Published online August 17, 2005

Interactions of Nitrogen, Weather, Soil, and Irrigation on Corn Yield

Nathan E. Derby,* Dean D. Steele, Jeff Terpstra, Raymond E. Knighton, and Francis X. M. Casey

ABSTRACT but no difference in yields with increased N when the

Numerous factors that affect corn yield in the Northern Great season was cool.
Plains are not considered in current N fertilizer recommendations. A Soil moisture status is also important in maintaining
linear model to describe the interactions of N fertility, weather, soil, maximum corn yields, and maintaining optimal soil mois-
Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved.

and irrigation on corn (Zea mays L.) yield was developed for a 6-yr ture is facilitated by irrigation in areas of coarse-textured
field study in southeastern North Dakota. Varying rates of postemer- soils. Stegman (1982) found that yields decreased with
gence fertilizer N were applied to plots in four quadrants of a sprinkler- increased water stress, but if the water stress is limited
irrigated field. Four irrigation-scheduling methods varied by year (Y) to one portion of the growing season and irrigation replen-
and quadrant (Q). Grain yield, stover yield, and N uptake response ishes the profile moisture, high yields can still be achieved.
to a covariate (soil N plus starter fertilizer N), N treatment (Nt), Q,
Some research has indicated that crop N requirements
Y, irrigation treatment (I), previous irrigation treatment (P), and
the two-way interactions that included Nt, were investigated using
are lower for irrigated versus rainfed soils (Oberle and
standard regression techniques. Yield increased significantly with Nt Keeney, 1990), unless rainfed condition results in water
up to a rate of 135 kg N ha⫺1 for most Y, Q, and I. Average yearly stress.
yield was adversely affected by cool climatic conditions in 2 yr. Soil Current recommendations for corn fertilization in
conditions in two quadrants resulted in a 1.25 Mg ha⫺1 average annual North Dakota indicate that 21.4 to 26.8 kg of N is needed
yield reduction compared with the other quadrants. An irrigation per Mg of corn grain yield, which is equivalent to 1.2
scheduling method utilizing a water balance algorithm resulted in to 1.5 lb of N per standard (56 lb) bushel of corn in
higher yields compared with the other scheduling methods. The inter- conventional units (Dahnke et al., 1992; Berglund and
actions of Nt with Q, Y, and I were significant. Optimized total N McWilliams, 1999). Typically, the N recommendations
rates (the sum of soil nitrate N, starter N, and postemergence Nt) derived
are adjusted for soil test N, date of sampling, and previ-
from quadratic-linear plateau analysis varied (232–374 kg N ha⫺1) with
Y, Q, and I and averaged 295 kg N ha⫺1. Soil properties, soil moisture
ous crop. Current nutrient recommendations, which are
availability, and yearly climatic factors should be considered when based on many site-years of yield response curves, make
making fertilizer recommendations. no adjustments for variation in weather conditions, soil
type, and management factors on a site-specific or year-
specific basis (Dahnke and Olson, 1990).
The objective of this study was to determine the yield
M aximum corn grain yields are influenced by a
number of factors, including N fertility, growing
season conditions, water availability, and soil conditions.
response of corn in the Northern Great Plains to the
combined effects of increased applied N fertilizer, irriga-
Studies investigating N fertility rates have shown that tion scheduling method, soil type, and yearly climatic
corn yields increase in a quadratic fashion with increased variation.
applied N to a plateau level (Cerrato and Blackmer,
1990). Maximum yields have been observed at N rates MATERIALS AND METHODS
less than the maximum rates applied in various studies
(Onken et al., 1985; Schlegel and Havlin, 1995). Albus Field Experiments
and Moraghan (1995) found marked yield responses to For the 1990 through 1995 growing seasons, a field-scale
N fertilizer in a number of short-season corn hybrids. study of irrigation and fertilizer N management effects on
In the Northern Great Plains, favorable growing con- groundwater quality was conducted near Oakes, ND, USA.
ditions are important due to the short growing season. Meteorological data were measured at an automated weather
Derby et al. (2004) found that yield was adversely af- station approximately 2.4 km from the site. The weather sta-
fected by cool growing season conditions. Similarly, tion was located at 46⬚04⬘ N, 98⬚06⬘ W, and 401 m elevation and
is part of the North Dakota Agricultural Weather Network
Swan et al. (1987) reported decreased corn yields with
(NDAWN) (Enz et al., 1997). Growing degree units were
insufficient cumulative growing degree days (CGDD). calculated from NDAWN data based on a lower limit of 10⬚C
Wienhold et al. (1995) found 60% greater yields with and an upper limit of 30⬚C (Gilmore and Rogers, 1958). Rain-
increasing N fertility when the growing season was warm fall and irrigation amounts were measured directly at the field
site. This site was 64 ha, of which 53.4 ha was irrigated by a
sprinkler (center pivot) irrigation system. The soil at the site
N.E. Derby and F.X.M. Casey, Dep. of Soil Sci., D.D. Steele, Dep. (Fig. 1) was primarily Hecla loamy fine sand (sandy, mixed,
of Agric. and Biosyst. Eng., and J. Terpstra, Dep. of Stat., North frigid Oxyaquic Hapludoll) on the south portion of the field
Dakota State Univ., Fargo, ND 58105; and R.E. Knighton, USDA-
CSREES, Washington, DC 20250. Received 14 Feb. 2005. *Corre-
and Wyndmere fine sandy loam (coarse-loamy, mixed, super-
sponding author ([Link]@[Link]). active, frigid Aeric Calciaquoll) and Stirum fine sandy loam

Published in Agron. J. 97:1342–1351 (2005).

Nitrogen Management Abbreviations: CGDD, cumulative growing degree days; ET, evapo-
doi:10.2134/agronj2005.0051 transpiration; I, irrigation treatment; NE, northeast; Nt, nitrogen treat-
© American Society of Agronomy ment; Ntot, total nitrogen; NW, northwest; P, previous irrigation treat-
677 S. Segoe Rd., Madison, WI 53711 USA ment; Q, quadrant; SE, southeast; SW, southwest; Y, year.

1342
 DERBY ET AL.: INTERACTIONS OF N, WEATHER, SOIL, AND IRRIGATION ON CORN YIELD 1343
Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved.

Fig. 1. Order 1 soil survey, plot locations, and quadrant boundaries on research quarter section. Soil electrical conductivity (dS m⫺1) is indicated
by dashed contour lines.

