SSSAJ Grow Your Career with SSSA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 1 January 2007
Published in Soil Sci Soc Am J 71:171-180 (2007)
DOI: 10.2136/sssaj2006.0057
© 2007 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (5)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Williams, J. D.
Right arrow Articles by Crouse, D. A.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Williams, J. D.
Right arrow Articles by Crouse, D. A.
Agricola
Right arrow Articles by Williams, J. D.
Right arrow Articles by Crouse, D. A.
Related Collections
Right arrow Nitrogen

SOIL FERTILITY & PLANT NUTRITION

Comparison of Soil Nitrogen Tests for Corn Fertilizer Recommendations in the Humid Southeastern USA

Jared D. Williams*

Dep. of Agribusiness, Sci. and Tech., Brigham Young Univ.-Idaho, Rexburg, ID 83460-1110

Carl R. Crozier

Dep. of Soil Science, Vernon James Research and Extension Center, 207 Research Rd., Plymouth, NC 27692

Jeffrey G. White

Dep. of Soil Science, North Carolina State Univ., Raleigh, NC 27695-7619

Ravi P. Sripada

Dep. of Crop Science, North Carolina State Univ., Raleigh, NC 27695-7620

David A. Crouse

Dep. of Soil Science, North Carolina State Univ., Raleigh, NC 27695-7619

* Corresponding author (williams{at}byui.edu).


   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Environmental concerns about increasing NO3 levels in watersheds in North Carolina and elsewhere indicate the need for better N fertilizer management. Nitrate levels might be reduced if N rates could be adjusted based on field- or site-specific knowledge of corn (Zea mays L.) response to N fertilization. Currently, there is no effective soil N test for the humid southeastern USA. This study was conducted to compare three soil N tests for practicality, precision, and ability to correlate with economic optimum N rate (EONR) and fertilizer response on southeastern U.S. soils. The soil N tests were the Illinois soil N test (ISNT), the gas pressure test (GPT), and the incubation and residual N test (IRNT). Soil samples were collected from the sites of 16 N-response trials from 2001 to 2003 where different mineralizable and residual N levels were expected. The ISNT was determined to be the most practical test because it was the easiest to perform and could be completed in 1 d. The ISNT and GPT had better precision (lower CV) than the IRNT (9 and 13 vs. 61%, respectively). All three tests were related to EONR; ISNT had the strongest linear relationship (r2 = 0.90) when consideration was restricted to sites on mineral soils. The ISNT and GPT were related to delta yield (maximum yield minus check yield; r2 = 0.49 and 0.60, respectively) and fertilizer response (r2 = 0.31 and 0.51, respectively). These results indicate the potential of the ISNT and GPT to account for mineralizable and residual soil N levels and thus improve current corn N recommendations in the humid southeastern USA.


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Currently recommended N management in North Carolina and elsewhere in the humid southeastern USA is based on realistic yield expectations (RYE). In North Carolina, N recommendations based on RYE are made from a database of RYE by soil map unit or based on a grower's documented historic yields (the average of the best three yields during a 5-yr period; North Carolina Nutrient Management Workgroup, 2003). The recommended N rate is calculated using the RYE and a soil- and crop-dependent N application factor (18–22.5 kg N Mg–1 corn grain yield). These N fertilizer recommendations are based on decades of field response trials (Kamprath et al., 1973; Kamprath, 1986), but can result in under- or overapplication, in part because they do not consider soil N (Mulvaney et al., 2001). Soils under certain management practices (e.g., manure applications or legume cover crops) or soils with high organic matter content may have sufficient residual and potentially mineralizable N for optimum corn growth and need little or no N fertilizer.

Overapplication of N fertilizer has been linked to high levels of NO3 in shallow groundwater underneath agricultural fields in North Carolina and elsewhere (Gambrell et al., 1974; Jacobs and Gilliam, 1985). The Neuse River Basin in North Carolina has experienced fish kills linked to nonpoint sources of NO3 pollution (North Carolina Division of Water Quality, 1996, p. 44). Agricultural N pollution could be reduced by more effectively determining accurate fertilizer rates, so less N is susceptible to leaching and runoff (Khan et al., 2001). A soil N test to predict corn N need or corn responsiveness to N fertilizer could be used to more accurately predict fertilizer recommendations. The challenge is to develop a soil N test that is accurate, timely, and cost effective for predicting corn yield response to N fertilizer or corn fertilizer N requirement.

