Field Crops Research 198 (2016) 215–225

Contents lists available at ScienceDirect

Field Crops Research

journal homepage: www.elsevier.com/locate/fcr

Maize grain yield components and source-sink relationship as

affected by the delay in sowing date
Lucas E. Bonelli a,b,∗ , Juan P. Monzon b , Anibal Cerrudo a,c , Roberto H. Rizzalli a ,
Fernando H. Andrade a,b,c
a
Facultad de Ciencias Agrarias (UNMdP), Ruta 226 km 73.5, 7620 Balcarce, Argentina
b
Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Argentina
c
Instituto Nacional de Tecnología Agropecuaria (INTA), Argentina

a r t i c l e i n f o a b s t r a c t

Article history: Delaying maize (Zea mays L.) sowing date can diminish grain yields through reductions in the number, size
Received 7 June 2016 and activity of growing grains (sink strength) and/or reductions in the assimilate supply (source capacity)
Received in revised form 30 August 2016 to grains during the grain filling period. Whether the source capacity or the sink strength is the limiting
Accepted 1 September 2016
factor for grain yield in late sown maize still remains unclear. Understanding source-sink relationships is
relevant to optimize crop management practices, to identify critical processes for crop modelling and to
Keywords:
develop breeding strategies. The objective of this work was to assess the effect of delays in maize sowing
Sowing date
date on grain yield components and on the source-sink relationship during the grain filling period. Three
Maize
Potential grain yield
well irrigated and fertilized maize field experiments were conducted at Balcarce, Argentina (37◦ 45 S,
Source-sink relationships 58◦ 18 W; 130 m a.s.l.) during 2009–10; 2010–11 and 2011–12 cropping seasons. Sowing dates ranged
Radiation use efficiency from October to January covering a broad range of the seasonal photo-thermal variation. Grain yield was
affected by sowing date and varied from 1680 g m−2 (early sowings) to 203 g m−2 (late sowings). Grain
number per unit area was reduced proportionally less than weight per grain as sowing date was delayed.
Variations in grain yield were related to the harvest index, and were closely associated with dry matter
accumulation during the post-silking period. The variation of source capacity was higher than that of
sink strength during the grain filling period and the source/sink ratio decreased from early to late sowing
dates. Results indicate that crop growth during the grain filling period was limited by the sink strength
in early sowing dates and by the photosynthetic source capacity in the late ones.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction and temperature change considerably around the beginning and

the end of the maize cropping season (Shaw, 1988; Wilson et al.,
Sowing date is one of the main management practices used to 1995).
adjust the timing of occurrence of crop phenological phases and Variations in grain yield can be analyzed in terms of the crop
therefore, to determine the environmental conditions under which carbon economy during the grain filling period. This approach, com-
crops develop and grow. Changes in maize sowing date alter crop monly described as source-sink relationship, aims to identify when
growth rate and the length of crop phenological phases which, in grain yield is limited by the supply (source capacity) or by the
turn, modify potential grain yield and its components (Cirilo and demand (sink strength) of assimilates during the grain filling period
Andrade, 1994a). These alterations are particularly significant in (Tollenaar, 1977). In maize, source capacity is mainly determined
cool-temperate short season environments, where solar radiation by assimilate production by crop photosynthesis during the grain
filling period. Sink strength is defined by the ability of the grow-
ing grains to accommodate these assimilates. Maize grain yield has
been generally reported as to be sink limited (Tollenaar and Lee,
Abbreviations: ECG, end of crop growth; RUE, radiation use efficiency; TT, ther-
mal time; PAR, photosynthetically active radiation; EGFP, effective grain filling 2011; Westgate et al., 2004). As a consequence, most recommended
period; CGR, crop growth rate; GGR, grains growth rate; HI, harvest index; fi, fraction management practices (Andrade et al., 2005) and major breeding
of intercepted PAR; RMSE, root mean square error. efforts in maize (Echarte et al., 2000; Tollenaar et al., 1992) have
∗ Corresponding author at: Facultad de Ciencias Agrarias (UNMdP), Ruta 226 km
been focused in maximizing the number of grains per unit area.
73.5, Balcarce 7620, Argentina.
E-mail address: [email protected] (L.E. Bonelli).

http://dx.doi.org/10.1016/j.fcr.2016.09.003
0378-4290/© 2016 Elsevier B.V. All rights reserved.
216 L.E. Bonelli et al. / Field Crops Research 198 (2016) 215–225

In temperate and cool-temperate environments, when maize two replications. Treatments consisted of seven sowing dates rang-
sowing is delayed, crop flowering occurs far after the summer ing from November to January. Plots were four rows wide by 10 m
solstice and grain filling period takes place close to the end of long. In Exp. 2 and C. Exp. I, the hybrid DK692 was tested in a ran-
the cropping season. Thus, as sowing date is delayed in these domized complete-block design with four replications. Treatments
environments, both the critical period for grain set and the grain consisted of four contrasting sowing dates ranging from October to
filling period are subjected to a progressive deterioration of photo- January. In Exp. 3, hybrids P39B77, I550 and DK692 were tested in
thermal conditions for crop growth. In accordance, delays in maize a split-plot randomized design with four replications. Treatments
sowing date can diminish grain yields through reductions in: i) the involved the factorial combination of four contrasting sowing dates
number, size and activity of growing grains and/or ii) the assimilate (main plot) and the three abovementioned hybrids (subplot). In C.
production by photosynthesis during the grain filling period (Cirilo Exp. II, a total of 16 commercial hybrids sown on a single sowing
and Andrade, 1996, 1994b; Tsimba et al., 2013a). date (Oct-3) were tested in a randomized complete-block design
Information about grain yield components and source-sink rela- with three replications. Plots in Exp. 2, Exp. 3, C. Exp I and C. Exp
tionships during the grain filling period helps in the selection and II were nine rows wide by 12 m long. Sowing dates and hybrids of
optimization of key crop management practices such as hybrid three main experiments are detailed in Table 1.
selection, plant density, irrigation strategies, crop protection, etc.
(Andrade et al., 2005). This information is also relevant to iden- 2.3. Measurements and estimations
tify critical processes for crop modelling and to develop breeding
strategies (Hall and Sadras, 2009). Previous investigations on maize 2.3.1. Crop phenology
sowing date have described grain yield components and, to some Crop phenological stages were determined according to Ritchie
extent, indicative variables of the source-sink relationship during et al. (1989). Number of expanded leaves was weekly recorded
the grain filling period. However, whether the source capacity or from emergence to VT in five plants per plot. Leaves number 5 and
the sink strength is the limiting factor to grain yields in late sown 10 were paint-marked to keep a leaf number reference. Duration
maize still remains unclear. The objective of this work was to assess of phenological stages was expressed in days and in thermal time
the effect of sowing date delays on maize grain yield components (TT) units. Daily TT degrees were calculated as follows (Eq. (1)) and
and on the source-sink relationship during the grain filling period. accumulated between phenological stages:
◦  Tmax + Tmin
2. Materials and methods TT Cd−1 = − Tb (1)
2

