Published May, 1990

Relative Sensitivity of Grain Yield and Biomass Accumulation to Drought

in Field-Grown Maize
T. R. Sinclair,* J. M. Bennett, and R. C. Muchow
ABSTRACT
Grain yields in maize (Zea mays L.) are decreased substantially
by water deficits. Since the impact of drought stress on grain yield
varies among growth stages, such responses are sometimes interpreted as indicators of varying drought sensitivities with crop growth
stage. However, varying leaf area and biomass accumulation rate
confound this interpretation because the use of water changes during
the season. This research examined whether a proportional decrease
in accumulated crop biomass of field-grown maize was sufficient to
account for the decrease in grain yield resulting from drought. A
total of 11 field experiments were conducted at Gainesville, FL on
an Arredondo fine sand (loamy, siliceous hyperthermic Grossarenic
Paleudult) or a Kendrick fine sand (loamy, siliceous, hyperthermic
Arenic Paleudult and Katherine, Australia on a Fenton clay loam
(Oxic or Rhodic Paleustalf) in which irrigation was withheld for
various periods to result in a range of severities in crop water deficits. At crop maturity, both grain yield and total accumulated biomass were measured and the relationship between these two parameters was examined. In all cases, high, linear correlations between
grain yield and accumulated biomass were obtained. At moderate
stresses, harvest index of the crop was stable. Under severe stress,
where accumulated biomass was less than about 1100 g nr2, grain
yield was decreased to a greater extent than was accumulated biomass. Nevertheless, under a wide range of conditions, grain yield of
field-grown maize crops was closely linked to the accumulation of
biomass.

AIZE GRAIN YIELD is sensitive to drought (Robins and Domingo, 1953; Denmead and Shaw,
M
1960; Claassen and Shaw, 1970; Musick and Dusek,
1980). The yield decrease is related to the developmental stage at which the water deficits are imposed,
with the greatest yield decreases resulting from
drought stress at or near anthesis. Such results have
been interpreted to indicate that physiological processes associated with anthesis and early grain development are especially vulnerable to water deficits in
the maize plant (Westgate and Boyer, 1986). Yield
prediction efforts have frequently identified anthesis
as an especially sensitive stage to drought, and consequently drought at anthesis has caused yield estimates to be discounted substantially (Shaw, 1974;
Slabbers et al., 1979).
Nevertheless, careful analysis of seed yield in response to drought for many crops tend to discount
large growth stage effects. For example, Muchow and
T.R. Sinclair, USDA-ARS, Agronomy Dep., Univ. of Florida, and
J.M. Bennett, Agronomy Dep., Univ. of Florida, Gainesville, FL
32611; and R.C. Muchow, CSIRO, Cunningham Lab., St. Lucia,
Queensland 4067, Australia. Florida Agric. Exp. Stn. Journal Ser.
no. R-00093. Received 16 Aug. 1989. ""Corresponding author.
Published in Crop Sci. 30:690-693 (1990).

SINCLAIR
ET AL.: MAIZESENSITIVITY
TODROUGHT
Sinclair (1986) using a simple model of crop growth
analyzed the water use, biomass accumulation, and
final yield of soybean [Glycine max (L.) Merr.] when
subjected to water deficits at various growth stages.
While the greatest yield decreases resulted from water
deficit at early reproductive stages, these decreases occurred because this was the stage of maximumrates
of biomass accumulation and water use, and hence,
the response was not specifically related to reproductive growth. Zinselmeier et al. (1988) found that water
deficits imposed at anthesis on maize plants grown in
pots in the field had little or no effect on kernel set.
To examine the relative influence of drought on grain
and biomass accumulation, Sinclair (1988) reviewed
published reports on harvest indices (ratio of seed
mass to total plant mass) of grain legumes and sorghum (Sorghum bicolor L. Moench) obtained from
drought experiments. In these studies, harvest index
decreased substantially
only when grain and biomass
accumulation were less than one-half of those of wellwatered treatments.
The objective of this research was to examine the
relative sensitivity in maize of grain and biomass accumulation to drought stress under field conditions.
To test the hypothesis that grain yield and total accumulated biomass at maturity are closely correlated
over a range of water deficits, data were obtained from
11 irrigation experiments performed in different environments and using different cultivars. The results
of these experiments were analyzed by comparing
grain yield against total biomass accumulation.
MATERIALS AND METHODS
Eleven data sets were obtained from irrigation experimentsin whichmaizecrops were subjected to varying levels
of water deficits. Four experiments were conducted at
Gainesville, FL and seven experiments were at Katherine,
Australia. Water managementtreatments usually ranged
fromfrequent, high-intensity irrigation to rainfed plots with
no irrigation. At maturity, plants were harvested from large
plot areas by cutting the plants at the soil, and the total crop
dry weight was obtained as a measure of accumulated crop
biomass. For each plot, seed was separated from the rest of
the plant parts to obtain grain dry weight. Analysis of the
relationship between grain yield (G) and accumulatedbiomass (B) of individual plots within experiments was done
by regression using the simple linear model,
G= a + b B