The climate in the southeastern USA complicates soil N test development because of warm temperatures and rainfall that exceeds evapotranspiration, resulting in rapid mineralization, immobilization, leaching, and denitrification. The preplant and presidedress NO3 tests have had limited success in estimating corn N need in the humid southern USA (Bundy et al., 1992; Grove, 1992). Soil N tests that measure mineralizable N under aerobic conditions are not practical for predicting yield response to fertilizer because they require a long incubation period (30–60 d; Bundy and Meisinger, 1994). Potential soil tests for predicting fertilizer response in corn would be an anaerobic IRNT (Bundy and Meisinger, 1994; Crozier et al., 2003), the ISNT (Khan et al., 2001), and the GPT (Picone et al., 2002).

The IRNT combines residual NO3, residual NH4, and NH4 mineralized from a short-term (7-d) waterlogged incubation (Waring and Bremner, 1964) to predict N that will be available for crop growth (Crozier et al., 2003). The waterlogged incubation measures KCl-extractable NH4 after a 7-d incubation from a soil sample taken before corn planting. Independently, the residual NO3 and waterlogged incubation tests have had limited success in predicting corn N need or fertilizer response (Bundy and Meisinger,1994) because N supplied to the crop by the soil comes from both the residual and mineralizable N pools (Bundy et al., 1992). Accounting for both inorganic and mineralizable N may better predict corn yield response to N fertilizer (Crozier et al., 2003).

The ISNT was developed by examining the different components of organic soil N liberated through acid hydrolysis followed by alkalization with NaOH (Mulvaney and Khan, 2001). The major components of the liberated organic N are total hydrolyzable N, hydrolyzable NH4, hydrolyzable amino acid N, and hydrolyzable amino sugar N (Mulvaney and Khan, 2001). A study of the components of hydrolyzable soil organic N showed that hydrolyzable amino sugar N was best related to corn yield response to N fertilization (Mulvaney et al., 2001). Determining hydrolyzable amino sugar N using a distillation method is tedious and not practical for routine laboratory analysis (Mulvaney et al., 2001; Khan et al., 2001). The ISNT method was developed as a quick and simple alternative for determining amino sugar N by performing the alkalization directly on the soil, rather than using the soil hydrolysate, and measuring alkali-hydrolyzable amino sugar N. Acid-hydrolyzable amino sugar N and alkali-hydrolyzable amino sugar N are well related (r2 = 0.82), and alkali-hydrolyzable amino sugar N, that is, the ISNT, was accurately able to identify responsive and nonresponsive sites in Illinois (Khan et al., 2001).

The GPT was developed for measuring NH4 content in manure slurries (Piccinini and Bortone, 1991). The GPT measures pressure developed in a closed vessel when the strong oxidizing reagent Ca(ClO)2 oxidizes NH4 to N2 gas and C to CO2. Nitrogen mineralization is estimated by the increase in pressure (Chescheir et al., 1985). The GPT has been used to determine the amount of available N in poultry litter and soils (Picone et al., 2002). The GPT used with soils in Georgia was well related with other plant-available N tests, for example, the 24-d aerobic N mineralization test (r2 = 0.77) and soil microbial biomass C test (r2 = 0.90; Picone et al., 2002). The relationship between the GPT and other plant-available N tests suggests that it may also be related to corn yield response to N fertilization. The greatest advantage of the GPT is its rapid analysis time (25 min), which would make this test simple to incorporate into a routine laboratory procedure (Picone et al., 2002).

The soil tests described above could be performed before corn planting to modify early season N fertilizer recommendations. The primary objectives of this study were to: (i) evaluate and compare the laboratory practicality and precision of the ISNT, GPT, and IRNT; and (ii) determine and compare the linear relationships of soil N test values with corn N response parameters on Piedmont and Coastal Plain soils. We also examined the relationships of the crop response parameters with soil humic matter (HM) (Mehlich, 1984) and NO3–N concentration ([NO3–N]) to compare with the soil N test relationships and determine if the soil N tests provided additional information.


   MATERIALS AND METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
This research was conducted in the major row crop areas of North Carolina, including the Lower and Middle Coastal Plain and Piedmont on soils that represent millions of hectares of land in agriculture production in the southeastern USA. Sites were located on research farms and as on-farm trials (Table 1). Although soil parent material and genesis differ among the three regions, the dominant soil order is Ultisol (Table 1). Soils in the Piedmont tend to have high clay content and are derived primarily from saprolite. Middle Coastal Plain soils tend to have a high sand content and are derived primarily from alluvial, marine, and marsh deposits. The Lower Coastal Plain, which is located near sea level, has a high water table and is predominantly poorly drained, resulting in development of mineral (Aquults and Aquepts) and organic soils (Histosols; Table 1) from parent materials similar to those in the Middle Coastal Plain.