(coarse-loamy, mixed, superactive, frigid Typic Natraquolls) Coe, 1984) that allowed 60 to 70% estimated depletion of
on the north portion of the field. plant-extractable water before the first irrigation, full evapo-
The field was divided into four quadrants (denoted NW ⫽ transpiration (ET) replacement from the first irrigation through
northwest, NE ⫽ northeast, SW ⫽ southwest, and SE ⫽ south- the blister kernel (R2) stage, followed by irrigation whenever
east) to accommodate differences in soil types and for four soil moisture depletion reached 60 to 70% for the period R2
different irrigation scheduling method purposes. Irrigation through maturity; (C) irrigation based on a water balance algo-
treatments were assigned to quadrants every year based on a rithm that replaced 80% of estimated crop ET, i.e., the same
statistical design (Cochran and Cox, 1957) in which years and criteria as Method B, but each irrigation was 20% smaller;
quadrants were the dimensions of a modified Latin square. and (D) irrigation based on CERES–Maize crop growth model
Steele et al. (2000) described the irrigation aspects of the (Jones and Kiniry, 1986) predictions of plant-extractible soil
experiment in detail, including complete descriptions of the water (PESW), i.e., irrigations occurred when the estimated
treatments, irrigation scheduling criteria, application depths PESW was at 60 to 70% depletion. Steele et al. (1994) de-
targeted for each irrigation event (32 mm per event for Method scribed the CWSI method used in this study. Analysis of irriga-
A, B, and D and 25 mm per event for Method C), and seasonal tion efficiency values showed no significant differences due
(1991–1995) irrigation totals (averaged 136, 125, 106, and to irrigation methods (Steele et al., 2000). Irrigation events
122 mm for Methods A, B, C, and D, respectively). The irriga- were applied by on-site research personnel.
tion scheduling methods were (A) irrigation when tensiometer Tillage, planting, cultivation, pesticide application, and starter
readings at 0.30-m depth reached 40 ⫾ 10 kPa and/or crop fertilization were done by the farmer cooperator. Spring seed-
water stress index (CWSI) values reached 0.25 ⫾ 0.05; (B) bed preparation tillage was done with a combination disk/
irrigation based on a water balance algorithm (Stegman and chisel plow implement, usually two times before planting. The
 1344 AGRONOMY JOURNAL, VOL. 97, SEPTEMBER–OCTOBER 2005

Table 1. Planting, harvest, and fertilizer split dates; average spring soil N; and starter fertilizer rates for plots.
Dates of split N
Planting Emergence Spring Harvest
Year date† date soil N Starter N Starter P V6 V15 R1 date
kg ha⫺1 kg ha⫺1 kg ha⫺1
1990 7 May 23 May 57 14 67‡ 22 June 17 July 1 Aug. 25 Sept.
1991 13 May 22 May 61 14 67‡ 27 June 15 July 22 July 18 Sept.
1992 30 Apr. 12 May 62 14 67 15 June 22 July 4 Aug. 30 Sept.
1993 5 May 18 May 38 14 67 22 June 2 Aug. 11 Aug. 21 Sept.
Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved.

1994 22 Apr. 16 May 55 48 58 14 June 12 July 25 July 28 Sept.

1995 5 May 23 May 34 11 56 20 June 20 July 2 Aug. 20 Sept.
† Corn variety Pioneer 3737 planted at a rate of 69 200 to 74 100 seeds ha⫺1.
‡ Additional fertility corrections applied with starter: 16.8 kg ha⫺1 ZnSO4 yielding 5.6 kg ha⫺1 Zn; 56 kg ha⫺1 IRON-SUL yielding 0.8 kg ha⫺1 K2O, 28 kg
ha⫺1 S, 11.2 kg ha⫺1 Fe, and 0.4 kg ha⫺1 Cu.

corn was planted in rows spaced 76.2 cm apart at a seed density ears were harvested, and the whole plants were weighed to
of approximately 74 000 seeds ha⫺1. Starter fertilizer was ap- calculate stover yield. The ears were dried at 60⬚C and then
plied with the seed during the planting operation. Weed con- shelled. The dry grain was weighed, and yields were calculated
trol was a combination of between-row cultivation at about on a 15.5% moisture content, wet basis. Grain and stover
V6 and spraying of selective herbicides at appropriate times. subsamples were ground and analyzed for Ntot by the Kjeldahl
Pesticide applications were done on an as-needed basis to method (Bremner and Mulvaney, 1982).
control insect infestations.
The research team scheduled the irrigations and managed
the postemergence fertilization of 72 small plots. These plots
Data Analysis
were established along two transects within the irrigated por- Linear model techniques were used to analyze the data.
tion of the field (Fig. 1). Each plot was 9.1 m (12 rows) wide The model for the main effects with two-way interactions
by 18.3 m long, and the plots were arranged in groups of six involving Nt is
plots with three six-plot groups per field quadrant. The loca-
tion of each plot area was referenced by angle and distance Gijklm ⫽ ␮ ⫹ ␤Xijklm ⫹ Nti ⫹ Ij ⫹ Pk ⫹ Yl ⫹ Qm ⫹
from fixed points for identification in subsequent years. The (NtI)ij ⫹ (NtP)ik ⫹ (NtY)il ⫹ (NtQ)im ⫹ εijklm,
fertilizer treatments within each block of six plots ranged from
a reference regime of zero applied N to what was considered i ⫽ 1, 2,…,6; j ⫽ 1, 2, 3, 4; k ⫽ 1, 2, 3, 4;
a nonlimiting regime, with four intermediate regimes. Treat- l ⫽ 1, 2,…,6; m ⫽ 1, 2, 3, 4 [1]
ments were designated as Nt1 ⫽ 0 kg N ha⫺1, Nt2 ⫽ 45 kg N
ha⫺1, Nt3 ⫽ 90 kg N ha⫺1, Nt4 ⫽ 135 kg N ha⫺1, Nt5 ⫽ 180 where the response G was determined for corn grain yield,
kg N ha⫺1, and Nt6 ⫽ 225 kg N ha⫺1, not including the starter stover yield, grain N content, grain N uptake, stover N content,
N. These N application treatments were randomly arranged stover N uptake, or total plant N uptake; ␮ is the overall
within each block at the outset of the study. Each Nt was effect; X is the covariate spring soil N plus starter fertilizer
applied to the same plot for each year of the study to facilitate N; Nt1, Nt2,..., Nt6 represent the effects of the six N treatments
a “mining down” of residual soil N at the lower, yield-limiting 0, 45, 90, 135, 180, and 225 kg N ha⫺1, respectively; I1, I2, I3,
rates, i.e., plots were not annually re-randomized. and I4 represent the effects of the four irrigation scheduling
Postemergence fertilizer Nt was applied in three splits, start- methods A, B, C, and D, respectively; P1, P2, P3, and P4 repre-
ing when the corn reached the six-leaf stage (V6) (Hanway, sent the effects of the previous year’s four irrigation scheduling
1966, p. 3–18). Dates of planting, spring soil test N, fertilizer methods A, B, C, and D, respectively; Y1, Y2,..., Y6 represent
application, and starter fertilizer rates are shown in Table 1. the effect of years 1990–1995, respectively; Q1, Q2, Q3, and Q4
The amount of starter fertilizer varied in 1994 and 1995 be- represent the effect of the four quadrants SW, SE, NW, and
cause the farmer cooperator planted the field. The fertilizer NE, respectively; (NtI) is the interaction effect of the six N
N amounts, forms, and application methods for each Nt are treatments and four irrigation treatments; (NtP) is the interac-
summarized in Table 2. The last two applications of N were tion effect of the six N treatments and the four previous-year
applied by hand as urea because of lack of equipment and irrigation treatments; (NtY) is the interaction effect of the six
the practicality of injecting urea ammonium nitrate (UAN) N treatments and the 6 yr; (NtQ) is the interaction effect of the
at an advanced stage of corn growth. Soil samples for N were six N treatments and the four quadrants; and εijklm represents
taken from each plot area at the four-leaf stage to a depth of random error. This model contains all of the possible two-
91 cm. way interaction terms involving N treatment. We note that
After the corn reached physiological maturity (R6), two some unforeseen linear dependencies in our design prevented
adjacent 4.6-m rows were hand-harvested from each plot. All us from fitting a model with other two-way (and higher order)