[1]

where a and b are the intercept and slope, respectively, of

the linear regression.
Florida. Experiments were conducted at the University
of Florida Agronomy
Farm(29, 38N lat) on an Arredondo
or a Kendrickfine sand soil, whichhave an available water
storage capacity of 40 to 50 mm.Due to the low waterholdingcapacity of these soils, visible signs of plant drought
stress usually developedwithin 5 to 6 d after the soil profile
had been at field capacity. Fertilization included application
-2 13.0 g K m
of approximately2.4 g N m-2, 3.5 g P m-2, and
to the soil prior to sowingin each experiment and two or
three side-dressings of fertilizer to provide approximatelyan
additional 20 g N m-2 and 7 g K m-2. Sowingwas generally
in early spring (Table l) to take full advantageof the usually
dry conditions that prevail in April and Mayat this location.

691

Table1. Description
of fourirrigationexperiments
at G~inesville,
FL(F1-F4)andsevenirrigationexperiments
at Katherine,Australia (A1-A7).
No.irrigationHarvest
Experiment
Maizehybrid Sowing
date
treatments area
2m

F1
F2
F3
F4
AI
A2
A3
A4
A5
A6
A7

Pioneer3851 2, 3 Mar.1987
Pioneer3851 11Mar.1988
McCurdy84AA 26Feb.1982
Pioneer3165 14Apr.1987
DekalbXL82 10Oct.1984
DekalbXL82 6 Feb.1985
DekalbXL82 20 Aug.1985
DekalbXL82 25 Nov.1983
DekalbXL82 30 Aug.1986
DekalbXL82 7 Feb.1984
DekalbXL82 1 Feb.1988

8
6
3
4
3
3
3
2
2
2
2

4.9
5.0
5.5
4.4
2.0
2.0
2.0
2.0
2.0
2.0
2.0

The crops were commonlysown in 0.6-m row widths and

thinned to a final plant population of about 7 to 10 plants
m-2. Eachirrigation treatmentconsisted of three or four replicates with each replicate plot being greater than 13.5 by
13.5 m.
At the time of canopyclosure but prior to tasseling for
Exp. FI and F2, eight and six irrigation treatments, respectively, wereinitiated. Irrigation was withheldfor differing
lengths of time amongtreatments so that crop drought
stresses of varying severity were imposedduring anthesis
and early grain development. Irrigation treatments were
continued until maturity whensamples were harvested from
the center of each plot. A similar experimental regimeincludinga well-wateredtreatment, twodeficit irrigation treatments, and a rainfed treatment was used in Exp. F4 (Nzeza,
1988). ExperimentF3 was slightly different in that it included a rainfed treatment that wasinitiated after crop establishment and a 10-d water deficit treatment that was imposed immediatelypreceding silking (Bennett et al., 1989).
Australia. Katherine is located in the Northern Territory
(14 lat., 29S) and has a monsoonalclimate with a welldefined hot, wet season from Novemberto March and a
warm, dry season the remainder of the year. The soil for
this experiment was Fenton clay loam and has an available
water storage capacity of 100 to 120 mm.Experiments were
sownat various dates throughthe year as indicated in Table
1. Before sowing12 g N m-~, 3 g P m-2, and 5 g K m-2 were
broadcast. Twoside-dressings of 6 g N m-2 each were applied
during crop growth. Seeds were sown in 0.5-m rows and
thinned after emergenceto 7 plants m-2. The subplots for
each treatment were 18 by 6 m. A complete description of
the Australian experiments is given by Muchow
(1989).
RESULTS
Irrigation treatments imposed in each of the 11 experiments resulted in a substantial range in grain yield
(Table 2). Generally, the most severe water deficits
decreased grain yields to 10 to 20%of the fully irrigated treatment. Intermediate irrigation treatments
provided intermediate grain yields.
Data from Exp. F1 and F2 (Fig. 1 and 2) illustrate
the observed range in grain yield and crop biomass
accumulation. The striking feature in these data is the
high correlation between grain yield and accumulated
biomass. Linear regression analysis of the data in Fig.
1 gave a highly significant correlation between grain
yield and accumulated biomass (r 2 --- 0.986, Table 2).
The irrigation treatments imposed in Exp. F2 resulted
in more severe stress than in Exp. F1 as evidenced by