2.1. Site description and crop management where Tmax and Tmin are daily maximum and minimum temper-
atures, respectively. The base temperatures (Tb) for TT estimations
Three experiments were conducted at INTA-Balcarce experi- were previously estimated following the methodology descripted
mental station, Argentina (37◦ 45 S, 58◦ 18 W; 130 m a.s.l.) during by Tsimba et al. (2013b). Briefly, a Tb-iterative procedure was per-
2009–10 (Exp. 1), 2010–2011 (Exp. 2) and 2011–2012 (Exp. 3) formed to find the lowest TT CV% for each of the main phenological
cropping seasons. Data of two complementary experiments con- phases of each hybrid across sowing dates and growing seasons. The
ducted during 2012–13 (C. Exp. I) and 2011–12 (C. Exp. II) were most accurate Tb did not differ from 10 ◦ C (for emergence to silk-
also included in an analysis that merged all observed grain yields. ing period) and 0 ◦ C (for silking to physiological maturity). These Tb
Experiments were established on a fine-loamy typic argiudol with were in agreement with those proposed by Tollenaar et al. (1979)
an effective depth of 1.5 m. Topsoil organic matter was 5.6% and for the pre-silking period and by Birch et al. (1998) and by Muchow
extractable phosphorous content was 32 ppm. The field was man- (1990) for the post-silking period.
aged under conventional tillage in Exp. 1 and under no-till in Exp. Grain samples were taken every 7–10 days starting at 10 days
2, Exp. 3, C. Exp. I and C. Exp II. Wheat (Triticum aestivum L.) was after 50% silking and ending at least three weeks after black layer
the previous crop in all instances. formation. At each sampling date 15 grains were taken from the
Plots were over-sown with hand planters and were thinned central portion of two ears per plot. A total of 8 different plants (two
to a uniform plant population immediately after V2 crop phe- plants by 4 blocks) were randomly selected at each sampling date.
nological stage (Ritchie et al., 1989). Final plant population was Grains were dried at 90 ◦ C for 7–10 days in an air forced oven and
8.7 plants m−2 in Exp.1 and 10 plants m−2 in Exp. 2, Exp. 3, C. Exp. I weighed. Mean growth rate per grain and duration of effective grain
and C. Exp II. Rows were 0.52 m apart. Soil water content was kept filling period (EGFP) were estimated following the lineal-plateau
above 65% of the maximum soil available water by complementing fitting method (in which the weight per grain is expressed as a
precipitations with sprinkler irrigation. In order to provide an ade- function of the days after silking) described by Egli (1998). Physi-
quate mineral nutrition, plots were fertilized before sowing with ological maturity1 was estimated as the day when the maximum
30 kg of P ha−1 (diammonium phosphate). A mix of 50 kg of S ha−1 weight per grain was achieved.
(calcium sulfate) and 400 kg N ha−1 (urea) was also applied in three
equal splits at i) emergence, ii) V5 and iii) R1 crop phenological 2.3.2. Dry matter accumulation, PAR interception and grain yield
stages. Potassium requirement was assumed to be covered by the Aboveground dry matter was periodically measured in Exp. 2
natural abundance of K+ in the illite-rich soils where experiments and Exp. 3 by taking samples of 8–10 plants (sample area of about
were conducted. Weeds, insects, and diseases were effectively con- 1 m2 ) from the five central rows. A total of 5–6 samples were taken
trolled. every 15–25 days during each crop cycle. Three to four plants were
left as border between successive samples. Aboveground dry mat-
ter in Exp. 1 was determined at harvest from 10 plants (sample area
2.2. Plant material and experimental design

Three commercial hybrids were used: P39B77 Herculex®

1
LibertyR® (relative maturity rating: 92; Pioneer, Arg.), I550 MGRR2 The “physiological maturity” term in the current work is used to refer to the
moment when grains stopped growing, even for those cases of premature cessation
(relative maturity rating: 102; Illinois, Arg.) and DK692 MGRR2 of grain growth. We also used the term “end of crop growth” (ECG) to refer to the
(relative maturity rating: 119; DeKalb, Arg.). In Exp. 1, the hybrid moment at which crop aboveground dry matter accumulation ceased, which may
P39B77 was tested in a randomized complete-block design with not be coincident with physiological maturity (Uhart and Andrade, 1991).
 L.E. Bonelli et al. / Field Crops Research 198 (2016) 215–225 217

Table 1
Duration of crop phenological stages, mean temperature and final number of leaves of contrasting maturity maize hybrids at different sowing dates in Balcarce (37◦ 45 S,
58◦ 18 W). Durations are expressed in days (d) and thermal time units (TT). Base temperatures for thermal time estimations were 10 ◦ C for emergence-silking period and
0 ◦ C for silking-physiological maturity period.

Exp. Sowing date Hybrid Emergence to silking period Silking to Phys. Maturity period Final
leaf
number

duration mean duration mean

temp. temp.
◦ ◦
d TT C d TT C

1 (2009–10) Nov-16 P39B77 56 553 19.9 50 1048 21.4 17.0

Dec-11 P39B77 48 568 22.4 59 1123 19.1 17.0
Dec-22 P39B77 47 561 22.6 54 974 18.0 17.5
Dec-29 P39B77 46 544 22.4 58 1020 17.6 17.0
Jan-05 P39B77 48 533 21.5 55 936 17.0 17.0
Jan-12 P39B77 49 540 21.4 45 756 16.8 17.0
Jan-20 P39B77 51 533 20.7 36 590 16.4 16.5
Sowing date *** *** – *** *** – **
(1.7) (19.6) (2.2) (34.9) (0.3)

1 (2010–11) Oct-4 DK692 92 770 18.1 58 1228 21.4 21.4

Nov-2 DK692 72 731 20.1 55 1139 21.0 21.9
Dec-1 DK692 66 728 21.4 60 1167 19.6 21.8
Jan-3 DK692 61 683 21.4 67 1070 16.0 20.2
Sowing date *** *** – *** *** – **
(1.9) (23) (1.9) (32) (0.6)