Table 2. Amounts, forms, and application methods for fertilizer N split applications.
N amount per fertility treatment
Growth stage Fertilizer form Application method Nt1 Nt2 Nt3 Nt4 Nt5 Nt6
kg N ha⫺1
Six leaf (V6) UAN† (28–0–0) liquid injection – 45 90 90 90 90
15 leaf (V15) urea (46–0–0) hand application‡ – – – 45 90 90
Silking (R1) urea (46–0–0) hand application‡ – – – – – 45
Total 0 45 90 135 180 225
† UAN, urea ammonium nitrate.
‡ Fertilizer was applied in the row and followed by 6 mm of irrigation to prevent volatilization losses.
 DERBY ET AL.: INTERACTIONS OF N, WEATHER, SOIL, AND IRRIGATION ON CORN YIELD 1345

interactions. However, since N was the main focus, two-way parameters will be limited. Discussion of Nt rate per-
interactions are fairly easy to interpret, the model R2 is high, tains to postemergence N application and does not in-
and no violations of statistical assumptions were identified clude the N applied as starter or soil test N. We note
(via a residual analysis), we feel that this is the most appro- that although the different Nts varied in timing, form
priate model for this data set. Analyses were performed using
S-Plus statistical software (MathSoft, 1999).
of fertilizer, and application method (Table 2), we feel
The yield response to Ntot (the sum of soil nitrate N, starter that this type of split application resulted in the best
N, and Nt) was modeled with the SAS NLIN procedure (Ihnen crop utilization of the applied N by reducing the risk
and Goodnight, 1985) and is described by the following qua- of leaching losses compared with if all of the N had
Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved.

dratic-plus-plateau model: been applied at once.

Analysis-of-covariance summary information is listed
Y ⫽ a ⫹ bX ⫹ cX if X ⬍ C [2]
in Table 3. The residual standard error for corn grain
Y ⫽ YP if X ⱖ C [3] yield was 1.06 Mg ha⫺1 on 340 degrees of freedom. The
where Y is grain yield (Mg ha⫺1);
X is the sum of applied and R2 value for the model was 0.91, i.e., 91% of the yield
soil N (kg ha⫺1); a, b, and c are coefficients obtained by fitting response is explained by Eq. [1]. The F statistic for the
this model to the LSMEAN yield values; C is the critical or model was 36.66 on 90 and 340 df, resulting in p ⬍ 0.001.
optimized rate of fertilization (kg ha⫺1); and YP is the plateau Upon further examination, all factors in Eq. [1] for corn
yield (Mg ha⫺1). The values of C and YP are obtained from grain yield were found to be significant at ␣ ⫽ 0.05 or
the fitted model. The model (Eq. [2] and [3]) was used to lower except the NtP interaction and the covariate soil
determine the optimized N rate (the N at which maximum test N plus starter fertilizer N. High R2 values and similar
yield was achieved) for years, quadrants, and irrigation sched- significance of terms was observed for the other re-
uling method. Since the plateau yield was considered to be
the maximum yield, the value of C for maximum yield was
sponse variables grain N content, grain N uptake, stover
calculated by equating the first derivative of the response yield, stover N content, stover N uptake, and total plant
equations (Eq. [2]) to zero and solving for X (Eq. [4]) N uptake. This indicates that the model was appropriate
for this study.
X ⫽ ⫺b/2c [4]
This is the point C at which the quadratic and plateau portions Main Effects
of the curve meet and represents the optimized N rate for
maximum yield but not necessarily maximum economic re- Corn grain yields, grain N content and N uptake, stover
turn, as will be discussed later. yield, stover N content and N uptake, and total plant N
uptake in this study were influenced significantly by N
treatment (Nt), soil properties or quadrant (Q), yearly
RESULTS AND DISCUSSION climatic variations (Y), irrigation scheduling method (I),
While corn grain yield is the focus of the discussion, and the previous year’s irrigation scheduling method
the effects on the response variables grain N content (P). Corn grain yield, stover yield, plant N content, and
and N uptake, stover yield, stover N content and N N uptake as affected by each main effect are shown in
uptake, and total plant N uptake are also presented in Table 4. Significant differences of yield means within
the tables to illustrate the robustness of the statistical response parameters Nt, I, P, Y, and Q were indicated in
model used. However, the discussion regarding those the instances where the two-way interactions of I, P, Y,

Table 3. Analysis-of-covariance summary for irrigated corn grain yields, grain N percentage, grain N uptake, stover yield, stover N
percentage, stover N uptake, and total plant N uptake.
Grain Grain N Grain N Stover Stover N Stover N Total plant
Model summary† yield‡ content uptake yield content uptake N uptake
RSE 1060 kg ha⫺1 0.730 g kg⫺1 13.1 kg ha⫺1 810 kg ha⫺1 0.9793 g kg⫺1 7.33 kg ha⫺1 15.2 kg ha⫺1
R2 0.91 0.86 0.91 0.81 0.89 0.84 0.90
F(90, 340) 36.66*** 23.15*** 36.26*** 15.8*** 29.52*** 20.34*** 34.28***
Source§ df F statistic
␤ 1 2.80 9.11** 6.92** 1.36 3.76 4.403* 8.605**
Nt 5 162*** 196*** 223*** 58.1*** 101*** 116*** 260***
I 3 7.43*** 4.75** 5.82*** 5.32** 1.88 3.58* 6.23***
P 3 12.8*** 5.96*** 13.1*** 9.42*** 2.19 5.45** 14.4***
Y 5 224*** 99.7*** 155*** 68.6*** 348*** 138*** 94.4***
Q 3 110*** 17.6*** 106*** 96.4*** 7.10*** 61.0*** 128***
NtI 15 1.90* 1.30 1.34 0.502 0.556 0.556 1.1
NtP 15 0.735 1.14 0.429 0.521 1.80* 1.37 0.424
NtY 25 10.7*** 3.24*** 10.0*** 4.49*** 4.19*** 2.38*** 6.90***
NtQ 15 1.99* 1.60 1.32 1.15 1.28 1.36 1.52
* Significant at the 0.05 level.
** Significant at the 0.01 level.
*** Significant at the 0.001 level.
† RSE ⫽ residual standard error, R2 ⫽ coefficient of multiple determination, and F(90, 340) ⫽ F statistic on 90 and 340 degrees of freedom.
‡ Adjusted to 15.5% moisture content, wet basis. Note: grain N content is on a dry matter basis.
§ ␤ ⫽ covariate of spring soil test N plus applied starter fertilizer N, Nt ⫽ N treatment, I ⫽ irrigation treatment, P ⫽ previous irrigation treatment, Y ⫽
year, Q ⫽ quadrant, NtI ⫽ N ⫻ irrigation interaction, NtP ⫽ N ⫻ previous irrigation interaction, NtY ⫽ N ⫻ year interaction, and NtQ ⫽ N ⫻
quadrant interaction.
 1346 AGRONOMY JOURNAL, VOL. 97, SEPTEMBER–OCTOBER 2005

Table 4. Corn grain yield (and other response variables) means and statistical groupings for main effects.
Corn grain Grain N Grain N Stover Stover N Stover N Total plant N
Treatment class yield† content uptake yield content uptake uptake
Mg ha⫺1 g kg⫺1 kg ha⫺1 kg ha⫺1 g kg⫺1 kg ha⫺1 kg ha⫺1
N treatment, kg N ha⫺1 (Nt)
0 4.93‡ 10.3 44.0 3903.2 4.17 16.9 60.8
45 7.21 11.0 68.6 5158.8 4.81 25.0 93.6
90 8.22 12.0 83.3 5660.1 5.61 31.4 114.7
135 9.24 13.2 102.8 5786.7 6.62 37.5 140.3
Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved.

180 9.50 13.4 107.6 6062.4 7.05 41.7 149.4

225 9.55 13.7 110.9 6037.3 7.54 44.9 155.8
Irrigation treatment (I)
A 7.86‡ 12.1a 83.5a 5262.7a 5.84a 31.0a 114.6ac
B 8.53 12.3a 90.3b 5687.6b 5.93ab 33.4b 123.8b
C 8.16 12.5b 87.6b 5505.3b 6.17b 34.6b 122.2b
D 7.88 12.2a 83.3a 5283.2a 5.92ab 32.5ab 115.8c
Previous irrigation treatment (P)
A 7.64a§ 12.2a 81.0a 5128.1a 5.85 31.0a 112.1a
B 8.11b 12.2a 85.1a 5447.6b 5.89 32.9a 118.0b
C 8.72c 12.6b 94.2b 5830.3c 6.22 35.7b 130.0c
D 7.96ab 12.2a 84.4a 5332.9ab 5.90 31.9a 116.3ab
Year (Y)
1990 9.95‡ 13.3 111.8 7181.1 5.53 40.2 152.0
1991 9.00 12.4 96.4 4759.6 6.09 29.4 125.8
1992 5.97 13.4 68.9 4703.2 4.13 20.0 88.9
1993 5.19 11.7 52.4 4776.5 10.27 50.1 102.5
1994 10.27 11.2 101.2 5545.1 5.19 30.1 131.3
1995 8.28 11.8 86.5 5642.9 4.59 27.5 114.0
Quadrant (Q)¶
SW 9.14‡ 12.5a 99.2a 6227.8a 6.05ac 37.8a 137.0a
SE 8.99 12.5a 96.2a 5993.2b 6.29a 37.8a 134.1a
NW 7.15 12.1b 74.8b 4790.8c 5.74b 28.1b 102.9b
NE 7.15 12.0b 74.6b 4727.1c 5.79bc 27.9b 102.4b
† Pairwise comparisons were done using LSMEANS values, not actual parameter estimates.
‡ Means are not compared because of statistically significant (␣ ⫽ 0.05) interactions with another variable.
§ For each response parameter (corn grain yield, grain N content, etc.), means in each treatment class (Nt, I, P, Y, or Q) followed by the same letter are
not significantly different at ␣ ⫽ 0.05.
¶ SW, southwest; SE, southeast; NW, northwest; NE, northeast.

and Q with Nt were not significant (Table 3). If there was soils on the northern portion of the field. Lower produc-
a significant interaction, response parameter means could tivity was a result of low water permeability and sodium
not be compared for that main effect. Further discussion effects on the Stirum soil and the higher salinity in the
about significant differences within interactions will follow. northern portion of the field [electrical conductivity
With regard to corn grain yield, previous irrigation (EC) ⫽ 3.0–3.5 dS m⫺1] compared with the south (EC ⫽
treatment (P) was the only main effect of which treat- 2.0–2.5 dS m⫺1) (Fig. 1). Although it would appear that
ment means could be compared since the NtP inter- the soil type is not the same for all plots within some
action was not significant (Table 3). Corn grain yield quadrants (Fig. 1), we maintain that it is very difficult
varied significantly by P with a previous year’s irrigation to delineate exactly where a change in soil type actually
treatment of C resulting in the highest yield. The effect occurs in the field. Experience has shown us that the
of P resulted in an average yield range of 1.08 Mg ha⫺1. productivity of the soil on the NE and NW quadrants
Comparison of response variable means (other than was much lower than the SE and SW. This was con-
grain yield due to significant interaction) for I treat- firmed by the statistically significant lower mean values
ments indicated that Irrigation Treatments B and C in the north versus the south Q treatments for all vari-
were not significantly different for all variables except ables, a result of reduced plant vigor. Lower plant N
grain N content and resulted in the highest values for uptake may result in excess nitrate leaching and ground-
grain N content and uptake and stover yield, N content, water contamination, not to mention the reduced eco-
and N uptake. Response variable means within previous nomic return due to overapplication of fertilizer. Yield
irrigation treatment (P) showed that Treatment C re- potential of different areas of a field needs to be consid-
sulted in significantly higher values for all variables ered when making N fertilizer recommendations and
(Table 4). Treatments B and C were based on the same may warrant variable-rate applications.
water balance algorithm with water applications being
20% smaller for C.
Two-Way Interactions
Comparison of means for grain N content and uptake
and stover yield, N content, and N uptake for main The interactions of Nt with the other main effects
effect Q showed a clear separation between the two were also investigated. The two-way interactions of Nt
north quadrants, NE and NW, and the two south quad- with Y, Q, and I are shown in Tables 5, 6, and 7, respec-
rants, SE and SW (Table 4). Quadrants SE and SW had tively. The interaction of Nt with P is not shown because
higher values for all variables. The quadrant effect was NtP was not significant for any of the response variables
basically a soil effect, with less productive salt-affected except stover N content.
 DERBY ET AL.: INTERACTIONS OF N, WEATHER, SOIL, AND IRRIGATION ON CORN YIELD 1347

Table 5. Treatment means and statistical groupings for the N ⫻ year interaction.
N treatment, kg N ha⫺1
Year 0 45 90 135 180 225
Corn grain yield†, Mg ha⫺1
1990 8.49a‡ 10.00b 10.03b 10.15b 10.62b 10.42b
1991 4.87a 8.91b 9.22bc 10.20cd 10.28cd 10.50d
1992 3.34a 5.07b 6.37c 6.87cd 7.32d 6.83cd
1993 3.73a 4.73b 5.64c 5.61c 5.70c 5.73c
Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved.

1994 5.99a 8.52b 10.09c 12.23d 12.42d 12.35d

1995 3.14a 6.06b 7.95c 10.38d 10.66de 11.47e
Grain N content, g kg⫺1
1990 12.0a 12.7b 12.7a 13.8b 14.0b 14.5b
1991 9.8a 11.1b 12.3c 13.6d 13.6d 13.8d
1992 11.7a 11.8a 13.3b 14.5c 14.5c 14.5c
1993 9.7a 10.6b 12.0c 12.4cd 12.5cd 12.9d
1994 8.7a 9.7b 10.9c 12.2d 12.8d 12.7d
1995 10.2a 10.3a 10.9b 12.5c 12.8c 13.9d
Grain N uptake, kg ha⫺1
1990 85.6a 105.0b 107.3b 118.4bc 126.3c 127.9c
1991 40.5a 84.6b 95.1b 116.9c 118.6c 122.8c
1992 33.3a 50.9b 71.7c 84.6d 89.5d 83.6d
1993 31.9a 43.2b 57.3c 59.1c 60.5c 62.5c
1994 44.9a 72.7b 94.0c 126.7d 135.0d 133.8d
1995 27.8a 55.0b 74.4c 111.0d 115.9d 134.7e
Stover yield, kg ha⫺1
1990 7155.3a 7075.1a 7194.5a 6943.0a 7594.3a 7124.4a
1991 3100.4a 4520.3b 4935.0bc 5266.0bc 5589.6c 5146.2bc
1992 2395.4a 4352.6b 5177.4c 5315.4c 5588.5c 5390.1c
1993 3615.2a 4685.2b 5229.9b 4911.9b 5043.1b 5173.5b
1994 4356.6a 5134.2b 5292.8bc 5936.8cd 6181.3d 6369.0d
1995 2796.3a 5185.2b 6131.1c 6346.8c 6377.3cd 7020.5d
Stover N content, g kg⫺1
1990 4.29a 5.12ab 4.88ab 5.72bc 6.20cd 7.00d
1991 3.83a 4.58a 5.76b 6.88c 7.51cd 7.97d
1992 3.17a 3.13a 3.72a 4.52b 4.83bc 5.38c
1993 6.83a 8.42b 10.38c 11.88d 12.17d 11.945d
1994 3.36a 4.07a 4.99b 5.93c 6.29c 6.50c
1995 3.56a 3.52a 3.91a 4.81b 5.29b 6.44c
Stover N uptake, kg ha⫺1
1990 30.5a 36.1ab 36.1ab 41.0bc 47.2cd 50.3d
1991 11.4a 19.7b 27.4b 35.7c 41.2c 41.0c
1992 7.9a 13.6ab 19.2bc 23.9cd 26.9d 28.6d
1993 25.7a 39.9b 53.9c 58.2cd 61.2d 61.8d
1994 15.0a 21.6b 27.0b 35.4c 39.6cd 42.1d
1995 10.7a 19.3b 24.6b 30.6c 34.3c 45.7d
Total plant N uptake, kg ha⫺1
1990 116.2a 141.1b 143.4bc 159.4cd 173.5de 178.3e
1991 51.8a 104.3b 122.6c 152.7d 159.8d 163.8d
1992 41.2a 64.5b 90.9c 108.4d 116.4d 112.2d
1993 57.5a 83.1b 111.2c 117.3c 121.7c 124.2c
1994 59.8a 94.3b 121.0c 162.1d 174.6d 175.9d
1995 38.6a 74.3b 99.0c 141.6d 150.2d 180.4e
† Pairwise comparisons were done using LSMEANS values, not actual parameter estimates.
‡ For each response parameter (corn grain yield, grain N content, etc.), means within each row (year) followed by the same letter are not statistically
different at ␣ ⫽ 0.05.

Table 6. Treatment means and statistical groupings for the N ⫻

quadrant (NtQ) interaction. Table 7. Treatment means and statistical groupings for the N ⫻
irrigation treatment (NtI) interaction.
N treatment, kg N ha⫺1
N treatment, kg N ha⫺1
Quadrant† 0 45 90 135 180 225
Irrigation treatment 0 45 90 135 180 225
Corn grain yield‡§, Mg ha⫺1
SW 5.89a¶ 8.54b 9.34c 10.22d 10.51d 10.34d Corn grain yield†‡, Mgha⫺1
SE 6.34a 8.48b 9.10bc 9.76cd 10.21d 10.03d A 4.28a§ 6.52b 7.80c 9.40d 9.56d 9.58d
NW 3.70a 6.08b 7.09c 8.39d 8.96d 8.67d B 5.42a 7.85b 8.71c 9.57d 9.86d 9.75d
NE 3.76a 5.75b 7.33c 8.59d 8.33de 9.16df C 5.19a 7.91b 8.27bc 9.20d 8.86cd 9.54d
D 4.80a 6.57b 8.09c 8.79cd 9.72e 9.33de
† SW, southwest; SE, southeast; NW, northwest; NE, northeast.
‡ Pairwise comparisons were done using LSMEANS values, not actual † Pairwise comparisons were done using LSMEANS values, not actual
parameter estimates. parameter estimates.
§ The NtQ interaction was not significant for grain N content, grain N ‡ The NtI interaction was not significant for grain N content, grain N
uptake, stover yield, stover N content, stover N uptake, and total plant uptake, stover yield, stover N content, stover N uptake, and total plant
N uptake. N uptake.
¶ Means within each row (quadrant) followed by the same letter are not § Means within each row (irrigation treatment) followed by the same letter
statistically different at ␣ ⫽ 0.05. are not statistically different at ␣ ⫽ 0.05.
 1348 AGRONOMY JOURNAL, VOL. 97, SEPTEMBER–OCTOBER 2005

There was a significant interaction between the Nt basically identical for the quadrants, it was the magni-
and Y treatments. For the NtY interaction (Table 5), tude of the response that differed between the south
the grain and stover yield increased significantly with and north quadrants. As discussed above, the SW quad-
increased Nt up to 135 kg ha⫺1 in most years, with grain rant had the highest yields overall, and the interaction
yield increases observed only to Nt of 45 and 90 kg ha⫺1 shows that the yield increased significantly until an Nt
in 1990 and 1993, respectively. There were only slight rate of 135 kg N ha⫺1. In comparison, the significant
increases in yield when the Nt was increased to 180 kg yield increases for the NW quadrant were also observed
ha⫺1. However, N accumulation in the plant did show until an Nt rate of 135 kg ha⫺1, but the yield was 1.83
Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved.

some increases at higher N rates. There was no signifi- Mg ha⫺1 less than on the SW quadrant.
cant difference between the yields at 135 and 180 or Mean yields for the N treatment–irrigation treatment
225 kg N ha⫺1 for any year except 1995. interaction (NtI) are shown in Table 7. Yield was again
In 1990, the first year of the study, grain and stover near maximum at 135 kg ha⫺1 with no significant differ-
yield had the least response to additional applied Nt, ence between the 135 and 180 kg ha⫺1 Nt rates except
and grain yield did not vary significantly for Nt higher for Irrigation Treatment D. The NtI interaction showed
than 45 kg ha⫺1. This was caused by increased availabil- that no significant increases in grain yields were mea-
ity of soil N resulting from mineralization from the initi- sured above Nt rates of 135 kg N ha⫺1, except for D
ation of irrigation in 1989 (Casey et al., 2002). A similar where the maximum yield was obtained at 180 kg N ha⫺1.
lack of response to fertilizer N was observed by Onken Irrigation Treatment B had the highest yields across all
et al. (1985) during a year when the soil had high residual Nt rates. This indicated that Irrigation B may have been
N. The greatest responses to Nt were observed in 1991, better suited to providing sufficient amounts of water
1994, and 1995, which were normal growing seasons during critical periods of corn growth when the rate of
(CGDD ⫽ 1245, 1255, and 1198⬚C, respectively). Yield nutrient uptake was also increasing. The NtP interaction
response to Nt was less in 1992 and 1993, which were was not significant for any of the response variables;
cool, wet years and had the lowest CGDD (1092 and hence, pair-wise comparisons within previous irrigation
982⬚C, respectively). Yield was highly correlated with scheduling methods are not presented.
CGDD from emergence to harvest (r ⫽ 0.94) and was
dramatically lower in 1992 and 1993. Yields in 1992 were Optimized Nitrogen Rates and Recommendations
depressed as a result of cooler-than-normal tempera-
tures while 1993 yields were low because of cool weather The discussion in this section focuses on Ntot rate at
and a frost in September before the crop matured maximum corn grain yields for Y, Q, and I. The Ntot
(Derby et al., 2004). Interestingly, stover N content and rate for these calculations and yield response curves is
stover N uptake were highest across all Nt in 1993. the sum of soil nitrate N, starter N, and the postemer-
Although early-season temperature was a very impor- gence Nt. In general, yield response to Ntot followed a
tant component of the yearly weather effect, Y was quadratic-plus-plateau relationship (Eq. [2] and [3]). This
likely a combination of many factors. Derby et al. (2004) typical yield response to applied N, often referred to as
considered both temperature and solar radiation when the Law of Diminishing Returns (Willcox, 1949), has
calculating CGDD and ET values for midseason N fertil- been reported extensively (Cerrato and Blackmer, 1990;
ity management. Variable precipitation patterns would Dahnke and Olson, 1990; Vanotti and Bundy, 1994;
have also affected crop growth because, even though the Schlegel and Havlin, 1995). The quadratic-plus-plateau
site was irrigated, the irrigation regimes were designed nonlinear models for each year, quadrant, and irrigation
to allow soil moisture deficits, so precipitation variation treatment indicated that there were wide ranges of Ntot
was still an important factor in determining corn yield. rates that optimized yield (Fig. 2). The average opti-
This analysis indicated that interannual variability caused mized Ntot rate for all observations for the study period
by adverse weather conditions can result in substantial was 295 kg N ha⫺1. The range of optimized Ntot rates
yield reductions by shortening the growing season and between years (Fig. 2a) was from 232 kg ha⫺1 in 1993
preventing maturity. Hence, the use of a static yield to 296 kg ha⫺1 in 1994. Note that the low-yielding years,
response curve for N recommendations that averages 1992 and 1993, did not vary much from the other years
over interannual variability could result in overfertiliza- in terms of optimized Ntot rate.
tion because of unrealized yield goals in unseasonably The highest-yielding quadrants (SE and SW) had the
cool years. If weather conditions warrant, reduced early- lowest average optimized Ntot rates of 291 kg ha⫺1 while
season fertilization with a midseason adjustment may be the lowest-yielding quadrants (NE and NW) had the
an alternative to current fertilizer management schemes, highest average optimized Ntot rate of 334 kg ha⫺1
and Derby et al. (2004) provide a model to adjust in- (Fig. 2b). Although the maximum yields varied little
season rates based on CGDD or ET for the Northern between irrigation treatments, Irrigation Treatment D
Great Plains. had a higher optimized Ntot rate than the other treat-
Although there was definite separation in yields be- ments even though all irrigation treatments had similar
tween the north and south portions of the field (NtQ maximum yields (Fig. 2c).
interaction), maximum yield was also obtained at the To reduce fertilizer N inputs while maintaining near
135 to 180 kg N ha⫺1 treatment range with no significant maximum yields, we can consider N application for 98%
difference in yields between those rates within each quad- of maximum yield (Dahnke and Olson, 1990). For exam-
rant (Table 6). Although response of yield to Nt was ple, the optimized Ntot rate considering all observations
 DERBY ET AL.: INTERACTIONS OF N, WEATHER, SOIL, AND IRRIGATION ON CORN YIELD 1349
Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved.

Fig. 2. Quadratic-plus-plateau response curves for (a) years, (b) quadrants, and (c) irrigation treatments. Optimized total N rates at 98% of
maximum yield are indicated by vertical dashed lines. Maximum yields are indicated by large circles. Heavy solid and dashed lines are for
all yield observations over the course of the study. The current North Dakota State University recommendation (dashed-dotted line) is based
on 21.4 kg of N per Mg of corn grain yield, which is equivalent to 1.2 lb of N per standard (56 lb) bushel of corn.
 1350 AGRONOMY JOURNAL, VOL. 97, SEPTEMBER–OCTOBER 2005

Table 8. Comparison of optimized N rates at maximum yields to N rates at 98% of maximum yield and corresponding change in per-
hectare income when fertilizing for 98% of maximum yield.
Optimized Yield at 98% of N at N savings at Change in
Variable† N rate‡ optimized N optimized yield 98% yield 98% yield per-hectare income
kg ha⫺1 Mg ha⫺1 kg ha⫺1 US$
All 295 9.93 9.73 251 44 $8.80
Year ⫽ 1990 281 10.78 10.57 225 56 $13.87
Year ⫽ 1991 268 10.38 10.18 231 37 $4.07
Year ⫽ 1992 279 7.37 7.22 237 42 $11.50
Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved.

Year ⫽ 1993 232 5.87 5.75 190 42 $13.76

Year ⫽ 1994 296 12.80 12.54 260 36 ⫺$0.02
Year ⫽ 1995 274 11.27 11.04 236 38 $2.96
Quadrant ⫽ NE 374 9.66 9.47 318 56 $16.09
Quadrant ⫽ NW 293 9.11 8.93 251 42 $8.50
Quadrant ⫽ SE 277 10.53 10.32 234 43 $6.70
Quadrant ⫽ SW 306 10.80 10.59 256 50 $10.83
Irrigation ⫽ A 283 10.17 9.97 240 43 $7.82
Irrigation ⫽ B 278 10.03 9.83 237 41 $6.90
Irrigation ⫽ C 296 9.67 9.48 250 46 $9.92
Irrigation ⫽ D 345 9.88 9.68 296 49 $11.70
† All ⫽ all observations for the study period averaged over three replicates by N treatment; NE ⫽ northeast; NW ⫽ northwest; SE ⫽ southeast;
SW ⫽ southwest.
‡ Includes soil N, N applied as starter, and postemergence applied N.

for the study was 295 kg N ha⫺1 at the maximum yield that maximum yield will be achieved every year. Based
of 9.93 Mg ha⫺1 (Table 8). Compare that to a yield of 9.73 on readily available information such as early- to mid-
Mg ha⫺1 (98% of 9.93 Mg ha⫺1) where the corresponding season CGDD and soil properties, producers can modify
Ntot rate for all observations in Fig. 2 is 251 kg ha⫺1. their practices to maximize profits in good years and
That is a 15% reduction in applied N for a 2% reduction minimize environmental degradation in poor years.
in yield. Economically, a producer would have a net
increase of approximately $8.80 ha⫺1 assuming a corn
price of $78.57 Mg⫺1 ($2.00 bu⫺1) and fertilizer N cost CONCLUSIONS
of $0.55 kg⫺1 ($0.25 lb⫺1). An even higher return could Nitrogen fertility has a major role in maintaining max-
be calculated if application costs are considered. Envi- imum corn grain yields; however, a number of other
ronmentally, the difference of 44 kg N ha⫺1 would not factors limit yields even when N fertility is optimal. In
be available for surface or groundwater contamination. this study of corn yield in the Northern Great Plains,
Current N recommendations are adequate some years USA, postemergence N rates greater than 135 kg N ha⫺1
because they were developed over many site-years. This did not significantly increase yield. Based on quadratic
is confirmed by the agreement of the current recommen- response curves, 98% of maximum yield was realized
dation to the yield response curve for 1994 in Fig. 2a. at a Ntot rate of 251 kg N ha⫺1 over the entire study
The current recommendation line [0.0214 kg of N per period but ranged between 190 and 318 kg N ha⫺1 de-
kg of corn grain yield, which is equivalent to 1.2 lb N pending on year, quadrant, and irrigation scheduling
per standard (56 lb) bushel of corn] does come close to method. Areas of the field with sodium- and salt-
the point of 98% of maximum yield for 1994. However, affected soils had significantly lower productivity. Simi-
the current recommendation fails by over-recommend- larly, cool, wet years had lower productivity than years
ing N in cool, wet years (1992 and 1993) when there is with warm growing seasons. Timely water applications
a need to optimize management to avoid environmental scheduled using a water balance algorithm and in-season
degradation. The comparisons are similar for quadrants soil moisture corrections also maintained high yields
(Fig. 2b) and irrigation treatments (Fig. 2c) where the compared with other irrigation scheduling methods. Re-
current recommendation fits the response curves ade- search of this nature should be used to modify existing
quately at lower yields but would result in overfertiliza- fertilizer recommendation methods to include adjust-
tion at higher yield goals. Given the current environ- ments for low-productivity soil, proper irrigation man-
mental issues, we should not continue to recommend agement, and midseason adjustments based on current
the highest N rates every year assuming the growing growing season conditions to minimize costs and water
conditions will allow for maximum yield potential. quality impacts.
It is clear that there was not simply one optimized
Ntot rate for maximum corn yield but a range, de- ACKNOWLEDGMENTS
pending on weather conditions for a given year, timely
water applications, and soil properties. Research of this The research was supported by North Dakota Agricultural
Experiment Station Project 5041 and by the U.S. Department
nature provides information that can be used to develop
of Interior, Bureau of Reclamation. Use of equipment and
a more robust fertilizer N recommendation scheme that trade names does not constitute endorsement by North Da-
takes into account yearly weather (Derby et al., 2004), kota State University or the U.S. Bureau of Reclamation;
soil properties, and soil moisture availability. It may not names are provided for information only. The authors grate-
be environmentally wise to continue N recommenda- fully acknowledge the assistance of Dale Coe, Walter Albus,
tions based on a static yield response curve and assuming James Moos, Wade Kuehl, and Dale Esser for field operations.
 DERBY ET AL.: INTERACTIONS OF N, WEATHER, SOIL, AND IRRIGATION ON CORN YIELD 1351

