Kyle W. Freeman, Daryl B. Arnall, Robert W. Mullen, Kefyalew Girma, Kent L. Martin, Roger K. Teal and William R. Raun

 

ABSTRACT

As research intensifies on developing precision agricultural practices for corn production, an important component will be to identify the scale at which these practices should be implemented.  It is hypothesized that differences in corn production exist on a by-plant basis. This 2-yr study was conducted to determine if corn forage biomass and nitrogen uptake could be determined on a by-plant basis.  Further study focused on the use of by-plant height measurements for improved prediction of plant biomass and nitrogen uptake.  Experimental locations were Efaw research farm, Stillwater, OK, Perkins research station, Perkins, OK, and Lake Carl Blackwell irrigated research station, near Perry, OK.  Optical sensor readings were collected on corn plants at various growth stages ranging from V8 to VT.  The average NDVI for each plant was calculated over the area occupied by the plant and correlated with plant biomass, forage yield based on the area occupied by the plant, and nitrogen uptake of that plant.  Plant height alone was a good predictor of plant biomass without accounting for area at all stages of growth sampled.  The index of NDVI x plant height provided the highest correlation with by-plant forage yield on an area basis.  Forage N uptake was most accurately predicted using NDVI alone.  The area occupied by each plant was not found to be related to plant height or plant biomass.

 

 INTRODUCTION

            Araus (1996) reported that indices based on red/near infrared ratios can yield estimates of leaf area index (LAI), green biomass, crop yield, and canopy photosynthetic capacity.  As noted by Filella et al. (1995), remote sensing can provide inexpensive, large-area estimates of nitrogen (N) status in wheat. They further reported that the use of reflectance at 430, 550, 680 nm, and red edge wavelengths offers potential for assessing N status of wheat.  Stone et al. (1996) demonstrated that N uptake of winter wheat and the normalized difference vegetative index (NDVI) are highly correlated.  Fox et al. (2001) compared late-season diagnostic tests for predicting N status of corn.  Their work showed that the stalk NO₃^- test was an excellent predictor of corn N status when samples are collected from one-fourth milk line growth stage (MLGS) to a few weeks after black-layer formation.  They also noted that chlorophyll meter (CM) readings are an accurate predictor of N sufficiency if drought-stressed fields are not included, and they added that the advantage of the CM test is that it gives on-site results.  GopalaPillai and Tian (1999) used high-resolution color infrared (CIR) images collected from an airborne digital camera to detect spatial variability of crop nutrient stress and spatial variability of grain yield.  These CIR images could easily delineate levels of nitrogen stress in poor areas of the field 75 days after planting, however, they could not differentiate between N levels in the areas of the field with higher fertility.  They also concluded that NDVI was a better indicator of N stress than uncalibrated image gray level values.        

Katsvairo et al. (2003) studied how biomass, N concentrations, and N uptake could be used to facilitate variable rate N management.  They found that these factors had no spatial variability at V6, R1, and R6 growth stages.  However, they did state that plant height showed significant spatial variability but did not consistently correlate with corn yields in a dry year, but they recognized that more research should be conducted on plant height measurements.  A study by Muchado et al. (2002) revealed that by using plant height, 90 and 61% of the variation in total dry matter and grain yield, respectively, could be explained in a dry year.  These data are supported by Sadler et al. (1995), who reported that differences in phenology, biomass, leaf area, and yield components were most pronounced under drought. 

As precision farming becomes accepted and adopted, delineating the proper field element size becomes more important.  Sadler et al. (2000) studied the effects of soil variation on crop phenology, biomass, and yield components of corn under drought.  Their experiment analyzed detailed soil maps at a scale of 1:1200 and extensive sampling of crop characteristics across an eight hectare field.  The results proved that grain yield variation within a soil map unit was too large for the soil survey alone to be used to create homogenous soil management zones for use in precision farming.  Sadler et al. (2000) went on to state that these results supported the need for on-the-go measurements of soil properties and plant response that could be used in conjunction with soil surveys to create management zones that can be used in models, or by themselves, to predict grain yield. 