View this table:
[in this window]
[in a new window]

  		Table 1. Site and soil information for research plots.

 
Crop management data including cultivar, seeding rate, row width, previous crop, irrigation, and tillage method for each site are presented in Table 2. Sites were irrigated when water was limited as determined by the research station manager (on-farm trials were not irrigated). Lime and fertilizer rates other than N were based on North Carolina Department of Agriculture and Consumer Services (NCDA & CS) Soil Testing Laboratory results and recommendations (Hardy et al., 2005). Soil samples were collected from N response trials conducted by several unrelated research projects for 16 site-years in 2001, 2002, and 2003 (Crozier et al., 2003; Sripada et al., 2005). Sites were categorized by soils being mineral or organic, with organic soils defined as either Histosols or mineral soils with a histic epipedon (Table 1). Soil samples for ISNT, IRNT, GPT, and HM were taken to a depth of 20 cm immediately before planting in April and May of each study year. Sampling depth was based on the NCDA & CS Soil Testing Laboratory protocol to conform to the standard sampling technique used in North Carolina. Soil samples were composites of six soil cores taken randomly from within each replication (approximately 9 by 18 m) of each trial. Data for each site-year were determined as the means of the replicate soil samples taken at each site-year. Samples were air dried and ground to pass a 2-mm sieve.


View this table:
[in this window]
[in a new window]

  		Table 2. Crop management information.

 
Nitrogen fertilizer was applied between the V4 and V10 growth stages (Ritchie et al., 1997) at four to seven rates in a randomized complete block design with three to six replications (Table 3). Nitrogen treatment plots were four rows wide (~3 m) and 10 m long. Nitrogen treatment rates at planting and at sidedress are shown in Table 3; total N applied was calculated by adding N applied at planting to N applied at sidedress. Grain yield data were collected by hand harvesting 3 m or machine harvesting 10 m, both from each of the middle two rows of the four-row plots (i.e., a total of 6 or 20 m of row, respectively). All yield data were adjusted to 155 g kg–1 moisture content.


View this table:
[in this window]
[in a new window]

  		Table 3. Sites, number of replications, N rates at planting, and N rates and growth stage at sidedress for N-response trials.

 
The method used to determine soil organic matter was the HM test, a routine method used in North Carolina by the NCDA & CS Soil Testing Laboratory. The soil HM test consists of an extraction with 0.2 M NaOH + 0.02 M DTPA (Mehlich, 1984) followed by a colorimetric determination of the alkali-extractable humic substance. Harrison et al. (1976) demonstrated that this HM determination has a strong linear relationship with total soil organic matter determined via the Walkley–Black method (Allison, 1965): soil organic matter = 1.22HM + 0.12; r2 = 0.95. Particle size distribution was determined using the hydrometer method to determine the percentages of sand, silt, and clay. Hydrometer readings were taken at 30 and 90 s to determine the sand fraction and at 6 and 16 h for the clay fraction (Gee and Bauder, 1979).

Illinois Soil Nitrogen Test
The ISNT was performed using a modification of the method of Khan et al. (2001). The modification was the use of an incubator instead of a hot plate or griddle, because of concerns of uneven heating with the latter when using these (University of Illinois, 2004). The ISNT was performed by placing 1.00 g of air-dried soil into a 0.47-L (1 pint) mason jar and adding 10 mL of 2 M NaOH. A 60-mm petri dish was filled with 5 mL of H3BO3 indicator solution (bromocresol green and methyl red) and attached to the jar lid so as to be suspended above the soil solution. The jar lid was immediately attached to the jar (air tight) and the whole assembly was heated to 48 to 50°C in an incubator for 5 h. Thermometers were positioned randomly inside the incubator to verify that the temperature was within ±1°C of 49°C. After the incubation, samples were allowed to cool to room temperature, petri dishes were removed from the jars, and the indicator solution was diluted with 5 mL of deionized H2O. The diluted indicator solution was titrated using a standardized H2SO4 solution (approximately 0.01 M) to an endpoint established on the basis of color. Soil test concentrations (mg kg–1) were calculated as S x T, where S is milliliters of H2SO4 used in titrating and T is the titer (µg N mL–1; T = 280 µg N mL–1 for 0.01 M H2SO4) of H2SO4 (Khan et al., 2001; University of Illinois, 2004).

