Published June 15, 2016
RESEARCH
Maize Stalk Lodging: Flexural Stiffness
Predicts Strength
Daniel J. Robertson, Shien Yang Lee, Margaret Julias, and Douglas D. Cook*
ABSTRACT
Late-season stalk lodging in maize (Zea mays L.)
is a major agronomic problem that has farreaching economic ramifications. More rapid
advances in lodging resistance could be
achieved through development of selective
breeding tools that are not confounded by
environmental factors. It was hypothesized
that measurements of stalk flexural stiffness (a
mechanical measurement inspired by engineering beam theory) would be a stronger predictor
of stalk strength than current technologies. Stalk
flexural stiffness, rind penetration resistance
and stalk bending strength measurements were
acquired for five commercial varieties of dent
corn grown at five planting densities and two
locations. Correlation analyses revealed that
stalk flexural stiffness predicted 81% of the variation in stalk strength, whereas rind penetration
resistance only accounted for 18% of the variation in stalk strength. Strength predictions based
on measurements of stalk flexural stiffness
were not confounded by hybrid type, planting
density, or planting location. Strength predictions based on rind penetration resistance were
moderately to severely confounded by such factors. Results indicate that stalk flexural stiffness
is a good predictor of stalk strength and that it
may outperform rind penetration resistance as a
selective breeding tool to improve lodging resistance of future varieties of maize.
D.J. Robertson, S.Y. Lee, M. Julias, and D.D. Cook, Division of
Engineering, New York Univ. Abu Dhabi, Abu Dhabi, United Arab
Emirates. Received 1 Nov. 2015. Accepted 20 Jan. 2016. *Corresponding
author ([email protected]).
Y
ield losses in maize due to stalk lodging (breakage of the
stalk below the ear) range from 5 to 20% annually worldwide
(Flint-Garcia et al., 2003). Crop losses negatively impact farmers
and affect society as a whole by creating instability in the overall
crop supply (Hagenbaugh, 2007). Selective breeding approaches
to improving lodging resistance have historically relied on counting the number of lodged plants at harvest. Unfortunately, this
approach is severely confounded by several uncontrolled environmental factors including insect damage, disease, and weather
patterns conducive to lodging (e.g., wind and rain storms) (FlintGarcia et al., 2003). For example, a severe wind storm may flatten
an entire field of maize regardless of the lodging resistance of
stalks within the field, whereas the absence of wind and rain
storms may leave the entire field standing regardless of the lodging resistance of any individual stalk, thus making it impossible
to distinguish stronger varieties from weaker varieties. More
rapid gains in lodging resistance could be achieved by developing
testing methodologies that can predict lodging resistance in the
absence of lodging related weather events (Cloninger et al., 1970;
Hu et al., 2012). The purpose of the current study is to investigate a testing methodology that does not rely on lodging-related
weather events and could potentially be used as a selective breeding tool to combat late-season stalk lodging in the future.
Regardless of weather, environment, chemical composition, or stalk morphology, the culminating event in stalk lodging
is structural failure (i.e., breakage) of the stalk. Stalk lodging
research has therefore focused on measuring and increasing stalk
Published in Crop Sci. 56:1711–1718 (2016).
doi: 10.2135/cropsci2015.11.0665
© Crop Science Society of America | 5585 Guilford Rd., Madison, WI 53711 USA
This is an open access article distributed under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
crop science, vol. 56, july– august 2016 www.crops.org1711
strength (i.e., a stronger stalk is more resistant to lodging
than a weaker stalk, regardless of morphology, chemical composition or weather). However, the corn stalk is
a multi-scale, anisotropic, nonlinear, nonprismatic, biological structure that is subject to large deformations
before breaking. Correctly measuring failure properties
of such structures and tissues typically requires specialized engineering expertise. Previously presented methods
of measuring stalk bending strength generally induce
premature stalk breakage and produce failure types and
patterns that are unrelated to the failure types and patterns
of naturally lodged corn stalk (Robertson et al., 2014,
2015a, 2015b). This may explain why some studies have
only found moderate correlations between stalk strength
and lodging. Furthermore, mechanical measurements of
stalk strength are typically time-consuming, load the stalk
in an unnatural manner, require specialized laboratory
equipment, and apply destructive testing methodologies
(physically break or crush the stalk). Consequently, after
more than a century of research, the counting of lodged
stalks at harvest is still the primary method used to quantify lodging resistance in selective breeding trials.