Solie et al. (1996) defined field element size as the area that provides the most precise measure of the available nutrient and where the level of that nutrient changes with distance.  This work went on to say that the fundamental field element size averages 1.5 m².  A microvariability study by Raun et al. (1998) found significant differences in surface soil test analyses when samples were <1m apart for both mobile and immobile nutrients.  Solie et al. (1999) stated that in order to describe the variability encountered in field experiments, soil, plant, and indirect measurements should be made at the meter or submeter level. 

            Identifying and understanding the variability among plant-to-plant spacing within the row is also crucial for precision farming techniques.  This variability is usually due to the combination of crowded plants (doubles, triples, etc.) and long gaps or skips.  It is possible that plants next to gaps can compensate and produce larger ears, but they generally cannot compensate enough for the smaller ears of the crowded plants that are competing for sunlight, water and nutrients.   A growth stage difference of two leaves or greater between adjacent plants in a row will almost always result in the later developing plant being barren at harvest (Nielson, 2001).  Nielson went on to quantify the variability between plants in a row by using plant spacing variability (PSV).  The PSV is simply the standard deviation of the plant spacing within a representative row in a field.  Nielson noted that in 350 production corn fields in Indiana and Ohio, 16% had a PSV of three inches or less, 60% had a PSV of three to five inches, and 24% of the fields had a PSV of six inches or greater.  Further research showed that for every one inch in PSV about 157 kg ha^-1 of yield loss occurred. 

The use of chlorophyll meters to measure N status in corn has been a very successful technique.  The success of this technique is due primarily to the high correlation between chlorophyll content and leaf N concentration (Schepers et al. 1992).  Nitrogen fertilization strategies using chlorophyll meters are now implemented in the corn belt.  Varvel et al. (1997) discussed the use of reference strips of nitrogen in corn fields.  They implemented the sufficiency index concept and used chlorophyll meters to measure crop health.  They applied nitrogen fertilizer when the crop had a chlorophyll meter reading less than 95% of the reference strip.  They stated that this concept of using chlorophyll meters and sufficiency index should result in greater N use efficiency and less N being available for leaching to the groundwater, since these applications are made when N uptake by corn is greatest. 

In 1996, Stone et al. investigated the use of hand-held sensors to detect and predict forage N uptake and grain yields in winter wheat.  These sensors measured red and near infrared irradiance from the crop.  These irradiance measurements were then used to calculate NDVI.  They found NDVI to be highly correlated with forage N uptake and grain yields of winter wheat.  Johnson and Raun (2001) developed a fertilizer response index (RI) that was calculated by dividing the average NDVI from a non-N limiting strip (created in each field by fertilizing a strip at a rate where N would not be limiting throughout the season) by the average NDVI in a parallel strip that was representative of the N availability across the field as affected by N fertilizer applied by the farmer.  This RI would then suggest how responsive the crop would be to added N fertilization in a given year.  Similarly, Raun et al. (2002) showed that their methods recognize that each 1m² area in wheat fields need to be sensed and managed independently and that the need for fertilizer N is temporally dependent.  The objective of this study was to relate by-plant sensor data and plant height with corn forage biomass, corn forage yield, and corn forage N uptake.

 

MATERIALS AND METHODS      

Two dryland field experiments were initiated in the spring of 2003 to evaluate the use of sensor readings for predicting by-plant total biomass and N uptake.  The locations included Efaw and Perkins research stations in 2003 and 2004.  In 2005, experiments were located at the Efaw research station and at the Lake Carl Blackwell (LCB) irrigated research farm.  All locations were planted with a row spacing of 0.76 m.  The soil at Efaw is classified as Easpur loam (fine-loamy, mixed superactive thermic Fluventic Haplustoll).  Perkins is classified as Teller sandy loam (fine, mixed, thermic Udic Argiustolls).  Soil classification of the Lake Carl Blackwell experiment is Pulaski fine sandy loam (coarse/loamy, mixed nonacid, thermic, Typic, Ustifluvent). 