3 (2011–12) Oct-4 P39B77 75 618 17.8 50 1109 23.3 17.2

I550 82 702 18.3 52 1133 22.7 19.2
DK692 86 752 18.6 57 1233 22.3 21.0

Nov-14 P39B77 54 555 20.6 49 1059 22.4 17.6

I550 60 634 21.0 53 1140 22.1 19.3
DK692 67 713 21.0 59 1220 21.2 21.0

Dec-15 P39B77 50 596 22.7 50 975 19.8 17.9

I550 57 678 22.8 60 1120 18.7 20.0
DK692 60 719 22.9 59 1085 18.4 21.8

Jan-4 P39B77 51 600 22.5 59 1012 17.2 17.4

I550 58 682 22.4 54 878 16.3 18.6
DK692 65 762 22.3 69 1006 14.6 20.8
Sowing date *** *** – *** *** – ***
(0.31)
Hybrid *** *** – *** *** – ***
(0.26)
Sowing date × Hybrid *** *** – *** *** – NS
(2) (24.5) (2.4) (40.6)

ANOVA p-values: *p < 0.05, ** p < 0.01, *** p < 0.001.

NS: not significant.
LSD values for mean comparison are presented in brackets.

of about 1.15 m2 ) of two central rows per plot. After each sampling, to obtain the total intercepted PAR by the crop. Mean radiation
plants were immediately oven dried (60–70 ◦ C) to constant weight. use efficiency (RUE) was calculated as the ratio between above-
Weather records (daily temperatures and solar radiation) were ground dry matter and intercepted PAR accumulated during the
obtained from a meteorological station sited at 500 m from the same period.
experimental site. Daily incident PAR was obtained by multiplying Grain yield was determined after grain black layer formation
solar radiation by 0.48 (Monteith, 1972). Photosynthetically active by hand harvesting all ears from a minimum area of 4 m2 per plot.
radiation (PAR) interception was registered every 10–15 days dur- Harvest index (HI) was estimated as the ratio between grain yield
ing each crop cycle. The fraction of intercepted PAR (fi) was and total aboveground dry matter at harvest. Final mean weight
calculated as follows (Eq. (2)): per grain was determined at harvest for each grain yield sample by
weighting a 500 grains sub-sample. Aboveground dry matter, grain
It
fi = 1 − (2) yield and mean weight per grain were expressed on a dry-weight
I0
basis (0% moisture content).
where, It is the incident radiation just below the lowest green layer
of leaves and I0 is the incident radiation at the top of the canopy. 2.3.3. Crop growth rate and source-sink relationship
The values for It and I0 were obtained with a linear 1 m-long PAR Actual and potential crop growth rate (CGR) were estimated
ceptometer (Cavadevices Bar-Rad, Arg.) horizontally and perpen- for the EGFP. Actual CGR was estimated from dry matter measure-
dicular to the rows. Five determinations per plot were made at ments following Eq. (3). Potential CGR was performed by adapting
each measurement date. PAR measurements were confined to the the net canopy photosynthesis model of Monteith (1972). Mon-
solar midday (1200–1300 h) on sunny days only. Daily fraction teith’s adapted model was based on the daily incident PAR (Mj m−2 ),
of PAR interception was obtained by linear interpolation within the measured daily fraction of intercepted PAR (fi), and the poten-
measurement dates. Daily incident PAR was multiplied by the cor- tial RUE (g of aboveground dry matter per Mj of intercepted PAR)
responding daily fraction of PAR interception and accumulated following Eq. (4). In Eqs. (3) and (4) DM indicates the aboveground
218 L.E. Bonelli et al. / Field Crops Research 198 (2016) 215–225

|Si  | and |Oi  | indicate the absolute distance between Si and Ō, and
between Oi and Ō, respectively. This coefficient can take values from
0 to 1 indicating the degree to which the observed data is accurately
estimated by simulation (a value of 1 indicates a perfect agreement
and a value of 0 a complete disagreement between observed and
simulated CGR).
An approximation to the source-sink relationship during the
EGFP was performed for treatments of Exp. 3. To enable the com-
parison among treatments that differed in the length EGFP, values
of source capacity and sink strength were expressed in a daily rate
basis (g m−2 d−1 ). Sink strength was calculated as the mean grain
growth rate (GGR) by multiplying the mean growth rate per grain
during the EGFP by the number of grains per unit area at harvest.
Source capacity was based on the potential CGR for the same period.
This methodological approach was carried out to avoid the poten-
tial autocorrelation between source and sink estimations caused by
the negative feedback of an insufficient sink demand on photosyn-
thesis (Paul and Foyer, 2001).
Fig. 1. Observed vs. simulated potential maize crop growth rates (CGR). Observed
values were taken from well irrigated and fertilized maize crops at Balcarce (37◦ 2.4. Data analysis
45 S, 58◦ 18 W) during the period around 30 days bracketing silking. Data are
from Exp. 2 and Exp.3 of present work (hybrids P39B77, I550 and DK692) and from Data were analyzed through descriptive statistics, regressions
sowing date maize experiments conducted by Cirilo and Andrade (1994a) (hybrid
and ANOVA procedures using the R statistical software (RDC Team,
DK636). All these observed CGR data came from crops with near full interception
(>95%) of photosynthetically active radiation. Root mean square error (RMSE) and 2008). Lineal model for Exp. 1 and Exp. 2 was as follows (Eq. (8),
Index of agreement (u) indicate simulation accuracy in absolute and relative terms, adapted from Kuehl (2000)):
respectively.
yij =  + i + j + eij (8)

dry matter in a specific moment, a and b indicate the beginning where yij is the observed response of the variable in the ith -sowing
and the end of the EGFP, respectively and n indicates the length (in date and jth -block,  is the general mean,  i is the effect of sowing
days) of the EGFP. date, j is the mean of the jth -block deviation from the general mean
  and eij is the experimental error. Lineal model for Exp. 3 was as
DMb − DMa
Actual CGR g m−2 d−1 = (3) follows (Eq. (9)):
n  
yijk =  + ˛i + k + dik + ˇj + ˛ˇ + eijk (9)
  1
b ij
−2 −1
Potential CGR gm d = PAR × fi × RUE (4)
n where yijk is the observed response of the variable in the ith -sowing
a
date of the jth -hybrid located in the kth -block,  is the general mean,
Potential RUE was estimated as a function of daily mean temper- ˛i is the sowing date effect, k is the mean of the kth -block deviation
ature (Tmean) according to Andrade et al. (1993) following Eq. (5). from the general mean, dik is the experimental error of the main
RUE estimations were confined to the range [6.7; 22.2] as lower and plot, ˇj is the hybrid effect, (˛ˇ)ij is the interaction effect between
upper temperature limits, respectively, according to the tempera- sowing date and hybrid and eijk is the experimental error of the sub-
ture range used in CERES-Maize (López-Cedrón et al., 2005). Thus, plot. Sowing dates and hybrids were always considered as fixed
for daily mean temperatures ≤6.7 ◦ C or ≥22.2 ◦ C constant values of effects. Additivity was assumed for all lineal effects. Appropriate
0 and 4.2 g Mj−1 were computed, respectively. standard errors according to Kuehl (2000) were calculated from
  residual sum of squares of each ANOVA. Estimated parameters were
RUE g MJ−1 = 0.27 × Tmean − 1.8 (5) compared using t-tests (␣ = 0.05).

