SSSAJ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 5 April 2007
Published in Soil Sci Soc Am J 71:735-744 (2007)
DOI: 10.2136/sssaj2006.0135
© 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 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 HighWire
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
Right arrow Plant and Soil Interactions
Right arrow Maize

SOIL FERTILITY & PLANT NUTRITION

Illinois Soil Nitrogen Test Predicts Southeastern U.S. Corn Economic Optimum Nitrogen Rates

Jared D. Williamsa,*, Carl R. Crozierb, Jeffrey G. Whitec, Ronnie W. Heinigerd, Ravi P. Sripadae and David A. Crousef

a Dep. of Agribusiness, Science, and Technology Brigham Young Univ.-Idaho Rexburg, ID 83460-1100
b Dep. of Soil Science Vernon James Research and Extension Center 207 Research Station Rd. Plymouth, NC 27692
c Dep. of Soil Science North Carolina State Univ. Raleigh, NC 27695-7619
d Dep. of Crop Science Vernon James Research and Extension Center 207 Research Station Rd. Plymouth, NC 27692
e Dep. of Crop Science North Carolina State Univ. Raleigh, NC 27695-7620
f Dep. of Soil Science North Carolina State Univ. Raleigh, NC 27695-7619

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


   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
An accurate and quick soil N test is needed for N fertilizer recommendations for corn (Zea mays L.) for the humid southeastern USA. The Illinois soil N test (ISNT) has been used to distinguish fertilizer-responsive from unresponsive sites in Illinois. We determined relationships between economic optimum N rates (EONR) and ISNT levels in representative southeastern soils in 35 N-response trials in the Piedmont (n = 4) and Middle (n = 8) and Lower (n = 23) Coastal Plains of North Carolina from 2001 to 2004. The ISNT was strongly correlated with EONR for well or poorly drained sites (r2 = 0.87 [n = 20] and 0.78 [n = 10], respectively); data were insufficient for establishing correlations for very poorly drained or severely drought-stressed sites. Expressing ISNT on a mass per unit volume basis vs. EONR improved the correlations slightly (r2 = 0.88 and 0.79 for well and poorly drained sites, respectively), but these improvements would not justify the necessary soil bulk density determinations. Regressions of ISNT vs. minimum, average, and maximum EONR based on different N-fertilizer cost /corn price ratios (11.4:1, 7.6:1, and 5:1, respectively) showed strong correlations with EONR for well-drained sites (r2 = 0.77, 0.87, and 0.87, respectively) and poorly drained sites (r2 = 0.84, 0.78, 0.70, respectively). The ISNT–EONR correlations were different among the cost/price ratios for well-drained sites, but not different for poorly drained sites. Because ISNT predicted EONR robustly to different cost/price ratios, ISNT has the potential to modify or replace current N recommendation methods for corn.

Abbreviations: EONR, economic optimum nitrogen rate • HM, humic matter • ISNT, Illinois soil nitrogen test • RYE, realistic yield expectation


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Nitrogen fertilizer management for corn is coming under increasing scrutiny because of environmental and economic concerns. Increasing levels of NO3 and fish kills in the hypoxia zone of the Gulf of Mexico and in North Carolina's Neuse River have been attributed to agricultural application of N (Gambrell et al., 1974; Jacobs and Gilliam, 1985; North Carolina Division of Water Quality, 1996, p. 44; Council for Agricultural Science and Technology, 1999). Nitrogen fertilizer management based on yield goals or realistic yield expectations (RYE) has reduced fertilizer inputs in North Carolina (Heiniger et al., 2002), but N inputs required for profitable corn production are still a concern (Mulvaney et al., 2006). In North Carolina, N recommendations based on RYE are made from a database of corn 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., 2006). Under- and overapplying N fertilizer can cause serious environmental concerns and have adverse economic consequences for corn producers. A diagnostic tool is needed to refine N recommendations based on an estimation of crop-available soil N.