 For each site and forage harvest, 13 to 17 m of row were identified that included exactly 50 corn plants.  Three forage harvests of 50 individual plants were taken at each location at various growth stages (Table 1).  Each plant was sensed using a GreenSeeker active, optical sensor that was mounted to a bicycle with a shaft encoder to log distance (1 reading per cm of linear distance traveled) with each NDVI reading collected. 

The GreenSeeker™ Hand Held Optical Sensor Unit (NTech Industries, Inc.) was used to collect normalized difference vegetative index (NDVI) measurements.  This device uses a patented technique to measure crop reflectance and to calculate NDVI.  The unit senses a 0.6 x 0.01 m spot when held at a distance of approximately 0.6 to 1.0 m from the illuminated surface.  The sensed dimensions remain approximately constant over the height range of the sensor.  The sensor unit has self-contained illumination in both the red (650 ± 10 nm full width half magnitude (FWHM)) and NIR (770 ± 15 nm FWHM) bands.  The device measures the fraction of the emitted light in the sensed area that is returned to the sensor (reflectance or ρ).  These fractions are used within the sensor to compute NDVI according to the following formula:

                                                                          

            Where:

ρ_NIR – Fraction of emitted NIR radiation returned from the sensed area(reflectance)  

ρ_Red – Fraction of emitted red radiation returned from the sensed area (reflectance)

 

The sensor samples at a very high rate (approximately 1000 measurements per second) and averages measurements between outputs (each cm).  The sensor was passed over the crop at a height of approximately 0.9 m above the crop canopy and oriented so that the 0.6 m sensed width was perpendicular to the row and centered over the row.  With advancing stage of growth, sensor height above the ground increased proportionally.  The mean NDVI was computed for each plant across growth stages and sensing dates.   Growth stages in corn were identified using the terminology developed at Iowa State University (1993).

Immediately after sensing, each plant was cut at ground level and wet weights recorded by plant.  Each plant was then dried at 75 °C for 4 days and dry weights subsequently recorded.  Dry plant material was then ground to pass a 240 mesh screen and analyzed for total N using a Carlo-Erba dry combustion unit (Schepers et al., 1989).

            To determine corn forage yield it was imperative to determine the area that each individual plant occupied.  Prior to sensing, the distance between each plant was measured at each site and harvest.  The area each plant occupied was calculated by taking the distance half way to the plant in front and behind it.  This determined the linear dimension and was then multiplied by the row width of 0.76 m to calculate the area for a given plant.  This process allowed us to determine a forage yield per unit area as a function of the linear distance between plants within a row.  Consistent with the method used to obtain by-plant forage yields, NDVI readings for each plant were determined in the same fashion whereby sensor readings ½ the distance to the neighboring plant in front and behind of the plant in question were averaged and subsequently paired with the dry matter data.  This was accomplished by employing the shaft encoder since distance and NDVI were written to the data file.  Because total distances and distances between plants were recorded previously, sensor data could be partitioned accordingly. 

            Plant heights were also recorded for each individual plant prior to harvest.  Plant height was determined by extending the last collared leaf upright.  For the 3^rd cutting the corn height was measured to the top of the tassel.

            An index was calculated by multiplying NDVI readings and plant height in order to assess a pseudo three dimensional image of total biomass.  Earlier work by Stone et al., (1996) showed that NDVI alone was an excellent predictor of wet and/or dry winter wheat biomass.  Regression analysis was done using plant height, NDVI, and NDVI x height index (the product of NDVI and plant height) as independent variables and corn forage biomass (g per plant not accounting for area occupied by the plant), corn forage yield (g of biomass accounting for area occupied by plant), and corn forage N uptake (forage yield*N concentration of forage) as dependent variables.

