Mid-Season Prediction of Wheat Grain
Yield Potential Using Plant,
Soil, and Sensor Measurements (J. Plant Nutrition)
Girma, Kefyalew, K.L. Martin, R.H. Anderson, D.B. Arnall, K.D. Brixey, M.A. Casillas, B. Chung, B.C. Dobey, S. Kamenidou, S.K. Kariuki, E. Katsalirou, J.C. Morris, J.Q. Moss, C.T. Rohla, B.J. Sudbury, B.S. Tubana, and W.R. Raun*.
The components that define cereal grain yield potential have not been well
defined. The objective of this study was to collect many differing biological
measurements from a long-term winter wheat (Triticum aestivum L.) study in an
attempt to better define yield potential. Four treatments were sampled that
annually received 0, 45, 90, and 135 kg N ha-1 at fixed rates of P (30 kg ha-1)
and K (37 kg ha-1). Mid-season measurements of leaf color, chlorophyll,
normalized difference vegetative index (NDVI), plant height, canopy temperature,
tiller density, plant density, soil moisture, soil NH4-N, NO3-N, organic C,
total nitrogen, pH, and nitrogen mineralization potential were collected. In
addition, soil texture and bulk density were determined to characterize each
plot. Correlations and multiple linear regression analyses were used to
determine those variables that can predict final winter wheat grain yield. Both
the correlation and regression analyses suggested mid-season NDVI, chlorophyll
content, plant height and total N uptake as good predictors of final winter
wheat grain yield.
In an attempt to better understand several
selected variables that influence yield potential, a succinct literature review
on each specific variable measured from a long-term winter wheat experiment is
presented. The methods used for collecting data on each of the variables are
Leaf color charts have been used in rice (Oryza
sativa L.) to determine nitrogen (N) status (1) and are related to chlorophyll
meter readings (2). Aerial photographs analyzing color have successfully been
used to predict sidedress N needs in corn (Zea mays L.) (3). Crop and turfgrass
color is often evaluated visually, which is a somewhat subjective measure. Leaf
color charts and standardized color charts may be effective in some situations,
but analysis differs by person and cultivar. Time of year, time of day, cloud
cover, etc. could alter an observation. Karcher and Richardson (4) used digital
image analysis to quantify turfgrass color. They found that digital image
analysis is a reliable and objective means to evaluate turf color, but only when
images are collected under equal lighting conditions. Digital photography and
image analysis software has also been used to study canopy coverage in wheat (5)
and soybeans [Glycine max L. (Merr.)] (6). Digital photography and image
analysis may potentially help farmers determine N requirements in crops by
determining crop color in an objective manner.
Chlorophyll Meter Readings
Leaf chlorophyll content can be closely
correlated with leaf N content due to the location of N in the leaf, which is
primarily a component of the chlorophyll molecule. Therefore, Wood et al. (7)
found that chlorophyll measurements correlate well with N concentration in the
plant tissue. Many methods of N determination in plants exist, but most are
destructive procedures. The ability to determine the chlorophyll content and the
related N status of a plant at different stages and in conjunction with other
factors would be extremely valuable in establishing how yield potential is
affected by the chlorophyll content and the related N status.
The Minolta Soil Plant Analysis Development
(SPAD-502) chlorophyll meter is one tool that enables researchers to determine
chlorophyll content by measuring leaf greenness (8). The SPAD uses a silicon
photodiode to derive the ratio of transmittance through the leaf tissue at 650
nm compared with transmittance at 940 nm, and a value is given based on that
ratio (9). Most of the research conducted in conjunction with the SPAD to date
has been aimed at utilizing it as an N management tool. Research conducted in
Nebraska has shown that the chlorophyll meter reading used with a sufficiency
index was successful as an in-season N management tool for irrigated corn (10).
In-season chlorophyll meter-based treatments used with a sufficiency index were
able to maintain the yield achieved with traditional fixed timing N treatments
while also reducing the amount of N applied in rice experiments conducted in the
Philippines (11). Other research with corn indicates that chlorophyll meter
readings as an estimate of leaf chlorophyll content correlate with yield as
accurately as leaf N concentrations (12) and similar results were reported
between SPAD values and wheat grain yield in India (1).
Species and variety differences will result in differing chlorophyll meter readings. Researchers with the United States Department of Agriculture (USDA) and the University of Nebraska documented considerable differences in chlorophyll meter readings between different varieties and hybrids of corn and sorghum (8). To utilize the meter as a management tool, however, the most reliable method involves the use of nitrogen rich reference strips within each field. These reference strips should accurately represent the field for all factors except the N treatment to insure that the yield potential of the field is not limited by a lack of N (8). The literature recommends 20 (9) to 30 (8) readings per field to establish an accurate representation; however, an independent sample experiment indicated that as few as 8 averaged samples accurately represented a plot of approximately 1 m2. In corn, readings should be collected from the uppermost, fully expanded leaf at a point one-half the distance from the leaf tip to the collar, and one-half the distance between the leaf margin and the midrib (8). After collecting average readings representing both the field and the reference strip, a sufficiency index can be calculated.
When utilizing the chlorophyll meter as a
management tool, N application is based on the calculated sufficiency index. For
corn, a SI lower than 95% indicates an N deficiency and additional N should be
added (10). An SI of 90% was found to be more efficient when applied to rice
grown in Asia (11).
Factors affecting SPAD values include radiation
differences between seasons, variety and species differences, plant and soil
nutrient status (including N and other nutrients), and biotic and abiotic
stresses (8). Excess soil water in subhumid regions with high organic matter
soils have also been shown to disrupt the relationship between chlorophyll meter
readings and plant N status (13). The chlorophyll meter has been proven to
accurately determine when fertilization is needed, but is limited in that it is
not able to estimate the amount of fertilizer needed (10).
Nitrogen Availability and Active NDVI Sensors
Nitrogen is an essential plant nutrient
required for high yield. Under ideal conditions, N availability to the crop is
one of the most critical and variable factors when predicting crop yields. Of
the N fertilizer used for crops, only 33% is actually removed in the grain (14).
The use of reflectance of red and near infrared wavelengths during early growth
stages has been greatly researched and indicates N status of wheat and can be
used to predict potential grain yield (15, 16).
Oklahoma State University (Stillwater, OK),
jointly with NTech Industries (Ukiah, CA) developed the GreenSeekerTM Hand Held
Optical Sensor. This sensor with the current algorithm has the ability to
determine the amount of N fertilizer to apply for each 0.4 m2 area. This allows
farmers to increase yield while reducing excess application of N fertilizer. It
was reported following extensive soil sampling, optical sensor measurements of
plants, and geostatistical analysis that the spatial scale of N availability was
at <1 m2 and that each square meter needed to be treated independently (17, 18,
19). When N management decisions are made on areas of <1 m2, the variability
that is present at that resolution can be detected using optical sensors and
treated accordingly with foliar application of N (14, 20, 21). Remote sensing
data is collected by a modified daytime-lighting reflectance sensor which is
used to estimate early-season plant N uptake. The estimate is based on a
relationship between NDVI and plant N uptake between Feekes physiological growth
stage 4 (leaf sheaths lengthen) and 6 (first node of stem visible) (20, 21, 22).
