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


     

Published online 2 December 2005
Published in Soil Sci Soc Am J 70:172-182 (2006)
DOI: 10.2136/sssaj2005.0034
© 2005 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
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 Mulvaney, R. L.
Right arrow Articles by Ellsworth, T. R.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Mulvaney, R. L.
Right arrow Articles by Ellsworth, T. R.
Agricola
Right arrow Articles by Mulvaney, R. L.
Right arrow Articles by Ellsworth, T. R.
Related Collections
Right arrow Nitrogen
Right arrow Soil Fertility and Productivity
Right arrow Maize Management

Nutrient Management & Soil & Plant Analysis

Need for a Soil-Based Approach in Managing Nitrogen Fertilizers for Profitable Corn Production

R. L. Mulvaney*, S. A. Khan and T. R. Ellsworth

Dep. of Natural Resources and Environ. Sci., Univ. of Illinois, 1102 S. Goodwin Ave., Urbana, IL 61801

* Corresponding author (mulvaney{at}uiuc.edu).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Nitrogen fertilization for corn (Zea mays L.) production has relied extensively on yield-based recommendations that were developed to represent regional averages, yet are routinely applied to individual fields, on the assumption that fertilizer N serves as the major supply for crop N uptake. Using data from 102 on-farm N-response studies, an evaluation was conducted of the Illinois proven-yield (PY) method for accuracy and economic profitability on a site-by-site basis. As additional objectives, the Illinois soil N test (ISNT) was evaluated for detecting whether N fertilization was economical, and for quantifying crop response to N fertilization relative to soil and management factors. For 18% of the site-years studied, N recommendations by the PY method were accurate to within 20 kg ha–1, whereas 13% were underfertilized by 25 to 129 kg ha–1 (60 kg ha–1 on average) at a current cost of $5 to $170 ha–1 ($75 ha–1 on average), and 69% were overfertilized by 21 to 235 kg ha–1 (103 kg ha–1 on average) at a cost of $12 to $130 ha–1 ($57 ha–1 on average). The latter group included 30 site-years that were completely nonresponsive to N fertilization, all but two of which were predicted by site-average ISNT values assuming a critical test level of 230 mg kg–1. This level was exceeded for 19 of 69 responsive site-years, mostly during 2001–2003 when corn followed soybean (Glycine max L. Merr.) with high plant populations. A higher critical test level would have been required under such conditions, owing to more extensive residue inputs that would promote microbial N immobilization, and increased crop uptake of mineralized soil N. The ISNT was significantly related to crop N requirement, and was the most powerful predictor of error in PY recommendations (P < 0.001).

Abbreviations: ANOVA, analysis of variance • EONR, economically optimum N rate • EOY, economically optimum yield • FNUE, fertilizer N uptake efficiency • ISNT, Illinois soil N test • PY, proven-yield • SD, standard deviation


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
SINCE THE 1970s, N fertilizer recommendations for midwestern corn production have relied on a yield-based system, whereby an expected yield goal is multiplied by a constant factor (typically 19.4–24.2 kg N Mg–1 or 1.1–1.4 pound N bushel–1), with adjustments to account for N credits from previous cropping or the recent use of manure (e.g., Illinois Agronomy Handbook, 2002). This system utilizes a mass balance approach that assumes constant efficiency in crop uptake of fertilizer and soil N (Stanford, 1973; Meisinger, 1984; Meisinger et al., 1992). Yield-based systems were originally intended as a first approximation in making generalized fertilizer N recommendations for long-term periods on a regional scale, but have been applied indiscriminately to fertilize individual fields in a particular growing season.

Implicit to yield-based N recommendations is the presumption that mineralization is a negligible source for crop N uptake, which would necessarily imply that yield in the absence of applied N supplies a fixed proportion of crop N uptake that is substantially less than that from fertilizer. Yet unfertilized (check) plot yields in N-response studies often exceed the yield increase obtained with fertilization (Lory and Scharf, 2003), and in many of these studies, sites have been detected where corn is completely nonresponsive to fertilizer N (e.g., Bundy and Malone, 1988; Blackmer et al., 1989; Fox et al., 1989; Schmitt and Randall, 1994). Such sites have often been excluded in averaging response data to evaluate yield-based N recommendations (e.g., Vanotti and Bundy, 1994; Brown, 1996; Lory and Scharf, 2003; Nafziger et al., 2003), but even so, the recommended rates tend to be excessive. This was the case, for example, with 96% of 193 responsive site-years analyzed by Lory and Scharf (2003), for which the recommended N rate exceeded the economically optimum N rate (EONR) by up to 227 kg ha–1 (90 kg ha–1 on average). More importantly, recommended and optimum N rates were not correlated significantly (r = 0.04) in the latter study, suggesting that yield-based N recommendations lack predictive value. The same concern has been raised previously by researchers in Iowa (Peterson and Corak, 1993; Blackmer et al., 1997), Wisconsin (Vanotti and Bundy, 1994; Bundy, 2000), Pennsylvania (Fox and Piekielek, 1995), and Ontario (Kachanoski et al., 1996).

The only hope for improving fertilizer N recommendations for corn production in a humid region such as Illinois is to account for a soil's capacity to supply plant-available N through mineralization. The usual approach has been to measure soil NO3–, either before or after planting. Some success has thereby been achieved in detecting nonresponsive sites (e.g., Bundy and Malone, 1988; Blackmer et al., 1989; Schmitt and Randall, 1994), although complications arise from the need for special sampling protocol and from spatial and temporal variability in soil NO3– concentrations, which depend on numerous N-cycle processes, including mineralization, immobilization, nitrification, denitrification, leaching, and plant uptake.

A better approach would focus on the soil's N-supplying capacity by estimating mineralizable organic N, which is subject to fewer N-cycle processes than NO3– and should thus be less dynamic. Research since the 1950s has provided growing support for the concept that soil organic matter is not uniformly mineralizable, but consists primarily of a passive fraction accompanied by a less extensive pool of biologically active organic N associated with microbial biomass (e.g., Jansson, 1958; Paul and Juma, 1981; Mengel, 1996). The latter constituents are identified largely as {alpha}-amino N and (amide + amino sugar)-N, both of which have been linked to net mineralization and/or crop N uptake in pot experiments (Mengel, 1996).

Several attempts were made during the 1960s and 1970s to provide a chemical basis for soil management effects on crop growth and fertilizer N response under field conditions; however, the results generally indicated little variation in the distribution of N, and the usual conclusion was that no particular fraction of hydrolyzable soil N is more labile than others (e.g., Keeney and Bremner, 1964; Khan, 1971; Meints and Peterson, 1977). This conclusion has been widely accepted, but must be questioned in light of recent evidence that conventional steam-distillation methods do not permit quantitative analyses for amino sugar N or amino acid N (Mulvaney and Khan, 2001). Based on the latter finding, simple diffusion methods were developed for N-distribution analysis of soil hydrolysates that are accurate, specific, and reliable.

In subsequent work by Mulvaney et al. (2001), the newly developed diffusion methods were applied to soil samples collected by Brown (1996), from sites that differed in whether corn had been responsive to N fertilization. The results showed a much higher concentration of amino sugar N for nonresponsive than for responsive soils, whereas no consistent difference was detected in their concentrations of total hydrolyzable N, hydrolyzable NH4+–N, or amino acid N. Upon incubation, mineral N production was found to be much more extensive by nonresponsive than by responsive soils, and to be accompanied by a net decrease in amino sugar N but not in amino acid N (Mulvaney et al., 2001). Based on these findings, a simple soil test, the so-called Illinois soil N test (ISNT), was developed that estimates amino sugar N without the need for hydrolysis (Khan et al., 2001). When this test was applied to 25 site-average soil samples collected by Brown (1996) to a depth of 30 cm, a critical range of 225 to 240 mg kg–1 completely resolved 12 nonresponsive from 13 responsive soils.

The present study originated with the objective of evaluating the effectiveness of the ISNT in differentiating responsive from nonresponsive site-years under a wide range of soil and cropping conditions. As additional objectives, the database thereby generated was used to assess (i) how these conditions might influence a quantitative relationship between ISNT values and crop responsiveness to N fertilization; and (ii) the accuracy and economic consequences of N recommendations by the PY method, primarily on a site-by-site basis. Very little peer-reviewed information is available on the latter issue, despite the fact that this method has been promoted for several decades in many states through university extension publications.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Field Plot Management
The work reported herein involved 102 N-response experiments located throughout Illinois, largely on farmer fields. Of these experiments, 51 were reported by Brown (1996), including 11 conducted in 1990, 18 in 1991, and 22 in 1992.An additional 51 experiments included 14 in 2001, 16 in 2002, and 21 in 2003. In each case, N rates were applied according to a randomized complete block design with four replicates, by sidedressing urea-NH4NO3 solution (360 g N L–1) when corn was 15 to 30 cm tall. In 1990, 1991, and 1992, plots measured approximately 4 m in width x 15 m in length, and N applications were based on a PY recommendation, assuming soil productivity under high management (Fehrenbacher et al., 1978). Nitrogen was applied at 0, 20, 40, 60, 80, or 100% of the recommended rates in 1990, and at 0, 25, 50, 75, 100, or 125% of these rates in 1991 and 1992. In 2001, 2002, and 2003, plots measured approximately 5 m in width x 15 m in length, and N applications ranged from 0 to 235 kg ha–1, in equal increments of 33.6 kg ha–1. At each site, an adapted corn hybrid was planted in rows spaced 76 cm apart in April or May. Thinning was done before N application at V3 to V6, so as to obtain a uniform population within the experimental area. At physiological maturity, grain yield was determined by hand-harvesting 9 m of the two middle rows, and was adjusted to a constant moisture content (155 g kg–1).

Soil Samples
Soil samples were collected from the experimental area at each site in late March or early April, including surface (0–18 cm) samples for routine soil fertility assessment (pH, P, and K) and profile samples for NO3– testing (1990–1992) or the ISNT (2001–2003). Surface samples were collected as a 5-core composite from the entire experimental area, using a 2.5-cm diam. probe. The cores were dried at room temperature (1990–1992) or in a forced-air oven at 40°C (2001–2003), crushed with a mechanical grinder to pass a 2-mm screen, and then mixed thoroughly before analyses for pH, available P, and exchangeable K as described by Mulvaney et al. (2001), and for organic C and total N as described by Khan et al. (2000). The data are summarized by Table 1, according to soil series for site-years identified as responsive or nonresponsive to N fertilization.


View this table:
[in this window]
[in a new window]
  		Table 1. Characterization of soils for site-years studied in N-response experiments.

 
Profile samples were collected to depths of 0 to 30 and 30 to 60 cm in 1990–1992, and from 0 to 15, 15 to 30, and 30 to 60 cm in 2001–2003. In each case, five soil cores were collected from each block using a 5-cm (for 0- to 30-cm samples) and then a 2.5-cm (for 30- to 60-cm samples) diam. probe, combined by depth, and subsequently frozen (–10°C) within 12 h after collection. Before use, core samples collected in 1990–1992 were allowed to thaw at room temperature, screened while still field-moist to <2 mm, and then air-dried at room temperature, with subsequent transfer for storage in polyethylene or paper bags. In 2001–2003, profile samples were dried in a forced-air oven at 40°C, and were then crushed in a hammer mill to <2 mm.

Analytical Methodology
The N test described by Khan et al. (2001) was performed as specified in a technical note (15N Analysis Service, 2004) concerning the ISNT, which describes three modifications to improve the uniformity of heating with the griddle employed (Model 76220; West Bend, West Bend, WI): (1) replacement of the original temperature controller with an electronic unit, (2) enclosure of the griddle within a polyethylene box as a draft shield, and (3) rotation of jar positions after heating for 1.5 and 3 h. To ensure the validity of ISNT data, care was taken that heating was always done at the same measured temperature (54°C), samples were analyzed in triplicate (for 2001–2003 samples) or quadruplicate (for 1990–1992 samples), and a reference soil sample was included on each griddle.

For 1990–1992 site-years, analyses by the ISNT were performed on a composite sample of air-dried soil (0–30 cm) prepared by combining an equal weight of soil collected from each block. For 2001–2003 site-years, block samples were analyzed individually, and ISNT data for 0 to 30 cm were generated by averaging values measured for 0 to 15 and 15 to 30 cm. Analyses were also performed on the 30- to 60-cm samples collected for the latter group, but are reported for only a single site-year to demonstrate an interaction with crop N response.

Experimental Site-Years
The 102 site-years studied are characterized by Tables 2 and 3, which show the soil series; the year when N response was studied; the previous crop; the tillage system in use; the source and amount of manure N applied for the growing season studied, as well as residual manuring within the previous 2 to 5 yr; plant population estimated from stand counts; a site-average ISNT value and the standard deviation (SD) computed from four (1990–1992) or 12 (2001–2003) replicate values; check-plot corn yield data; and the magnitude of the error in the PY recommendation and the corresponding economic cost. For each site-year, a recommended N rate was determined as described in the Illinois Agronomy Handbook (2002), using productivity indices reported by Fehrenbacher et al. (1978) for high management (1990–1992 site-years) or by Olson and Lang (2000) for optimum management (2001–2003 site-years). In the case of nonresponsive site-years (Table 2), the magnitude of error in the PY recommendation was obtained as the recommended N rate, and the economic cost was calculated on the assumption that fertilizer N costs $0.55 kg–1.


View this table:
[in this window]
[in a new window]
  		Table 2. Characterization of nonresponsive site-years.{dagger}

 

View this table:
[in this window]
[in a new window]
  		Table 3. Characterization of responsive site-years.{dagger}

 
Site-years were identified as nonresponsive (Table 2) or responsive (Table 3) for N fertilization of corn, based on regression analyses performed by fitting a linear, linear plateau, quadratic, or quadratic plateau model to N-rate and mean yield data using Microsoft Excel 2003 software (Microsoft Corp., Redmond, WA) configured with Solver as an add-in. A response model was selected for each site-year on the basis of residual analysis and the Chi-Square test for goodness of fit. In the case of responsive site-years, economically optimum yield (EOY) and EONR were computed by the resulting regression equation assuming a value of 0.1 for the N/corn price ratio; delta yield (Kachanoski et al., 1996) was obtained as the difference between the EOY and check-plot yield [which in some cases was achieved with a fall application of 7–78 kg N ha–1 as (NH4)2HPO4]; fertilizer N uptake efficiency (FNUE) was estimated as delta yield/EONR; and error in the PY recommendation was calculated from the difference, recommended N– EONR. To estimate an economic cost for overfertilized site-years, the latter difference was multiplied by $0.55 kg–1 N, whereas with an insufficient N recommendation, a cost was computed assuming $98.42 Mg–1 for yield loss estimated by regression and a credit of $0.55 kg–1 for unused N.

Analysis of Variance
After excluding 18 site-years involving an obvious limitation to crop growth or fertilizer N response (site-years 2, 10, 11, 17, 21, 34, 41, 43, 50, 52, 54, 57, 77, 91, 92, 99, 101, and 102), plus two others where corn followed wheat (site-years 35 and 43), the Proc MIXED procedure within SAS (SAS Institute, 1998) was employed to examine the effects of both categorical [year (treated as a random effect); previous crop; tillage; manuring (current and residual or residual only)] and continuous (population and ISNT value) variables on soil organic C, sidedressed N, EOY, delta yield, EONR, FNUE, and PY error (kg ha–1) as dependent variables. Tukey-Kramer tests were performed in carrying out pairwise comparisons of treatment means.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Evaluation of Yield-based Nitrogen Management
The importance of fertilizer N management in corn production is clearly evident from the economic costs that often exceed $100 per hectare. The rationale for this investment resides in the PY method, whereby crop N uptake is ascribed largely to N fertilization. The resulting recommendations have been widely adopted on the premise that yield must not be limited by inadequate N supply, yet have seldom been evaluated relative to grain yield with a lower (or higher) rate of N fertilization, or for accuracy in fertilizing individual sites where N-response studies have been conducted to determine an EONR. Such studies provide the basis for the evaluations reported in Tables 2 and 3, which indicate the magnitude of the error in the PY recommendation for each site-year studied and the corresponding economic cost. Table 4 summarizes the latter information for site-years under different crop rotations or with current manuring.


View this table:
[in this window]
[in a new window]
  		Table 4. Effectiveness of the proven-yield (PY) method for site-years under different management.

 
Tables 2 and 3 provide good reason to question the validity of the PY method, as check-plot yields ranged from 1.9 to 13.1 Mg ha–1 (8.2 Mg ha–1 on average, representing 77% of the average EOY) for the 52 site-years that received no input of N from sidedressing, manuring, or fall-applied (NH4)2HPO4. The implicit capacity of the soil to provide plant-available N, as reflected in these values, cannot be adequately accounted for in making PY recommendations, which assumes that the soil supplies a constant proportion of crop N uptake.

Considering all 102 of the N-response trials reported herein, PY recommendations were accurate to within 20 kg ha–1 for only 18 of the site-years studied, while 69% of these recommendations were excessive, involving the majority of both nonresponsive and responsive site-years. Underfertilization occurred with 13% of the latter group, and accounted for the most serious economic loss observed. There were four cases where N utilization was limited by a prolonged moisture stress; the usual and expected result was overfertilization.

In 22 of the N-response experiments reported, manure had been applied for the growing season studied, so the PY recommendation was adjusted to incorporate standardized credits for manure N (Illinois Agronomy Handbook, 2002). The adjustment proved inadequate, except for identifying three nonresponsive site-years where the manure credit exceeded the N requirement estimated for the yield goal. Of the 19 remaining currently manured site-years, 15 were completely nonresponsive to N fertilization, but would have received 38–159 kg N ha–1 by the PY method at a cost of $21 to $88 ha–1. Fertilization also would have been recommended for the four additional site-years where a yield response was observed. Three of the latter cases involved a corn–soybean rotation, and the combined N credits would have led to underfertilization. In contrast, the PY method would have overfertilized a responsive site-year under continuous corn, for which the manure credit was inadequate. The implication is that a credit approach cannot provide a reliable basis for quantifying manure N availability, as has been reported previously (Hansen et al., 2004). This would indeed be expected given the inherent complications associated with such factors as manure C and N concentrations, N losses through NH3 volatilization, and inaccuracies in manure application.

A further problem arises because the PY method does not account for residual availability of manure N, which can persist for several years after application (Eghball and Power, 1999; Eghball et al., 2004). The resulting impact on soil N availability was verified in the present project by using Fisher's Exact Test to evaluate the effect of manure history on crop N response for site-years under continuous corn or in a corn–soybean rotation without current manuring. A significant difference at P < 0.01 was thereby found, in which 82% of the nonresponsive site-years in this group had a history of manuring, as compared with 16% of those that were responsive. Residual manure was a more common occurrence when corn was grown continuously than in rotation with soybean. The latter difference is particularly apparent for nonresponsive site-years that were not currently manured, among which were all seven of those under continuous corn but only two of four that were in a corn–soybean rotation. The PY recommendations were always excessive for continuous corn (by 49–235 kg N ha–1, at a cost of $27–$130 ha–1), with or without a response to N fertilization, whereas either under- or overfertilization occurred when there was a manure history for corn in rotation with soybean.

As with current manuring, fixed N credits are utilized in PY recommendations when corn is grown after a legume. The present project involved 54 such site-years that had not been manured for at least 1 yr before the growing season studied, including 49 in a corn–soybean rotation and five where first-year corn followed alfalfa. Of the latter group, four site-years were nonresponsive to N fertilization, but would have been fertilized with 105 to 123 kg N ha–1 by the PY method at a cost of $58 to $68 ha–1, even after maximizing the alfalfa credit (112 kg N ha–1). The error was more extensive in magnitude (162–193 kg N ha–1) and cost ($89–$106 ha–1) for four nonresponsive site-years where corn followed soybean, involving either no-till (site-years 6 and 22) or residual manuring (site-years 15 and 33). In contrast, underfertilization often occurred when a yield response was obtained with soybean as the previous crop, whereas no such occurrences were observed with continuous corn, suggesting a greater need for N fertilization when corn follows soybean. The latter difference was substantiated, after excluding manure and tillage effects, by an ANOVA that showed significant (P < 0.0001) increases in EONR and delta yield when the previous crop was soybean rather than corn. The EONR estimated for corn after soybean was significantly (P < 0.0001) greater for 2001–2003 (153 kg N ha–1) than for 1990–1992 (96 kg N ha–1) site-years, suggesting that current production practices have increased the fertilizer N requirement of corn within this rotation. Such an increase is likely attributable to greater nutrient demand by improved hybrids selected for maximal yields with high planting rates.

These findings raise serious questions about the use of standardized credits for estimating the fertilizer value of legume-derived N, which ranges widely with species and environmental conditions (Heichel and Barnes, 1984). An inherent difficulty arises, for example, because plant uptake of mineral N reduces symbiotic fixation (e.g., Giöbel, 1926; Thornton, 1946), and thus a single legume credit cannot suffice for soils that differ considerably in their capacity for mineralization (Kurtz et al., 1984). In the case of soybean, a positive credit may often be inappropriate, because the grain has a higher N concentration than with corn, and soil N removal can be much more extensive (Gentry et al., 1998). In the present project, a soybean credit was inappropriate for nonmanured site-years under a corn–soybean rotation, as one-third of this group would have been underfertilized by the PY method, at an average cost of $57 ha–1.

Lacking any N credit for management history, PY recommendations were excessive for all but one of the 23 site-years under continuous corn that had not received manure for the growing season studied (although in almost 50% of these cases, manure had been applied within the previous 2–5 yr). Almost one-third of this group was nonresponsive to N fertilization, as compared with <10% of the 49 site-years in a corn–soybean rotation with no manure credit. While on average both groups were overfertilized by the PY method, the error was much more extensive (128 versus 46 kg N ha–1 as calculated using actual errors rather than the magnitudes reported in Table 4) when corn was the previous crop (P < 0.01), with no instances of underfertilization. These findings may be explained in part by a more extensive occurrence of residual manure and a larger input of fertilizer N applied annually to continuous corn, which promotes residue decomposition with microbial production of labile soil N that would reduce fertilizer N response (Shen et al., 1989; Stevens et al., 2005).

Based on Tables 2 to 4, the preceding discussion raises serious questions about the practical value of the PY method for fertilizing individual sites, as does the fact that EOY (data not shown, but readily calculable as yield without sidedressed N plus delta yield) was not related to EONR (r = 0.08). This method likewise proved to be inaccurate when fertilizer N recommendations were averaged for the 102 site-years studied, contrary to the usual justification for yield-based N management. A value of 154 kg N ha–1 was thereby obtained, as compared with 90 kg N ha–1 for the average EONR. The difference was reduced but not eliminated by excluding the 33 nonresponsive site-years, in which case the PY recommendation averaged 38 kg N ha–1 higher than did the EONR (131 kg ha–1). The latter strategy has often been employed in reporting N-response trials but cannot be justified, as the PY method provides no a priori basis for identifying nonresponsive site-years.

Evaluation of Soil-based Nitrogen Management
The recurring evidence of serious inaccuracy in fertilizer N recommendations by the PY method has obvious economic implications for individual farmers, and also raises concern about environmental pollution. Extrapolating from the average error in these recommendations for the site-years studied ($50 ha–1), the annual cost to Illinois agriculture would exceed $220 million, which does not include additional expenses associated with excessive N fertilization, such as the loss of Ca2+, Mg2+, and K+ that serve as counterions during the leaching of NO3–. Such estimates emphasize the need to account for a soil's capacity to supply plant-available N through mineralization, which is the key to improving fertilizer N management in a humid region such as Illinois. The ISNT was developed precisely for this purpose, and is designed to estimate an alkali-1abile fraction of soil N, nominally referred to as amino sugar N, which has been related to net N mineralization (Mulvaney et al., 2001). The same relationship has been observed in several previous investigations to evaluate various alkaline reagents as a chemical index of soil N availability (e.g., Cornfield, 1960; Gianello and Bremner, 1988; Vanotti et al., 1995; Mengel, 1996), whereas soil organic matter measurements are of limited value for the latter purpose (e.g., Schmidt et al., 2002; Walley et al., 2002).

As originated, the ISNT is employed to identify sites where N fertilization is ineffective for increasing corn yield, although an obvious potential also exists for estimation of fertilizer N requirements, in lieu of yield-based N management. Assuming the same critical test value determined by Khan et al. (2001) for 25 site-years in 1990 to 1992 (230 mg kg–1) as a first approximation in evaluating soil N availability without considering management history, the ISNT was 94% effective in identifying site-years characterized in the present project by the lack of an economic yield response to N fertilization. The majority of this group had been manured for the growing season studied, or had received manure within the previous 2 to 5 yr while cropped to continuous corn. As shown by Table 5, all but two such site-years were detected successfully by the ISNT, the only exceptions occurring when yield data were erratic within and among replicate plots, which tended to show a similar pattern of variability in soil test values. Moreover, the ISNT was completely effective in predicting 8 site-years that were nonresponsive to fertilizer N following previous cropping to soybean or alfalfa.


View this table:
[in this window]
[in a new window]
  		Table 5. Effectiveness of the Illinois soil N test (ISNT) for differentiating responsive from nonresponsive site-years for N fertilization of corn.

 
The present project also involved 69 site-years that were responsive to N fertilization. Of these, 50 were correctly identified on the basis of the same critical test value noted previously, while the remaining 19 would have been identified as nonresponsive because this value was exceeded. The latter group raises an apparent concern about the utility of the ISNT for fertilizer N management, but also provides an opportunity to gain valuable insight about interactions that affect crop N requirement, and thereby affect the critical level for interpreting test values. Both topics are addressed by the following discussion, although necessarily through scientific inference concerning factors known to affect crop growth and nutrient requirements, since the work reported was not designed to quantify specifically such effects.

In order for the ISNT to be utilized successfully, conditions must be conducive to soil N mineralization, as well as crop N uptake and utilization. This requirement was not satisfied with four of the 19 site-years incorrectly identified as nonresponsive by the ISNT, owing to serious moisture stress that occurred for most (site-years 101 and 102) or some (site-years 76 and 79) of the growing season. The effect of this stress on interpretation of the ISNT is clearly demonstrated from a comparison of yield data for site-years 22 and 76, which involved the same location with ISNT values above the critical level, but different growing conditions. Rainfall was adequate to promote mineralization throughout the 2001 growing season, whereas a 6-wk period occurred without appreciable rainfall during May and June of 2003, which drastically decreased check-plot yield and led to a dramatic yield response to N fertilization.

Interpretations of ISNT data can also be vitiated if fertilizer N requirement is increased by other factors that reduce soil N availability or crop N utilization, such as weed competition or a soil fertility limitation. This was the case with one responsive site-year (no. 95) where weed competition would have decreased crop uptake of soil and fertilizer N, and with six others for which the critical test value was exceeded with a pH of 5.0 to 5.2 (site-years 57 and 92), Bray-1 P at 14 to 21 mg kg–1 (site-years 34, 54, and 57), or exchangeable K at 103 to 132 mg kg–1 (site-years 55 and 59). Soil acidity would have impeded mineralization, thereby reducing the availability of labile soil N estimated by the ISNT, whereas a deficiency of P or K would have decreased the physiological efficiency of plant N utilization for grain production. This decrease was clearly reflected in a high EONR (Table 3) that was larger when the limitation involved K instead of P. The latter difference may be related to mineralization, which served as a supplemental source of P but not K.

Although often overlooked in fertilizer recommendations promoted during the past three decades, and generally neglected in the scientific literature on soil fertility, plant population has been recognized as a crucial factor in the successful use of soil testing (Bray, 1948; Melsted and Peck, 1973). A fundamental interaction thereby arises, such that a certain critical soil test level could become inadequate if the planting rate were increased. This is exactly what has been observed with the ISNT. In the original work by Khan et al. (2001), a test level of 225 to 235 mg kg–1 was completely effective in identifying 12 of 25 site-years as nonresponsive to N fertilization, based on N-response trials conducted between 1990 and 1992 with 47 400 to 68 900 (60 700 on average) plants ha–1. When the same critical ISNT range was applied in the present project, eight failures occurred that are of particular interest because the test value exceeded the critical range, yet a crop N response was obtained with no apparent limitation in growing conditions. Two of these failures involved continuous corn in 1991 or 1992 with a plant population of 60 300 or 64 600 (62 400 on average) plants ha–1 (site-years 85 and 90), but delta yield was quite limited (0.6 Mg ha–1 on average) with a marginal economic return ($9 ha–1 on average). The failure rate increased substantially in 2001–2003 with six site-years in a corn–soybean rotation that were planted to a higher density of 66 000 to 77 500 (70 800 on average) plants ha–1 (site-years 56, 61, 66, 87, 88, and 98), in which case a six-fold increase also occurred in delta yield (average of 3.7 Mg ha–1). Table 3 provides many examples of greater N response with higher plant populations, for site-years having similar ISNT values, as further evidenced by an ANOVA that showed a significant (P < 0.05) increase in EONR with population for responsive site-years. These findings would be expected if fertilizer N requirement is subject to a fundamental interaction between soil N availability and plant demand, as is readily apparent from studies by Lang et al. (1956). A higher planting rate would thereby increase the critical level for interpreting the ISNT.

The interaction of the ISNT with plant population is clearly demonstrated by Table 6, which summarizes the magnitude of crop N response for site-years broadly grouped into two population classes representing only 1990–1992 (<61 000 plants ha–1) or largely 2001–2003 (>61 000 plants ha–1) site-years, and three ISNT classes consisting of site-years identified as highly responsive (<190 mg kg–1), moderately responsive (190–229 mg kg–1), or nonresponsive (≥230 mg kg–1) to N fertilization. The data in Table 6 leave little doubt about the need for soil-based N management, as fertilizer N requirement decreased with increase in the ISNT, while an increase occurred with plant population, reflecting higher crop N demand. The latter trend adds a new dimension to fertilizer N management with the ISNT, whereby planting rate can be adjusted to fully exploit soil N availability, provided that productivity is not limited by other soil properties (e.g., moisture). Moreover, Table 6 suggests that the ISNT may provide valuable input for optimizing planting rate, although economic factors must also be considered.


View this table:
[in this window]
[in a new window]
  		Table 6. Interaction of the Illinois soil N test (ISNT) and plant population in affecting crop N response.

 
Soil testing is inherently more complicated for N than for P or K, due in part to the fact that mineralization and immobilization are highly dependent on the quantity and quality of C available for microbial utilization. This interaction would necessarily affect the predictive value of the ISNT for estimating soil N availability following a recent incorporation of carbonaceous residues, such as bedding-laden manure or a nonleguminous cover crop. The latter explanation applies to one of the ISNT failures observed (site-year 66), in which case rye (Secale cereale L.) had been grown as an overwinter catch crop following soybean and would have promoted immobilization after being incorporated by spring tillage.

The same effect of C is implicated for five other site-years under a corn–soybean rotation, which were characterized by a high soil content of organic C (21–27 mg kg–1) when incorrectly identified as nonresponsive by the ISNT in 2001–2003 (site-years 56, 61, 87, 88, and 98). The test values thereby obtained suggest a considerable capacity for mineralization that would have promoted uptake of soil N by the previous soybean crop, thereby reducing the role of symbiotic fixation in meeting the high N requirement of this legume. The resulting decrease in soil N availability, combined with the absence of annual N fertilization, would have impeded decomposition of corn stover, potentially contributing to microbial competition for available N, and hence promoting a crop N response, during the growing season studied. This possibility would be enhanced by the growing trend toward high planting rates for corn and soybean, and is consistent with previous work by Studdert and Echeverría (2000) showing a longer half-life for soil organic C with a corn–soybean rotation, as compared with continuous corn. When these two rotations were compared by an ANOVA that removed tillage and manure effects, a significant difference was obtained at P < 0.05, involving a greater organic C content for site-years where corn followed soybean (22.7 g kg–1), as opposed to corn (17.8 g kg–1). The implication is a higher critical level for the ISNT when corn is grown in rotation with soybean.

As has generally been observed, ISNT values decreased with depth of sampling, for each of the 2001–2003 site-years listed in Tables 2 and 3. The magnitude of this decrease was especially marked for site-years 87 and 88, both of which were located within a field where corn had repeatedly been grown in the past for silage production. The low test values thereby obtained for the 30- to 60-cm depth (data not reported) suggest a limitation in subsoil fertility, which may have contributed to fertilizer N responsiveness that would not have been predicted by testing the surface 30 cm. Any such limitation would have been exacerbated for these site-years because of their high plant populations that would have increased competition for uptake of water and nutrients. The result would have been deeper and more extensive rooting, as observed in several field studies by Pavylochenko (1937).

Utilizing the traditional approach for yield-response trials, the work reported involved replicate field plots in a balanced statistical design, so as to define a site-average relationship between ISNT values and crop N response. The availability of within-site data for 2001–2003 studies revealed that yields were often erratic in comparing different N rates within a block, as well as among blocks. In such cases, ISNT values for different blocks tended to reflect a similar pattern of spatial variation. This association is clearly evident from Table 7, for a site-year identified as nonresponsive from averaged data when one block was actually responsive. The latter response coincided with the lowest ISNT values, particularly for the 30- to 60-cm depth, and test values were generally consistent with block differences in check-plot yield. There is an obvious implication that the effectiveness of the ISNT will depend on sampling scale, which should be adequate to characterize the area sampled for yield measurement. Equally obvious is a potential for site-specific N management.


View this table:
[in this window]
[in a new window]
  		Table 7. Spatial variability among replicate blocks for site-year 17 in Illinois soil N test (ISNT) values and crop N response.

 
If the ISNT measures a mineralizable form of soil N, then a positive relationship should exist between test values and grain yield without incremental N fertilization (i.e., check-plot yield). This relationship was confirmed by carrying out an ANOVA, and was found to be highly significant (P < 0.001). The same level of significance was observed in differentiating responsive from nonresponsive site-years by the ISNT and in relating test values to delta yield, EONR, and FNUE. The latter effects were negative, as would be expected because crop N response and fertilizer requirement would be reduced by more extensive mineralization. The mineral N thereby generated is ignored in making yield-based fertilizer N recommendations, and as noted previously for the PY method, the result is often overfertilization. The potential value of the ISNT for fertilizer N management was clearly demonstrated by an ANOVA to compare different parameters in accounting for the errors we observed in PY recommendations. Significant effects were observed for only two of the parameters evaluated. Besides the previous crop (P < 0.05), the ISNT was the most powerful predictor (P < 0.001).


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
A yield-based approach for fertilizer N management is inherently flawed by the underlying assumption that soil N provides a constant proportion of crop N uptake, and by the use of fixed credits to estimate the input of N from recent manuring or a previous legume. The usual result is overfertilization, although as demonstrated herein, the PY method can lead to underfertilization with a corn–soybean rotation. The economic and environmental consequences can be alleviated in a humid region by adopting a soil-based approach that quantifies the effect of mineralization on plant N availability.

The ISNT far surpassed the PY method in identifying sites where N fertilization was completely ineffective, and also proved to be sensitive to quantitative differences in soil N availability, suggesting that yield-based N rates can be reduced when test values are high. In order for this test to be utilized successfully, soil sampling must be done on an appropriate scale and to a depth consistent with ISNT calibration, and interpretations must account for crop rotation, planting density, and any occurrence of a soil fertility limitation.


   ACKNOWLEDGMENTS
 
We thank L. C. Gonzini and J. J. Warren for conducting the field-plot research reported, T. J. Smith for assistance with data collection and soil series identifications, Dr. S. Aref for the statistical analyses presented, Dr. C. S. Mulvaney for prayerful support and suggestions, and the farmers whose cooperation provided the site-years studied. Partial support for this research was obtained from the Fertilizer Research and Education Council.

Received for publication January 26, 2005.


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




This article has been cited by other articles:


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 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 Mulvaney, R. L.
Right arrow Articles by Ellsworth, T. R.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Mulvaney, R. L.
Right arrow Articles by Ellsworth, T. R.
Agricola
Right arrow Articles by Mulvaney, R. L.
Right arrow Articles by Ellsworth, T. R.
Related Collections
Right arrow Nitrogen
Right arrow Soil Fertility and Productivity
Right arrow Maize Management

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