 

RESULTS AND DISCUSsION

In an attempt to more accurately predict corn forage biomass, the data were divided into two groups based on growth stage.  The first group consisted of corn plants harvested between the V8 and V10 growth stages (less than 65 days from planting to harvest).  The second group consisted of corn plants harvested between growth stages V11 and R1 (greater that 65 days from planting).  Table 1 describes growth stage and days from planting to harvest.  Plant populations were variable across harvest, locations and years (Table 1).  The Efaw location had consistently higher populations than the Perkins locations in 2003 and 2004.  In 2005, the LCB site was under sprinkler irrigation.

 Over years, locations, and stages of growth, the correlation of NDVI and height with wet biomass (either determined using determinate area, or not) was improved when compared to dry biomass.  However, for purposes of reporting the findings of this work, the focus has remained on dry biomass, despite the lower resultant correlation due to the errors associated with moisture determination (wet weights, dry weights, etc.). 

 

Biomass

            Across years and locations, NDVI was a poor predictor of dry plant biomass.  As noted above, NDVI was calculated and measured for each corn plant by averaging sensor readings from half the distance to the preceding plant and half the distance to the following plant in a row.  With unequal spacing often incurred by mechanized corn planting, the area occupied by individual corn plants varied.  This variation in plant spacing affected the ability of NDVI alone to predict dry corn biomass at early and late stages of growth.  There was a significant relationship between NDVI and plant biomass for data collected from corn growth stages V8 to V10 (Table 2).  

Height and biomass were highly correlated independent of the area the plant occupied (Figures 1 and 2), and the correlation with forage biomass was much better at earlier stages than later stages.  This is important because it indicates that height can be used by itself to estimate plant biomass without having to compensate for the area occupied by the plant.  Using data from the V8-V10 growth stages, the area occupied per plant was not related to either dry biomass per plant or plant height (Table 2).  These combined results (no correlation of area with either height or biomass) indicate that area was not an important variable in the prediction of dry biomass using the populations and hybrids employed in this trial.  When areas were partitioned into the following categories (area <0.2 m² or >0.2 m²), the resultant correlations were nearly identical, again suggesting the independence of these relationships as a function of area (graphs not reported).

The index of NDVI x height was a good predictor of plant biomass.  However, plant height by itself was more accurate in predicting dry plant biomass (Figures 3 and 4).   A linear regression performed between the NDVI x height index and plant biomass at early growth stages resulted in an r² value of 0.66.  At the later growth stage the NDVI x height index was not as good in predicting plant biomass compared to early growth stages.

 

Forage Yield

            Across years and locations, NDVI more accurately predicted forage yields accounting for area at earlier stages of growth (Figures 5 and Table 2).   This improved relationship between NDVI and corn forage yield at earlier growth stages is explained by increased sensitivity of NDVI.  When corn is younger and smaller, the sensor has the ability to detect more soil area of lower yielding plants compared to higher yielding plants.  Conversely, at later stages of growth, corn plants were taller which required increased elevation of the sensor and subsequently soil background had a diminished influence on NDVI.  The lower plant populations and poor growing conditions at Perkins consistently produced lower yielding plants and lower NDVI values than the Efaw location.

            Plant height measurements were also used to predict corn forage yield accounting for area.  At growth stages ranging from V8 to V10, plant height predicted forage yield similar to NDVI.  However, at later growth stages, plant height was a better predictor of forage yield than NDVI (Figures 6 and Table 2).  For the duration of this experiment, location, growth stage, and year tended to produce distinct data clusters when plant height and forage yield were plotted.  This observation was not noted when plant height was correlated with by-plant dry biomass at the early growth stages (Figure 1).  This further explains the ability of plant height to predict biomass, and the finding that there is little benefit in considering the area that the plant occupies across locations and growth stages of this experiment (Figure 1). 