Plant height factors have not recently been studied to a great extent. The difficulty of utilizing plant height as a factor to better predict yield potential is that it must be coupled with some other parameter that takes into account the spatial variability of the plants. Mallarino et al. (23) noted that it is possible that a variable is not related to the yield in a field because the range of variation within that field is above or below the range in which it influences yields. However, Machado et al. (24) found that plant height explained 61% of the variation in grain yield in corn. Studies conducted at Iowa State University showed that in three of five corn fields, a good relationship was found between yield and height (23). Current work in corn at Oklahoma State University suggests that plant height and NDVI can be reliable predictors of forage biomass and final grain yield on a by-plant basis. Recognizing the difference in corn and wheat, but acknowledging the concept employed here, plant height in wheat will likely play an effective role in predicting final grain yield, but may be improved by adding another dimension of plant characteristics.
Early work by Millard et al. (25) reported a correlation between canopy temperature and plant water stress. Remotely sensed infrared canopy and hand-held infrared thermometers provide a new, easy, non-destructive method for canopy temperature measurements (26, 27). Canopy temperature is used as a factor to predict water stress in several crops as a screening tool for breeding drought tolerant varieties and canopy temperature depression (CTD) as a potential indirect criteria for yield (26, 28, 29).
A modified crop water stress index that
includes environmental factors like canopy temperature was proposed by Clawson
et al, (28). A study by Rashid et al. (26), reported grain yield variation
related to plot-to-plot differences in canopy temperature under water stressed
conditions. For spring wheat, a positive correlation between grain yield and
cooler canopies was found (29).
Seeding density affects grain yields and plant
density (30). In addition, high seeding density can result in higher grain
yields and plant density as compared to low seeding density (31). High seeding
density has been reported to improve grain yield of wheat under conventional
cropping systems (32, 33). Robert and Xue (34) found that high seeding density
increased yield 12% to 18% when compared to low seeding density in Montana. Work
in Canada showed that a 60 kg ha-1 seeding rate had 61% higher plant density
than the 30 kg ha-1 seeding rate (30). High seeding density can control weed
biomass and reestablishment of weeds in ensuing seasons and can compensate for
poor plant establishment (33).
Tillering is the product of added stem formation on the mainstem or crown of the plant. According to Peterson et al. (35), the ability of a plant to produce tillers can play a role in wheat plant survival because the plant can produce a secondary crown which could prevent the plant from death if the main crown were to be damaged.
At three of four locations, Flowers et al. (36)
found a relationship between NIR, NDVI and tiller density to be significant. It
has also been found that it could be possible to use tiller density to give an N
recommendation for application during the growing season. Flowers et al. (37)
found that they could use NIR sensing during Feekes physiological growth stage 3
to predict tiller density and subsequent N recommendation using a soft red
winter wheat variety.
Several studies have shown the importance of
optimum soil moisture in wheat production. Robinson et al. (38) used a crop
model and historical climate records (1960-1993) to produce a long-term record
of yield and grain protein responses to N fertilizers. The study proved that
fertilizer N application was most profitable if used when measurements indicate
that the plant available soil moisture content before sowing is above average.
Another study by Major et al. (39) was conducted to investigate yield and yield
components of winter wheat as affected by soil water and N at five locations. It
was confirmed that moisture stress was the most limiting factor in the Great
Plains. The increase in wheat yield under optimum soil moisture was attributed
to increased number of spikes. Hoogenboom et al. (40) noted that the processes
affected by water stress include vegetative development, reproductive
development, photosynthesis, biomass partitioning, pod and seed addition, grain,
ear or head addition, leaf area growth and expansion, root growth,
transpiration, senescence and various other processes.
Precipitation is the source of all soil
moisture in rainfed cropping systems (41), thus crop production may be
susceptible to water stress once soil water drops below a critical level brought
about by fluctuating rainfall. Using computer simulation models, crop yield can
be predicted as a function of weather and soil condition, thus data on rainfall
and precipitation are one of the key inputs (41). Using historical data on
monthly rainfall and monthly evaporation, weekly and daily rainfall and
simulated weekly soil moisture, Huda (42) developed simple risk management
principles in evaluating alternate strategies to increase and stabilize wheat
yield in low rainfall areas of southern Australia. The study by Arora and Gajri
(43) on the evaluation of a crop growth-water balance model for analyzing wheat
responses to climate and water-limited environment suggested that computer
models can be applied for optimizing water use at the field scale.
Soil Color and Soil Texture
Soil color is important because it is related to organic-matter content, climate, soil drainage, and soil mineralogy (44). The Munsell® Soil Color Chart is used to determine and report soil color (45) This standardized system uses color chips to enable one to determine the hue, value, and chroma of soils.
Soil texture is recorded as the ratio of the
three soil components (sand, silt, clay) based on their physical dimensions.
This ratio is most frequently presented as a textural triangle which readily
enables one to determine the basic soil type (46). Soil texture contributes
significantly to physical properties (including pore structure, aeration, water
drainage) (47) and to chemical properties (including permanent and pH dependent
charge, cation exchange capacity (CEC), and soil organic material and nutrient
composition) (48). A table comparing the particle size limits in particle size
classification used by four different agencies has recently been republished
(49). Soil texture determinations are most often performed using the pipette and
the hydrometer methods (50).
Soil Bulk Density
Bulk density affects plant growth through its effect on soil strength and soil porosity (51). Increased bulk density increases soil strength and reduces soil porosity. Both of these factors limit root growth at some critical value. The mechanical effects of high bulk density are the direct effects on root growth. The indirect effects are a reduction in aeration and soil water conductivity and retention (52). Lower bulk density means less soil mass for a given depth of soil, therefore lower CEC and less nutrients.
Droogers and Bouma (53) compared four fields
that were under different management for 70 years. Two were cultivated with
organic farming and two with conventional farming. The loam soil in the
conventional fields had higher bulk density (1.61 Mg m-3) than the organic
fields (1.50 Mg m-3) and 7% lower simulated potato yield. Ishaq et al. (54)
observed that when subsoil bulk density was artificially increased from 1.65 to
1.93 Mg m-3, the grain yield of the next crop decreased by 38%. This study shows
the negative effects of subsoil compaction on crop yield. It also shows that the
effect of compaction is reduced over time, possibly because the soil structure
improves. Lal and Ahmadi (55) report that tillage treatment affects soil bulk
density but not grain yield in a consistent way season after season. However,
looking at the means of 11 seasons for one site, there is an inverse
relationship between bulk density and grain yield. The soil with the highest
bulk density (1.31 Mg m-3) had 13% less yield than the soil with the lowest bulk
density (1.25 Mg m-3).