Gas Pressure Test
The GPT was conducted according to the method of Picone et al. (2002) by placing 5.00 g of air-dried soil and 5 mL of distilled H2O into a 125-mL serum bottle. An apparatus was created to suspend Ca(ClO)2 above the soil slurry by placing a short piece of Tygon tubing in a septum, then filling the Tygon tubing with 0.3 g of Ca(ClO)2 and stopping it with a plastic cap. A hypodermic needle was used to push the plastic cap out of the Tygon tubing and allow the Ca(ClO)2 to mix with the soil slurry. The hypodermic needle was left in the septum for 3 s to allow pressure inside the bottle to equalize with the atmosphere. Since the needle sometimes became plugged with Ca(ClO)2 and did not allow pressure to equalize, we modified the method of Picone et al. (2002) by measuring the initial pressure with a tensimeter (Soil Measurement Systems, Tucson, AZ) immediately after removing the hypodermic needle. The bottles were then placed on an orbital shaker at 250 revolutions min–1 and agitated for 25 min. The pressure was measured immediately after shaking and pressure change was calculated by subtracting the initial pressure from the final pressure.

The GPT did not work on soils with a high HM content (i.e., ≥10%) because the soils absorbed the water and did not form a slurry and thus did not allow the reagent to mix with the soil. Adjusting the ratio by increasing the amount of water might be appropriate for organic soils, but would decrease the sensitivity of the test because some of the gas released by the reaction would be dissolved in the water (Picone et al., 2002).

Incubation and Residual Nitrogen Test
The IRNT is a combination of residual NO3–N, residual NH4–N, and NH4–N from a 7-d waterlogged incubation test. The residual NO3–N and NH4–N were determined by adding 5.00 g of soil, 20 mL of distilled H2O, and 20 mL of 2.0 M KCl to a plastic extraction cup. The samples were then agitated on an oscillating shaker for 1 h and filtered into plastic scintillation vials. The vials were frozen until analysis for NO3 and NH4 was conducted colorimetrically on an automated ion analyzer (QuikChem 8000, Lachat Instruments, Milwaukee, WI) which used a Cd reduction method for determining NO3 (Bundy and Meisinger, 1994). The waterlogged incubation was performed by adding 5.00 g of soil and 20 mL of distilled H2O to a plastic extraction cup. The cups were capped and placed in an oven to incubate for 7 d at 35 to 40°C. After the incubation period, the samples were extracted and NH4 concentration determined by the same procedure used for residual NH4 concentration. The IRNT values were calculated by adding the NH4–N from the incubated samples (mg kg–1) to the residual NO3–N and NH4 (mg kg–1).

Determining the Precision of the Tests Using a Modified Coefficient of Variation
Soil N test precision was determined by split replication of each individual soil sample. A modified CV was developed to compare the precision of soil N tests across different numbers of replications. The CV calculation was modified by replacing the observed mean (µxn, if n < 6) with the mean for the highest number of replications, which was six. The modified CV was computed as follows:

Formula 1[(1)]
where xi is the individual observation; n is the number of observations (i.e., 2, 3, 4, 5); and µx6 is the mean of six observations, used in place of the mean of two, three, four, or five observations. For example, a modified CV for three replications would be determined by selecting three observations from the total pool of six observations. The three-replication modified CV would be calculated where xi would be the three selected observations, µx6 would be the mean of all six observations, and n would be the number of observations (i.e., 3). The modified CV estimates the variation associated with each number of replications (two, three, four, or five) compared with the mean of the six replications, which was presumed to best represent the true population mean. A modified CV was calculated for every possible combination of two through five replications for each soil sample. A mean modified CV value was determined for each number of replications (two, three, four, and five) for each soil N test. For n = 6, the modified CV equals the actual CV.