692

CROP SCIENCE, VOL. 30,

Table 2. Rangein maize grain yield for 11 irrigation experiments.

Results of linear regression analyses of grain yield vs. total biomass within each experimentare given by the slope, standard error
(SE), intercept, and correlation coefficient (r2).

24
24
12
16
11
12
12
8
8
8
8

80O

Regressionof grain vs. biomass

2r
Slope (_+ SE) Intercept

Range of
Exp. n grain yieldS"
2g m
FI
F2
F3
F4
AI
A2
A3
A4
A5
A6
A7

MAY-JUNE1990

252-792
168-860
236-1423
0-986
0-876
521-894
122-829
637-871
563-921
373-906
270-930

600

2g m

g g-~
0.561
0.655
0.597
0.604
0.743
0.523
0.646
0.355
0.408
0.673
0.468

(0.014)
--6
0.986
(0.026)
-208**
0.966
(0.025)
-113"*
0.973
(0.028)
-365**
0.971
(0.040)
- 500** 0.974
(0.061)
-46
0.879
(0.101)
-362*
0.803
(0.049)
233*
0.896
(0.032)
62
0.964
(0.046)
--299** 0.973
(0.039)
16
0.960
*,** Significantlydifferentfromzero at the 0.05 and0.01 level of probability,
respectively.
Yieldrangeis for individual plots.

400

tn 200

400
I

800

~ 600
.~/
._~ 400

~ 200

i
400

Biomass

1200
(g -2)

1600

Fig. 2. Maize grain yield vs. crop biomass for individual plots in
Florida Exp. F2. Solid line was obtained from linear regression
analysis.

FI

800

~.
800

Biornass

~
1200
(g -2)

1600

Fig. 1. Maize grain yield vs. crop biomass for individual plots in
Florida Exp. FI. Solid line was obtained from linear regression
analysis.

moredata in the lower yielding range (Fig. 2). Again,

the variation in grain yield in Exp. F2 due to drought
stress was proportional to accumulatedbiomass(r 2 =
0.966, Table 2).
Linear regression analysis of grain yield on accumulated biomass was performed for each of the 11
data sets (Table 2). In all cases, there was a highly
significant correlation betweengrain yield and accumulated biomass. The lowest r 2 amongall experiments was 0.803 for Exp. A3. Only three of the 11
experimentshad a r 2 less than 0.95. Clearly, decreases
in .grain yield induced by water deficits were closely
correlated with decreases in biomass accumulation.
Sevenof the 11 experiments, including Exp. F2 (Fig.
2), resulted in a significant nonzerointercept as determinedfrom linear regressions of grain yield vs. total crop biomass (Table 2). Those experiments where
severe water deficits were imposed, resulting in very