The model was evaluated by comparing observed (Oi ) vs. simu-

3. Results
lated (Si ) CGR values of crops growing under contrasting radiative
and thermal conditions during the period of about 30 days brack-
3.1. Meteorological conditions and crop development
eting silking (Fig. 1). Model accuracy was described using the root
mean square error (RMSE) and the index of agreement (u) defined
The occurrence of crop phenological stages along the cropping
by Willmott (1981). RMSE was calculated following Eq. (6) and was
seasons is shown in Fig. 2. The length of the sowing-emergence
used to describe the average error of simulations in absolute units
period was between 10–12 d for October, 7–9 d for November and
(g m−2 d−1 ).
4–6 d for December and January sowing dates, in all experiments.

 n Crops developed from mid-October to late April covering a broad
1
RMSE =  2 range of thermal and radiative conditions. Frost free periods were
(Si − Oi ) (6)
n 187, 187 and 169 days long for 2009-10; 2010–11 and 2011–12
i=1
seasons, respectively.
Index of agreement (u) was used to describe the model accuracy Frost events that occurred early in the cropping seasons dur-
in relative units and was calculated as follows (Eq. (7)): ing Exp. 2 and Exp. 3 damaged the leaves of October sowing dates
(Fig. 2), affecting PAR interception only temporarily. Frost events
n
i=1 (S i
− Oi )2 that occurred late in the cropping season determined the cessation
u=1− n  2 (7) of crop growth during the reproductive period for most December
|Si | + |Oi |
i=1 and January sowing dates. These autumn frosts essentially acted as
 L.E. Bonelli et al. / Field Crops Research 198 (2016) 215–225 219

Fig. 2. Time-course of mean air temperature, incident photosynthetically active radiation (PAR) and photoperiod during maize cropping season at Balcarce (37◦ 45 S, 58◦
18 W). Vertical grey lines for temperature and PAR indicate the interquartile distance (i.e. the range between the 25th and 75th percentiles) of daily meteorological data
(1982–2011 period). Boxes represent the length of maize phenological periods from sowing to emergence (S-VE ), emergence to silking (VE -R1 ) and silking to the end of crop
growth moment (R1 -ECG). Left side numbers in each box indicate different hybrids (*1 for P39B77, *2 for I550 and *3 for DK692). Cross marked boxes indicate frost-free
periods for each cropping season obtained from empirical observation in the field. Recorded minimum temperature in the weather station (at 1.5 m in the meteorological
shelter) varied from 2.8 to 3.4 ◦ C for each frost event in the field. Boxplot diagrams indicate the distribution of extreme frost dates for each year in the period 1982–2011
(daily minimum temperature <3 ◦ C).

a defoliation since leaves died and grain filling continued, probably to the duration of the emergence-silking (p < 0.001; R2 = 0.82) and
at the expense of sugar translocation from stems. silking-physiological maturity periods (p = 0.002; R2 = 0.37). The
Average values of daily mean temperature and solar radiation length of the emergence-silking period expressed in TT units was
during the reproductive period of each sowing date treatment were related to the final number of leaves (p < 0.001; R2 = 0.87) (Table 1).
plotted together with those corresponding to representative maize With the exceptions of treatments in which the grain filling period
systems of contrasting environments around the world (Fig. 3). was prematurely interrupted, time (in days) from silking to phys-
Average values of early sowing dates of maize at Balcarce resem- iological maturity of late sowings was similar or longer than that
ble those at temperate and high productive maize regions whereas, of early sowings (Table 1). Changes in sowing date exposed the
the average values of late sowing dates at Balcarce were similar to photoperiod-sensitive stages of maize (estimated to be from V4 to
those at cool-temperate and less productive maize regions. V6 ) to contrasting photoperiods (Fig. 2) which ranged from around
Table 1 shows the duration of the main phenological phases for 13.9–15.8 h d−1 . Despite differences in the number of leaves in
each crop. The length of the EGFP (not directly shown in Table 1) response to photoperiod (Table 1), TT from emergence to silking
can be calculated by subtracting the length of the “lag phase” to was not significantly affected by sowing date (p > 0.34).
the silking-physiological maturity period. Duration of the lag phase
varied as a function of the flowering time and lasted 10, 13 and
3.2. Grain yield components and crop growth determining factors
15 days when crops flowered during January, February and March,
respectively. The emergence-silking period expressed in days was
Grain yields ranged from 1680 to 203 g m−2 and were signifi-
progressively reduced with the delay in sowing date but it was rela-
cantly affected by hybrid and sowing date (Table 2). Grain yields
tively constant for each hybrid when expressed in TT units (Table 1).
were directly associated with hybrid cycle length for October
Total cycle length (in days) differed among hybrids and was related
and November sowings (Exp. 3; p < 0.001; R2 = 0.86). This positive
 220
Table 2
Grain yield components and crop growth determining factors of contrasting maturity maize hybrids at different sowing dates in Balcarce (37◦ 45 S, 58◦ 18 W).