Soil pH is responsible for various soil chemical reactions and processes. According to Sparks (46), acid rain is one of the major sources of low pH. This is a result of a reaction of atmospheric emission of industrial gases, sulfur dioxide (SO2) and nitrogen dioxide (NO2) forming sulfuric and nitric acid respectively. Evans (56) observed that a decrease in soil pH in Australian soils was largely influenced by agricultural practices.
At low pH, phosphorus availability may be
reduced and calcium and magnesium levels are likely to be inadequate. Aluminum
hydrolyses under acidic conditions forms hydroxyl aluminum which can be fixed on
the cation exchange capacity (CEC). Hydroxyl aluminum is said to be
non-exchangeable which reduces the CEC of clays and soils (48). This may result
in deficiencies of these nutrients restricting optimal crop production.
Nitrogen Mineralization Potential
Many factors are involved in the process of N
mineralization including soil moisture, temperature, texture, structure,
aeration, microbial biomass, and the substrate chemical constituency (57, 58,
59). All of these interact to ultimately govern the amount of N available via
mineralization from year to year. Rasmussen et al. (60) suggests management is
the primary factor responsible for regulating N dynamics in agricultural systems
and should be taken into consideration when estimating N fertilizer needs. They
found the highest N mineralization potential under uncultivated pasture with
increasing rates of N application, reduction in tillage intensity, and a higher
frequency of cropping as determinant factors enhancing the mineralizable pool of
organic-N in the cultivated fields they studied.
A study from Canada conducted by Walley et al. (61) assessed the relationships between soil N availability indices (SNAI), yield, and plant N uptake in wheat from fields positioned at various locations on the landscape. The SNAIs included 2 and 16-week aerobic incubations, NO3 sorption onto anion-exchange membranes, N extraction and hydrolysis with hot KCl in laboratory under optimal microbial temperature and moisture regimes. These researchers reported a slight correlation among all SNAIs with yield and for the most part with plant N uptake. The best variability explained was 40% (62). They noted that SNAIs must be combined with field scale variability in order to accurately predict the size of the potentially mineralizable-N pool in glacial till semi-arid agricultural systems.
Organic C and Total N in Soils
It is common to find a net mineralization of N in soil when residuals are added that have high C:N ratios (63). It has been found that a C:N ratio of 25 results in a highly stable product, good fertilizer value and low potential for possible environmental pollution (64). However, looking at the C:N ratio may not result in an accurate prediction of resulting mineralization or immobilization (65).
With decreasing particle size, mineralization
of C and N decreased resulting in a more stable C:N ratio (66). In one
experiment, C was added with the expectation of increased biomass C and N along
with a decrease in inorganic N. The additions of C and N to the soil environment
result in a far more complex than a simple C:N ratio (63). Resulting changes in
C and N is effected by multiple factors. This includes but is not limited to
form of C, form of N, C:N ratio, microbial activity, soil texture, age of
environment, soil nutrients present, and tillage practices (67).
Soil, equipment, seed, and environmental factors all affect wheat stand establishment. Oklahoma farmers commonly expect 77% of their viable seed to emerge, however research has shown that the percentage emergence may be closer to 57% (68). Stockton et al. (68) found that stands in individual fields ranged from less than 30% emergence to better than 80%. Weisz et al. (69) reported that as plant stand or tiller density increases, grain yield tends to increase and the variation within the field decreases. These studies show the presence of stand variability in wheat fields and are an indication of the need to detect this variability to predict yield.
Coefficient of Variation (CV) is defined as the standard deviation divided by the mean (70). Steel et al. (70) describes CV as a quantity of use to the experimenter in evaluating results from different experiments of the same unit of measurements that are possibly conducted by different persons. Little and Hills (71) suggest that CVs can be used to compare experiments involving different units of measurements and/or plot sizes. The CV is a relative measure of variation and varies with every comparison on what is considered large or small, and only experience with similar data can determine its magnitude (70).
In an evaluation of sixty-two, wheat field
research projects, Taylor et al. (72) observed that mean yield and CV were
negatively correlated. Taylor’s work also showed that CVs decreased with
corresponding decreases in plot size. Washmon et al. (73) suggested that if
within field CVs could be predicted, the potential response to added nutrients
may also be established and in-season nutrient additions adjusted accordingly.
Furthermore, mid-season CV of a field can be equated to the response index,
which is currently used by various researchers to determine top-dress fertilizer
needs. Lukina et al. (74) observed that as the vegetation coverage increased,
the CV of spectral radiance (NDVI) values, decreased. This suggested that the CV
of NDVI readings could be used to estimate stand density. Raun et al. (61)
showed that NDVI values from mid-season sensor readings could be used to predict
yield. Combining NDVI and CV may result in improved prediction of yield
potential and improved mid-season prediction of the response to applied
MATERIALS AND METHODS
Data from a long-term (30 year old) winter wheat experiment established at Agronomy Research Station in Stillwater, OK were used for this report. The soil at the experimental site is a deep well-drained Kirkland Silt Loam (fine, mixed, thermic Udertic Paleustolls (75)). The soil texture of this site was loam (32.36% sand, 20.28% clay and 47.36% silt). The experiment employed a randomized complete block experimental design with 4 replications. The treatments used for this report were 0, 45, 90 and 135 kg N ha-1 with fixed levels of 29 and 37 kg ha-1 P and K, respectively. One additional control plot (no fertilizer) was also included. Plot size was 12.2 m by 18.3 m. Ammonium nitrate (34-0-0), triple super-phosphate (0-20-0), and potassium chloride (0-0-50) were broadcast and incorporated prior to planting. Winter wheat was planted in 25-cm rows at a seeding rate of 67 kg ha-1 and grown under conventional tillage (disk incorporation of wheat straw residues following harvest and prior to planting).
Leaf color was measured from 4 randomly selected shoots within each plot at Feekes physiological growth stages 5 (stem elongation), 7 (two nodes), 10 (late boot stage) using the color chart developed at the International Rice Research Institute (http://www.irri.org). Similarly, chlorophyll readings were collected using a Minolta 502 SPAD meter from 15 randomly selected fully extended leaves in each plot at Feekes physiological growth stages 5, 7, 10. Sensor readings were collected from each plot using a GreenSeeker hand-held sensor developed by NTech Industries that measures NDVI. Canopy temperature was measured at 5 randomly selected positions within each plot using a handheld infrared thermometer. Similarly, plant height was determined by measuring the height of 15 randomly selected extended leaves in each plot.
Seeding density was determined by calculating the number of seeds per kilogram planted on each plot. Tiller density was evaluated by counting the number of tillers in 10 plants in 6 randomly selected rows. The tiller density was evaluated at Feekes physiological growth stage 5, 7, and 10. The values reported in Table 2 represent the average number of tillers per plant.