            The NDVI x plant height index was also correlated with corn forage yield accounting for area.  This index proved to be a better predictor of corn forage yield than either NDVI or plant height alone (Figures 7 and 8).  Further investigations showed that this index performed similarly at early and later stages of growth. 

The relationship between forage yield and NDVI, plant height, and their product was investigated across all years, locations, and growth stages.  When growth stages were combined, there was an improved correlation between plant height and the NDVI x height index and by-plant forage yields (data not reported).  Alternatively, NDVI was a much poorer predictor of corn forage biomass across all growth stages compared to separate evaluation of NDVI and corn forage yield by two growth stage ranges (V8-V10, and V11-R1).  Although there was an increase in the ability of plant height and the NDVI x height index to predict forage yield when plotted across all stages of growth, they were partitioned separately to better understand if forage yields could be predicted at early stages of growth.

 

N uptake

            The amount of N taken up in corn forage was highly correlated with NDVI (Figures 9 and 10).  At early stages of growth, NDVI explained 64% of the variation in N uptake.  This correlation was slightly lower at later growth stages.  In both cases, NDVI proved to be a better predictor of N uptake than forage yield or plant biomass (data not shown).  This increase in correlation with N uptake could be explained by the ability of NDVI to detect differences in red absorption and variation in chlorophyll content.  Thomas and Oernther (1972) noted similar finding in sweet peppers, as N-deficiency symptoms became more pronounced reflectance in the visible portion of the spectrum (500 to 700 nm) increased.  However, No relationship was noted between NDVI and tissue N concentration in corn forage across years, locations, and growth stages (data not reported).

At early and later stages of growth, plant height was not as accurate a predictor of N uptake in the forage as NDVI (Table 2).  Bt, at later growth stages, there was a significant correlation between plant height and N uptake than earlier growth stages.  A relationship was present between plant height and N uptake at early growth stages, however this relationship differed based on growth stage and location.  Forage harvested between V8 and V9 at Efaw in 2004 and LCB in 2005 took place following early irrigations.  This may have allowed for favorable growing conditions that led to increased N uptake compared to the other locations that were harvested at early growth stages, but where moisture was limiting.  This difference in growing conditions resulted in decreased correlation between plant height and N uptake at growth stages V8 to V10.  At later growth stages, the relationship of plant height and NDVI was much improved than at earlier growth stages. 

The NDVI x plant height index was also a good predictor of N uptake in corn forage.  Similar to plant height, this index had a much stronger relationship with N uptake at later growth stages compared to earlier stages of growth.  The V8-V10 growth stages did show a correlation with N uptake, but this relationship was not consistent across locations (Table 2).  The NDVI x height index expressed a strong relationship with N uptake (r² = 0.77) using an exponential model for corn forage harvested from V11-R1 growth stages (Figure 11).

 

CONCLUSIONS

The objective of this experiment was to determine if corn forage biomass, corn forage yield, and corn forage N uptake could be accurately predicted using by-plant sensor data and plant height collected at various stages of corn development.  Results showed that forage biomass, forage yield, and forage N uptake could be accurately predicted using indirect measures.  By-plant forage yields, accounting for area occupied by the plant, were accurately predicted using the index NDVI x plant height.  Forage yields were also correlated with NDVI and plant height individually.  These relationships with forage yields were consistently better at early stages of growth.  The best predictor of forage N uptake was NDVI alone when compared to plant height and the index of NDVI x height at early growth stages.  Plant height, NDVI and their product had no relationship with forage N uptake.  Sensor NDVI was not as good a predictor of plant biomass as was plant height alone, without accounting for the area the plant occupied.   There was a better relationship with plant height and plant biomass, without accounting for the area occupied by the plant than when forage yield was calculated using area occupied.  This suggests that plant height was independent of the area occupied by the plant and was influenced more likely by day of emergence than any thing else.  Area occupied by a corn plant was shown to be unrelated to plant height or plant biomass for the in-row variability encountered in these experiments. 