An ideal breeding tool for combating stalk lodging
should (i) accurately predict stalk strength, (ii) not be
confounded by environmental factors (i.e., be universal
in its application), (iii) require minimal effort and time
to use, and (iv) not permanently damage the stalk, thus
allowing analysis of individual plants throughout their
lifecycle (i.e., collection of temporal data). The consensus
among some researchers is that rind penetration resistance
best adheres to the above mentioned requirements (Hu
et al., 2012; Twumasi-Afriyie and Hunter, 1982). Rind
penetration resistance measurements are acquired by forcing a small needle or spike through the rind of the stalk
and measuring the maximum applied force. The method
is very fast, allowing breeders to phenotype hundreds of
plants in a minimal amount of time. Furthermore, it does
not require breaking the stalk or killing the plant, thus
enabling collection of temporal data. Numerous studies
have demonstrated that rind penetration resistance negatively correlates with stalk lodging in the field (Gou et
al., 2007; Hu et al., 2012; McRostie and MacLachlan,
1942). However, despite much research rind penetration
resistance is still severely limited in its potential as a selective breeding tool to improve late-season stalk lodging
resistance. Rind penetration measurements can readily
identify weak stalk varieties but are generally not able
to distinguish varieties with moderate lodging resistance
from varieties with high lodging resistance (Pedersen
and Toy, 1999). Its utility in late stage breeding trials is
therefore severely hindered. Consequently, development
of ultra-high yielding crop varieties that demonstrate
superior lodging resistance while being tolerant of high
1712
planting densities remains unrealized (Gou et al., 2010;
Hu et al., 2012).
The goal of the current study was to investigate an
alternative approach to predicting stalk strength that could
possibly be used as a selective breeding tool to develop
superior crop varieties in the future. In particular, it was
hypothesized that stalk flexural stiffness may be an accurate predictor of stalk strength. Stalk flexural stiffness is a
mechanical measurement inspired by structural engineering beam theory. Stalk flexural stiffness measurements can
be obtained rapidly without damaging the stalk, and engineering theory predicts such measurements to be strongly
related to structural strength (Beer and Johnston, 1981). To
test the flexural stiffness hypothesis the bending strength,
stalk flexural stiffness, and rind penetration resistance of
two replicates of five commercial varieties of dent corn
were sown at five planting densities, in two locations, were
measured and compared. Rind penetration resistance measurements were acquired to provide a benchmark against
which to compare stalk flexural stiffness predictions. All
stalks were removed from the field before testing so that
strength measurements could be acquired in a controlled
laboratory setting to limit sources of experimental noise.
Future research will focus on developing field based tools
and techniques to measure stalk flexural stiffness and stalk
strength. These tools will be validated against the controlled
laboratory methods developed as part of the current work.
MATERIALS AND METHODS
Plant Materials
Five commercial varieties of dent corn were grown during
the 2013 season at Monsanto facilities in Iowa in a randomized block design which included five planting densities of
119000, 104000, 89000, 74000, and 59000 plants ha–1 (48000,
42000, 36000, 30000, and 24000 plants ac­–1), two locations,
and two replicates. Plants were allowed to reach full maturity,
and remained in the field until just before harvest. At this time
ten consecutive stalks from the middle of each plot were cut just
above ground level and just above the ear. To reduce the chance
of stalk spoilage, ears and leaves were removed and stalks were
placed on forced air dryers to reduce moisture to approximately
13% w/w. Only stalks that were found to be free of disease and
pest damage were included in the current study.
All mechanical measurements were obtained in a laboratory setting using the same universal testing machine (details
below). Every stalk included in the study was subjected to the
three mechanical tests: stalk flexural stiffness, bending strength,
and rind penetration resistance.
Stalk Flexural Stiffness
Engineering beam theory is commonly used by structural
engineers to quantify the bending stiffness and strength of
buildings, beams, and structures. In the current work, the
authors sought inspiration from the equations of engineering
beam theory to develop a nondestructive means of predicting corn stalk strength. It was hypothesized that stalk strength
www.crops.org
crop science, vol. 56, july– august 2016
could be predicted by slightly flexing stalks to obtain a composite bending stiffness measurement which was termed ‘stalk
flexural stiffness’. This quality can be calculated as follows:
Stalk flexural stiffness =Æ
a 2b 2
[1]
L
where ∅ is the slope of the force-displacement curve obtained
by slightly flexing the stalk in three point bending, a and b are
the distances to the applied load as measured from the supports
on the left and right sides of the stalk respectively, and L is the
distance between supports. In the current study slope measurements (∅) were obtained by inducing less than 6 mm of stalk
deflection. This procedure ensured the test did not induce any
permanent damage to the stalk.