Percent soil moisture was determined within each plot at Feekes physiological growth stages 5 and 7 from depths of 0-15, 15-30, and 30-46 cm (0-6, 6-12, and 12-18 in) using weights of wet and dry soil samples to calculate the percent moisture content.
Soil color was determined by comparing the dry soil color to the Munsell® soil color chart under full sunlight. The soil texture was determined by using a hydrometer (48) on one sample collected from each plot.
The bulk density was measured using soil cores sampled from the different treatments (core method) (76). Four cores were taken from each of the treatments at a depth of 0-15 cm. The bulk density values are calculated as the average of four samples taken from each of the treatments.
N mineralization was studied using the method described by Stanford and Smith (77). Four 20 g dry weight equivalent (DWE) aliquots of each bulked field moist soil sample was brought to 55% water filled pore space and incubated at 25ºC in a dark constant temperature room in 100 ml specimen cups. Samples were weighed weekly to maintain moisture content and supplied with demonized water accordingly. Every 30 days, a set of subsamples was removed and analyzed for NH4-N and NO3-N using a Lachat Quickchem Autoanalyzer (Milwaukee, WI). The following nonlinear regression equation described by Smith et al. (78) was used to calculate the mineralizable pool or organic N (No) and the first-order rate coefficient (k):
Nm = No [1-exp (-kt)]
Where Nm = amount of N mineralized at a specific time (t). All results reported are the mean of 4 replications and are on a moisture free basis. Moisture was determined after drying at 95 ºC for 24 hours.
From each plot, 15 soil cores from a depth of 0-15 cm were collected. Soil Samples were air-dried at ambient temperature and ground to pass a 20 mesh screen. Samples were extracted using 2-M KCl (79) and analyzed for NH4 N and NO3 N using an automated flow injection analysis system (Lachat Quickchem Autoanalyzer). Total N and organic C were determined using a Carlo-Erba NA 1500 dry combustion analyzer (80). Soil pH was determined by collecting 10 g of soil from each plot using a 1:1 soil:water ratio from the 0-15 cm soil depth.
The correlation coefficients between final
grain yield and plant and soil variables showed a good impression of which
variables to consider. The challenge of identifying suitable variables to
explain final grain yield however, is evaluating the contribution of the
variables jointly. A multiple linear regression approach was used to address
this issue. Since there were several variables correlated with final grain
yield, a stepwise regression was carried out using two criteria, namely, R2 and
Cp (These criteria are presented in details in (81) and (82)). Decision for
entering and removing variables was done using two significant levels (0.15 and
0.10). Once the variables were selected, models containing various number of
predictor variables were tested for their ability to recover the variability in
final grain yield based on the F-test and the partial regression sum of squares
for each variable.
RESULTS AND DISCUSSION
Leaf Color, Chlorophyll, and NDVI
Mean leaf color, chlorophyll meter readings (SPAD), and NDVI measurements by treatment are reported in Table 1. In general, all three were well correlated with final wheat grain yield at Feekes physiological growth stages 5, 7, and 10. However, there was a tendency for all to be better correlated with final grain yield at Feekes physiological growth stage 7 (simple correlation coefficients, r > 0.85). NDVI readings were better correlated with leaf color chart readings than SPAD readings. This would suggest some disparity in the information collected within the NIR band (NDVI uses reflectance at 650 and 770 nm while the Minolta SPAD unit measures transmittance at 650 and 940 nm).
Because of the ease associated with collecting
NDVI readings using hand-held sensors, and because the current GreenSeeker
sensor averages readings from the entire surface area (versus 1 spot on a leaf
using the chlorophyll meter), NDVI readings are vastly superior to chlorophyll
meter readings and/or leaf color charts. Furthermore, because readings are
calibrated (repeatable over space and time and not influenced by ambient light
or temperature), it is likely that their use in agronomic research will increase
substantially over the next few years.
Plant Height and Canopy Temperature
As plant height increased, so did wheat grain yield for this experiment that only included one variety (Table 2). Plant height was highly correlated with final grain yield at all growth stages (simple correlation coefficients ranged from r = 0.73 to 0.90) and tended to increase as the season progressed. This makes sense because measuring the most fully extended leaf at Feekes physiological growth stage 5 is somewhat subjective, while at later stages of growth, the main stem and associated leaves are more easily identifiable.
Canopy temperature was not highly correlated
with wheat grain yield at any stage of growth. Canopy temperature was lower
where N fertilizer was applied (treatments 2, 3, 4) and higher under no N
(treatments 1 and 10) (Table 2). This was likely due to the increased foliage in
the N fertilized plots and increased soil coverage, thus decreased soil
temperatures as a result of increased light interception by the growing plants.
However, there was not a lot of treatment separation using canopy temperature,
likely because there were simply not large measurable differences, or the IR
thermometer used was not highly sensitive.
Similar to other variables included in this study, tillers were well correlated with final grain yield at all the three growth stages (r = 0.55, 0.89, and 0.24, respectively). Further, the relationship was much stronger at Feekes 7. Discernable differences were much more difficult to detect at Feekes 10, largely due to tiller abortion that takes place between jointing and boot stages (Table 2).
Mean soil moisture values determined at Feekes
physiological growth stages 5 and 7 are reported by depth and treatment in Table
3. Soil moisture in the surface 0-15 and 15-30 cm depths at Feekes physiological
growth stage 5 was negatively correlated with wheat grain yield (r = -0.68 and
-0.53, respectively). By Feekes physiological growth stage 7, no relationship
between soil moisture and wheat grain yield was found at any depth. Under
rainfed conditions where moisture is generally limiting, having a negative
correlation between surface soil moisture and final grain yield is plausible.
This is generally because the increased growth in the fertilized plots
(treatments 2-4) would have an increased demand for water and as a result, more
rapid depletion (Table 3).
Soil Bulk Density, Resistance and pH
Soil bulk density (0-15cm) is reported by
treatment in Table 4. Bulk density values decreased with increasing applied N.
This is consistent with results reported by Lal and Ahmadi (55), who studied
soils within the 1.2 to 1.4 g cm-3 range. Similarly, surface soil bulk density
values were negatively correlated with final grain yield (r = -0.35). Soil
resistance measured to a depth of 15 cm tended to be higher in those plots where
N had been applied continuously over this 30 year experiment (Figure 1).
However, there were many exceptions to this finding when evaluating single
measurements in specific treatments (data not reported). Similar to observations
for bulk density, soil pH decreased with increasing N applied (Table 4), and was
highly correlated with final grain yield (r = -0.74). This was expected
considering that the continuous N applications have led to lower pH, but not to
the extent where soil pH has had a negative influence on wheat grain yield
(lowest values at 5.6).