 

REFERENCES

Araus, J.L.  1996.  Integrated physiological criteria associated with yield potential.  p. 150-166.  In M.P. Reynolds, S. Rajaram, and A. McNab (eds) Increasing Yield Potential in Wheat: Breaking the Barriers.  Mexico, D.F.: CIMMYT.

 

Filella, I., L. Serrano, J. Serra, and J. Penuelas. 1995. Evaluating wheat nitrogen status with canopy reflectance indices and discriminate analysis. Crop Sci. 35:1400-1405.

 

Fox, R.H., W.P. Piekielek, and K.E. Macneal.  2001.  Comparison of late-season diagnostic tests for predicting nitrogen status of corn. Agron. J. 93:590-597.

 

GopalaPillai, S. and L. Tian. 1999. In-field variability detection and spatial yield modeling for corn using digital aerial imaging.  Trans. ASAE 42(6): 1911-1920.

 

Iowa State University.  1993.  How a Corn Plant Develops, Special Report No. 48.  Cooperative Extension Service, Ames, IA.  (http://maize.agron.iastate.edu/corngrows.html#v9mg) (Verified – 02/07/2006).

 

Johnson, G.V. and W.R. Raun. 2003. Nitrogen response index as a guide to fertilizer management. J. Plant Nutr. 26:249-262.

 

Katsvairo, T.W., W.J. Cox., and H.M. Van Es. 2003.  Spatial growth and nitrogen uptake variability of corn at two nitrogen levels.  Agron. J. 95:1000-1011.

 

Machado, S., E.D. Bynum, Jr., T.L. Archer, R.J. lascano, L.T. Wilson, J. Bordovsky, E. Segarra, K. Bronson, D.M. Nesmith, and W. Xu.  2002.  Spatial and temporal variability of corn growth and grain yield:  implications for site-specific farming. Crop Sci. 42:1564-1576.

 

Mullen, R.W., Kyle W. Freeman, William R. Raun, G.V. Johnson, M.L. Stone, and J.B. Solie. 2003.  Identifying an in-season response index and the potential to increase wheat yield with nitrogen. Agron. J. 95:347-351.

 

National Agriculture Statistics Service.  2003. Track records, United States crop production.  United States Department of Agriculture.  Available at http://www.usda.gov/nass/pubs/trackrec/track03c

 

Nielson, R. L.  2001.  Stand establishment variability in corn.  Dept. of Agronomy publication # AGRY-91-01.  Purdue University, West Lafayette, IN.

 

Norwood, C.A. and R.S. Currie.  1996.  Tillage, planting date, and plant population effects on dryland corn. J. Prod. Agric. 9:119-122.

 

Sadler, E.J., P.J. Bauer, and W.J. Busscher. 1995.  Spatial corn yield during drought in the SE coastal plain.  P. 365-381. In P.C. Robert et al. (ed.) Site-specific management for agricultural systems. ASA, CSSA, SSSA, Madison, WI.

 

Sadler, E.J., P.J. Bauer, and W.J. Busscher.  2000.  Site-specific analysis of a droughted corn crop: Ι. Growth and grain yield. Agron. J. 92:395-402.

 

Schepers. J.S., D.D. Francis, and M.T. Thompson. 1989. Simultaneous determination of total C, total N and 15N on soil and plant material. Commun. Soil Sci. Plant Anal.  20:949-959.

 

Schepers. J.S., D.D. Francis, M.F. Vigil, and F.E. Below. 1992. Comparisons of corn leaf nitrogen and chlorophyll meter readings.  Commun. Soil Sci. Plant Anal.  23:2173-2187.

 

Solie, J.B., W.R. Raun, R.W. Whitney, M.L. Stone, and J.D. Ringer.  1996.  Optical sensor based field element size and sensing strategy for nitrogen application. Trans. ASAE.       39(6):1983-1992.

