| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Texas A&M Univ. Texas Agricultural Experiment Station, RR 3, Box 219, Lubbock, TX 79403
* Corresponding author (k-bronson{at}tamu.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Abbreviations: ET, evapotranspiration • GNDVI, green normalized difference vegetative index • GVI, green vegetative index • LEPA, Low energy precision application irrigation • NDVI, normalized difference vegetative index • NIR, near infrared • RNDVI, red normalized difference vegetative index • RVI, red vegetative index
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Previous studies on proximal sensing of cotton biomass and leaf N are few. Proximal reflectance of other crops to assess biomass and N status include corn (Zea mays L.), wheat (Triticum aestivum L.), rice (Oryza sativa L.), and soybean [Glycine max (L.) Merr.] (Bausch and Duke, 1996; Osborne et al., 2002; Stone et al., 1996; Raun et al., 2002; Takebe et al., 1990; Cassanova et al., 1998; Ma et al., 2001). Maas (1998) estimated canopy cover of cotton with a hyperspectral radiometer above the canopy. Huete et al. (1985) and Huete (1987) studied proximal spectral response of varying amounts of cotton cover to separate plant signal from soil background. Typically, ratio indices of near infrared (NIR) to red (or green) reflectance are calculated from multispectral reflectance data. The red normalized difference vegetative index or RNDVI is calculated as (RNIR - RRed)/(RNIR + RRed), where RNIR and RRed are reflectance in the NIR and in the red regions, respectively (Tucker, 1979). Shanahan et al. (2001) reported that GNDVI estimated corn yield better than RNDVI. A GVI was calculated by Bausch and Duke (1996) as RNIR/RGreen (RNIR and RGreen are reflectance in the NIR and reflectance in the green regions, respectively) from reflectance of corn under a center pivot at a 10-m height, and was related to leaf N. Li et al. (2001) reported that RNDVI and RVI (Jordan, 1969) calculated from in-season proximal sensing (3 m above the ground) of multispectral reflectance were positively correlated with cotton biomass, N accumulation, but not leaf N. In addition to canopy measurements of spectral reflectance, recent studies have demonstrated the use of measuring spectral reflectance of harvested cotton leaves to estimate N content (Saranga et al., 1998; Tarpley et al., 2000). This approach however, has the disadvantage of not estimating crop biomass and N accumulation that canopy reflectance provides (Li et al., 2001).
The hand-held chlorophyll meter has been used to rapidly determine greenness of cotton leaves, and in-directly N status (Wood et al., 1992; Wu et al., 1998; Bronson et al., 2001). Both chlorophyll meter readings and petiole NO-3 were positively affected by N fertilizer rate, but petiole NO-3 was more variable (Bronson et al., 2001). Leaf chlorophyll content estimated by chlorophyll meter readings correlated with corn yield just as well as leaf N concentration (Schepers et al., 1992). With both remote sensing approaches and the chlorophyll meter, normalizing data to well-fertilized reference plots or strips has been employed to account for variability due to varieties, growth stages, and environmental conditions in corn, rice, and wheat studies (Peterson et al., 1993; Varvel et al., 1997; Shanahan et al., 2001; Scharf and Lory, 2002; Hussain et al., 2000; Raun et al., 2002).
The first hypothesis of this study, therefore, is that proximal sensing of multispectral reflectance of cotton can be used to assess cotton leaf N and biomass from early squaring to peak bloom. We also hypothesized that the chlorophyll meter can accurately assess cotton leaf N status.
The objectives of the study were to: (i) determine the effect of N management on multispectral reflectance and chlorophyll meter readings of cotton and (ii) determine if reflectance and chlorophyll meter readings can assess cotton leaf N concentration and biomass.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The experiments were conducted in a randomized complete block design with split-plots and three replicates. There were two irrigation regimes in the main plots and five N management treatments in the subplots. The two water regimes were surface and subsurface drip irrigation at 75% estimated evapotranspirtation (ET) replacement for Lubbock; and LEPA (Lyle and Bordovsky, 1981) at 80 and 105% estimated ET replacement at Ropesville. Nitrogen management treatments consisted of well fertilized, soil test-based, reflectance-based, chlorophyll meter-based, and zero N. Rates of N fertilizer in these treatments ranged from 0 to 202 kg N ha-1. Further details of the soil properties, treatments and crop management are described in Chua et al. (2003).