Nitrogen diagnostic tools can be categorized as plant or soil based. Plant-based tools include plant tissue analyses and crop color tests. Early season plant tissue tests that measure NO3 or total N concentration have not been consistent enough to be used to develop fertilizer recommendations (Fox et al., 1989; Bundy and Andraski, 1995). Crop color tests include chlorophyll meters or aerial photographs, but neither tool can be used to detect luxury consumption of N in corn or accurately predict corn N need early in the season (Blackmer and Schepers, 1995; Sripada et al., 2005). Soil-based diagnostic tools include soil NO3 tests such as the preplant NO3 test and the presidedress NO3 test (Bundy et al., 1992; Magdoff et al., 1984), and potentially mineralizable N tests such as aerobic and anaerobic incubations that seek to quantify N that will become available to the crop from soil organic matter (Bundy and Meisinger, 1994).

Soil NO3 tests have not been widely used in the humid southeastern USA because warm temperatures and high rainfall result in soil NO3 losses through leaching and denitrification. Soil NO3 tests also do not measure N that may be mineralized after sampling, resulting in underestimation of the soil N supply. Excluding fertilizer, the largest source of N taken up by a corn crop in high-rainfall areas like the humid Southeast tends to be N mineralized from organic matter (Crozier et al., 1994, 1998). Soil N tests that measure potentially mineralizable N may be better for predicting corn N requirements than residual NO3 tests.

Estimating the fraction of potentially mineralizable N that will become available for crop uptake during the growing season is difficult because of the effects of moisture and temperature on the N cycle (El-Sadek et al., 2002; Delgado et al., 2001). Soil N mineralization tests have been developed based on how they determine potentially mineralizable N and can be divided into biological and chemical tests. Biological N mineralization tests attempt to imitate in the laboratory the biological processes that make soil organic N available under field conditions (Bundy and Meisinger, 1994), but because these tests require an incubation period, they are not very practical for N management in corn (Keeney, 1982). Chemical N mineralization tests are designed to quantify specific chemical fractions present in soils, but these tests have generally shown low correlation with N mineralization and crop N need (Khan et al., 2001).

Recently, a chemical N mineralization test was developed by separating the different components of organic N liberated through acid hydrolysis: hydrolyzable NH4, amino acid N, and amino sugar N (ASN; Mulvaney and Khan, 2001). A study of the hydrolyzable soil organic N components showed that ASN was related to corn yield response to N fertilization in Illinois (Mulvaney et al., 2001; Khan et al., 2001). The acid hydrolysis method for liberating ASN is laborious, and the ISNT was developed as a quick method for estimating ASN (Khan et al., 2001). Williams et al. (2007) developed response functions for corn yield response factors (e.g., delta yield, fertilizer response, and economic optimal N rate) vs. ISNT that could be used to refine yield-based N recommendations. The results of the Illinois (Khan et al., 2001) and North Carolina (Williams et al., 2007) studies showed potential for developing relationships to determine EONR recommendations for corn using the ISNT.

Economic optimum N rates can be used to adjust fertilizer N recommendations based on a cost/price ratio. The use of monetary values allows producers to calculate optimum N rates for any specific combination of fertilizer costs and corn grain prices to maximize profits (Kelling and Bundy, 2001). Several different response models have been used to create optimum N rates, but a quadratic plateau model has been shown to best describe yield response (Cerrato and Blackmer, 1990).

The objectives of this study were to: (i) examine the correlations between ISNT and EONR for developing a soil test-based N recommendation under conditions absent severe water stress; (ii) evaluate the impact of soil, year, and seasonal temperature on the prediction of EONR from ISNT; and (iii) compare the effects of different fertilizer cost/grain price ratios on EONR predicted by ISNT.


   MATERIALS AND METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
This research was conducted on 35 sites in the major row crop areas of North Carolina including the Lower and Middle Coastal Plain and Piedmont. These sites were representative of millions of hectares in agricultural production in the southeastern USA. The predominant soil order in these areas is Ultisols (Table 1), but the Lower Coastal Plain sites, which are located near sea level, have poorly drained sandy soils with high organic matter and include some Histosols, Alfisols, and Inceptisols (Table 1). Sites were categorized as well (n = 23), poorly (n = 10), and very poorly (n = 2) drained based on soil survey drainage classification, the presence or absence of artificial subsurface or surficial drainage structure (tiles, ditches, and crowns), and field observations during the growing season. Well-drained sites were defined as having soils that were naturally well drained or tile drained. Poorly drained sites were defined as having either poorly drained soils without adequate artificial drainage (no tile, ditch, or crown drainage, or observed poor drainage) or organic soils. Very poorly drained sites were defined as having a very poorly drained soil survey drainage classification and no artificial drainage.