Exp. Sowing Hybrid Grain Yield Grain number Mean weight Emergence to silking period Silking to ECG period Harvest
Date at harvest per grain at index
harvest

Accumulated Mean fraction RUE Accumulated Mean fraction RUE

incident PAR of PAR incident PAR of PAR
interception interception
g m−2 GN m−2 mg Mj m−2 g Mj−1 Mj m−2 g Mj−1

1 (2009–10) Nov-16 P39B77 1089 4313 253 599 0.41 – 464 0.88 – 0.53
Dec-11 P39B77 1019 3974 256 510 0.32 – 461 0.87 – 0.55
Dec-22 P39B77 904 3457 261 495 0.46 – 401 0.86 – 0.47
Dec-29 P39B77 759 3398 223 483 0.46 – 393 0.90 – 0.44
Jan-5 P39B77 608 3301 184 475 0.48 – 349 0.92 – 0.39
Jan-12 P39B77 417 3059 136 458 0.48 – 281 0.94 – 0.32

L.E. Bonelli et al. / Field Crops Research 198 (2016) 215–225

Jan-20 P39B77 203 2773 73 452 0.49 – 214 0.94 – 0.19
Sowing date * *** *** *** *** *** – ***
(867) (15) (11) (0.03) (9) (0.03) (0.02)

2 (2010–11) Oct-4 DK692 1680 6522 258 918 0.42 3.6 575 0.92 3.8 0.51
Nov-2 DK692 1498 6157 243 826 0.44 3.8 517 0.93 3.6 0.49
Dec-1 DK692 1368 4933 278 745 0.56 3.8 431 0.96 3.5 0.44
Jan-3 DK692 579 4376 132 633 0.56 3.5 338 0.95 2.7 0.29
Sowing date *** *** *** *** NS *** NS ** ***
(536) (9) (26) (0.05) (14) (0.5) (0.01)

3 (2011–12) Oct-4 P39B77 1250 4682 267 744 0.29 3.5 544 0.94 2.7 0.55
I550 1350 5604 240 832 0.38 3.6 536 0.96 3.0 0.49
DK692 1490 5570 268 875 0.39 3.6 558 0.95 3.2 0.50

Nov-14 P39B77 1240 4866 254 590 0.46 3.5 492 0.95 2.9 0.50
I550 1320 6102 216 654 0.54 3.6 499 0.97 2.8 0.47
DK692 1440 5538 260 726 0.55 3.8 524 0.95 2.8 0.50

Dec-15 P39B77 1110 4583 242 561 0.35 5.0a 414 0.94 2.8 0.49
I550 1120 5252 214 634 0.43 4.8a 407 0.97 2.8 0.48
DK692 1100 5160 213 654 0.43 4.7a 388 0.96 2.9 0.45

Jan-4 P39B77 770 4219 182 511 0.53 3.8 339 0.95 2.2 0.46
I550 720 4959 145 570 0.54 3.7 279 0.95 2.1 0.44
DK692 630 4361 144 623 0.58 3.7 226 0.95 2.0 0.36
Sowing date *** *** *** *** *** *** NS *** ***
(281) (0.06) (0.3) (0.01)
Hybrid *** *** *** *** NS *** NS NS ***
(190) (0.03)
Sowing date × Hybrid NS *** *** NS NS *** NS *** ***
(14) (22) (15) (0.4) (0.02)

ANOVA p-values: * p < 0.05, ** p < 0.01, *** p < 0.001.

NS: not significant.
LSD values for means comparison are shown in brackets.
a
These RUE values may be overestimated. A heavy storm damaged temporarily canopy structure of these crops at V12 –V14 phenological stage (35 d after emergence). Consequently, some of the midday PAR interception
measurements before silking did not adequately estimate real PAR interception.
 L.E. Bonelli et al. / Field Crops Research 198 (2016) 215–225 221

Fig. 3. Mean solar radiation and mean temperature from silking to physiological maturity of: i) maize crops sown at different sowing dates in Balcarce Exp. 1, Exp. 2 and
Exp. 3 (empty symbols) and ii) representative maize crops of typical agricultural systems located in contrasting environments around the world (full star symbols). All these
data correspond to temperate and cool temperate environment except Brazil. Safrina cropping season in Mato Grosso (Brazil) involves delayed plantings that do not result
in a deterioration of radiation and temperature during the reproductive period. Meteorological data were obtained from http://power.larc.nasa.gov).

relationship was reduced and even reversed as sowing date was Table 3
Set of candidate explanatory variables for actual post-silking dry-mater
delayed. Grain number per unit area was reduced proportionally
accumulation.
less than weight per grain as sowing date was delayed.
The delay in sowing date reduced the amount of incident PAR Variable Significance ra
from emergence to silking. Differences in the accumulated inter- Potential post-silking dry-matter p < 0.0001 0.95
cepted PAR during that period (not directly shown on Table 2), Mean Potential CGR (S-ECG) p < 0.0001 0.89
however, were smaller than those observed for incident PAR Mean incident PAR p < 0.0001 0.87
Mean temperature p < 0.0001 0.85
because of increases in the fraction of PAR interception as sowing
Harvest Index p < 0.001 0.70
date was delayed (Table 2, Fig. 4). Accumulated dry matter differed Time from silking to ECG p < 0.001 0.65
among hybrids in October and November sowings but only at the Photo-thermal quotientb p < 0.01 0.63
end of the crop cycle (Exp. 3; Fig. 4). Delaying sowing date reduced Time from silking to Phys. Mat. p < 0.46 0.16
accumulated incident PAR by shortening the reproductive period a
Simple correlation coefficient of the variable with actual dry matter accumula-
(Table 1) and by placing this period toward the end of the season, tion during the post-silking period.
b
when daily incident PAR had decreased (Fig. 2). Calculated as the quotient between average mean PAR and mean temperature
during the period from silking to ECG.
Harvest index decreased progressively with the delay in sowing
date (Table 2). Grain yield was related to aboveground dry matter
accumulated during the post-silking period (p < 0.0001; R2 = 0.94). silking dry matter. However, as sowing date was delayed, grain
Thus, major changes in HI resulted from a disproportionate varia- yields approached these estimated dry matter values (see 1:1 line
tion in the post-silking dry matter accumulation. and calculated RMSE and u for sowing dates prior and after Nov-15
Since crops reached and maintained similar fractions of PAR in Fig. 5).
interception after silking (Table 2), variations in the above-
ground dry matter accumulated during the post-silking period
resulted from differences in RUE and in accumulated incident 3.3. Source-sink relationships
PAR (Table 2) (multiple regression analysis p < 0.0001; R2 = 0.99).
Radiation use efficiency was positively associated with mean tem- The source-sink relationships during the EGFP are shown in
perature (p < 0.0001; R2 = 0.32) and, within each crop, RUE was Fig. 6. The variation of potential source capacity was higher than
lower during the post-silking vs. the pre-silking periods. Variation that of sink strength (Fig. 6a) and the source/sink ratio decreased
in RUE during the post-silking period was mainly accounted for from early to late sowing dates. In late sowings, the sink strength
by sowing date and differed among hybrids only for the October was higher than the potential source capacity (Fig. 6a). Actual and
sowing date (Exp. 3, Table 2). potential source capacity were similar only at late sowings (Fig. 6b).
Actual post-silking dry matter accumulation was tested against In early sowings, actual source capacity was lower than its cor-
a set of candidate variables (Table 3). The purpose of this analy- responding potential source capacity and, to some extent, it was
sis was to find predictor variables for the observed variation in related to its sink strength.
grain yields. With the exception of time from silking to physio-
logical maturity, all the evaluated variables showed a significant 4. Discussion
association with the target variable (Table 3). The most signifi-
cant association was observed for potential post silking dry matter, This work reappraised the effect of sowing date on maize grain
which results from the product of the mean potential CGR and the yield and its components. The experimental set up at Balcarce pro-
time from silking to ECG. vided a range of temperature and radiative conditions as wide as
The potential post silking dry matter values were then com- that found across temperate and cool-temperate maize environ-
pared with the observed grain yields across all the experiments ments (Fig. 3). The results of this work and the innovative approach
(Fig. 5). Grain yields were, in general, lower than potential post- to describe source-sink relationships can contribute to a better
222 L.E. Bonelli et al. / Field Crops Research 198 (2016) 215–225