N mineralization potential
Results from total inorganic N mineralized at 1, 30, 60, 90, and 120 days as a function of treatment is reported in Table 5. Total mineralized inorganic N was highly correlated with final grain yield, with the highest value found at 60 days (r = 0.54, 0.49, 0.79, 0.75, 0.64, respectively). N mineralized at 60 and 90 days were found to be highly correlated with NDVI at Feekes physiological growth stages 5 and 7 (all r values > 0.57). Similarly, N mineralized at 60 and 90 days was highly correlated with chlorophyll meter readings collected at Feekes physiological growth stages 5 and 7 (r values > 0.71). Also, N mineralized at 90 days was highly correlated with plant height (Feekes 7 and Feekes 10, r > 0.74).
Surface NH4-N and NO3-N
Surface soil (0-15cm) NH4-N and NO3-N collected once in March (Feekes physiological growth stage 5) were both highly correlated with final grain yield (r > 0.51). This was expected since these long-term plots have received the same amount of N each year, and that span a range of rates (0 to 135 kg N ha-1 yr-1). As the annual N rate increased, so did surface NH4-N and NO3-N (Table 4). However, these quantities are dynamic in that the amounts depend on moisture and temperature, thus a specific amount is not necessarily correlated with a specific yield level.
Organic C and Total N
Surface soil organic C and total N determined via dry combustion was positively correlated with final grain yield (r > 0.62). Organic C and total N were also highly correlated with NDVI sensor readings at Feekes physiological growth stages 5 and 7 (r values all > 0.72). Consistent with work by Davis et al. (83) soil organic C and total N increased with increasing applied N (Table 4). In this long-term experiment, organic C would be expected to be correlated with final grain yield, especially when including plots that have not received N for this same period of time, and that would be subjected to increased N mining from the soil organic matter component.
The coefficient of variation from NDVI readings determined at Feekes physiological growth stages 5, 7, and 10 was negatively correlated with final grain yield at all stages (r > -0.80). As the CVs increased, grain yields decreased (Table 2). In general, we would expect that CVs should not be correlated with yield in large population of data. However, in this long-term trial, a negative and highly correlated relationship was expected since applied N increased tillers and thus coverage which would decrease the variability in NDVI sensor readings (Table 2). Also, as was expected, the CVs from NDVI sensor readings at Feekes physiological growth stages 5 and 7 were negatively correlated with NDVI readings (r > -0.87), further suggesting that as NDVI increased, cover increased and the amount of soil (NDVI values < 0.20) readings would decrease thus decreasing CV. Coefficient of variation readings were also negatively correlated with number of tillers (r >-0.86) at these same respective growth stages.
Multiple Linear Regressions
The Stepwise variable selection using significance level of 0.15 resulted in 11 predictor variables (Table 6) while significance level of 0.1 ended up with selection of two variables. For the first selection scenario after fitting a regression model, those variables contributing insignificant part of the variability were removed and a six variable model was fitted. Interestingly, the predictor variables in this model were mostly plant measurements taken between Feekes physiological growth stage 5 and 7. This supports the previous findings by Lukina et al. (75) who reported that in season measurements such as NDVI were reliable for predicting final grain yield.
The second stepwise model selection removed all variables but leaf chlorophyll content at Feekes physiological growth stage 7 and total N in soil as predictors of final grain yield with associated R2 of 0.94. On the other hand, the best single predictor variable model was the model containing leaf chlorophyll content at Feekes 7 explaining 87% of the variability in grain yield explained by the full model containing 11 variables. NDVI measured at Feekes 7 also explained about 77% of the variability in yield explained by the full model.
The four, three and two predictor variable
models from the first selection scenario had very close R2 and any of the models
can be used for prediction of final grain yield. Obviously, more variables in a
model enable to capture the variability in the predicted final grain yield
better than few predictor variables. However, the models above clearly showed
that most of the variability in final grain yield could be easily recovered by
inclusion of at most four predictor variables. Therefore, NDVI, chlorophyll
content and height at Feekes physiological growth stage 7, and total N could be
used to reliably predict final grain yield of winter wheat.
Correlation and regression analyses results
suggested that variables such as canopy temperature, NH4-N, NO3-N and bulk
density were found to be poorly associated with final grain yield and were also
not good predictors of final grain yield. Interestingly, mid season measurements
(Feekes 5 to 7) of NDVI, chlorophyll content, leaf color, CV and plant height
were found to be strongly associated with final grain yield. Variables such as
tillers per plant and canopy temperature had strong association with final grain
yield only at a specific growth stage which narrows the prediction window as a
function of growth stage. Both the correlation and regression analyses generally
suggested that several variables can be used for predicting yield potential. It
is also important to give attention to the limitation of each variable from
biological, physical and engineering perspectives. The linear multiple
regression equation suggested that NDVI, chlorophyll content, plant height and
total N uptake were good predictors of final winter wheat grain yield and the
literature review presented in this report also support this conclusion.
1. Singh, B.; Singh, Y.; Ladha, J.K.; Bronson, K.F.; Balasubramanian, V.; Singh, J.; Khind, S. Chlorophyll Meter and Leaf Color Chart-based Nitrogen Management for Rice and Wheat in Northwestern India. Agron. J. 2002, 94, 821-829.
2. Yang, W.H.; Peng, S.; Huang, J.; Sanico, A.L.; Buresh, R.J.; Witt, C. Using Leaf Color Charts to Estimate Leaf Nitrogen Status. Agron. J. 2003, 95, 212-217.
3. Scharf, P.C.; Lory, J.A. Calibrating Corn Color from Aerial Photographs to Predict Sidedress Nitrogen Need. Agron. J. 2002, 94, 397-404.
4. Karcher, D.E.; Richardson, M.D. Quantifying Turfgrass Color Using Digital Image Analysis. Crop Sci. 2003, 43, 943-951.
5. Lukina, E.V.; Stone, M.L.; Raun, W.R. Estimating Vegetation Coverage in Wheat Using Digital Images. J. Plant Nutr. 1999, 22, 341-350.
6. Purcell, L.C. Soybean Canopy Coverage and Light Interception Measurements Using Digital Imagery. Crop Sci. 2000, 40, 834-837.
7. Wood, C.W.; Reeves, D.W.; Duffield, R.R.; Edmisten, K.L. Field Chlorophyll Measurements for Evaluation of Corn Nitrogen Status. J. Plant Nutr. 1992, 15, 487-500.
8. Peterson, T.A.; Blackmer, T.M.; Francis, D.D.; Schepers, J.S. Using a Chlorophyll Meter to Improve N Management. Nebguide G93-1171A. Coop. Ext. Serv., Univ. of Nebraska, Lincoln. 1993.
9. Schepers, J.S.; Francis, D.D. Precision Agriculture – What’s in Our Future. Commun. Soil Sci. Plant Anal. 1998, 29, 1463-1469.
10. Varvel, G.E.; Schepers, J.S.; Francis, D.D. Ability for In-season Correction of Nitrogen Deficiency in Corn Using Chlorophyll Meter. Soil Sci. Soc. Am. J. 1997, 61, 1233-1239.
11. Hussain, F.; Bronson, K.F.; Singh, Y.; Singh, B.; Peng, S. Use of Chlorophyll Meter Sufficiency Indices for Nitrogen Management of Irrigated Rice in Asia. Agron. J. 2000, 92, 875-879.
12. Schepers, J.S.; Francis, D.D.; Vigil, M.; Belows, F.E. Comparison of Corn Leaf Nitrogen Concentrations and Chlorophyll Meter Readings. Commun. Soil Sci. Plant Anal. 1992, 23, 2173-2187.
13. Dinnes, D.L.; Karlen, D.L.; Jaynes, D.B.; Kaspar, T.C.; Hatfield, J.L.; Colvin, T.S.; Cambardella, C.A. Nitrogen ManagementStrategies to Reduce Nitrate Leaching on Tile-Drained Midwestern Soils. Agron. J. 2002, 94, 153-171.
14. Raun, W.R.; Johnson, G.V. Improving Nitrogen Use Efficiency for Cereal Production. Agron. J. 2001, 93, 131-138.
15. Raun, W.R.; Solie, J.B.; Johnson, G.V.; Stone, M.L.; Lukina, E.V.; Thomason, W.E.; Schepers, J.S. In-season prediction of potential grain yield in winter wheat using canopy reflectance. Agron. J. 2001, 93, 131-138.
16. Filella, R.A.; Ress, D.; Sayre, K.D.; Lu, Z.M.; Condon, A.G.; Saavedra, A.L. Wheat yield progress associated with stomatical conductance. Crop Sci. 1995, 35, 1400-1405.
17. Raun, W.R.; Solie, J.B.; Johnson, G.V.; Stone, M.L.; Whitney, R.W.; Lees, H.L.; Sembiring, H.; Phillips, S.B. Micro-variability in Soil Test, Plant Nutrient, and Yield Parameters in Bermudagrass. Soil Sci. Soc. Am. J. 1998, 62, 683-690.
18. Solie, J.B.; Raun, W.R.; Whitney, R.W.; Stone, M.L.; Ringer, J.D. Submeter Spatial Variability of Selected Soil and Bermudagrass Production Variables. Soil Sci. Soc. Am. J. 1999, 63, 1724-1733.
19. Raun, W.R.; Solie, J.B.; Johnson, G.V.; Stone, M.L.; Mullen, R.W.; Freeman, K.W.; Thomason, W.E.; Lukina, E.V. Improving Nitrogen Use Efficiency in Cereal Grain Production With Optical Sensing and Variable Rate Application. Agrin. J. 2002, 94, 815-820.
20. Solie, J.B.; Raun, W.R.; Whitney, R.W.; Stone, M.L.; Ringer, J.D. Optical Sensor Based Field Element Size and Sensing Strategy for Nitrogen Application. Trans. ASAE 1996, 39, 1983-1992.
21. Stone, M.L.; Solie, J.B.; Raun, W.R.; Whitney, R.W.; Taylor, S.L.; Ringer, J.D. Use of Spectral Radiance for Correcting In-season Fertilizer Nitrogen Deficiencies in Winter Wheat. Trans. ASAE. 1996, 39, 1623-1631.
22. Large, E.C. Growth Stages in Cereals. Plant Pathol. 1954, 3, 28-129.
23. Mallarino, A.P.; Oyarzabal, E.S.; Hinz, P.N. Interpreting Within-field Relationships Between Crop Yields and Soil and Plant Variables Using Factor Analysis. Precision Agric. 1999, 1, 15-25.
24. Machado, S.; Bynum, E.D.; Archer, T.L.; Lascano, R.J.; Wilson, L.T.; Bordovosky, J.; Segarra, E.; Bronson, K.; Nesmith, D.M.; Xu, W. Spatial and Temporal Variability of Corn Growth and Grain Yield: Implications for Site-specific Farming. Crop Sci. 2002, 42, 1564-1576.
25. Millard, J.P.; Jackson, R.D.; Goettleman, R.C.; Reginato, R.J. Crop Water-stress Assessment Using an Airborne Thermal Scanner. PE&RS. 1978, 44(1), 77-85.
26. Rashid, A.; Stark, J.C.; Tanveer, A.; Mustafa, T. Use of Canopy Temperature Measurements as a Screening Tool for Drought Tolerance in Spring Wheat. J. Agron. and Crop Sci. 1999, 182, 231-237.
27. Nielson, D.C.; Halvorson, A.D. Nitrogen Fertility Influence on Water Stress and Yield of Winter Wheat. Agron. J. 1991, 83, 1065-1070.
28. Clawson, K.L.; Jackson, R.D.; Pinter, P.J. Evaluating Plant Water Stress With Canopy Temperature Differences. Agron. J. 1989, 81, 858-864.
29. Fisher, R.A.; Ress, D.; Sayre, K.D.; Lu, Z.M.; Condon, A.G.; Saavedra, A.L. Wheat Yield Progress Associated With Stomatical Conductance. Crop Sci. 1998, 38, 1467-1475.
30. McLeod J.G.; Campbell, C.A.; Gan, Y.; Dyck, F.B.; Vera, C.L. Seeding Depth, Rate and Row Spacing for Winter Wheat Grown on Stubble and Chemical Fallow in the Semiarid Prairies. Can. J. Plant Sci. 1995, 76, 207-214.
31. Johnson, A.; Evans, M.; Wansink, K. Challenging Sowing Rates for Wheat to Achieve Target Plant Densities. http://www.regional.org.au/au/asa/2001/p/10/johnson.htm. (accessed Feb 2004).
32. Conventry, D.R.; Cann, D.K.; Brooke, H.D.; Reeves, T.G. Influence of Genotype, Sowing Date, and Seeding Rate on Wheat Development and Yield. Aust. J. Exp. Agric. 1993, 33(6), 751-757.
33. Rurnett, V.F.; Schneider, J.H.; Dempsey, F.W. Effect of Seeding Rate on Organic Wheat Production and Ryegrass Density in Northeast Victoria. http://www.regional.org.au/au/asa/2003/c/7/burnett.htm#TopOfPage. (accessed Feb 2004).
34. Stougaard, R.N.; Xue, Q.; Spring Wheat Seed Size and Seeding Rate Effects on Yield Loss Due to Wild Oat (Avena fatua) Interference. Weed Sci. 2003, 52(1), 133-141.
35. Peterson, C.M.; Klepper, B.; Rickman, R.W. Tiller Development at the Coleoptiler Node in Winter Wheat. Agron. J. 1982, 74, 781-784.