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

  		Table 1. Detailed study site and soil information for corn N-response trials.

 
Soil samples were collected from corn N-response trials conducted during several unrelated research projects for 35 site-yr from 2001 to 2004. Crop management data including cultivar, seeding rate, row width, previous crop, irrigation, and tillage method for each site are presented in Table 2. Some research station sites were irrigated when water was limited as determined by the research station manager. Lime and fertilizer rates other than N were based on North Carolina Department of Agriculture and Consumer Services (NCDA & CS) Soil Testing Laboratory soil test results and recommendations (Hardy et al., 2005).


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

  		Table 2. Crop management information.

 
Soil samples were taken immediately before planting in April and May to a depth of 15 cm for no-till sites and 20 cm for conventionally tilled sites in accordance with the North Carolina Department of Agriculture recommendations for sampling depth (Hardy et al., 2005). Soil samples were collected for each replication of each N-response trial by compositing six to eight soil cores taken randomly from within each N treatment. The soil test value for each site-year was determined as the mean of the soil test values of the replicate soil samples taken at each site-year. Samples were air dried and ground to pass a 2-mm sieve. Nitrogen fertilizer was applied as either NH4NO3 or aqueous urea [CO(NH2)2]–NH4NO3 (30% N) between the V4 and VT growth stages (Ritchie et al., 1997), but typically at V7, at four to seven rates in a randomized complete block design with four to six replications (Table 3). Nitrogen treatment plots were four rows wide (~3 m) and 10 m long. Some sites received 12 to 224 kg N ha–1 at planting, which was added to the side-dress N to calculate total N rates (Table 3). Grain yield data were collected by hand (3 m) or machine harvesting (10 m) the middle two rows of each four-row plot. 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, replications, N rates at planting, sidedress N rates, and growth stage at sidedress for N-response trials.

 
Soil bulk density samples were taken from 2- to 12-cm depth using a 3-cm-diameter soil core sampler from each replication of each study before corn planting. At conventionally tilled sites, samples were taken in the shoulder of the row to avoid traffic areas, and the samples were assumed to be representative of the plow layer (0–20-cm depth). Samples were oven dried and weighed to determine dry bulk density (Blake and Hartge, 1986). Soil particle size distribution was determined using the hydrometer method to determine the sand, silt, and clay percentages. 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).

Soil humic matter (HM) was determined by the NCDA & CS Soil Testing Laboratory by extraction with 0.2 M NaOH + 0.02 M DTPA and determining the HM percentage colorimetrically (Mehlich, 1984). Soil HM was determined instead of soil organic matter (SOM) because it is part of routine soil analyses performed by the NCDA & CS Laboratory. Harrison et al. (1976) showed that soil HM and SOM are highly correlated (SOM = 1.22HM + 0.12, r2 = 0.95) in North Carolina soils.

Temperature and precipitation data were collected by the State Climate Office of North Carolina (Raleigh), which has weather stations located on research stations and in most North Carolina counties. Data from the nearest weather station were used for sites not located on research stations. Average maximum, minimum, and mean temperatures and precipitation for the growing season (VE–R5) were calculated from 15 April to 15 August. Severe drought stress sites were determined as unirrigated sites that received <50% of the normal (30-yr average) rainfall. Precipitation data were not considered in the statistical analysis because some sites were irrigated and amounts of irrigation water were not measured (Table 2).

Illinois Soil Nitrogen Test
Each composite soil sample was analyzed for ISNT-N in triplicate using a modification of the method reported in Khan et al. (2001). The modification consisted of using an incubator to provide more even heating of the samples (Williams et al., 2007; Klapwyk and Ketterings, 2005). Soil samples were thoroughly mixed and 1.00 g of air-dried soil was placed in a 0.47-L (1 pint) mason jar. The soil was treated with 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 entire assembly heated to 49°C (±1°) in an incubator for 5 h. After the incubation, 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 (approximately 0.01 M) H2SO4 solution to an endpoint established on the basis of color. Soil test (ISNT) concentrations (in milligrams per kilogram) were calculated as ST, where S is milliliters of H2SO4 used in titrating and T is the titer (micrograms N per milliliter) of H2SO4 (Khan et al., 2001; Univ. of Illinois at Urbana-Champaign, 2004). Test levels were converted to kilograms ISNT per hectare furrow slice with a depth of 0.2 m using bulk density data at each site.