Fig. 4. Canopy fraction of photosynthetically active radiation (PAR) interception and accumulated aboveground dry matter as a function of time after emergence for October
(a, b), November (c, d), December (e, f) and January (g, h) sowing dates of three maize hybrids: DK692, I550 and P39B77. Data from Exp. 3 (2011–12 growing season at
Balcarce 37◦ 45 S, 58◦ 18 W). Arrows in graphs b, d, f and h indicate silking date for P39B77 (1), I550 (2) and DK692 (3). Vertical bars indicate the standard error of mean.

understanding of the effects of photo-thermal environmental con- EGFP (Fig. 6a). Grain growth of these source limited crops was
ditions on maize grain yield determination. not necessarily limited by assimilate availability since, in maize,
sugar remobilization from the stem can contribute to grain filling
(Daynard et al., 1969). The herein proposed source-sink analy-
4.1. Grain yield components and source-sink relationships sis estimates net plant dry-matter stocks and does not consider
internal translocation of assimilates. Hence, not all these source
The reduction in number of grains per unit area was proportion- limited crops would increase grain yields in response to increases
ally lower than that in weight per grain as sowing date was delayed. in assimilate availability. The method for assessing potential source
In contrast, previous studies (Cirilo and Andrade, 1994a,b) indi- capacity (Eqs. (4) and (5)) may also overestimate crop growth
cated that grain number reduction was the main cause of grain yield capability since it does not consider the progressive deteriora-
decreases with the delay in sowing date. Recent investigations evi- tion of canopy photosynthetic capacity during the grain filling
denced that breeding efforts have steadily risen maize grain yield period caused by leaf aging and senescence processes (Crafts-
mainly through increases in reproductive sink strength. Thus, final Brandner and Poneleit, 1987; Dwyer and Stewart, 1986). Despite
weight per grain in current hybrids have become more prone to be these concerns, the method for assessing potential source capac-
source limited during the grain filling period (Cerrudo et al., 2013; ity is useful since it provides a simple and robust estimation of
Echarte et al., 2006). the upper threshold of CGR. From this reference value, assim-
As sowing date was delayed, crops maintained a relatively high ilate limitations to crop growth during the EGFP are promptly
number of grains per unit area (Table 2) with a growth rate (GGR) evident.
that exceeded their respective potential source capacity during the
 L.E. Bonelli et al. / Field Crops Research 198 (2016) 215–225 223

Fig. 5. Observed maize grain yield in relation to the simulated potential above-
ground dry matter accumulated during the post-silking period. Simulation based
on the potential crop growth rate (CGR) and the length of the reproductive period
(time from silking to the end of crop growth (ECG)). Potential CGRs were calculated
from daily incident PAR and mean temperature. Data of five maize experiments con-
ducted at Balcarce (37◦ 45 S, 58◦ 18 W) during 2009–10 (Exp.1), 2010–11 (Exp. 2),
2011–12 (Exp.3 and C. Exp. II) and 2012–13 (C. Exp. I) growing seasons. Root mean
square error (RMSE) and index of agreement (u) are indicated for sowing dates prior
Nov-15 (RMSE1 and u1 ) and after Nov-15 (RMSE2 and u2 ).

4.2. Radiation use efficiency

An interesting question arises from the observed RUE in Exp.

3 (Table 2): why did hybrids, with similar pre-silking RUE, differ
in their post-silking RUE at early sowing dates but not at late sow-
ing dates? Sink limitations can restrict photosynthesis when plants
cannot allocate the photoassimilate production (Paul and Foyer, Fig. 6. Source-sink relationship during the effective grain filling period (EGFP) of
2001). As shown in Table 2, hybrids with contrasting RUE during the maize in Balcarce (37◦ 45 S, 58◦ 18 W) as affected by four contrasting sowing dates
post-silking period also differed significantly in their grain number (Oct-4 (1); Nov-14 (2); Dec-15 (3); Jan-4 (4)) and three contrasting maturity hybrids
per unit area. Hence, a possible hypothesis is that RUE differences (DK692, I550 and P39B77). Observed sink strength in relation to the potential source
capacity (a) and actual source capacity in relation to the potential source capacity
among hybrids resulted from contrasting reproductive sink limita-
(b). Sink strength is indicated as the mean grain growth rate (GGR) during the EGFP
tions during the grain filling period. The progressive reduction in expressed on a community basis (g m−2 d−1 ). Source capacity (both potential and
radiation and temperature during the reproductive period as sow- observed) is indicated as the mean crop growth rate (CGR) during the EGFP. Dotted
ing date was delayed, turned the source more limiting than the sink line in both graphs indicates 1:1 relation. Data from Exp. 3, see material and methods
for crop growth (Fig. 6a). When the source capacity became the lim- for details about estimations of GGR and potential and actual CGR.

iting factor, crop growth was independent of the sink strength and
all hybrids showed similar post-silking RUE. durations and lower CGRs in late vs. early sowing dates reduced
The analysis of RUE revealed that the different current hybrids markedly the dry matter allocated to grains. These unbalanced
had similar canopy photosynthesis capability. Moreover, pre- changes in dry matter accumulation explained the observed vari-
silking RUE values did not differ from those observed by Andrade ation in HI. Previous investigations in maize (Maddonni, 2012;
et al. (1993) for older hybrids. These observations are consistent Otegui et al., 1996) carried out grain yield simulations by multi-
with those reported by Tollenaar and Aguilera (1992) who stud- plying an estimated aboveground dry mater (accumulated between
ied hybrids from different breeding eras and found RUE differences emergence and physiological maturity) by a fixed value of HI = 0.46.
only for the post-silking period, during which, contrasting sink lim- A similar modelling approach was also proposed by Muchow et al.
itations among hybrids were manifested. Sink limitations of early (1990) but including a model algorithm that progressively modified
sown maize can be reduced or removed through increases in plant the HI as a function of grain development.
density (Sarlangue et al., 2007). Conversely, grain yield responses The present work not only evidenced an inconsistency in using
to increases in plant density would not be expected to occur in any a constant value of HI, but also, found a simple alternative to sim-
case when source capacity is the limiting factor (e.g. late sowing ulate the effect of sowing date on maize grain yield (Fig. 5). This
dates), except for situations where crop canopy does not reach full modeling approach can be useful, for example, to assess the rel-
PAR interception. evance of the grain yield interaction between sowing date and
hybrid cycle length (Exp. 3; Table 2). The most suitable combina-
4.3. Harvest index and modelling approaches tion of hybrid maturity class and sowing date depends mainly on
the length of the cropping season (Olson and Sander, 1988; Shaw,
Delaying sowing resulted in similar (or slightly increased) 1988) and, for any particular environment, its assessment requires
amounts of dry matter at silking (Fig. 4), indicating that dry mat- an analysis based on a long term series of climate data. It should be
ter allocation to vegetative structures was practically unchanged noticed that a simulation model based on the potential post-silking
by sowing date. In contrast, for the post-silking period, shorter dry matter accumulation assumes by definition that grain yield is
224 L.E. Bonelli et al. / Field Crops Research 198 (2016) 215–225