Bending Strength
Bending strength was measured using three-point bending,
following the guidelines outlined in (Robertson et al., 2014,
2015b). Supports were placed at the initial and terminal nodes
of each stalk sample, and the load was applied to the node closest to the center of the stalk. This approach minimizes artificial
deformation of the stalk’s cross-section and produces failure
patterns in agreement with those observed in naturally failed
(i.e., lodged) specimens (Robertson et al., 2015a).
Three-point bending tests were performed using an
Instron universal testing machine (Model 5965, Instron Corp.,
Norwood, MA). A 500-N load cell was used to collect force
data every 100 ms, and the stalk was displaced at a constant rate
of 10 cm min–1 until breakage of the stalk occurred. Bluehill
software (Instron, Norwood, MA) was used to collect force
and displacement data during the test procedure. Force and
displacement data were used to calculate the induced bending
moment (M) according to the following equations for nonsymmetric three-point bending (Howell, 2001)
M (x ) =
Fbx
, x £a
L
Fa
(L - x ), x > a [2]
L
where x is the distance along the stalk measured from its left
side, a and b are the distances to the applied load as measured
from the supports on the left and right sides of the stalk respectively, L is the distance between supports (i.e., L = a + b), and F
is the applied load. Bending strength was defined as the maximum moment applied to the stalk at the location of stalk failure.
M (x ) =
Rind Penetration Resistance
To eliminate both intra- and inter-user variation in rind penetration resistance results (Gibson et al., 2010), measurements were
taken in a laboratory setting with the universal testing machine
described above. This setup ensured a constant insertion angle
and insertion velocity of the impinging needle. During each test
the stalk was placed on a flat horizontal surface and oriented such
that the minor axis of the stalk cross-section was facing up. The
middle section of the third above ground internode of each stalk
was then punctured by a steel needle at a constant rate of 30
mm s–1 (Gou et al., 2010). The needle was displaced until it had
completely punctured the rind and entered the pith tissues. A
force gauge attached to the needle measured the force of contact
crop science, vol. 56, july– august 2016 between the stalk and the needle. The needle was 1.5 mm in
diameter and tapered to a sharp point over a distance of 5 mm.
As in previous studies, rind penetration resistance was defined as
the maximum load achieved during each test.
Statistics
Correlation analyses were used to assess and compare the effectiveness of stalk flexural stiffness and rind penetration resistance
as predictors of stalk strength. As a first assessment, measurements
from all hybrids, planting densities, locations, and replicates
were combined and a univariate regression was performed with
stalk flexural stiffness and rind penetration resistance as independent variables and stalk strength as the dependent variable
(i.e., strength vs. stalk flexural stiffness and strength vs. rind
penetration resistance). Coefficients of determination (R 2) and
associated p-values were calculated for both stalk flexural stiffness and rind penetration resistance regression lines.
A second assessment of the effectiveness of stalk flexural
stiffness and rind penetration resistance as predictors of stalk
strength was performed by conducting correlation analyses on
grouped data. Data were grouped according to the five hybrid
types, five planting densities, two locations, and two replicates
investigated in this study. Therefore, a total of 16 regression
relationships (for five hybrids, five planting densities, two
locations, and two replicates per location) were determined for
stalk flexural stiffness and another 16 were determined for rind
penetration resistance. R 2 values, p-values, and regression line
coefficients were computed for each regression. Results from
rind penetration resistance regressions were then compared to
results from stalk flexural stiffness regressions.
Finally a third analysis was conducted to determine the
consistency or robustness of regression relationships for rind
penetration resistance and stalk flexural stiffness. Can the relationships between rind penetration resistance/stalk flexural
stiffness and stalk strength obtained from one set of stalks be
used to predict the strength of a second unique set of stalks?