Statistical Analysis
Regression, multiple regression, and stepwise (best R2) regression analysis were performed using PROC REG in SAS Version 8.3 (SAS Institute, Cary, NC). Pairwise correlations were calculated using JMP 5.1 (SAS Institute, Cary, NC). Linear regression models for EONR vs. ISNT by drainage class, year, and price ratio were compared using a multiple regression model with an indicator variable for testing differences among intercepts and an interaction variable for testing differences among slopes.

Economic optimum N rates were calculated using linear-plateau and quadratic-plateau functions in PROC NLIN in SAS Version 8.3 (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 the better fit compared with the linear-plateau model (Williams et al., 2007). 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 EONR were calculated using the first derivative of the quadratic-plateau model and a cost/price ratio (Cerrato and Blackmer, 1990). In two cases (Sites 27 and 31), 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. In any case where a response did not fit a quadratic- or a linear-plateau function ({alpha} = 0.05), treatment means were compared using Tukey's means test to determine the optimum N level. Since future cost/price fluctuations cannot be predicted, correlations between ISNT and EONR were determined across a reasonable range of fertilizer cost/corn price ratios based on prices during the study period (Table 4). Three different EONR were calculated to represent inexpensive fertilizer and high-price grain (maximum EONR), expensive fertilizer and low-price grain (minimum EONR), and an average-cost fertilizer and average-price grain (average EONR).


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

  		Table 4. Economic optimum N rate (EONR) categories, cost/price ratios, and N fertilizer costs and corn prices in metric and English units.

 
Economic optimum yield (EOY) for sites where the linear plateau was the best fit model was determined as the plateau yield. The EOY for sites where the quadratic plateau was the best model was determined as the yield at average EONR. For sites that did not fit a quadratic- or a linear-plateau function, EOY was determined as the average yield of all N treatments receiving N rates equal to and greater than the average EONR. The N application factor was calculated as average EONR (N rate) divided by EOY (yield). Realistic yield expectation data for each site (EOY, N application factor, and average EONR) were compared with RYE data from the North Carolina Nutrient Management Workgroup (2003) database.


   RESULTS AND DISCUSSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Illinois Soil Nitrogen Test to Predict Economic Optimum Nitrogen Rates
A comparison of yield data for individual sites that received preplant and sidedress N applications showed no significant effect of N application timing (data not shown). The mean ISNT concentration for the site-years in this study was 93 mg kg–1 with a range of 14 to 488 mg kg–1 (Table 5). Site-year mean average EONR ranged from 27 to 336 kg ha–1, with a mean of 180 kg ha–1. Regression analysis of average EONR vs. ISNT including all site-years showed no significant correlation (all points in Fig. 1 ). A study by Williams et al. (2007) indicated that the models to predict EONR from ISNT concentration would be different for non-muck (lower ISNT concentration) and muck (higher ISNT concentration) soils. Muck and non-muck soils drain differently, which influences N mineralization and denitrification and probably resulted in the need to create separate models for these data. Fox and Piekielek (1984) showed that correlations of N availability to corn with anaerobically mineralized N and with chemical indexes improved when sites were divided into drainage classes, so we reanalyzed the data by the three site drainage types.


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

  		Table 5. Illinois soil N test results on a concentration basis (ISNTc) and on a hectare furrow slice basis (ISNThfs), soil bulk density, humic matter (HM), economic optimum yield (EOY), minimum, average, and maximum economic optimum N rate (EONR) for the three cost/price ratio categories (Table 4), and realistic yield expectation (RYE) parameters. Data were classed by drainage type and ranked within classes by ISNTc.

 

Figure 1
View larger version (27K):
[in this window]
[in a new window]

  		Fig. 1. Economic optimum N rate for the average fertilizer cost/corn price ratio (EONRavg) vs. the Illinois soil N test concentration (ISNTc) for all sites. Regression models were fit to the well and poorly drained sites.