36. Flowers, M.; Weisz, R.; Heiniger, R. Remote Sensing of Winter Wheat Tiller Density for Early Nitrogen Application Decisions. Agron. J. 2001, 93, 783-789.
37. Flowers, M.; Weisz, R.; Heiniger, R.; Tarleton, B.; Meijer, A. Field Validation of a Remote Sensing Technique for Early Nitrogen Application Decisions in Wheat. Agron. J. 2003, 95, 167-176.
38. Robinson, J.B.; Freebairn, D.M.; Dimes, J.P.; Dalal, R.C.; Thomas, G.A.; Weston, E.J. Modeling Wheat Production from Low-rainfall Farming Systems in Northern Australia. Environmental International. 1999, 25(6-7), 861-870.
39. Major, D.J.; Blad, B.L.; Bauer, A.; Hatfield, J.L.; Hubbard, K.G.; Kanemasu, E.T.; Reginato, R.J. Winter Wheat Grain Yield Response to Water and Nitrogen on the North American Great Plains. Agr. Forest Meteorol. 1988, 44(2), 141-149.
40. Hoogenboom, G.; Jones, J.W.; Wilkens, P.W.; Batchelor, W.D.; Bowen, W.T.; Hunt, L.A.; Pickering, N.B.; Singh, U.; Godwin, D.C.; Baer, B.; Boote, K.J.; Ritchie, J.T.; White, J.W. Crop Models. G.Y. Tsuju DSSAT; Vehara, G., Balas, S., Eds; Version 32; University of Hawaii: Honolulu, Hawaii, 1994; 95-244.
41. Heinemann, A.B.; Hoogenboom, G.; Chojnicki, B. The impact of Potential Errors in Rainfall Observation on the Simulation of Crop Growth, Development and Yield. Ecological Modelling. 2002, 15(1), 1-21.
42. Huda, A.K.S. Management Strategies to Minimize Climatic Risk to Wheat Production in Low Rainfall Areas of Southern Australia. Agricultural and Forest Meteorology. 1994, 69(1-2), 125-146.
43. Arora, V.K.; Gijri, P.R. Evaluation of a Crop Growth-water Balance Model for Analyzing Wheat Responses to Climate and Water-limited Environments. Field Crop Research. 1998, 59(3), 213-224.
44. Troeh, F.R.; Thompson, L.M. Soils and Fertility, 5th Ed.; Oxford University Press: New York, 1993; 55.
45. Munsell. Munsell® Soil Color Charts, 1988 Ed.; MacBeth Division of Kollmorgen Instruments Corporation: Baltimore, Maryland, 1988.
46. Sylvia, D.M.; Fuhrmann, J.J.; Hartell, P.G.; Zuberer, D.A. Principles and Applications of Soil Microbiology: Prentice Hall, Upper Saddle River, New Jersey, 1999; 24-25.
47. Brady, N.C.; Weil, R.R. The Nature and Properties of Soils, 13th Ed.; Prentice Hall Publishing Co, New York, 2002.
48. Sparks, D.L. Environmental Soil Chemistry, 2nd Ed.; Academic Press, New York, 2003; 268-270.
49. Jones, J.B. Jr. Laboratory Guide for Conducting Soil Tests and plant Analysis; CRC Press, Boca Raton, Florida, 2001; 158.
50. Gee, G.W.; Bauder, J.W. Particle Size Analysis. Methods of Soil Analysis, Part 1 Physical and Mineralogical Methods, 2nd Ed.; Klute, A.; American Society of Agronomy, Inc., Madison, Wisconsin, 1986; 383-411.
51. Chan, K.Y. Bulk Density. Encyclopedia of Soil Science; Lal, R., Eds; Marcel Dekker: New York, 2002; 128-130.
52. Zoebisch, M.A.; Dexter, A.R. Encyclopedia of Soil Science; Lal, R., Eds; Marcel Dekker: New York, 2002; 311-316.
53. Droogers, P.; Bouma, J. Biodynamic vs. Conventional Farming Effects on Soil Structure Expressed by Simulated Potential Productivity. Soil Sci. Soc. Am. J. 1996, 60, 1552-1558.
54. Ishaq, M.; Hassan, A.; Saeed, M.; Ibrahim, M.; Lal, R. Soil Compaction Effects on Crops in Punjab, Pakistan. I. Soil Physical Properties and Crop Yield. Soil Tillage Res. 2001, 59, 57-65.
55. Lal, R.; Ahmadi, M. Axle Load and Tillage Effects on Crop Yield for Two Soils in Central Ohio. Soil Tillage Res. 2000, 54, 111-119.
56. Evans, G. Acidic soils in Australia. Bureau of Rural Resources, Department of Primary Industries and Energy: Canberra. 1991; 38.
57. Ringuelet, A.; Bachmeier, O.A. Kinetics of soil nitrogen mineralization from undisturbed and disturbed soil. Comm. Soil Sci. and Plant Anal. 2002, 33, 3703-3721.
58. Drury, C.F.; Zhang, T.Q.; Kay, B.D. The non-limiting and least limiting water ranges for soil nitrogen mineralization. Soil Sci. Soc. Am. J. 2003, 67, 1388-1404.
59. Vigil, M.F.; Kissel, D.E. Equations for estimating the amount of nitrogen mineralized from crop residues. Soil Sci. Soc. Am. J. 1991, 55, 757-761.
60. Rassmussen, P.E.; Douglas, C.L., Jr.; Collins, H.P.; Albrecht, S.L. Long-term cropping system effects in mineralizable nitrogen in soil. J. Soil Bio. Biochem. 1998, 30, 1829-1837.
61. Walley, F.; Yates, T.; van Groenigen, J.W.; Kessel, C. Relationships between soil nitrogen availability indices, yield, and nitrogen accumulation of wheat. Soil Sci. Soc. Am. J. 2002, 66, 1549-1561.
62. Malhi, S.S.; Nyborg, M.; Heaney, D.J. Potential for nitrogen mineralization in central Alberta soils. J. Fert. Resch. 1992, 32, 321-325.
63. Groffman, P.M. Carbon additions increases nitrogen availability in northern hardwood forets soils. Biol. Fert. Soils. 1999, 29, 430-433.
64. Ndegwa, P.M.; Thompson, S.A. Effects of C-to-N ratio on vermicomposting of biosolids. Bioresource Tech. 2000, 75, 7-12.
65. Risai, E.L.; Tian, G.; Khang, B.T.; Opuwaribo, E.E. Nitrogen mineralization of roots of maize and selected woody species. Comm. Soil Sci. Plant Anal. 1999, 30 (9&10), 1431-1447.
66. Rovira, P.; Vallejo, V.R. Mineralization of carbon and nitrogen from plant debris size and depth of burial. Soil Biol. Bioch. 2002, 34, 327-339.