low grain yields, were mostlikely to havea significant

negative intercept.
Theimpact of the severe stresses on the relationship
between grain yield and accumulated crop biomass is
illustrated by combining.all the data obtained for the
same maize cultivar at Katherine (Fig. 3). The severe
drought-stress treatments at Katherine resulted in premature senescenceof the crop and this early maturity
was associated with a lowered ratio of grain yield to
biomass. Excludingthe data in Fig. 3 from the severe
drought-stress treatments that resulted in premature
senescence, a linear regression analysis resulted in an
insignificant intercept and a slope of 0.475 g g-L
DISCUSSION
Results from these 11 irrigation experiments with
maize were uniformly consistent in demonstrating a
high degree of correlation betweengrain yield and accumulated crop biomass at maturity. Over the wide
range of environments and treatments used in these
experiments, the loss of ability to accumulategrain
under water deficits was closely linked to the loss in
the ability to accumulatecrop biomass. Therefore, it
is speculated that a major reason for the observation
that maize grain yield is most severely limited by
drought stress occurring at or near anthesis maywell
be that this growthstage is also the period for maximumrates of biomass accumulation and water use.
Waterdeficits at anthesis result in large decreases in
accumulatedbiomass, whichin turn result directly in
losses in grain yield.
The results of these experiments do indicate that
severe drought stresses result in enhanceddepressions
in grain yield. Asillustrated by Fig. 3 for droughtconditions sufficiently severe to result in premature senescence and prevent accumulated biomass from
exceeding about 1100 g m-2, the grain yield was

SINCLAIR ET AL.: MAIZE SENSITIVITY TO DROUGHT

1000

750

o>

;p
50O
>
-a
o

250

0
500 ~~

1000

1500

2000

Biomass
(g.rrf )
Fig. 3. Maize grain yield vs. crop biomass for individual plots from
all seven experiments done at Katherine, Australia. Excluding the
circled data from severely stressed treatments, the relationship
obtained from linear regression analysis was y*= 7.03 4- 0.475 x
(i2 = 0.87).

depressed and a negative intercept can be obtained

from linear regression analyses. Stress treatments were
sufficiently severe in six of the 11 experiments to result
in significant negative intercepts. Consequently, harvest index usually was found to be decreased under
severe drought treatments when2 accumulated biomass
was less than about 1100 g nr .
Nevertheless, the results of these experiments indicate that, for many practical applications, it may be
reasonable to assume that harvest index is stable. This
is illustrated by rearranging Eq. [1] to solve for harvest
index (HI):
ffl

-|-*+i

[2]

In cases where a is much less than B, which occurred

frequently in these experiments, the harvest index is
predicted to be stable. As B decreases as a result of
drought stress, the harvest index is predicted to decrease for cases where a is negative. Yet, unless grain
yields have been severely decreased by water deficits,
it may be possible to obtain reasonable grain yield
predictions by knowing accumulated biomass and
multiplying by a constant value of harvest index obtained from a well-watered treatment.
The observation that harvest index is fairly stable
under moderate drought stress and is decreased only
under severe drought stress is consistent with previous
studies. The experiments of Denmead and Shaw
(1960) were done in pots so that wilting generally was
observed on the second day after terminating irrigation. Severe stress was induced through three successive drying cycles. While drought stress at the silking
stage had the greatest yield decrease (50%), the harvest
index was decreased by a comparatively small amount
(16%). In the experiments of Claassen and Shaw
(1970), the pots were four-times larger but three plants

693

were grown in each pot, again resulting in rapid development of severe water deficits. In their first year,
grain yield was decreased greatly by drought stress
(grain yield for drought-stressed plants at silking was
47% of well-watered plants) as was harvest index (0.29
compared with 0.49). However, in the second year,
the greatest grain-yield decrease resulted from drought
stress during grain growth. In this case, grain yield was
decreased to about 65% of the well-watered plants,
while harvest index decreased slightly (0.42 compared
with 0.49). Insufficient data were presented by Robins
and Domingo (1953) and by Musick and Dusek (1980)
to allow a comparison of grain yield and accumulated
biomass. Recently, Wolfe et al. (1988) found in a 2year experiment with fertilized mai/e that lack of irrigation in one year caused no change in harvest index, but harvest index decreased by 17% in a second
year when accumulated biomass decreased to 60% of
the irrigated crop.
In conclusion, the results of our studies demonstrate
that although drought stress results in decreased grain
yield, the decrease is proportional to the decrease in
accumulated biomass. In fact, moderate drought
stresses do not result in a change in the harvest index
of the crop. Only under severe drought stresses where
accumulated biomass totals less than about 1100 g nr2
did additional influences of stress cause harvest index
to decrease. For many practical applications, these results indicate that the effects of drought stress on grain
yield can be approximated by assuming a stable harvest index and knowing accumulated crop biomass.