 
Earlier analysis of HM vs. ISNT in North Carolina by Williams et al. (2007) found only a weak correlation (r2 = 0.38). Regression analysis of all our data except Sites 12 and 13 (HM content for these sites were above the assay's saturation level of 1 Mg m–3) showed a relationship between HM and ISNT (HM = 0.036ISNT– 0.19, r2 = 0.69), but these results appeared to be highly influenced by drainage class (Fig. 2 ). When separating sites by drainage class, the well-drained sites showed no relationship existed between HM and ISNT, but HM and ISNT for the poorly drained sites were related (HM = 0.035ISNT + 0.29, r2 = 0.66).


Figure 2
View larger version (24K):
[in this window]
[in a new window]

  		Fig. 2. Humic matter (HM) vs. the Illinois soil N test concentration (ISNTc) for all sites. Regression models were fit to the well and poorly drained sites.

 
The only very poorly drained sites were 12 and 13, which had the highest ISNT concentration (Table 5). Based on ISNT concentration, Sites 12 and 13 were expected to have low average EONR, but both sites had average EONR slightly above the average for all sites (Table 5, Fig. 1) and had high HM values (>1 Mg m–3). Thus, these results may reflect a phenomenon similar to that described by Klapwyk and Ketterings (2006) wherein the ISNT threshold value for expected nonresponse to N increased as soil organic matter content increased. The high HM and very poorly drained soils at Sites 12 and 13 suggest frequent saturation and a tendency for slow mineralization and rapid denitrification. Sites 12 and 13 may have had relatively high average EONR despite high ISNT concentration because N mineralized during the growing season may have been denitrified before plant uptake. The elimination of these two very poorly drained sites from further analyses improved correlations.

Severe drought conditions in 2002 (Fig. 3 ) resulted in water stress at three well-drained, unirrigated sites (Table 5). These three drought sites had much lower average EONR per ISNT concentration when compared with similar non-water-stressed sites (Fig. 1). The lower average EONR per ISNT concentration would be expected because water stress limits corn yield potential and N uptake. Additionally, the dry soil conditions may have decreased microbial activity and mineralization rates, potentially resulting in less soil N becoming available for plant uptake. The drought sites were removed from further analysis because the goal of the research was to establish models between average EONR and ISNT concentration under relatively water-stress-free conditions that allowed yield potential to be realized.


Figure 3
View larger version (54K):
[in this window]
[in a new window]

  		Fig. 3. Growing season (a) total precipitation and (b) mean temperature for 2001, 2002, 2003, 2004, and 30-yr average for the Lower Coastal Plain, Middle Coastal Plain, and Piedmont regions.

 
Comparison of Well and Poorly Drained Models
Regression models for both well and poorly drained sites showed average EONR was negatively related with ISNT concentration (Fig. 1). Regression parameters for the well and poorly drained soils were compared, and the analysis showed different slopes and intercepts between the drainage classes (data not shown). The average EONR vs. ISNT concentration model had an intercept SE of 22 kg N ha–1 for the well-drained and 27 kg N ha–1 for the poorly drained site models. The well-drained sites were more sensitive to changes in ISNT concentration than the poorly drained sites (Fig. 1). For example, an increase of 1 mg ISNT kg–1 resulted in a decrease in EONR of 2.9 kg ha–1 for the well-drained sites, but only 1.2 kg ha–1 for the poorly drained sites. The steep response of the EONR to ISNT at well-drained sites raises concerns regarding the precision of the test, as a small difference in the ISNT equates to a large difference in EONR. Williams et al. (2007) showed, however, that the ISNT is relatively precise (CV = 10%). For the well-drained sites this corresponds to a standard deviation of ~5 mg kg–1, which in terms of the EONR translates to approximately 16 kg N ha–1. The strong correlations and low standard error for the ISNT concentration vs. average EONR models showed potential for developing N recommendations.