67. Cote, L.; Brown, S.; Pare, D.; Fyles, J.; Bauhus, J. Dynamics of carbon and nitrogen mineralization in relation to stand type, stand age, and soil texture in the boreal mixedwood. Soil Biol. Bioch. 2000, 32, 1079-1090.
68. Stocton, R.D.; Krenzer, E.G., Jr.; Solie, J.; Payton, M.E. Stand establishment of winter wheat in Oklahoma: A survey. J. Prod. Agric. 1996, 9(4), 571-575.
69. Weiz, R.; Crozier, C.R.; Heiniger, R.W. Optimizing nitrogen application timing in no-till soft red winter wheat. Agron. J. 2001, 93, 435-442.
70. Steel, R.G.D.; Torrie, J.H.; Dickey, D.A. Principles and Procedures of Statistics A Biometrical Approach. 3rd Ed. McGraw-Hill Book Co.: New York, NY, 1997.
71. Little, T.M.; Hills, F.J. Agriculture Expansion. John Wiley and Sons: New York, 1978.
72. Taylor, S.L.; Johnson, G.V.; Raun, W.R. A field exercise to acquaint students with soil testing as a measure of soil fertility status and field variability. J. Nat. Res. Life Sci. Educ. 1997, 26, 132-135.
73. Washmon, C.N.; Solie, J.B.; Raun, W.R.; Itenfisu, D.D. Within field variability in wheat grain yields over nine years in Oklahoma. J. Plant Nutr. 2002, 25, 2655-2662.
74. Lukina, E.V.; Raun, W.R.; Stone, M.L.; Solie, J.B.; Johnson, G.V.; Lees, H.L.; LaRuffa, J.M.; Phillips, S.B. Effect of row spacing, growth stage, and nitrogen rate on spectral irradiance in winter wheat. J. Plant Nutr. 2000, 23, 103-122.
75. Lukina, E.V.; Freeman, K.W.; Wynn, K.J.; Thomason, W.E.; Mullen, R.W.; Johnson, G.V.; Elliott, R.L.; Stone, M.L.; Solie, J.B.; Raun, W.R. Nitrogen fertilization optimization algorithm based on in-season estimates of yield and plant nitrogen uptake. J. Plant Nutr. 2001, 24, 885-898.
76. Blake, G.R.; Hartge, K.H. Bulk Density. Methods of Soil Analysis, Part 1-Physical and Mineralogical Properties, 2nd Ed.; Klute, A. Ed.; ASA and SSSA: Madison, WI, 1986; 363-382.
77. Stanford, J.L.; Smith, S.J. Nitrogen mineralization potentials of soils. Soil Sci. Soc. Am. Proc. 1972, 36, 465-472.
78. Smith, J.L.; Schnobel, R.R.; McNeal, B.L.; Campbell, G.S. Potential errors in the first order model for estimating soil nitrogen mineralization potentials. Soil Sci. Soc. Am. J. 1980, 44, 996-1000.
79. Bremner, J.M. Methods of Soil Analysis; Part 2; Black, C.A., Ed.; Agron. Monogr. 9. American Society of Agronomy, Madison, WI.
80. Schepers, J.S.; Francis, D.D.; Thompson, M.T. Simultaneous determination of total C, total N, and 15N on soil and plant material. Commun. Soil Sci. Plant Anal. 1989, 20, 949-959.
81. Mallows, C.L. Some Comments on Cp. Technometrics 1973, 15, 661 -675.
82. Kutner, M.H.; Nachtsheim, C.J.; Neter, J. Chapter 9 building the regression model I. Model selection and validation. Applied linear regression models, 4th Ed.; McGraw Hill Comp., Inc.: New York, 2004; 294-305.
83. Davis, R.L.; Patton, J.J.; Teal, R.K.;
Tang, Y.; Humphries, M.T.; Mosali, J.; Girma, K.; Lawles, J.W.; Moges, S.M.;
Malpati, A.; Si, J.; Zhang, H.; Deng, S.; Johnson, G.V.; Mullen, R.W.; Raun,
W.R. Nitrogen balance in the Magruder Plots following 109 years in continuous
winter wheat. J. Plant Nutr. 2003, 26(8), 1561-1580.
Table 1. Response of NDVI, CV (from NDVI) and leaf color chart readings as influenced by treatment, long-term fertility experiment #222, Stillwater, OK, 2004.
N, P, and K applied preplant and incorporated as ammonium nitrate (34%N), triple super phosphate (20%P), and potassium chloride (49%K).
NDVI – normalized difference vegetative index.
SPAD – soil plant analysis development developed by Minolta Co. as an indirect measure of Chlorophyll
LC, leaf color chart, (range of 1-5)
F5 – Feekes growth stage 5 (leaf sheaths strongly erected)
F7 – Feekes growth stage 7 (second node of stem visible)
F8 – Feekes growth stage 10 (in boot)
SED – standard error of the difference between two equally replicated means
Table 2. Response of canopy temperature as influenced by treatment, long-term fertility experiment #222, Stillwater, OK, 2004.
N, P, and K applied preplant and incorporated as ammonium nitrate (34%N), triple super phosphate (20%P), and potassium chloride (49%K).
F5 – Feekes growth stage 5 (leaf sheaths strongly erected)
F7 – Feekes growth stage 7 (second node of stem visible)
F8 – Feekes growth stage 10 (in boot)
collected at 12:30 p.m.
r – simple correlation coefficient for each variable versus final grain yield
Table 3. Soil moisture content at different depths, Long-term Fertility Experiment #222, Stillwater, OK, 2004.
Mean – mean gravimetric moisture % determined by averaging 4 cores.
SED – standard error of the difference between two equally replicated means, = √2*s2/n
Table 4. Response of bulk density (0-15 cm), NH4-N, NO3-N, total N, organic C, and pH as influenced by treatment, long-term fertility experiment #222, Stillwater, OK, 2004.
N, P, and K applied preplant and incorporated as ammonium nitrate (34%N),
triple super phosphate (20%P), and potassium chloride (49%K).
SED – standard error of the difference between two equally replicated means, = √2*s2/n
r – simple correlation coefficient for each variable versus final grain yield
Table 5. Cumulative N mineralized in 120 days of aerobic incubation as influenced by treatment, Experiment #222, Stillwater, OK, 2004.
Table 6. Identification of predictor variables and best linear regression models for explaining variation in final grain yield in winter wheat
TN and OC - total soil N and organic carbon determined by dry combustion
NDVI_F5, 7 and 10 – normalized difference vegetative index sensed at Feekes 5, 7 and 10
SPAD_F7 - leaf chlorophyll content measured at Feekes 7
HT_F7 - plant height measured at Feekes 7
pb - Bulk density
CV_F5 & 7 - NDVI coefficient of variation at Feekes 5 & 7
Figure 1. Soil resistance by depth as influenced by annually applied N-P-K (kg ha-1) in Experiment 222, Stillwater, OK