance: breeding crops to
reduce pesticide dependence. AgAccess/International
Development Research Centre, Davis, p 569
Robinson HF, Comstock RE, Harvey PH (1951) Genetic and
phenotypic correlations in corn and their implication in
selection. Agron J 43:282–287
Robinson DC, Comstock RE, Harvey PH (1955) Genetic variances in open pollinated corn. Genetics 40:45–60
Robson P, Donnison I, Thorogood D, Cowan S, Ougham H,
Thomas H (2001) New ways to stay green. IGER innovations, vol 5. Sports Turf Research Institute, Bingley,
pp 12–15
Shashkumar RT, Pichaimuthu MR, Rawal D (2010) Generation
mean analysis of resistance to downy mildew in adult
muskmelon plants. Euphytica 173:121–127
Smith CW, Braden CA, Hequet EF (2009) Generation mean
analysis of near-long-staple fibre length in TAM 94L-25
upland cotton. Crop Sci 49:1638–1646
Sprague GF, Russell WA, Penny LH, Horner TW, Hanson WD
(1962) Effects of epistasis on grain yield in maize. Crop Sci
2:205–208
Steel GD, Torrie JH (1980) Principles and procedures of statistics: a biometrical approach. McGraw-Hill Book Co.,
New York, pp 56–78
Subedi KD, Ma L (2005) Nitrogen uptake and partitioning in
stay-green and leafy maize hybrids. Crop Sci 45:740–747
Thomas H, Smart CM (1993) Crops that stay green. Ann Appl
Biol 123:193–219
Tollenaar M, Daynard TB (1978a) Leaf senescence in short
season maize hybrids. Can J Plant Sci 58:869–874
Tollenaar M, Daynard TB (1978b) Relationship between
assimilate source and reproductive sink in maize grown in a
short season environment. Agron J 70:219–223
Viana JMS (2006) Generation mean analysis in relation to
polygenic systems with epistasis and fixed genes. Pesq
Agropec Bras 35:1159–1167
Warner JM (1952) A method for estimating heritability. Agron J
44:427–430
123
122
Weiland R, Ta TC (1992) Allocation and retranslocation of 15N
by maize (Zea mays L.) hybrids under field conditions of
low and high N fertility. Can J Plant Physiol 19:77–88
Willman MR, Below FE, Lambert RJ, Howey AE, Mies DW
(1987) Plant traits related to productivity of maize.
I. Genetic variability, environmental variation, and correlation with grain yield and stalk lodging. Crop Sci 27:
1116–1121
123
Euphytica (2013) 189:111–122
Worku M, Bänziger M, auf’Erley GS, Friesen D, Diallo AO,
Horst WJ (2007) Nitrogen uptake and utilization in contrasting nitrogen efficient tropical maize hybrids. Crop Sci
47:519–528
Wright S (1968) The genetics of qualitative variability. In:
Wright S (ed) Evolution and genetics of populations.
Genetic and biometric foundations, vol I. University of
Chicago Press, Chicago III, pp 371–420