The differences in the regression parameters for the well and poorly drained sites may be a reflection of the different amounts and types of organic matter between the two site drainage classes. The poorly drained sites had almost three times more HM (550 kg m–3) than the well-drained sites (200 kg m–3). Based on prior studies of Kamprath et al. (1958) and Dolman and Buol (1968), the organic matter in the soils at the well-drained sites probably had a lower C/N ratio and thus greater net N mineralization (Kamprath et al., 1958) than the soils at the poorly drained sites, which probably had a C/N ratio resulting in net N immobilization (Dolman and Buol, 1968). The combination of high HM content with a high C/N ratio could result in a higher N fertilizer demand at the poorly drained sites to meet and surpass the microbial N demand. Additionally, the poorly drained sites have a potential for higher denitrification caused by the soils being wetter and more frequently saturated than the well-drained sites (Fox and Piekielek, 1984; Delgado et al., 2001).

Effect of Year and Temperature on Economic Optimum Nitrogen Rate Prediction by the Illinois Soil Nitrogen Test
Limitations of soil N tests include changes in corn N demand among years and within growing seasons (Magdoff, 1991; Mulvaney et al., 2001). Corn N demand is influenced by the effects of weather and temperature on crop growth. Ideally, fertilizer recommendations based on ISNT concentration would accurately predict average EONR independent of environment and temperature conditions. Drought conditions would be an exception because water-stressed corn takes up less N and has a suppressed response to N fertilizer. As one might expect, weather conditions varied from year to year in this study (Fig. 3). The year effect for poorly drained sites could not be tested because of insufficient sites to fit models to individual years. Linear regression models of ISNT concentration vs. average EONR for the well-drained sites were compared, however, for a year effect between 2003 and 2004 (Fig. 4 ). The 2001 and 2002 growing seasons were not included in this analysis because of insufficient data for fitting regression models. The slopes were not statistically different between years, but the intercepts were. The difference in intercepts between the 2 yr might have been a result of different rainfall amounts and patterns. In 2003, rainfall was higher than the 30-yr average and evenly dispersed, resulting in adequate moisture throughout the growing season (Fig. 3). In 2004, rainfall was near the 30-yr average (Fig. 3), but lower than average during the early part of the growing season (data not shown). The influence of growing season temperature on ISNT concentration prediction of the average EONR was tested using a stepwise model that included site-year average growing-season maximum, minimum, and mean temperatures. Temperature was little different among years (Fig. 3b) and was not significant in the stepwise model (data not shown), thus prediction of average EONR using ISNT appeared not to be influenced by growing season temperature. While these results show some year-to-year consistency, additional years of data are needed to better elucidate the potential effects of seasonal rainfall and temperature on the ISNT's ability to predict EONR.


Figure 4
View larger version (28K):
[in this window]
[in a new window]

  		Fig. 4. Economic optimum N rate for the average fertilizer cost/corn price ratio (EONRavg) vs. the Illinois soil N test concentration (ISNTc) for 2003 and 2004 well-drained sites. Regression models are shown for 2003, 2004, and all well-drained sites in all years.

 
Comparison of Illinois Soil Nitrogen Test Concentration and Mass Per Unit Volume to Predict Economic Optimum Nitrogen Rate
Relationships between ISNT and average EONR across drainage classes might be improved by calculating ISNT on a mass per unit volume basis. This would account for differences in soil bulk density among sites, and thus might result in a common slope and intercept for both drainage classes. For example, Sites 34 and 21 both had ISNT concentrations of 93 mg kg–1 (Table 5), but Site 34 had a greater bulk density and thus a greater per-hectare-furrow-slice ISNT than Site 21 (279 vs. 246 kg ISNT ha–1, respectively). Regression models of average EONR vs. ISNT per hectare furrow slice for well and poorly drained sites had lesser slopes than the average EONR vs. ISNT concentration models and showed trivial improvements in coefficients of determination and RMSE (Table 1, Fig. 1 and 5 ). Statistical analysis of the average EONR vs. ISNT per hectare furrow slice models showed that the intercepts and slopes for the well and poorly drained sites were different (data not shown). The use of mass per unit volume appeared to have accounted for some differences in bulk densities between mineral and organic soils, as the slopes for the well and poorly drained models were more similar for the ISNT per hectare furrow slice model than for the ISNT concentration model (Fig. 5 and 1, respectively). The coefficients of determination, however, for the regression models of average EONR vs. ISNT concentration and ISNT per hectare furrow slice were nearly identical, thus there appeared to be little or no benefit to the additional effort required to determine soil bulk density. The decrease in the regression slope as a result of incorporating bulk density, however, would allow a somewhat greater resolution in determining the EONR; this effect was somewhat greater for the well-drained than the poorly drained sites.