Cotton (Paymaster Round-up Ready 2326, Delta and Pine land Co., Scott, MS) was seeded in 1-m beds at rates of 70000 and 100000 seed ha-1 for Ropesville and Lubbock, respectively. Planting dates were 6 May for Ropesville 2000, 2 May for Ropesville in 2001, 25 May for Lubbock 2000, and 1 May for Lubbock 2001. The 2001 Ropesville cotton crop was destroyed by hail, therefore we have no data to report for that site-year.
At the two-leaf stage, 34 kg N ha-1 as urea ammonium nitrate (320 g N kg-1) was dribbled about 10 cm away from the seed row and hand-incorporated using a hoe on all N fertilized plots, except the well-fertilized plots, which received 67 kg N ha-1. The amounts of and timing of N fertilizer applied with or just before irrigations at early squaring, early bloom, and peak bloom for each of the three site-years are detailed in Chua et al. (2003). Growth stages were established visually when approximately 50% of the plants in the field were at the growth stage of interest. In nearly all cases the growth stages determined in this manner were within a few days of the growth stage suggested by Oosterhuis (1990) based on the cumulative heat units (base 15°C).
Leaf N status and biomass were monitored at early squaring, early bloom, and peak bloom using a hand-held multispectral radiometer and chlorophyll meter. One chlorophyll meter (Model SPAD 502, Minolta Camera Co., Ltd., Japan) reading was taken from the uppermost, fully expanded leaf of 20 random plants per subplot. Chlorophyll meter readings expressed in SPAD (Soil Plant Analysis Development) units are light transmittance through the leaf blade at 650 nm compared with transmittance at 940 nm (Schepers and Francis, 1998). One plant canopy reflectance reading was taken on each of the four center rows of every subplot using a Cropscan model MSR16R radiometer (CropScan, Inc. Rochester, MN). The radiometer has a sensor with 16 pairs of upward and downward facing interference filters centered at the following wavelengths: 450, 470, 500, 530, 550, 570, 600, 630, 650, 670, 700, 780, 820, 870, 1600, and 1700 nm. The width of each waveband increases from 6.5 to 17.0 nm for the 450- to 1700-nm waveband centers. The sensor was adjusted approximately 50 cm above the canopy and centered on the row. Reflectance readings were taken between 2 h before solar noon and 20 min before solar noon. Readings were not taken when the shadow of the sensor fell within its field of view (28°), usually within 20 min of the solar noon. Overcast sky was avoided during reflectance data collection. The radiometer was periodically calibrated by using an opal glass to provide the same irradiance alternatively to the upward sensors and to the downward sensors, both at a 45° angle to the sun. Percentage of reflectance was calculated for each waveband as reflected irradiance/incoming irradiance. The post-processing program that calculated percentage of reflectance applied sun angle cosine corrections to the millivolt (mV) readings of each sensor, based on time, latitude, and longitude, as well as sensor temperature corrections (Cropscan Inc., 1998). Using percentage reflectance (R) data, the following vegetative indices were calculated: GVI = (RNIR/Rgreen) (Bausch and Duke, 1996), GNDVI = (RNIR - Rgreen)/(RNIR + Rgreen) (Gitelson et al., 1996), RVI = (RNIR/Rred) (Jordan, 1969), and RNDVI = (RNIR - Rred)/(RNIR + Rred) (Tucker, 1979). There were twelve different vegetative indices that resulted from combining the percentage reflectance data of one of the three NIR wavebands used (780, 820, and 870 nm) and one of the four red (630, 650, 670, and 700 nm); and twelve from combining reflectance of one of three NIR wavebands and one of the four green (530, 550, 570, and 600 nm) wavebands.
Sufficiency index for the chlorophyll meter plots was calculated for each growth stage as the ratio of the average chlorophyll meter reading of the plots to the average chlorophyll meter reading of the well-fertilized plots (Varvel et al., 1997) within each irrigation treatment. Sufficiency indices for the reflectance plots were calculated by growth stage for each vegetative index, by dividing the index of the reflectance plots by the average vegetative index of the well-fertilized reference plots, within each irrigation treatment. When the sufficiency index was < 0.95, 34 kg N ha-1 was applied just before or during irrigation to reflectance or chlorophyll meter plots within 24 h of the time of the readings. The vegetative index used for basing N applications to the reflectance plots was the GVI (R820/R550, R is percentage of reflectance at waveband indicated in subscript).
Aboveground biomass was sampled from 1.0 m of a row at the time the chlorophyll meter and reflectance readings were taken at early squaring, early bloom, and peak bloom. Plants were separated into leaves, stems, squares, and bolls and were dried to constant weight at 65°C. Leaves were ground to 0.5 mm, and analyzed for N concentration with a LECO FP-528 Protein N analyzer (LECO Corp., St. Joseph, MI).
Statistical Analyses
The fixed effects of irrigation, N management, and irrigation x N management were determined for leaf N concentration (g kg-1), leaf N accumulation (kg ha-1), biomass, chlorophyll meter readings, and vegetative indices for each growth stage site-year, and for lint yield by site-year using PROC MIXED in SAS for the split-plot design (SAS Institute Inc., 1999). Means were separated using Fisher's protected least significant difference (LSD) at the 0.05 probability (P) level. Leaf N and biomass was correlated to reflectance at each waveband using PROC CORR (SAS Institute Inc., 1999) for each growth stage site-year. Correlations of biomass, leaf N, leaf N accumulation, and lint yield with chlorophyll meter readings and the vegetative indices were also calculated using PROC CORR for each growth stage site-year.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
Water Management Effects
Irrigation management had essentially no effect on the three site-years of leaf N, biomass, lint yields, spectral reflectance, or chlorophyll meter readings. There were also nearly no interactions of water and N management for these parameters. For this reason, water management will not be further discussed, and all subsequent results of N management are averaged across water management within each site-year.
Nitrogen Management Effects
Ropesville 2000
Biomass was small at Ropesville 2000, and was not affected by N management at any growth stage (Table 1). Low biomass and slow growth may have been because of low soil P (11 mg Mehlich 3-P kg-1), high root-knot nematode (Meloidogyne incognita) numbers in soil (greater than damaging level of 1000 eggs 500 cm-3 suggested by Wheeler et al., 1999), and early season wind damage (Chua et al., 2003). Two weeks of high insect pressure in the reproductive stages at Ropesville in August 2000 may have also contributed to low lint yields. Leaf area was reduced about 10% by cabbage loopers (Trichoplusia ni, Hubner), and developing squares and bolls were damaged from beet army worms (Spodoptera exigua, (Hubner), which briefly reached the economic threshold of 5600 small larvae ha-1 (Muegge et al., 2001). Lint yields averaged 680 kg ha-1, with no affect of N management (Chua et al., 2003).
|
Lubbock 2000
Biomass was greater in Lubbock than in Ropesville 2000, but was only affected by N management at early bloom (Table 2). The same insects that affected the Ropesville crop were present at Lubbock in 2000, but the level of infestation and resulting damage were less than at Ropesville. Lint yields averaged 1015 kg ha-1, and N-fertilized plots were significantly greater than zero-N plots (Chua et al., 2003). Similar to Ropesville, leaf N and chlorophyll meter readings were affected by N management at all growth stages. By the Bell et al.'s (1998) guidelines, zero-N plots had deficient leaf N at early squaring. All of the treatments were deficient in leaf N at early bloom. Leaf N was greatest in the well-fertilized plots, and lowest in the zero-N plots. Nitrogen accumulation in leaf was only influenced by N treatment at early bloom. All N-fertilized treatments had significantly greater lint yields than the control, with no differences among the N-fertilized plots (Chua et al., 2003). The green-based vegetative indices were affected by N management at all growth stages, but the red-based vegetative indices were not (Table 2).
|
|
Correlation with Leaf N Concentration and Biomass
Ropesville 2000
There was no correlation between reflectance at individual wavebands and leaf N or biomass at early squaring (data not shown). Leaf N at early bloom in Ropesville was also not correlated with reflectance. Early bloom biomass however, was negatively correlated with reflectance in the visible (450–700 nm) and NIR at 1600- and 1700-nm wavebands, and positively correlated with NIR reflectance at 780 to 870 nm. At peak bloom leaf N at Ropesville 2000 was positively correlated (r = 0.42 to 0.46) with NIR reflectance from 780 to 870 nm. Biomass at this stage was negatively related to reflectance at 450, 470, 570, 600, 1600, and 1700 nm.
Leaf N was positively correlated related with chlorophyll meter readings at early bloom and peak bloom (Table 4). However, there was no significant relationship observed between biomass and chlorophyll meter readings at any growth stage. Green-based vegetative indices, GNDVI and GVI were correlated with leaf N, expect at early bloom. The red-based indices were not related to leaf N at any growth stage. All four vegetative indices were positively related with biomass at early squaring and early bloom. Stronger correlations were found between biomass and any of the four vegetative indices at early bloom (r = 0.69 to 0.73) than at early squaring (r = 0.41–0.45).
|
Leaf N at peak bloom at Lubbock 2000 was negatively related to reflectance from 450 to 650 nm (data not shown). Correlation was maximum (r = -0.64) at 550 and 570 nm. Biomass at peak bloom was not correlated with reflectance.
Strong positive correlations existed between chlorophyll meter readings and leaf N at all three growth stages (Table 5). Biomass, on the one hand, was weakly correlated with chlorophyll meter readings at early bloom and peak bloom only. All four vegetative indices were positively correlated with leaf N at all three growth stages. Stronger relationships with leaf N were observed for GVI or GNDVI at early bloom and peak bloom than at early squaring. The GVI and GNDVI had a positive relationship with biomass at early and peak bloom. Except at peak bloom, the RVI and RNDVI revealed strong positive correlations (r = 0.48 to 0.83) with biomass.
|
Chlorophyll meter readings were not related to biomass in any of the three growth stages (Table 6). Except at early squaring, leaf N was positively correlated with chlorophyll meter readings and all four vegetative indices. Lint yield did not correlate significantly with any vegetative index at early squaring. Stronger positive correlations of lint yield with the four vegetative indices were observed at peak bloom than early bloom.
|
|
Sufficiency indices calculated for the chlorophyll meter treatments are also presented in Table 7. In Ropesville 2000, the chlorophyll treatment in all three growth stages had sufficiency indices >0.95. In both 2000 and 2001 at Lubbock, the sufficiency index of the chlorophyll treatment was <0.95 in three out of six irrigation-growth stage combinations. The number of cases when the sufficiency index for the chlorophyll meter plots agreed with the sufficiency index in the reflectance plots (based on R820/R550) was 12 out of 18 cases (Table 7).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Leaf N concentration was correlated with GVI (and GNDVI) and RVI (and RNDVI) in seven and five out of nine cases, respectively. Leaf N accumulation was correlated with GVI and GNDVI in seven of nine cases and with RVI and RNDVI in six of nine cases.
The observation that the green-based indices resulted in better and more frequent correlation than the red-based indices for leaf N and leaf N accumulation is similar with recent reports. Gitelson et al. (1996) and Shanahan et al. (2001) have suggested that under moderate to high N status, green-based indices may be more sensitive to chlorophyll content and to final corn grain yield than red-based indices. In terms of sufficiency indices, however, the number of cases when sufficiency index <0.95 was similar between the red- and green-based indices was 14 out of 18 cases irrigation-growth stage site-year combinations.
Correlation with biomass was observed in six and five out of nine cases with red- and green-based indices, respectively. Lint yield was correlated with red- and green-based indices in six and four out of nine cases, respectively. The more frequent correlation between red-based indices and lint yield may have been related to the correlation with biomass. However, using proximal sensing to estimate lint yield in cotton has the limitation that biomass and lint yield may not correlate, especially if excessive vegetative growth and insect infestation of fruits occurs (Oosterhuis, 1990).