Descriptive statistics were performed using PROC MEANS in SAS (SAS Institute, 2002). The modified CV (n = 2, 3, 4, or 5) and the CV (n = 6) were compared among the soil N tests and among the different number of replications within each test using Tukey's means separation test in SAS PROC GLM with {alpha} = 0.05. Linear and quadratic regression analysis was performed using PROC REG in SAS (SAS Institute, Cary, NC)

Crop Response Parameters
Corn grain yield response to N fertilizer was modeled as linear-plateau and quadratic-plateau functions in PROC NLIN in SAS (Cerrato and Blackmer, 1990). If the quadratic term in the quadratic-plateau model was significantly different from zero as determined by a t-test at the {alpha} = 0.05 level, then the quadratic-plateau model was considered to be a better fit than the linear-plateau model. When using a linear-plateau function, if the slope of the linear portion is such that the response to N is profitable, both the agronomic and economic optimum N rates are at the inflection point, the point beyond which there is no further increase in yield with increased applications of N fertilizer. When using a quadratic-plateau function, the economic optimum N rates (EONR) were calculated using the first derivative of the quadratic plateau model and a price ratio of US$0.66 kg–1 N to US$80 Mg–1 corn (US$0.30 lb–1 N to US$2.00 bushel–1 corn; Cerrato and Blackmer, 1990). In one case (PBI3), the first derivative test resulted in an EONR greater than the highest N rate applied, and the highest applied N rate was determined as the EONR. Check yield was determined as the corn yield from plots that did not receive N fertilizer; for sites that did not have a 0 kg N ha–1 treatment, check yield was extrapolated using the quadratic- or linear-plateau function (Lory and Scharf, 2003). Economic optimum yield was determined as the yield at the EONR.

Linear regression models of soil N test levels vs. EONR were calculated using PROC REG in SAS. Percentage corn yield response to N fertilizer (fertilizer response) was calculated as 100(optimum yield – check-plot yield)/check-plot yield (Khan et al., 2001). Delta yield (yield response) was calculated by subtracting check yield from maximum yield. The N sufficiency threshold was represented by the x intercept of EONR or delta yield (0 kg ha–1 EONR or 0 Mg ha–1 delta yield) plotted as a function of a soil N test.

Regression analysis of soil N tests vs. crop response parameters was performed using PROC REG in SAS. Because the GPT could not be used on organic soils, analyses of soil N tests vs. crop response parameters were done for all sites (n = 16) and for mineral soil sites alone (n = 13).


   RESULTS AND DISCUSSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Laboratory Practicality
Soil tests were compared for laboratory practicality based on the time required to analyze one sample, the special equipment needed, hazardous waste generated, and whether the tests could be used for both mineral and organic soils (Table 4). The quickest test was the GPT (1 h). The time to process one sample for the ISNT was 5.5 h which was longer than the GPT, but still less than 1 d. The IRNT had the longest time to process one sample (7.5 d) because of the 7-d incubation period.


View this table:
[in this window]
[in a new window]

  		Table 4. Comparison of the laboratory practicality of the Illinois soil N test (ISNT), the gas pressure test (GPT), and the incubation and residual N test (IRNT).

 
The ISNT and GPT required apparatus to be constructed and other special equipment. The special apparatus for the GPT [the Ca(ClO)2 holding assembly] was disassembled as part of the test procedure and had to be reassembled each time the test was performed. The special apparatus for the ISNT (boric acid trap) was assembled once as part of the laboratory setup. Nitrate and NH4, the analytes of the IRNT, were determined colorimetrically using an analyzer commonly found in soil testing laboratories. All the tests produced some waste (Table 4), but Ca(ClO)2 had to be handled under a fume hood because it is hazardous. The hypodermic needles used in the GPT were an additional hazard and had to be handled carefully and disposed of properly.

The most practical test was the ISNT, because its apparatus needed to be constructed only once, it could be completed in 1 d, and its waste was easy to manage (Table 4). The GPT could be performed in the shortest amount of time, but the special apparatus had to be reconstructed for each batch of samples and the test could not be used for organic soils (Table 4). We considered the IRNT the least practical because of the 7.5 d needed to complete it.

Precision of Soil Nitrogen Tests
The ranges of modified CV for different numbers of replications (two, three, four, five, and six) for each test were 9 to10% for ISNT, 13 to 15% for GPT, and 61 to 70% for IRNT (Fig. 1 ). The ISNT was the only test to show statistical difference in modified CV among the number of replications (Fig. 1); however, the difference between two and six replications for the ISNT was only one percentage point. We concluded that duplicate analyses of each sample seemed sufficient for all tests. Among the tests, the modified CV for two replications ranged between 10 and 70%, with the ISNT having the lowest modified CV (10%) of the tests studied (Fig. 1). The CV for the ISNT was within the range of CV values (0.5–11.6%) reported by Klapwyk and Ketterings (2005). The GPT had a CV of 15%, substantially lower than the CV of 44% reported by Picone et al. (2002). Our procedure of measuring the initial pressure may have helped reduce the variability, but we did not test this. Potential sources of variation for the GPT include leakage due to repeated piercing of the septa or failure of the oxidizing reagent [Ca(ClO)2] to mix properly with the soil slurry. If the Ca(ClO)2 did not completely mix with the soil slurry, less reagent would react with the soil, resulting in less pressure (Picone et al., 2002). The IRNT had the highest modified CV values because it included the associated errors from determining residual NO3–N, residual NH4–N, and waterlogged-incubation NH4–N. The waterlogged incubation itself could be variable because it is dependent on microbial activity within each sample.