Figure 5
View larger version (26K):
[in this window]
[in a new window]

  		Fig. 5. Economic optimum N rate for the average fertilizer cost/corn price ratio (EONRavg) vs. the Illinois soil N test on a hectare furrow slice basis (ISNThfs) for well and poorly drained sites modeled using regression analysis.

 
Effect of Different Cost Ratios on Predicting Economic Optimum Nitrogen Rate with the Illinois Soil Nitrogen Test
For the well and poorly drained sites combined (n = 30), the means for the three EONR were 170 kg ha–1 for minimum EONR, 184 kg ha–1 for average EONR, and 194 kg ha–1 for maximum EONR (Table 5). Regression models of the three EONR vs. ISNT concentration for well and poorly drained sites were all significant, with coefficients of determination ranging from 0.70 to 0.87 (Table 6). For the poorly drained sites (n = 10), there were no statistical differences among the regression parameters for the three EONR models; minimum EONR had the highest coefficient of determination (0.84), and maximum EONR the lowest (0.70) (Table 6). For the well-drained sites (n = 20), the three EONR vs. ISNT concentration models had different intercepts, but there were no differences among the slopes (Table 6). The coefficient of determination for the minimum EONR vs. ISNT concentration model was slightly lower than for the average EONR and maximum EONR vs. ISNT models. These results for the well-drained sites show that, as fertilizer costs increase or corn prices decrease, a different model with a smaller intercept would be needed to recommend less N fertilizer to achieve economic optimum.


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

  		Table 6. Linear regression model parameters for economic optimum N rate (EONR) vs. the Illinois soil N test on a concentration basis, for the maximum, average, and minimum EONR resulting from three fertilizer N cost/corn price ratios for the well and poorly drained sites.

 
One concern with making N recommendations based on the average EONR vs. ISNT concentration models developed in this study is that some sites had very high predicted EONR. For example, the average EONR were 304 and 336 kg N ha–1 for Sites 26 and 27, respectively (Table 5); however, these sites also had the lowest ISNT concentration, consistent with the high EONR. The high EONR for these sites and other sites studied in the Lower Coastal Plain of North Carolina may have been the result of continuous grass crop rotations (wheat [Triticum aestivum L.]–sorghum [Sorghum bicolor (L.) Merr.] before corn) or other high-N-demanding crop rotations on sandy soils that receive high rainfall. Additionally, corn yields on small plots tend to be higher than in production fields because of better management (e.g., irrigation). For our sites, the experimental mean economic optimum corn yield (9.5 Mg ha–1) and average EONR (180 kg ha–1) were higher than the mean RYE (8.1 Mg ha–1) and N rate (159 kg ha–1) for the study sites (Table 5) as determined from the state RYE database (North Carolina Nutrient Management Workgroup, 2003). The experimental mean N application factor (i.e., ratio of average EONR to economic optimum corn grain yield) for our sites was 19.5 kg N Mg–1 corn (1.09 lbs N bu–1 corn grain yield), which was equivalent to the RYE database N application factor of 19.6 kg N Mg–1 (1.10 lbs N bu–1; Table 5; North Carolina Nutrient Management Workgroup, 2003). According to the RYE database, N application factors for these sites should range between 18.9 and 20.4 kg N Mg–1. Our sites showed a much wider range (Table 5), with high calculated N application factors at Sites 35 (36.5 kg N Mg–1) and 27 (33.9 kg N Mg–1) and low N factors at Sites 34 (3.0 kg N Mg–1) and 9 (6.8 kg N Mg–1). The wide range of N application factors calculated from actual N response data demonstrates one weakness of RYE-based fertilizer recommendations and the need for a soil-based N test like the ISNT for making accurate fertilizer applications.


   CONCLUSIONS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
The EONR was negatively and strongly correlated to ISNT concentration when separate models were fit to well and poorly drained sites. The different models based on site drainage characteristics probably reflect differences in organic matter character (C/N ratio); rates of N mineralization, denitrification, and immobilization; and N-uptake and N-use efficiency. The ISNT concentration was able to predict average EONR despite moderate weather variation between 2003 and 2004, although model intercepts were different between these years. The difference in intercepts between the years was probably the result of different rainfall amounts and patterns, which complicate the prediction of EONR regardless of method (i.e., soil test or RYE). The use of bulk density data to calculate ISNT per hectare furrow slice (mass per volume basis) did not improve the prediction of EONR. Our data indicated that prediction of EONR from ISNT concentration for the well-drained sites required a different model for the lowest fertilizer cost/corn price ratio (5:1) compared with the average and high cost/price ratios (7.6:1 and 11.4:1), but that the same model could be used for all cost/price ratios for the poorly drained sites. The EONR vs. ISNT concentration models predicted very high EONR (>300 kg ha–1) for some sites, but these sites had the lowest measured ISNT concentration of our study and some of the highest yields. The mean N application factor (ratio of applied N to corn grain yield) was similar to the North Carolina RYE database N application factor mean for the experimental soils. The high predicted N rates for specific fields appeared related to cropping history and soil characteristics (i.e., repeated grass crops on irrigated coarse soils).

The results indicate that the well and poorly drained soil ISNT concentration models for predicting EONR were relatively robust and show promise as a tool for N management. Further research is needed to calibrate and validate the average EONR vs. ISNT concentration relationships under grower conditions. The future application of the ISNT could be to modify current yield-based N fertilizer recommendations or to develop new recommendations based on ISNT. A possible strategy for using ISNT to predict corn N need could be to take a preplant soil sample for ISNT analysis, apply a small amount of starter N at planting, then apply sidedress N between V4 and V10 with total N rates based on the ISNT. Ruffo et al. (2005) studied the spatial variability of ISNT within Illinois fields in the context of predicting areas likely to be unresponsive to N fertilization. Additional research is needed, however, to determine whether the ISNT can be used to develop spatially variable EONR for variable-rate N applications within fields.


   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, Cleveland County, Lake Wheeler, Lower Coastal Plain Research Stations, and the entire on-farm support at Guy Davenport and Tankard Farms.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
This research was supported in part by USDA Initiative for Future Agricultural and Food Systems (IFAFS) Grant no. 00-52103-9644.

All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Permission for printing and for reprinting the material contained herein has been obtained by the publisher.

Received for publication March 28, 2007.


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




This article has been cited by other articles:


Home page
 Soil Sci.Home page
T. L. Roberts, R. J. Norman, N. A. Slaton, and C. E. Wilson Jr.
Changes in Alkaline Hydrolyzable Nitrogen Distribution with Soil Depth: Fertilizer Correlation and Calibration Implications
Soil Sci. Soc. Am. J., October 21, 2009; 73(6): 2151 - 2158.
[Abstract] [Full Text] [PDF]

Home page
 Soil Sci.Home page
H.-Y. Kwon, R. J. M. Hudson, and R. L. Mulvaney
Characterization of the Organic Nitrogen Fraction Determined by the Illinois Soil Nitrogen Test
Soil Sci. Soc. Am. J., April 21, 2009; 73(3): 1033 - 1043.
[Abstract] [Full Text] [PDF]

Home page
 Soil Sci.Home page
J. T. Spargo, M. M. Alley, W. E. Thomason, and S. M. Nagle
Illinois Soil Nitrogen Test for Prediction of Fertilizer Nitrogen Needs of Corn in Virginia
Soil Sci. Soc. Am. J., February 6, 2009; 73(2): 434 - 442.
[Abstract] [Full Text] [PDF]

Home page
 Agron. J.Home page
C. A. M. Laboski, J. E. Sawyer, D. T. Walters, L. G. Bundy, R. G. Hoeft, G. W. Randall, and T. W. Andraski
Evaluation of the Illinois Soil Nitrogen Test in the North Central Region of the United States
Agron. J., June 16, 2008; 100(4): 1070 - 1076.
[Abstract] [Full Text] [PDF]

Home page
 Soil Sci.Home page
J. T. Spargo and M. M. Alley
Modification of the Illinois Soil Nitrogen Test to Improve Measurement Precision and Increase Sample Throughput
Soil Sci. Soc. Am. J., May 1, 2008; 72(3): 823 - 829.
[Abstract] [Full Text] [PDF]

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 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 HighWire
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
Right arrow Plant and Soil Interactions
Right arrow Maize

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