source-limited. This implies that reproductive sink strength does References

not limit the photoassimilate allocation to grains during the post-
silking period. Accordingly, as shown in Figs. 5 and 6, observed grain Andrade, F.H., Uhart, S.A., Cirilo, A., 1993. Temperature affects radiation use
efficiency in maize. Field Crops Res. 32, 17–25, http://dx.doi.org/10.1016/0378-
yields would approach simulated potential post-silking dry-matter 4290(93)90018-I.
as long as the sink strength reaches or exceeds the source capacity. Andrade, F.H., Sadras, V.O., Vega, C.R.C., Echarte, L., 2005. Physiological
determinants of crop growth and yield in maize, sunflower and soybean. J.
Crop Improv. 14, 51–101, http://dx.doi.org/10.1300/J411v14n01 05.
4.4. Agronomic prospects Birch, C.J., Rickert, K.G., Hammer, G.L., 1998. Modelling leaf production and crop
development in maize (Zea mays L.) after tassel initiation under diverse
conditions of temperature and photoperiod. Field Crops Res. 58, 81–95.
Late sowings of maize have become a frequent practice in rain- Cerrudo, A., Di Matteo, J., Fernandez, E., Robles, M., Pico, L.O., Andrade, F.H., 2013.
fed systems of Argentina. Farmers have implemented this practice Yield components of maize as affected by short shading periods and thinning.
to reduce the risk of water deficit during the critical flowering Crop Pasture Sci. 64, 580–587.
Cirilo, A.G., Andrade, F.H., 1994a. Sowing date and maize productivity: I. Crop
period (Maddonni, 2012) despite the deterioration of environmen-
growth and dry matter partitioning. Crop Sci. 34, 1039–1043.
tal conditions (radiation and temperature) for crop growth during Cirilo, A.G., Andrade, F.H., 1994b. Sowing date and maize productivity: II. Kernel
the reproductive period. Maize grain yield in delayed sowings is number determination. Crop Sci. 34, 1044–1046.
Cirilo, A.G., Andrade, F.H., 1996. Sowing date and kernel weight in maize. Crop Sci.
less variable across years and closer to its corresponding poten-
36, 325–331, http://dx.doi.org/10.2135/cropsci1996.
tial grain yield when compared with that of early sowings (Mercau 0011183x003600020019x.
et al., 2014). Hence, some agronomic prospects can be derived from Crafts-Brandner, S.J., Poneleit, C.G., 1987. Carbon dioxide exchange rates, ribulose
our results to the management strategies for late sown maize of bisphosphate carboxylase/oxygenase and phosphoenolpyruvate carboxylase
activities, and kernel growth characteristics of maize. Plant Physiol. 84,
rainfed systems. 255–260, http://dx.doi.org/10.1104/pp.84.2.255.
Crop growth during the EGFP of late sown maize was limited by Daynard, T.B., Tanner, J.W., Hume, D.J., 1969. Contribution of stalk soluble
photosynthetic source capacity because actual grain growth rate carbohydrates to grain yield in corn (Zea mays L.). Crop Sci. 9, 831–834.
Dwyer, L.M., Stewart, D.W., 1986. Effect of leaf age and position on net
exceeded the potential crop growth rate (Fig. 6a). This extra grain photosynthetic rates in maize (Zea Mays L.). Agric. For. Meteorol. 37, 29–46,
growth had to be supported by remobilization of carbohydrates http://dx.doi.org/10.1016/0168-1923(86)90026-2.
(Daynard et al., 1969; Uhart and Andrade, 1991). Hence, for late Echarte, L., Luque, S., Andrade, F.H., Sadras, V.O., Cirilo, A., Otegui, M.E., Vega, C.R.C.,
2000. Response of maize kernel number to plant density in Argentinean
sowings, farmers must take into account the root and the stem hybrids released between 1965 and 1993. Field Crops Res. 68, 1–8.
strength of hybrids to avoid agronomic difficulties associated with Echarte, L., Andrade, F.H., Sadras, V.O., Abbate, P., 2006. Kernel weight and its
lodging and stem breakage. response to source manipulations during grain filling in Argentinean maize
hybrids released in different decades. Field Crops Res. 96, 307–312.
In addition, the source limitation in late sown maize indicates
Egli, D.B., 1998. Seed Biology and the Yield of Grain Crops. CABI, Wallingford, UK.
that the physiological condition of maize plants during the grain Hall, A., Sadras, V., 2009. Chapter 21—whither crop physiology? In: Sadras, V.,
filling period has become as relevant as the critical flowering Calderini, D. (Eds.), Crop Physiology. Academic Press, San Diego, pp. 545–570.
Kuehl, R.O., 2000. Design of Experiments: Statistical Principles of Research Design
period. Thus, those recommendations based on the optimization of
and Analysis, second edition. Duxbury, Pacific Grove, CA.
the crop physiological condition around flowering (Andrade et al., López-Cedrón, F.X., Boote, K.J., Ruíz-Nogueira, B., Sau, F., 2005. Testing
2005) should be reviewed for late sowings of current maize hybrids. CERES-maize versions to estimate maize production in a cool environment.
Our results suggest that crop management for late sown maize Eur. J. Agron. 23, 89–102.
Maddonni, G.A., 2012. Analysis of the climatic constraints to maize production in
needs to be focused, also, in preserving optimal physiological con- the current agricultural region of Argentina—a probabilistic approach. Theor.
ditions for crop growth during the grain filling period. Appl. Climatol. 107, 325–345.
Mercau, J.L., Otegui, M.E., Ahuja, L.R., Ma, L., Lascano, R.J., 2014. A modeling
approach to explore water management strategies for late-sown maize and
5. Conclusions double-cropped wheat-maize in the rainfed pampas region of Argentina. In:
Advances in Agricultural Systems Modeling. American Society of Agronomy,
Inc., Crop Science Society of America, Inc., and Soil Science Society of America,
Based on the progressive deterioration of the photo-thermal Inc., Madison, WI.
conditions for maize growth during the last part of the cropping Monteith, J.L., 1972. Solar radiation and productivity in tropical ecosystems. J.
season, this work attempted to elucidate the factors that limit grain Appl. Ecol., 747–766.
Muchow, R.C., Sinclair, T.R., Bennett, J.M., 1990. Temperature and solar radiation
yields when sowing date is delayed. We explored a broad range of effects on potential maize yield across locations. Agron. J. 82, 338–343.
sowing dates in a temperate maize region where solar radiation and Muchow, R.C., 1990. Effect of high temperature on grain-growth in field-grown
temperature vary widely along the cropping season. A consistent maize. Field Crops Res. 23, 145–158, http://dx.doi.org/10.1016/0378-
4290(90)90109-O.
response of grain yield components and source-sink relationships
Olson, R.A., Sander, D.H., 1988. Corn production. In: Sprague, G.F., Dudley, J.W.
indicated that, as sowing date is delayed, the supply of assimilates (Eds.), Corn and Corn Improvement. American Society of Agronomy, Madison
to grains (source capacity) becomes more limiting than the demand WI, pp. 639–686.
Otegui, M.E., Ruiz, R.A., Petruzzi, D., 1996. Modeling hybrid and sowing date effects
for assimilates by grains (sink strength) during the grain filling
on potential grain yield of maize in a humid temperate region. Field Crops Res.
period. These results prompt further research to review strategies 47, 167–174, http://dx.doi.org/10.1016/0378-4290(96)00031-7.
and recommendations for late sown maize. Paul, M.J., Foyer, C.H., 2001. Sink regulation of photosynthesis. J. Exp. Bot. 52,
1383–1400.
RDC Team, 2008. R: A language and environment for statistical computing. R
Acknowledgements Found. Stat. Comput.
Ritchie, S.W., Hanway, J.J., Benson, G.O., 1989. How a Corn Plant Develops. Iowa
State University of Science and Technology, Cooperative Extension Service.
The authors wish to thank two anonymous reviewers for their Sarlangue, T., Andrade, F.H., Calviño, P.A., Purcell, L.C., 2007. Why do maize hybrids
constructive comments, Emiliano Veliz and Diego Gaitán for field respond differently to variations in plant density? Agron. J. 99, 984, http://dx.
work assistance, Mariana Robles for writing advices and to all the doi.org/10.2134/agronj2006.0205.
Shaw, R.H., 1988. Climate requirement. In: Sprague, G.F., Dudley, J.W. (Eds.), Corn
members of Crop Ecophysiology Group in Balcarce who contributed and Corn Improvement. American Society of Agronomy, Madison WI, pp.
in several stages of this work. This work was supported by INTA 609–638.
(the National Institute for Agricultural Technology in Argentina) Tollenaar, M., Aguilera, A., 1992. Radiation use efficiency of an old and a new maize
hybrid. Agron. J. 84, 536–541, http://dx.doi.org/10.2134/agronj1992.
and Lucas E. Bonelli holds a scholarship from CONICET (the National 00021962008400030033x.
Scientific and Technical Research Council in Argentina). The present Tollenaar, M., Lee, E.A., 2011. Strategies for enhancing grain yield in maize. Plant
work is part of a thesis by Lucas E. Bonelli in partial fulfillment for Breed. Rev. 34, 37–82.
Tollenaar, M., Daynard, T.B., Hunter, R.B., 1979. Effect of temperature on rate of leaf
the Doctor’s degree at UNMdP (the National University of Mar del
appearance and flowering date in maize. Crop Sci. 19, 363–366.
Plata, Argentina).
 L.E. Bonelli et al. / Field Crops Research 198 (2016) 215–225 225

Tollenaar, M., Dwyer, L.M., Stewart, D.W., 1992. Ear and kernel formation in maize Uhart, S.A., Andrade, F.H., 1991. Source-sink relationships in maize grown in a
hybrids respresenting three decades of grain yield improvement in Ontario. cool-temperate area. Agronomie 11, 863–875.
Crop Sci. 32 (2), 432–438. Westgate, M.E., Andrade, F.H., Otegui, M.E., 2004. Physiology of the corn plant. In:
Tollenaar, M., 1977. Sink-source relationships during reproductive development in Wayne, S., Betrán, J., Runge, E.C.A. (Eds.), Corn: Origin, History, Technology, and
maize. A review. Maydica 22, 49–75. Production. John Wiley & Sons IncHoboken, New Jersey, pp. 235–271.
Tsimba, R., Edmeades, G.O., Millner, J.P., Kemp, P.D., 2013a. The effect of planting Willmott, C.J., 1981. On the validation of models. Phys. Geogr. 2, 184–194, http://
date on maize grain yields and yield components. Field Crops Res. 150, dx.doi.org/10.1080/02723646.1981.10642213.
135–144. Wilson, D.R., Muchow, R.C., Murgatroyd, C.J., 1995. Model analysis of temperature
Tsimba, R., Edmeades, G.O., Millner, J.P., Kemp, P.D., 2013b. The effect of planting and solar radiation limitations to maize potential productivity in a cool
date on maize: phenology: thermal time durations and growth rates in a cool climate. Field Crops Res. 43, 1–18.
temperate climate. Field Crops Res. 150, 145–155.