Figure 1
View larger version (19K):
[in this window]
[in a new window]

  		Fig. 1. The influence of the number of laboratory analysis replications on the mean modified CV (n = 2–5) and the CV (n = 6) for the incubation and residual N test (IRNT), gas pressure test (GPT), and Illinois soil N test (ISNT). Lowercase letters denote statistical difference ({alpha} = 0.05) among the number of replications for each test and capital letters denote statistical difference among tests with all six replications.

 
Soil Nitrogen Tests to Predict Crop Response Parameters
Yield responses to N fertilizer are presented graphically in Fig. 2 . The mean delta yield for the study was 6.0 Mg ha–1, with a range of 2.2 to 9.8 Mg ha–1 (Table 5). The mean yield response to N fertilization for the study was 169% and ranged between 30 and 395% (Table 5). In contrast to results reported from an Illinois study (Khan et al., 2001), no direct observation of threshold level could be made due to the absence of unresponsive sites. When all sites were considered together, delta yield and fertilizer response were not related to the soil tests, but they were weakly related to HM (Table 6). For mineral sites only, delta yield was related to GPT, ISNT, and HM (Fig. 3 ), and fertilizer response was related to GPT and ISNT. The GPT and ISNT explained 49 to 60% (Table 5) of the variability in delta yield, which may be sufficient for developing a fertilizer recommendation by calculating an N rate based on delta yield, as did Lory and Scharf (2003), but additional studies are warranted to fill gaps in the data (e.g., no data for GPT between 2.6 and 6.2 kPa; Fig. 3). The advantage of a delta yield approach would be its focus on a crop response parameter that directly affects fertilizer need (Lory and Scharf, 2003). The delta yield and fertilizer response relationships with GPT and ISNT suggest that thresholds (i.e., the N test values at which the crop does not respond to fertilizer N) may exist for these tests, but additional studies including unresponsive sites would be needed to determine the thresholds. If thresholds could be determined, current N recommendations based on RYE could be modified such that sites with a GPT or ISNT at or above threshold would receive no N fertilizer, as suggested by Khan et al. (2001).


Figure 2
View larger version (36K):
[in this window]
[in a new window]

  		Fig. 2. Corn yield response to total N rate in 16 small-plot experiments.

 

View this table:
[in this window]
[in a new window]

  		Table 5. Illinois soil N test (ISNT), gas pressure test (GPT), incubation and residual N test (IRNT), NO3–N concentration, humic matter (HM), and crop response parameter results.

 

View this table:
[in this window]
[in a new window]

  		Table 6. Coefficients of determination for linear regression models of soil N tests (Illinois soil N test, ISNT; gas pressure test, GPT; NO3–N concentration; incubation and residual N test, IRNT; humic matter, HM) vs. crop response parameters.

 

Figure 3
View larger version (13K):
[in this window]
[in a new window]

  		Fig. 3. Delta yield as a function of (a) the Illinois soil N test (ISNT), (b) the gas pressure test (GPT), and (c) humic matter (HM) for mineral soil sites only.

 
The mean check yield for this study was 4.0 Mg ha–1 with a range of 0 to 7.4 Mg ha–1 (Table 5). Check yield was not related to soil tests, HM percentage, or soil [NO3–N]. The lack of linear relationships of check yield with the soil N tests might have been the result of the large diversity of soil productivity among sites and soil variability in retention and mineralization of N. For example, the ranges of corn RYE according to the North Carolina Nutrient Management Workgroup (2003) for soils used in this study overlapped for the Piedmont (7.5–8.0 Mg ha–1) and Middle Coastal Plain (7.2–8.2 Mg ha–1) soils, but the range for the Lower Coastal Plain soils was greater (8.5–10 Mg ha–1). The range of N fertilizer recommendations based on RYE overlapped for all regions (Piedmont, 148 to160 kg ha–1; Middle Coastal Plain, 148 to 166 g ha–1; Lower Coastal Plain, 155 to 190 kg ha–1). These inherent differences in soil productivity among regions probably resulted in different check yields that were not reflective of N availability as measured by the soil N tests studied.