REFERENCES 575–606. In SAS user’s guide: Statistics. 5th ed. SAS Inst., Inc.,
Cary, NC.
Albus, W.L., and J.T. Moraghan. 1995. Responses of three early ma- Jones, C.A., and J.R. Kiniry (ed.) 1986. CERES–Maize: A simulation
turing corn hybrids to nitrogen fertilizer. J. Prod. Agric. 8:581–584. model of maize growth and development. Texas A&M Univ. Press,
Berglund, D.R., and D.A. McWilliams. 1999. Corn production for College Station.
grain and silage. Bull. A-834 (Rev.). North Dakota State Univ. MathSoft. 1999. S-Plus software ver. 5.1 Rel.1 for Sun SPARC, Sun
Ext. Serv., Fargo. OS 5.5. MathSoft, Cambridge, MA.
Bremner, J.M., and C.S. Mulvaney. 1982. Nitrogen—total. p. 559–624. Oberle, S.L., and D.R. Keeney. 1990. Factors influencing corn fertil-
In Methods of soil analysis. Part 2. Chemical and microbiological izer N requirements in the northern U.S. Corn Belt. J. Prod.
properties. 2nd ed. SSSA and ASA, Madison. Agric. 3(4):527–534.
Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved.

Casey, F.X.M., N.E. Derby, R.E. Knighton, D.D. Steele, and E.C. Onken, A.B., R.L. Matheson, and D.M. Nesmith. 1985. Fertilizer
Stegman. 2002. Initiation of irrigation effects on temporal nitrate nitrogen and residual nitrate-nitrogen effects on irrigated corn
leaching. Vadose Zone J. 1:300–309. yield. Soil Sci. Soc. Am. J. 49:134–139.
Cerrato, M.E., and A.M. Blackmer. 1990. Comparison of models for Schlegel, A.J., and J.L. Havlin. 1995. Corn response to long-term
describing corn yield response to nitrogen fertilizer. Agron. J. 82: nitrogen and phosphorus fertilization. J. Prod. Agric. 8:181–185.
138–143. Steele, D.D., E.C. Stegman, and B.L. Gregor. 1994. Field comparison
Cochran, W.G., and G.M. Cox. 1957. Experimental designs. 2nd ed. of irrigation scheduling methods for corn. Trans. ASAE 37(4):
John Wiley & Sons, New York. 1197–1203.
Dahnke, W.C., C. Fanning, and A. Cattanach. 1992. Fertilizing corn Steele, D.D., E.C. Stegman, and R.E. Knighton. 2000. Irrigation man-
grain, popcorn, silage corn, and sweet corn. Bull. SF-722 (Rev.). agement for corn in the northern Great Plains, USA. Irrig. Sci.
North Dakota State Univ. Ext. Serv., Fargo. 19:107–114.
Dahnke, W.C., and R.A. Olson. 1990. Soil test correlation, calibration, Stegman, E.C. 1982. Corn grain yield as influenced by timing of evapo-
and recommendation. p. 45–71. In R.L. Westerman et al. (ed.) transpiration deficits. Irrig. Sci. 3:75–87.
Soil testing and plant analysis. 3rd ed. SSSA Book Ser. 3. SSSA, Stegman, E.C., and D.A. Coe. 1984. Water balance irrigation schedul-
Madison, WI. ing based on Jensen–Haise equation: Software for Apple II, II⫹,
Derby, N.E., F.X.M. Casey, R.E. Knighton, and D.D. Steele. 2004. and IIE Computers. Res. Rep. 100. North Dakota Agric. Exp. Stn.,
Midseason nitrogen fertility management for corn based on weather North Dakota State Univ., Fargo.
and yield prediction. Agron. J. 96:494–501. Swan, J.B., E.C. Schneider, J.F. Moncrief, W.H. Paulson, and A.E.
Enz, J.W., G.P. Brockberg, R. Egeberg, R. Carcoana, D. Rice, and Peterson. 1987. Estimating corn growth, yield, and grain moisture
T. Scherer. 1997. Weather: North Dakota Agricultural Weather from air growing degree days and residue cover. Agron. J. 79:53–60.
Network (NDAWN). NDSU Ext. Serv. Software Users Guide 11 Vanotti, M.B., and L.G. Bundy. 1994. An alternative rationale for
(Rev.). North Dakota State Univ. Ext. Serv., Fargo. corn nitrogen fertilizer recommendations. J. Prod. Agric. 7:243–249.
Gilmore, E.C., Jr., and J.S. Rogers. 1958. Heat units as a method of Wienhold, B.J., T.P. Trooien, and G.A. Reichman. 1995. Yield and
measuring maturity in corn. Agron. J. 50:611–615. nitrogen use efficiency of irrigated corn in the northern Great
Hanway, J.J. 1966. How a corn plant develops. Iowa Coop. Ext. Serv. Plains. Agron. J. 87(5):842–846.
Spec. Rep. 48. Iowa State Univ., Ames. Willcox, O.W. 1949. Verification of the Mitscherlich effect law. Agron.
Ihnen, L.A., and J.H. Goodnight. 1985. The NLIN procedure. p. J. 41:225–229.