Simple correlations between the vegetative indices and leaf N and biomass assumes linear relationships. Ma et al. (2001) reported that NDVI was a non-linear function of soybean yield. In our study, only at early bloom in Lubbock 2000 did quadratic functions give a marginally better fit to linear functions between the vegetative indices and leaf N, leaf N accumulation and biomass (data not shown).
The frequent and high correlation we observed between cotton leaf N and chlorophyll meter readings are similar to published results in Alabama (Wood et al., 1992) and in China (Wu et al., 1998). However, calculating vegetative indices with spectral reflectance with a green or red waveband and with NIR reflectance resulted in more correlations with biomass (five of nine cases with GVI and six of nine cases with RVI) than with chlorophyll meter readings (two of nine cases). Biomass estimation, therefore is an important advantage that proximal sensing has over chlorophyll meter readings. Spectral radiometers can also be installed on ground-based applicators (Stone et al., 1996), while chlorophyll meters cannot.
The correlations presented in this study between vegetative indices or chlorophyll meter readings and cotton leaf N and biomass, demonstrate the potential of in-season canopy or leaf sensing of cotton. However, to use these tools most effectively in a wide range of situations, we strongly encourage the inclusion of well-fertilized reference plots or strips so that sufficiency indices can be calculated. The sufficiency indices from vegetative indices and chlorophyll meter readings successfully estimated little or no need of N in Ropesville 2000, that is, sufficiency index generally remained >0.95 between squaring and peak bloom. Indeed we observed no response to in-season N in the low-yielding Ropesville crop of 2000. The Lubbock 2001 crop was the highest yielding among the three site-years, and the sufficiency index <0.95 with both surface and subsurface drip irrigation at early bloom and peak bloom. An exception to this was the chlorophyll meter plots in surface drip irrigation at early bloom. The Lubbock 2000 crop was intermediate to the other crops both in terms of the number of times sufficiency indices <0.95 and in N response and lint yield levels.
It is encouraging that the vegetative index sufficiency indices and chlorophyll meter sufficiency indices were strongly correlated with both leaf N and biomass, and leaf N alone, respectively, and that the sufficiency indices successfully predicted need of in-season N fertilizer.
|
ACKNOWLEDGMENTS |
|---|
Received for publication September 5, 2002.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. T. Gomez-Casero, F. Lopez-Granados, J. M. Pena-Barragan, M. Jurado-Exposito, L. Garcia-Torres, and R. Fernandez-Escobar Assessing Nitrogen and Potassium Deficiencies in Olive Orchards through Discriminant Analysis of Hyperspectral Data J. Amer. Soc. Hort. Sci., September 1, 2007; 132(5): 611 - 618. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. F. Bronson, J. D. Booker, J. W. Keeling, R. K. Boman, T. A. Wheeler, R. J. Lascano, and R. L. Nichols Cotton Canopy Reflectance at Landscape Scale as Affected by Nitrogen Fertilization Agron. J., April 27, 2005; 97(3): 654 - 660. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. L. Ma, K. D. Subedi, and C. Costa Comparison of Crop-Based Indicators with Soil Nitrate Test for Corn Nitrogen Requirement Agron. J., March 1, 2005; 97(2): 462 - 471. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. T. Chua, K. F. Bronson, J. D. Booker, J. W. Keeling, A. R. Mosier, J. P. Bordovsky, R. J. Lascano, C. J. Green, and E. Segarra In-Season Nitrogen Status Sensing in Irrigated Cotton: I. Yields and Nitrogen-15 Recovery Soil Sci. Soc. Am. J., September 1, 2003; 67(5): 1428 - 1438. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Vadose Zone Journal | Journal of Plant Registrations | ||||
| Journal of Natural Resources and Life Sciences Education |
Journal of Environmental Quality |
||||
with reflectance at different green and near infrared wavebands of the reflectance and chlorophyll meter
treatments.