The mean EONR for this study was 193 kg ha–1, with a range of 68 to 336 kg ha–1 (Table 5). Linear and quadratic regression analyses were used to assess if EONR was related to soil N test (ISNT, GPT, and IRNT), [NO3–N], or HM. When all sites were considered together, HM and soil [NO3–N] were determined to be linearly related with EONR because quadratic terms from regression analysis of HM and [NO3–N] with EONR were not significant. Additionally, the HM and [NO3–N] relationships with EONR for all sites were greatly influenced by the TOI2 site, which had twice as much HM and NO3–N as the other sites (Tables 5 and 6, Fig. 4 ). The three soil N tests of primary concern were not related to EONR when all sites were analyzed. When the regression analysis was restricted to mineral soil sites, EONR was significantly related to all the soil tests studied and to [NO3–N] (Table 6, Fig. 5 ). Although soil [NO3–N] and HM were the only measurements able to predict EONR regardless of soil type (Table 6, Fig. 4), the relative weakness of the linear relationships and the influence of TOI2 on the relationships would limit their usefulness for developing fertilizer recommendations based on a 20-cm-depth soil sample. The IRNT and [NO3–N] also showed a relationship with EONR for the mineral sites, but had relatively low coefficients of determination (r2 = 0.33 and 0.41, respectively).


Figure 4
View larger version (16K):
[in this window]
[in a new window]

  		Fig. 4. Economic optimum N rate (EONR) as a function of (a) NO3–N concentration and (b) humic matter (HM) for all sites.

 

Figure 5
View larger version (25K):
[in this window]
[in a new window]

  		Fig. 5. Economic optimum N rate (EONR) as a function of (a) the Illinois soil N test (ISNT), (b) the gas pressure test (GPT), (c) the incubation and residual N test (IRNT), (d) NO3–N concentration, and (e) humic matter (HM) for mineral soil sites. Regression line fitted when model was significant ({alpha} = 0.05).

 
The best tests for predicting EONR for mineral soil sites were ISNT and GPT (Table 6, Fig. 5). The strong relationship between EONR and ISNT (r2 = 0.90) for mineral soils shows the potential for ISNT to be used to make N recommendations, but additional data is needed to confirm and calibrate these findings. The relationship between EONR and GPT (r2 = 0.62) shows some promise for its use in making N recommendations, but additional research is needed to determine if and how the relationship might be strengthened. The ISNT and GPT showed stronger linear relationships with crop response parameters for mineral sites than did [NO3–N] and HM (Table 6). The ISNT was weakly related with [NO3–N] and HM (r2 = 0.43 and 0.38, respectively; not shown). The GPT was strongly related to HM (r2 = 0.78, not shown), but was not significantly related to [NO3–N]. These results suggest that the ISNT and GPT provided additional or unique information than was acquired from HM or [NO3–N].

The relatively strong relationships of ISNT and GPT with EONR appeared to have two separate clouds of data. The plots of ISNT and GPT vs. EONR showed that these data groupings were associated with regions (Fig. 5a and 5b). In the ISNT vs. EONR relationship, a gap existed between the lower ISNT values of the Middle Coastal Plain sites and the higher ISNT values of the Piedmont and Lower Coastal Plain sites. The high coefficient of determination (r2 = 0.90) of the regression of ISNT versus EONR for all mineral sites, however, indicated that the model fitted the data well, suggesting similar trends for the lower ISNT group (Middle Coastal Plain) and the higher ISNT group (Lower Coastal Plain and Piedmont). The graph of GPT vs. EONR showed potential differences in the relationship between GPT and EONR among the regions. Means separation tests showed that mean GPT values from the Piedmont, Middle Coastal Plain, and Lower Coastal Plain sites were all different. Comparison of EONR data from each site showed that the Piedmont and Lower Coastal Plain sites did not have different EONRs, but that the Middle Coastal Plain sites had a significantly higher EONR. These findings suggest that regional differences (e.g., soil types) may require region-specific models, but additional research is needed to investigate the potential existence of unique relationships of ISNT and GPT with EONR within each region.