 

Solie, J.B., W.R. Raun, and M.L. Stone. 1999. Submeter spatial variability of selected soil and plant variables.  Soil Sci. Soc. Amer. J. 63:1724-1733.

 

Stone, M.L., J.B. Solie, R.W. Whitney, W.R. Raun, and H.L. Lees.  1996.  Sensors for the detection of nitrogen in winter wheat.  SAE Technical paper series.  SAE Paper No. 961757.  SAE, Warrendale, PA.

 

Stone, M.L., J.B. Solie, W.R. Raun, R.W. Whitney, S.L. Taylor, and J.D. Ringer.  1996.  Use of spectral radiance for correcting in-season fertilizer nitrogen deficiencies in winter wheat.  Trans. ASAE. 39(5):1623-1631.

 

Raun, W.R., J.B. Solie, G.V. Johnson, M.L. Stone, R.W. Whitney, H.L. Lees, H. Sembiring, and S.B. Phillips. 1998.  Micro-variability in soil test, plant nutrient and yield parameters in bermudagrass. Soil Sci. Amer. J. 62(2):683-690.

 

Raun, W.R., J.B. Solie, G.V. Johnson, M.L. Stone, R.W. Mullen, K.W. Freeman, W.E. Thomason, and E.V. Lukina.  2002.  Improving nitrogen use efficiency in cereal grain production with optical sensing and variable rate application.  Agron. J. 94:815-820.

 

Thomas, J.R. and G.F. Oerthner.  1972.  Estimating Nitrogen Content of Sweet Peppers Leaves by Reflectance Measurements.  Agron. J. 64(1):11-13.

 

Varvel, G.E., J.S. Schepers, and D.D. Francis.  1997.  Ability for in-season correction of nitrogen deficiency in corn using chlorophyll meters.  Soil Sci. Soc. Am. J. 61:1233-1239.

Table 1.  Planting and harvest dates, growth stages (stage), days from planting to forage harvest (DFP), and populations (Popu, plants ha^-1) of rows used for harvesting at each of the three corn biomass harvesting dates in by-plant corn experiment at Efaw, Perkins, and Lake Carl Blackwell, OK, 2003-2005.

^† month-day-yr  format;  ^‡  corn growth stages as defined by Iowa State University (1993)

 

 

Table 2.  Relationship between independent and dependent variables at two ranges of corn growth stages for by-plant corn experiment data averaged over locations and seasons, OK.  

 

Growth stage of corn†	Variable		Model type	r2 (prob.)
	independent	dependent
V8 to V10	NDVI	Biomass, kg	exponential	0.31 (*)
V11 to R1	NDVI	Biomass, kg	exponential	0.20 (ns)
V8 to V10	Area, m2	Biomass, kg	-‡	-
V11 to R1	Area, m2	Biomass, kg	-	-
V8 to V10	Area, m2	Height, cm	linear	0.03 (ns)
V11 to R1	Area, m2	Height, cm	linear	0.08 (ns)
V11 to R1	NDVI§	Forage yield, Mgha-1	linear	0.37 (*)
V11 to R1	Height, cm	Forage yield, Mgha-1	linear	0.43 (*)
V8 to V10	Height, cm	N uptake, kgha-1	linear	0.26 (ns)
V11 to R1	Height, cm	N uptake, kgha-1	linear	0.45 (**)
V8 to V10	NDVI X Height	N uptake, kgha-1	Exponential	0.46 (**)

^†  corn growth stages as defined by Iowa State University (1993); ^‡ no relationship between dependent and independent variables was observed; ^§ normalized Difference vegetative Index; r² denotes the proportion of variability in the dependent variable explained by the independent variable by the selected model; * and ** model significant at the 0.05, and 0.01 levels of probability, respectively; ns denotes not significant