That is, can the regression model obtained from one hybrid be
used to predict the strength of other hybrids, or can the relationship between stalk flexural stiffness and strength obtained
at one planting density be used to predict the strength of stalks
grown at other planting densities? To answer this question, the
experimental data were parsed into training sets and validation
sets. The 16 regression relationships described in the preceding paragraph were used as training sets. The 16 corresponding
validation datasets consisted of all the remaining experimental
data not included in the training set. For example, with hybrid 1
data as the training set, the validation set consisted of data from
hybrids 2 to 5, for a 20:80 split between training and validation
data. For each training and validation dataset, a coefficient of
determination was calculated between the regression line of the
training set and the data of the validation set. This coefficient of
determination (R 2valid) was used to assess the predictive power
of rind penetration resistance and stalk flexural stiffness for each
grouping factor in the randomized block design. To avoid any
ambiguity, the coefficient of determination of the training set
data is referred to as R 2 while the coefficient of determination
between the regression line of the training set and the data of
the validation set is referred to as R 2valid.
www.crops.org1713
RESULTS
Stalk failure modes and failure locations produced by laboratory measurements of bending strength in this study
were analogous to failure modes and locations observed in
naturally lodged stalks. In particular, consistent with previously reported field data (Robertson et al., 2015a), 98%
of all stalks broke due to a creasing type failure, and 95%
of all failures occurred within 3 cm of a node. In addition,
artificial deformations of stalk cross-sections were not
observed during bending tests (Robertson et al., 2014).
Univariate Regression Analysis
Univariate regression analysis revealed that both rind penetration resistance and stalk flexural stiffness are correlated
with stalk strength. However, stalk flexural stiffness was a
much stronger predictor of stalk bending strength than rind
penetration resistance. Only 18% of the variation in stalk
strength was predicted by rind penetration resistance while
stalk flexural stiffness predicted 81% of the variation in stalk
strength. Furthermore, stalk flexural stiffness accurately
predicted stalk strength across the entire range of measured
stalk flexural stiffness values whereas the accuracy of rind
penetration resistance varied considerably with rind penetration resistance values. Scatterplots, R2 values, p-values,
and the resulting lines of linear best fit for rind penetration
resistance vs. strength and stalk flexural stiffness vs. strength
are provided in Figure 1. To ensure regression analyses were
not being negatively affected by outliers, the top 5% of data
points with the largest “Cook’s distances” were removed
and the regressions were repeated. After removing outliers,
the R 2 for rind penetration resistance was 0.23 and the R 2
for stalk flexural stiffness was 0.84.
Regression Analyses of Grouped Data
Rind penetration resistance was an inferior predictor of
stalk strength as compared with stalk flexural stiffness when
data were grouped by factors of the experimental design.
In addition, rind penetration resistance predictions were
negatively affected by experimental factors whereas stalk
flexural stiffness predictions were fairly consistent across all
experimental factors. R 2 values of grouped data for rind
penetration resistance ranged from a minimum of 0.04
to a maximum of 0.32 with a mean of 0.19. In contrast,
the minimum R 2 for stalk flexural stiffness was 0.63, with
a maximum of 0.87 and a mean of 0.81. In other words,
the worst flexural stiffness relationship predicted twice as
much variation in stalk strength as the best rind penetration
resistance relationship. Regression line coefficients for rind
penetration resistance were also observed to vary substantially between groups whereas regression line coefficients
for stalk flexural stiffness remained fairly constant between
groups. Scatter plots and regression lines computed from
each level of each experimental factor investigated in this
study are presented in Figure. 2. The R 2 values and regression line coefficients are presented in Table 1.
Predictive Capabilities (Robustness)
of Regressions
The predictive potential of stalk flexural stiffness and
rind penetration resistance was assessed with training
and validation datasets (see methods above for a complete
explanation). In particular, the regression line of each
training dataset was compared to data from a validation
set to calculate R 2valid, the coefficient of determination
between the regression line of the training set and the
data of a validation set. The value of R 2valid represents the
predictive power of the training set.
This analysis demonstrated that stalk flexural stiffness
was much better at predicting the stalk strength values
of validation datasets than rind penetration resistance
was. For example, the regression between rind penetration resistance and stalk strength of hybrid 1 (training set)
Fig. 1. Scatterplots, linear regression lines, and coefficients of determination (R2) for univariate correlation analyses of (a) rind penetration
resistance vs. strength, and (b) stalk flexural stiffness vs. strength, demonstrating the superior predictive power of stalk flexural stiffness.
1714
www.crops.org
crop science, vol. 56, july– august 2016
Fig. 2. Regression lines of rind penetration resistance vs. stalk strength and stalk flexural stiffness vs. stalk strength. (a–d) Rind penetration
resistance data. (e–h) Stalk flexural stiffness data. Data are grouped according to hybrid (a, e), planting density (b, f), location (c, g) and
replicate (d, h). Stalk flexural stiffness was consistently a better predictor of stalk strength than rind penetration resistance.
accounted for 32% of the variation in stalk strength of
hybrid 1 (R 2 = 0.32). However, using the regression relationship from hybrid 1 to predict strength values of hybrids
2 to 5 based on their measured rind penetration resistance
values only predicted 7% of the variation in strength of
hybrids 2 to 5 (R 2valid = 0.07). In contrast, the R 2 value
for stalk flexural stiffness of hybrid 1 (training set) was
0.87, while R 2valid was 0.76 (i.e., the stalk flexural stiffness
regression line for hybrid 1 predicted 76% of the variation in strength observed in hybrids 2 to 5). The ability
of stalk flexural stiffness to predict outside its own dataset
was consistent across all factors and levels. This was not
the case for rind penetration resistance, which exhibited
a substantially lower and more variable capacity for prediction outside of the training dataset. These trends are
shown in the descriptive statistics provided in the lower
section of Table 1. For rind penetration resistance, R 2valid
values ranged from −1.0 to 0.23 with a median value of
0.12. Stalk flexural stiffness demonstrated a very different pattern, with R 2valid values ranging from 0.74 to 0.83,
crop science, vol. 56, july– august 2016 indicating that stalk flexural stiffness is well-suited for
assessing stalk strength, even for data based on different hybrids, planting densities, etc. Relative differences
between R 2 and R 2valid are also provided in Table 1,
showing that the relative difference for rind penetration
resistance data were substantially higher and more variable
than the relative differences for stalk flexural stiffness.
Box plots of all of the R 2 and R 2valid values for stalk
flexural stiffness and rind penetration resistance (Fig. 3)
illustrate systematic differences between measures of rind
penetration resistance and stalk flexural stiffness as predictors of stalk strength. All R 2valid values for stalk flexural
stiffness are >0.74, indicating a strong pattern of intergroup predictive power. In addition, the variance in both
R 2 values and R 2valid values was smaller for stalk flexural
stiffness than for rind penetration resistance. As expected,
R 2valid values which were computed from validation datasets tend to be lower than the R 2 values calculated from
the training datasets.
www.crops.org1715
Table 1. Regression statistics from each level of each experimental variable investigated in the study. Summary statistics are
provided at the bottom of the table.
Slope
Rind penetration resistance
Y-int
R2
R2valid
Diff†
Slope
Stalk flexural stiffness
Y-int
R2
R2valid
Diff†
Hybrid 1
Hybrid 2
Hybrid 3
Hybrid 4
Hybrid 5
59,000 plants ha–1
74,000 plants ha–1
89,000 plants ha –1
104,000 plants ha–1
119,000 plants ha –1
Location 1
Location 2
Rep 1, Location 1
Rep 2, Location 1
Rep 1, Location 2
Rep 2, Location 2
0.48
0.33
0.28
0.37
0.23
0.10
0.23
0.13
0.26
0.21
0.28
0.35
0.37
0.34
0.38
0.20
3.02
6.29
4.86
5.82
6.85
12.21
7.06
7.63
4.28
4.21
6.20
5.28
5.44
5.07
4.57
7.19
0.32
0.15
0.28
0.17
0.15
0.04
0.17
0.09
0.24
0.28
0.13
0.25
0.26
0.28
0.18
0.09
0.07
0.13
0.01
0.05
0.16
−1.00
0.07
0.13
0.04
−0.39
0.23
0.12
0.12
0.16
0.16
0.17
%
−79
−14
−97
−68
7
−2940
−56
49
−82
−239
80
−53
−54
−45
−10
84
´10­–4
1.58
1.29
1.56
1.62
1.63
1.30
1.39
1.44
1.50
1.45
1.58
1.29
1.46
1.47
1.39
1.73
0.31
2.30
1.38
1.62
1.22
3.64
2.66
1.85
1.28
1.40
0.31
2.30
1.80
1.91
2.12
−0.04
0.87
0.84
0.80
0.84
0.80
0.63
0.79
0.77
0.84
0.78
0.87
0.84
0.81
0.86
0.82
0.80
0.76
0.76
0.81
0.74
0.79
0.75
0.80
0.82
0.80
0.77
0.83
0.80
0.81
0.80
0.81
0.79
%
−13
−9
1
−12
−1
18
1
7
−6
−1
−5
−6
0
−7
−1
−2
Max.
Min.
Median
Mean
SD
0.48
0.10
0.28
0.28
0.10
12.21
3.02
5.63
6.00
2.08
0.32
0.04
0.18
0.19
0.08
0.23
−1.00
0.12
0.01
0.30
84
−2940
−54
−220
730
1.73
1.29
1.47
1.48
0.13
3.64
−0.04
1.71
1.63
0.93
0.87
0.63
0.81
0.81
0.06
0.83
0.74
0.80
0.79
0.03
18
−13
−2
−2
7
† The percentage difference between R2 and R2valid.
DISCUSSION
An ideal breeding tool for combating stalk lodging should
(i) accurately predict stalk strength, (ii) not be confounded
by environmental factors or covariates such as hybrid,
(iii) require minimal effort and time to use, and (iv) not
permanently damage the stalk, thus allowing analysis of
individual plants throughout their lifecycle (i.e., collection of temporal data). The consensus among previous
researchers is that rind penetration resistance best adheres
to these ideals and is therefore a good benchmark against
which to compare future breeding tools.
In the current study, stalk flexural stiffness outperformed rind penetration resistance with regards to the
aforementioned ideals. In particular, stalk flexural stiffness
was shown to be (i) a better predictor of stalk strength than
rind penetration resistance, (ii) not affected by hybrid type,
planting density, location, or replicate, whereas rind penetration resistance was moderately to strongly confounded
by such factors, and (iii) a method that does not induce any
permanent damage to the stalk whereas rind penetration
resistance requires piercing the stalk rind and pith. This
weakens the structural integrity of the plant and leaves
it susceptible to disease. Note that in the current study
stalk flexural stiffness measurements and rind penetration
resistance measurements were acquired in a controlled
laboratory setting to minimize measurement error. Therefore, time to phenotype or implement a rind penetration
resistance tool in the field vs. time to implement a tool for
1716
Fig. 3. Box plots of coefficients of determination for (a) rind penetration resistance and (b) stalk flexural stiffness.
measuring stalk flexural stiffness in the field was not evaluated in this study. Development of new measurement tools,
which can acquire bending strength and flexural stiffness
measurements of stalks in the field, will enable future investigations of time to phenotype. The authors expect stalk
flexural stiffness and rind penetration resistance to be similar with regards to difficulty of use and time to phenotype.
Stalk flexural stiffness can be acquired in the field by measuring the force response of stalks that have been deflected
www.crops.org
crop science, vol. 56, july– august 2016
by 5-6 mm. Rind penetration resistance can likewise be
obtained in the field using a handheld penetrometer.
The bending strength of any structure (including corn
stalk) is determined by two governing factors, namely the
structure’s material properties and the structure’s morphology (i.e., anatomical geometry). For most structures, shape
(morphology) dominates the bending response. For example, a recent engineering analysis of corn stalk structure and
strength demonstrated that changes in stalk morphology
are on average 18-fold more influential on stalk mechanical
stresses than are changes in material properties (Von Forell
et al., 2015). Selective breeding tools which account for
both morphologic and material contributions to stalk bending strength will likely outperform tools which focus solely
on material properties or stalk chemistry. Stalk flexural
stiffness measurements are influenced by both morphologic and material contributions to stalk strength, whereas
rind penetration resistance primarily measures the material properties of the rind (Twumasi-Afriyie and Hunter,
1982). Furthermore, rind penetration resistance only measures material properties of the stalk at a single point (the
point of needle insertion). In contrast, stalk flexural stiffness is influenced by both morphologic and material effects
throughout the entire portion of the flexed stalk. These
insights explain why stalk flexural stiffness is a better predictor of strength than rind penetration resistance. They
also explain why rind penetration resistance is strongly to
moderately affected by the covariates (grouping factors)
investigated in this study. Rind penetration resistance does
not accurately account for morphologic contributions to
stalk strength; therefore, covariates which affect stalk morphology will negatively impact rind penetration resistance
predictions. Because stalk flexural stiffness accounts for
morphologic contributions to stalk strength, it is not negatively affected by covariates which affect stalk morphology.
In support of this explanation is the fact that rind penetration resistance regression line coefficients varied the most
and rind penetration resistance R 2valid values were lowest
when analyzing different planting densities (the covariate
which most strongly influences stalk morphology).
Many prevalent problems in crop science are governed
by mechanical principles that require engineering expertise to correctly address. Top scientific academies, academic
organizations, and policymakers unanimously agree that
including engineers in plant research will accelerate research
progress and lead to development of breakthrough technologies (National Research Council, 2009; White House,
2012; Plant Science Research Summit, 2013). This study
and others have incorporated the advice of senior leadership with great success and have produced promising results
for addressing the long-standing problem of stalk lodging.
For example, in the past two years, engineering collaborations have revealed remarkably consistent yet previously
unrecognized failure patterns in naturally lodged corn
crop science, vol. 56, july– august 2016 stalk (Robertson et al., 2015a), illuminated the dominating effects of morphology on stalk strength (Von Forell et
al., 2015), discovered several structural weakness in corn
stalk architecture (Robertson et al., 2015b; Von Forell et
al., 2015), improved testing methodologies for measuring
stalk strength (Robertson et al., 2014, 2015a) and identified
a potentially transformative breeding tool for developing
lodging resistant crop varieties (i.e., stalk flexural stiffness).
These advances could not have been achieved by either
plant scientist or engineers working in isolation. Similar
collaborations will likely lead to rapid advances on numerous fronts in the future.
Limitations
All hybrids investigated in the current study were commercially available and demonstrate acceptable lodging
resistance. Thus, they represent a fairly narrow range of
possible stalk strength values. Previous studies investigating
rind penetration resistance have generally used commercial
and non-commercial maize varieties which represent an
extremely broad range of stalk strength values (e.g., Cloninger et al., 1970; Twumasi-Afriyie and Hunter, 1982). The
use of broad sampling designs of previous studies tends to
inflate R 2 values. For example, some studies have reported
correlation values of up to 0.98 for rind penetration resistance and stalk strength (Twumasi-Afriyie and Hunter,
1982). The current study included only lodging-resistant
varieties so that the utility of rind penetration resistance and
stalk flexural stiffness as breeding tools in late stage breeding trials could be determined.
All stalks investigated in the current study were mature
and stalk moisture had been reduced to stable levels. This
study design was chosen as the authors were primarily
interested in the problem of late-season stalk lodging which
occurs after stalks have matured and are left to dry in the
field. However, results from this study may not be applicable to green stalks. Turgor pressure could change regression
line coefficients for both stalk flexural stiffness and rind
penetration resistance. The ability of stalk flexural stiffness
to determine the strength of green stalks remains the subject of future research. In other words, even though stalk
flexural stiffness is nondestructive it remains to be determined if it can be used to predict the strength of corn stalk
at different growth stages.
All mechanical tests performed in the current study
(strength, stalk flexural stiffness, and rind penetration
resistance) were conducted under controlled laboratory
conditions with an Instron universal testing system. In addition, all stalks had approximately the same moisture content.
Such optimal conditions do not exist in field-based experiments. Field-based mechanical measurements are generally
more difficult to replicate than laboratory-based measurements and the moisture content of stalks in the field may be
more variable than those tested in the current study. These
www.crops.org1717
factors typically increase measurement error and therefore
decrease measurement reliability. However, a field-based
tool for measuring stalk strength and stalk flexural stiffness
also presents some advantages compared to laboratory-based
measurements. In particular, the natural boundary conditions of the stalk (e.g., the attachment of the stalk to the soil,
leaf and ear weight) can be maintained in a field-based test.
This enables the test to more faithfully represent the loading conditions of the stalk in a wind or rain storm and may
therefore serve to strengthen the correlation between stalk
flexural stiffness, stalk strength, and stalk lodging. Unfortunately such field tools have yet to developed and validated.
This remains the subject of future work.
CONCLUSIONS
The purpose of this study was to investigate the possibility of using stalk flexural stiffness as a predictor of stalk
strength. Results demonstrated that stalk flexural stiffness is
a good predictor of stalk strength. Furthermore, stalk flexural stiffness may outperform rind penetration resistance as
a selective breeding tool to improve stalk lodging resistance
in the future. In particular, results show that stalk flexural
stiffness can predict strength across various growing conditions (i.e., predictions are not confounded by different
hybrid types, planting densities, or planting locations). The
univariate stalk flexural stiffness regression line presented
in this paper will therefore likely be able to predict the
strength of future varieties of corn stalk in the absence of
stalk strength measurements. Accurate methods for obtaining measures of stalk flexural stiffness have been presented
so that plant scientist may begin investigating the utility of
stalk flexural stiffness as a selection index in selective breeding trials. Tools for measuring stalk bending strength and
stalk flexural stiffness in the field are currently being developed and will be evaluated in the future.
Rind penetration resistance appears to be a less promising breeding tool as it did not accurately predict stalk
strength across different growing conditions. Rather, the
relationship between rind penetration resistance and stalk
strength was highly dependent on planting density, hybrid
type, and location, thus potentially limiting its use in
selective breeding trials. In summary, collaborative efforts
between plant scientists and engineers have produced
promising results and may show stalk flexural stiffness to
be an effective breeding tool for combating late-season
stalk lodging in the future.
Acknowledgments
This research was supported by the Core Technology Platform
Resources at New York University–Abu Dhabi, by funding from
the National Institute of Food and Agriculture Award No. 201667012-24685, and by funding from the National Science Foundation Award No. 1400973. Corn stalk samples were generously
donated by Monsanto Company.
1718
References
Beer, F.P., and E.R. Johnston, Jr. 1981. Mechanics of materials.
McGraw-Hill, New York.
Cloninger, F., M. Zuber, H. Calvert, and P. Loesch, Jr. 1970.
Methods of evaluating stalk quality in corn. Phytopathology
60:295–300. doi:10.1094/Phyto-60-295
Flint-Garcia, S.A., C. Jampatong, L.L. Darrah, and M.D. McMullen. 2003. Quantitative trait locus analysis of stalk strength
in four maize populations. Crop Sci. 43:13–22. doi:10.2135/
cropsci2003.0013
Gibson, B., C. Parker, and F. Musser. 2010. Corn stalk penetration
resistance as a predictor of southwestern corn borer (Lepidoptera: Crambidae) survival. Midsouth Entomologist 3:7–17.
Gou, L., J. Huang, R. Sun, Z. Ding, Z. Dong, and M. Zhao. 2010.
Variation characteristic of stalk penetration strength of maize
with different density-tolerance varieties. Trans. Chinese Soc.
Agric. Engineering 26:156–162.
Gou, L., J. Huang, B. Zhang, T. Li, R. Sun and M. Zhao. 2007.
Effects of population density on stalk lodging resistant mechanism and agronomic characteristics of maize. (In Chinese with
English abstract). ACTA Agronomica Sinica 33(10):1688–1695.
Hagenbaugh, B. 2007. Corn has deep economic roots as high
prices create ripple effect. USA Today 24 Jan. 2007. http://
usatoday30.usatoday.com/money/industries/food/2007-0124-corn_x.htm (verified 17 Mar. 2016).
Howell, L.L. 2001. Compliant mechanisms. John Wiley & Sons,
New York.
Hu, H., Y. Meng, H. Wang, H. Liu, and S. Chen. 2012. Identifying quantitative trait loci and determining closely related stalk
traits for rind penetrometer resistance in a high-oil maize
population. Theor. Appl. Genet. 124:1439–1447. doi:10.1007/
s00122-012-1799-5
McRostie, G., and J. MacLachlan. 1942. Hybrid corn studies. Sci.
Agric. 22:307–313.
National Research Council. 2009. Committee on a new biology
for the 21st century: Ensuring the United States leads the
coming biology revolution. US National Academies Press,
Washington DC.
Pedersen, J., and J. Toy. 1999. Measurement of sorghum stalk
strength using the Missouri-modified electronic rind penetrometer. Maydica 44:155–158.
Plant Science Research Summit. 2013. Unleashing a decade of
innovation in plant science: A vision for 2015–2025. American Society of Plant Biologists, Rockville, MD. http://plantsummit.wordpress.com/ (verified 17 Mar. 2016).
Robertson, D., M. Julias, B. Gardunia, T. Barten, and D. Cook.
2015a. Corn stalk lodging: A forensic engineering approach
provides insights into failure patterns and mechanisms. Crop
Sci. 55(6):2833–2841. doi:10.2135/cropsci2015.01.0010
Robertson, D., S. Smith, B. Gardunia, and D. Cook. 2014. An
improved method for accurate phenotyping of corn stalk strength.
Crop Sci. 54:2038–2044. doi:10.2135/cropsci2013.11.0794
Robertson, D.J., S.L. Smith, and D.D. Cook. 2015b. On measuring the bending strength of septate grass stems. Am. J. Bot.
102:5–11. doi:10.3732/ajb.1400183
Twumasi-Afriyie, S., and R. Hunter. 1982. Evaluation of quantitative methods for determining stalk quality in short-season corn
genotypes. Can. J. Plant Sci. 62:55–60. doi:10.4141/cjps82-008
Von Forell, G., D. Robertson, S.Y. Lee, and D.D. Cook. 2015.
Preventing lodging in bioenergy crops: A biomechanical
analysis of maize stalks suggests a new approach. J. Exp. Bot.
66(14):4093–4095.
White House. 2012. National Bioeconomy Blueprint. US White
House, Washington, DC.
www.crops.org
crop science, vol. 56, july– august 2016