The relatively strong relationships of ISNT and GPT with EONR were limited to mineral soils, which is a concern in North Carolina where some of the most productive soils are high in organic matter (Histosols or minerals soils with histic epipedons). The current GPT method cannot be used with organic soils, but the test might be modified by altering the soil/water ratio. A different soil/water ratio specifically for use on organic soils would probably require that a separate calibration be established between GPT and EONR (Picone et al., 2002). For the ISNT, a relationship may exist between organic soil ISNT values and EONR, but the paucity of organic sites in this study (three) did not allow this to be investigated. Additional study of the organic soils of the southeastern U.S. Coastal Plain is needed to determine if relationships of EONR with ISNT or GPT can be used to develop a soil-test-based N recommendation.

The lack of a significant relationship between ISNT and EONR when organic soils were included may be due to differences in immobilization, denitrification, and mineralization rates and organic matter type among the mineral and organic soils. The organic soils had higher amounts of HM (>10%) and greater ISNT values (220–248 mg kg–1) than the mineral sites (HM = 0.3–4.3%, ISNT = 21–73 mg kg–1), but similar ranges of EONR (Table 5). The organic soil sites (TOI2, TCF, and GDA) were poorly drained Histosols or very poorly drained Inceptisols with histic epipedons (Table 1). Although these soils have some artificial drainage, they are susceptible to high denitrification and highly variable mineralization rates. Additionally, the organic soils consist of older organic matter that would be more resistant to mineralization than the younger organic matter in the mineral soils (Fox and Piekielek, 1984).

The mean ISNT in this study was 105 mg kg–1, with a range of 14 to 488 mg kg–1 (Table 5). These values were lower than those reported by Khan et al. (2001), which for responsive sites only, had a mean of 171 mg kg–1 using the open-bench method, and by Klapwyk and Ketterings (2005), which had a mean of 223 mg kg–1 using the enclosed-griddle method. The lower ISNT values in our study may be a result of the use of an incubator for heating samples and different soil organic matter types and levels. Klapwyk and Ketterings (2005) reported 22 mg kg–1 lower ISNT values when using an enclosed-griddle vs. the open-bench method.


   CONCLUSIONS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
The most promising soil N tests among those studied were the ISNT and GPT. The ISNT may be a better soil N test than GPT because the relationship of ISNT with EONR for mineral soils had a greater coefficient of determination than did that of GPR, and ISNT was more precise. In addition, the ISNT seemed to be the most practical to perform in the laboratory relative to the other tests evaluated. The GPT was related with EONR for mineral soils, but the coefficients of determination may be too low for GPT to be used to make N recommendations. The ISNT and GPT were related with delta yield, and an N fertilizer recommendation might be developed by calculating an N rate from delta yield using the formula developed by Lory and Scharf (2003). None of the four soil N tests were related with check yield, neither for all sites nor for mineral sites alone. The relationships of the soil N tests with the crop response parameters compared with the relationships of HM and [NO3–N] with the crop response parameters indicated the GPT and ISNT were better predictors of corn N need and EONR. The lower ISNT values that we found vs. those reported elsewhere (Khan et al., 2001) might have resulted from the method that we used (incubator vs. enclosed or open griddle heating; Klapwyk and Ketterings, 2005) and from different soil organic matter types and amounts.


   ACKNOWLEDGMENTS
 
We thank Alan Meijer, John Burleson, Don Davenport, and Brian Roberts for their field support. We thank Dan Israel, Peggy Longmire, and their lab staff for assistance in the laboratory. We also acknowledge the support staff at the Peanut Belt, Tidewater, Piedmont, and Cleveland County Research Stations, and the entire on-farm support at Guy Washington and Tankard Farms.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Abbreviations: EONR, economic optimum nitrogen rate; GPT, gas pressure test; HM, humic matter; IRNT, incubation and residual nitrogen test; ISNT, Illinois soil nitrogen test; [NO3–N], nitrate-nitrogen concentration; RYE, realistic yield expectations.

This research was supported in part by USDA Initiative for Future Agricultural and Food Systems (IFAFS) grant no. 00-52103-9644.

Received for publication February 10, 2006.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (5)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Williams, J. D.
Right arrow Articles by Crouse, D. A.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Williams, J. D.
Right arrow Articles by Crouse, D. A.
Agricola
Right arrow Articles by Williams, J. D.
Right arrow Articles by Crouse, D. A.
Related Collections
Right arrow Nitrogen

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome