Illinois Fertilizer
Conference Proceedings
January 22-24, 2001
| Home | 2001 Index | Search |
|
|
| Home | 2001 Index | Search |
S.A. Khan, R.L. Mulvaney, and R.G. Hoeft1
In Illinois, N fertilizer recommendations for corn production are estimated on the basis of a realistic yield goal, which is multiplied by 1.2 to calculate the pounds of N required per acre, less any adjustment for N derived from other sources, such as manure or legumes. A yield-based recommendation may have merit on a long-term basis, but under- or overfertilization is apt to occur in any given growing season since soil N availability is not taken into account. Insufficient application of N can have serious economic consequences for the farmer, whereas excessive fertilization increases the risk of environmental pollution.
Public concern that excessive N fertilization may contribute to NO3- enrichment of ground and surface water has stimulated interest in soil testing to improve the accuracy of N fertilizer recommendations for corn. This concern may well be justified, since crop responsiveness to N fertilization can vary widely even within the same field (Harrington et al., 1997), and nonresponsive sites have been detected throughout the north-central and northeastern U.S. (e.g., Bundy and Malone, 1988; Fox et al., 1989; Meisinger et al., 1992; Brown et al., 1993; Schmitt and Randall, 1994). For many years, a preplant NO3- test (PPNT) has been used in western Canada and the Great Plains region of the U.S. to account for carryover of mineral N from previous cropping (Dahnke and Johnson, 1990; Bundy and Meisinger, 1994). Though originally developed for use in semihumid areas where leaching is limited, the PPNT has recently been applied in humid regions of the north-central U.S. to detect residual NO3- in the surface 60 cm (2 feet) of medium- to fine-textured soils (Bundy and Malone, 1988; Schmitt and Randall, 1994). To improve the reliability of NO3- testing as a basis for fertilizer recommendations in humid regions, a presidedress NO3- test (PSNT) was developed by Magdoff et al. (1984), in which soil sampling is postponed until corn is 15 to 30 cm (6-12 inches) tall, so as to estimate plant-available NO3- as closely as possible to peak uptake by the crop. If the test indicates a low concentration of soil NO3- in the surface 30 cm (< 20-30 mg N kg-1), supplemental N is applied as a sidedressing. The PSNT has been recommended more widely than the PPNT in the eastern U.S., but usage has been limited by the need to collect soil samples during the growing season, at a time when farmers are occupied with many other tasks that cannot be neglected. Fertilization must be postponed until after testing, and this can lead to crop N deficiency if adverse weather conditions prevent sidedressing. Neither test is currently recommended in Illinois.
Besides logistical problems, an inherent limitation with soil testing for NO3- arises from the dynamic nature of N-cycle processes, most of which affect soil NO3- concentrations. Ideally, a soil test for N would estimate a labile organic fraction that supplies the plant through mineralization. This approach would have the major advantage over NO3- testing that soil test levels would depend on fewer N-cycle processes; therefore, these levels would be less prone to temporal and spatial variability, so that N availability could potentially be predicted on the basis of a one-time test, regardless of soil type or management. Moreover, the time of soil sampling would be much less critical than with NO3-, and samples could be stored for later analysis.
Numerous chemical methods have been proposed to estimate the availability of soil organic N (Bundy and Meisinger, 1994), but these have been based on an empirical approach, and their use has been very limited due to low correlations with crop N uptake and/or the production of mineral N during soil incubations. A more rational approach would require chemical fractionation of soil organic N to identify a labile pool; however, little progress has been made in this respect, due largely to fundamental defects in the available methodology that vitiated analyses for amino sugar-N and amino acid-N. These defects were identified and eliminated through a substantial effort that ultimately led to simple diffusion methods of fractionating the N in soil hydrolysates (Mulvaney and Khan, 2001).
As part of a current FREC project, the newly developed diffusion methods were applied in comparing N-distribution analyses for soil samples from sites that differed in crop responsiveness to N fertilization in a previous N-response study funded by FREC (Brown et al., 1993). 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 (Mulvaney et al., 2000). In subsequent incubation studies, nonresponsive soils produced a much larger quantity of mineral N than did responsive soils, and mineralization was accompanied by a net decrease in amino sugar-N but not in amino acid-N (Khan et al., 2000a).
The methods employed by Mulvaney et al. (2000) to differentiate between responsive
and nonresponsive soils require hydrolysis with 6 M HCl for 12 hours,
followed by filtration and neutralization of the hydrolysate, and are therefore
unsuitable for routine soil testing. The purpose of the work reported here
was to develop a much simpler technique, whereby amino sugar-N can be readily
estimated to detect sites that do not require N fertilization.
The soils used in the present project were PPNT samples (0-30 cm) that had been collected in late March or early April of 1990, 1991, or 1992 from 25 of the 75 sites studied by Brown et al. (1993). At each site, N was subsequently applied at six rates according to a randomized block design with four replications, by sidedressing urea-NH4NO3 solution (28 percent N) when corn was 15 to 30 cm tall. The particular samples used, representing a wide range of textural classes and management practices, were selected from sites receiving normal rainfall, and included 12 sites that were nonresponsive and 13 sites that were responsive to N fertilization. In each case, a composite sample of air-dried soil was prepared by combining equal weights of the replicate samples, so as to integrate block effects in comparing different sites. The entire sample was ball-milled to pass through a 0.15-mm screen and was then stored in a Mason jar sealed with an air-tight lid.
The 25 sites studied herein are characterized by Table 1, which shows the soil series; the county from which the sample was collected; selected soil chemical properties; the amount of manure applied; PPNT, PSNT, and check-plot yield data from Brown (1996); and an index of N-fertilizer response. Applications of manure N were estimated on the basis of the quantity of material applied as indicated by farmer records, and the average N concentration according to the Illinois Agronomy Handbook (1998). The percentage response of corn to N fertilization was calculated as 100 H (optimum yield B check-plot yield)/check-plot yield, using data reported by Brown (1996) for check-plot yield (i.e., the yield without sidedressing, as reported in Table 1) and optimum yield. The latter value was determined by fitting N-rate and corresponding yield data to a quadratic plateau model by nonlinear regression (SAS Institute, 1993).
To prepare soil hydrolysates, 5-g samples of soil (four replicates) were heated under reflux for 12 h in 125-mL Erlenmeyer flasks fitted with a 24/40 ground-glass joint for attachment to a 40-cm Liebig condenser, after treatment with 20 mL of 6 M HCl and two drops of octyl alcohol. The hydrolysis mixture was filtered through Whatman no. 50 filter paper under vacuum. Replicate hydrolysates were combined, and subsequently neutralized by addition of NaOH (Stevenson, 1996). The hydrolysates were stored under refrigeration (5°C).
Using the diffusion methods described by Mulvaney and Khan (2001), the soil
hydrolysates were analyzed for total hydrolyzable N, NH4+-N,
(NH4+ +
amino sugar)-N, and amino acid-N. Amino sugar-N was determined as the difference
between (NH4+ +
amino sugar)-N and NH4+-N.
Apparati
Diffusion unit. The diffusion unit used consists of a 1-pint (473-mL) wide-mouth Mason jar equipped with a lid that has been modified to support the bottom of a 60-mm (dia.) Pyrex petri dish. The necessary modifications are described in detail by Mulvaney (1996), Khan et al. (1997), and Mulvaney et al. (1997). Further information about Mason-jar diffusion methodology can be found therein, including the appropriate cleaning procedures.
Electric hot plate. A commercial griddle (West Bend Model 76212) was used. Before use, the heat control was adjusted so that a temperature of 48 to 50°C was obtained when a thermometer was immersed in 100 mL of deionized water in a Mason jar placed in the center of the griddle.
Microburette or automatic titrator. Titrations were performed using a 5-mL microburette or a Metrohm Model 678 EP/KF Processor equipped with a Model 665 Dosimat (Metrohm, Herisau, Switzerland) and a combination electrode designed for flat-surface measurements (Model 13-620-289; Fisher Scientific, Pittsburgh, PA).
Reagents
Sodium hydroxide solution (2 M). Reagent-grade NaOH pellets (80 g) were dissolved in approximately 800 mL of deionized water in a 1-L volumetric flask. After cooling, the solution was diluted to 1 L and mixed thoroughly. The flask was kept tightly stoppered to minimize absorption of atmospheric CO2 during storage of the NaOH solution. Alternatively, 2 M NaOH is available from Fisher Scientific (cat. no. LC24380).
Boric acid-indicator solution. This reagent was obtained from Fisher Scientific (cat. no. LC11750).
Dilute sulfuric acid (0.01 M standard). This reagent was prepared
by adding 5.6 mL of concentrated (18 M) H2SO4 to
10 L of deionized water in a 10-L Pyrex solution bottle. After thorough mixing
with a motorized stirrer, the solution was standardized by titrating several
5-mL aliquots of a THAM solution that was prepared by dissolving 0.2430 g of
dried, certified THAM (Sigma, St. Louis, MO) in 100 mL of deionized water in
a volumetric flask. The endpoint for these titrations was determined as described
in the procedure that follows. The molarity of the titrant was calculated as
0.025/V, where V is the mean value for the milliliters of H2SO4 required
to reach the endpoint. The calculated molarity was multiplied by 28000 to obtain
the titer (µg N mL-1). Alternatively,
standard 0.01 M (0.02 N) H2SO4 may
be purchased from Fisher Scientific (cat. no. SA226).
Procedure
A 1-g sample of air-dried soil (< 2 mm) was weighed into a Mason jar. A
petri dish was attached to the jar lid with a cable tie, and 5 mL of H3BO3-indicator
solution was dispensed into the dish. The soil sample was then treated with
10 mL of 2 M NaOH, and the jar was swirled to mix the contents, while
taking care to minimize soil adherence to the wall of the jar. Within 15 to
30 s, the lid was placed on the jar and sealed with a screw band, and the jar
was transferred to the hot plate. After 5 h, the jar was removed from the hot
plate and opened, and the petri dish was released from the jar lid. The H3BO3 solution
was diluted with 5 mL of deionized water, and subsequently titrated with 0.01 M H2SO4.
Prior to titration, 5 mL of H3BO3 solution
was dispensed into a petri dish containing 5 mL of deionized water, and the
endpoint was established on the basis of the resulting color (for manual titrations)
or pH (for automatic titrations). The micrograms of N liberated by diffusion
was calculated as S x T, where S is the volume of H2SO4 used
in titration of the sample and T is the titer of the titrant (for 0.01 M H2SO4, T =
280 µg N mL-1).
To optimize speed and convenience while maintaining adequate sensitivity and resolution, comparative studies were done involving different concentrations of NaOH (1, 2, or 5 M) and different diffusion periods (1-24 h), using the responsive and nonresponsive soils that had the highest and lowest concentrations of amino sugar-N. Studies were also conducted to ascertain whether soil N-test values are affected by sampling depth or aggregate size.
Data from replicate determinations were characterized by computing means and standard deviations, and mean values were compared on the basis of a least significant difference (LSD) at the 0.001 probability level. Simple correlation or regression analyses were performed to quantify the relationship of soil N-test values to soil chemical properties, check-plot yield, and N-fertilizer response.
Nitrogen fertilization is generally considered essential for corn production,
although a profitable yield response is not always obtained. This was the case
with 12 of the 25 soils used in the present study, despite the absence of an
obvious factor that may have limited yield, such as drought, soil acidity,
or a deficiency of P or K. Of this group, eight soils had been manured, but
the rate of manure-N application varied from 88 to 2240 lb acre-1 and
far exceeded crop N requirements at three sites that had been used for manure
disposal (soils 1, 2, and 5). One of the remaining nonresponsive soils was
from a site that had been cropped to alfalfa prior to corn (no. 3), and the
lack of response can be attributed to mineralization of alfalfa residue (El-Hout
and Blackmer, 1990; Bundy and Andraski, 1993; Schmitt and Randall, 1994).
Soil testing for NO3-, either
before (PPNT) or after (PSNT) planting, is currently considered the best option
for identifying sites where N fertilization will be ineffective in producing
a yield response (Bundy and Meisinger, 1994). Both the PPNT and the PSNT were
performed by Brown et al. (1993), and the data thereby obtained for the 25
soils studied in our work are included in Table 1.
Assuming a critical value of 16 mg N kg-1 for
the PPNT (Schmitt and Randall, 1994) and 21 mg N kg-1 for
the PSNT (Fox et al., 1989; Bundy and Andraski, 1993), both tests correctly
identified all 13 of the responding soils but often failed with the nonresponsive
soils. Examination of the data in Table 1 reveals
that, of the latter group, a lack of response was detected in four of 12 cases
by the PPNT, and in six of 12 cases by the PSNT. Interestingly, neither test
was effective with the majority of manured sites, including one used to dispose
of liquid swine manure (soil 2). The presence of available N from alfalfa (soil
3) was detected by the PSNT, but not by the PPNT.
The failure of the PPNT and PSNT in identifying many of the nonresponsive sites listed in Table 1 is no doubt due to the transient nature of mineral N in soils. Nitrate concentrations, for example, depend on numerous N-cycle processes, including mineralization, immobilization, nitrification, denitrification, leaching, and plant uptake. As a result, soil NO3- levels tend to be highly dynamic in a humid region, and therefore a one-time test such as the PPNT or PSNT is apt to be of little value for predicting crop N availability throughout the growing season.
Given the wide variety of soil types and management practices represented by the 25 samples studied in our work, there was good reason to expect a difference in their content and distribution of hydrolyzable N. This is exactly what was observed, as the data (Table 2) show a fivefold range in total hydrolyzable N, a seventeenfold range in amino acid-N, a threefold range in hydrolyzable NH4+-N, and an elevenfold range in amino sugar-N. For most of these fractions, the highest value was obtained with the most heavily manured soil that also had the highest content of organic C and total N (no. 5), and the lowest value was obtained with an unmanured soil that had the lowest content of organic C and total N (no. 25).
If a particular form of soil organic N is highly labile, then the concentration
of this form should be inversely related to crop responsiveness to N fertilization,
and a distinct difference should exist between responsive and nonresponsive
soils. Based on the data in Table 2, total hydrolyzable
N, amino acid-N, and hydrolyzable NH4+-N
cannot be employed to estimate soil N availability, since a considerable amount
of overlap occurred between data for responsive and nonresponsive soils. In
contrast, complete resolution was achieved on the basis of amino sugar-N, as
was initially reported by Mulvaney et al. (2000). The lowest value for any
nonresponsive soil exceeded the highest value for any responsive soil by more
than 30 percent, and on average, the difference was more than threefold.
Although responsive and nonresponsive soils can be distinguished on the basis of hydrolyzable amino sugar-N, the techniques involved are too laborious and time-consuming for use in routine soil testing. These limitations arise largely from the need to prepare a neutralized soil hydrolysate, whereas the analysis to determine hydrolyzable (NH4+ + amino sugar)-N is readily carried out by diffusion with strong alkali, which effects chemical deamination of free amino sugars (Mulvaney and Khan, 2001). Based on previous work to develop direct-diffusion methods for inorganic-N analysis of soil (Khan et al., 2000b), a soil test was developed to estimate amino sugar-N, in which alkalization is performed directly on the soil itself, rather than on a soil hydrolysate.
Heating is essential to promote the alkaline decomposition of amino sugars
when diffusions are performed to recover hydrolyzable (NH4+ +
amino sugar)-N (Mulvaney and Khan, 2001), so a hot plate is used in the soil
test described to provide a temperature of 48 to 50°C. A higher temperature
would be more effective in liberating amino sugar-N, but would reduce the capacity
of H3BO3-indicator
solution for absorption of gaseous NH3 (Khan
et al., 1997).
To optimize reaction conditions for liberating amino sugar-N from soil, studies were conducted to compare different concentrations of NaOH and different diffusion periods. The results are reported by Fig. 1, which shows data obtained with the two responsive and the two nonresponsive soils that had the highest and lowest concentrations of amino sugar-N. As expected, a larger amount of N was liberated when diffusion was performed with a higher concentration of NaOH or for a longer period. A 2 M reagent and a 5-h diffusion period were adopted in the soil test described as the best compromise in terms of speed, convenience, sensitivity, and resolution. A longer diffusion period was required with 1 M NaOH to clearly differentiate responsive from nonresponsive soils, whereas 5 M NaOH is a less convenient and more expensive reagent, and the diffusion period was much more critical for correct interpretation of soil-test values. With 2 M NaOH, a 5-h diffusion period was adequate to easily detect nonresponsive soils, and even provided sufficient resolution to clearly distinguish among all four soils. Moreover, the data in Fig. 1 show that soil-test values with 2 M NaOH do not increase appreciably if diffusion is continued beyond the 5-h period specified, although this practice should be avoided, since prolonged heating may lead to drying of the H3BO3 solution used to absorb gaseous NH3 and thereby vitiate the analysis. The latter problem does not occur if heating is discontinued after 5 h, as very little, if any, change has been observed in soil-test values by leaving the jar unopened overnight at room temperature (25°C). This could be a valuable option in processing a large number of soil samples, as is often necessary in soil testing laboratories.
Besides amino sugar-N, the soil test described recovers any exchangeable NH4+-N
that may be present, and is therefore of no value if employed following ammoniacal
fertilization. Nitrate-N is not recovered, so as to ensure that test values
will be unaffected by temporal variability in soil NO3- concentrations.
Soil sampling for routine assessment of pH, P, and K is normally done to a depth of six or seven inches (15-18 cm). The soil samples used in our work were collected from the surface foot (30 cm); however, a study to compare N-test values for different profile depths (Table 3) showed that the highest values were obtained for the surface 15 cm (6 inches), and that a decrease occurred with greater depth. This is exactly what would be expected for an organic fraction of soil N, whereas profile sampling to 30 or 60 cm is necessary with NO3- tests to account for leaching. Table 3 suggests that standard sampling techniques would be appropriate for the soil test described, and that testing could be done with any tillage system.
The soil samples used in the present project were much more finely ground than samples for routine soil testing, which usually involves screening to < 2 mm to remove organic detritus or inert rock fragments. Preliminary work to compare different aggregate sizes showed that fine grinding (to < 0.15 mm) had no appreciable effect on soil test values (data not reported), which can be attributed to the fact that soil aggregates quickly form a slurry when treated with 10 mL of 2 M NaOH. This suggests that the soil test described will require no special processing of soil samples.
The validity of the soil test described was evaluated using the complete set
of 25 soil samples, 12 of which had been correctly identified as being nonresponsive
to N fertilization on the basis of hydrolyzable amino sugar-N. The results
(Table 4) show that higher values were obtained for
the latter group than for the 13 responsive soils, and that the two groups
were completely resolved, even at the 0.001 probability level. Table
4 leaves no doubt that nonresponsive soils can be successfully detected
by a simple soil test that eliminates the time-consuming steps involved in
acid hydrolysis. Moreover, comparison of Table 2 and Table
4 reveals that soil test values tended to follow the same order as analyses
for hydrolyzable amino sugar-N. The close quantitative relationship between
these parameters is confirmed by Fig. 2, which shows a very high correlation
between soil test-N and amino sugar-N ® = 0.90***).
Besides providing an unprecedented capability to detect sites where there is no need for N fertilization, the soil test described is unsurpassed in simplicity and convenience. No specialized or expensive equipment is required, and all of the reagents are commercially available at reasonable cost. The test procedure is performed directly on the soil sample, without the need for extraction, as is required in conventional soil tests for available P or K. This will be an important advantage in processing large numbers of samples in routine soil testing.
A simple soil test was developed to estimate amino sugar-N, as a means of
identifying sites where corn does not respond to N fertilization. This test
avoids the need for acid hydrolysis or chemical extraction, and is ideally
suited for routine soil testing. By design, NO3--N
is not recovered, so as to reduce soil test variability and eliminate the need
for profile sampling and special care in sample processing. When applied to
25 Illinois soils that differed widely in properties and management, higher
test values were obtained for 12 nonresponsive than for 13 responsive soils,
and the two groups were completely resolved at the 0.001 level of significance.
On the basis of soil test-N, Fig. 3 shows that all 25 soils were classified
correctly as responsive (< 225 mg kg-1)
or nonresponsive (> 225 mg kg-1) to N
fertilization.
Table 1. Chemical properties of soil from N-response sites.
Table 2. Concentrations of hydrolyzable N in soil from study sites.
Table 3. Effect of profile depth on soil N-test values.
Table 4. N-test values for soils differing in N-fertilizer responsiveness.
Figure 1. Effectiveness of different additions of NaOH for determination of available soil N during diffusion at 48-50°C.
Figure 2. Relationship between soil test-N and hydrolyzable amino sugar-N for surface (0-30 cm) samples of 25 Illinois soils.
Figure 3. Relationship between soil test-N and N-fertilizer response for surface (0-30 cm) soils from 25 N-response sites.
1S.A. Khan is a Research Specialist in Agriculture and R.L. Mulvaney is a Professor, Department of Natural Resources and Environmental Sciences, University of Illinois. R.G. Hoeft is a Professor, Department of Crop Sciences, University of Illinois, Urbana, IL.
Brown, H.M. 1996. Evaluation of nitrogen availability indices. Ph.D. thesis, University of Illinois, Urbana-Champaign, IL.
Brown, H.M., R.G. Hoeft, and E.D. Nafziger. 1993. Evaluation of three N recommendation systems for corn yield and residual soil nitrate. In: 1993 Illinois Fertilizer Conference Proceedings (R.G. Hoeft, ed.). pp. 43-49.
Bundy, L.G., and T.W. Andraski. 1993. Soil and plant nitrogen availability tests for corn following alfalfa. Journal of Production Agriculture, 6:200-206.
Bundy, L.G., and E.S. Malone. 1988. Effect of residual profile nitrate on corn response to applied nitrogen. Soil Science Society of America Journal, 52:1377-1383.
Bundy, L.G., and J.J. Meisinger. 1994. Nitrogen availability indices. In: Methods of Soil Analysis. Part 2. Biological Methods (R.W. Weaver et al., ed.). SSSA Book Ser. 5. Soil Science Society of America, Madison, WI. pp. 951-984.
Dahnke, W.C., and G.V. Johnson. 1990. Testing soils for available nitrogen. In: Soil Testing and Plant Analysis. 3rd edn. (R.L. Westerman et al., ed.). SSSA Book Ser. 3. Soil Science Society of America, Madison, WI. pp. 127-139.
El-Hout, N.M., and A.M. Blackmer. 1990. Nitrogen status of corn after alfalfa in 29 Iowa fields. Journal of Soil and Water Conservation, 45:115-117.
Fox, R.H., G.W. Roth, K.V. Iversen, and W.P. Piekielek. 1989. Soil and tissue nitrate tests compared for predicting soil nitrogen availability to corn. Agronomy Journal, 81:971-974.
Harrington, P.D., E.D. Nafziger, and R.G. Hoeft. 1997. On-farm evaluation of variable-rate nitrogen fertilizer responses using farmer-owned equipment. In: 1997 Illinois Fertilizer Conference Proceedings (R.G. Hoeft, ed.). pp. 57-64.
Illinois Agronomy Handbook 1999-2000. 1998. University of Illinois, Urbana-Champaign, IL. pp. 93-94.
Khan, S.A., R.L. Mulvaney, and R.G. Hoeft. 2000a. Mineralization potential and changes in hydrolyzable nitrogen for soils that differ in nitrogen fertilizer responsiveness. Agronomy Abstracts. pp. 285.
Khan, S.A., R.L. Mulvaney, and R.G. Hoeft. 2000b. Direct-diffusion methods for inorganic-nitrogen analysis of soil. Soil Science Society of America Journal, 64:1083-1089.
Khan, S.A., R.L. Mulvaney, and C.S. Mulvaney. 1997. Accelerated diffusion methods for inorganic-nitrogen analysis of soil extracts and water. Soil Science Society of America Journal, 61:936-942.
Magdoff, F., D. Ross, and J. Amadon. 1984. A soil test for nitrogen availability to corn. Soil Science Society of America Journal, 48:1301-1304.
Meisinger, J.J., V.A. Bandel, J.S. Angle, B.E. O=Keefe, and C.M. Reynolds. 1992. Presidedress soil nitrate test evaluation in Maryland. Soil Science Society of America Journal, 56:1527-1532.
Mulvaney, R.L. 1996. Nitrogen – Inorganic forms. In: Methods of Soil Analysis. Part 3. Chemical Methods (D.L. Sparks et al., ed.). SSSA Book Ser. 5. Soil Science Society of America, Madison, WI. pp. 1123-1184.
Mulvaney, R.L., and S.A. Khan. 2001. Diffusion methods to determine different forms of nitrogen in soil hydrolysates. Soil Science Society of America Journal (in press).
Mulvaney, R.L, S.A. Khan, and R.G. Hoeft. 2000. A soil organic nitrogen fraction that reduces the need for nitrogen fertilization. In: 2000 Illinois Fertilizer Conference Proceedings (R.G. Hoeft, ed.). pp. 21-33.
Mulvaney, R.L., S.A. Khan, W.B. Stevens, and C.S. Mulvaney. 1997. Improved diffusion methods for determination of inorganic nitrogen in soil extracts and water. Biology and Fertility of Soils, 24:413-420.
SAS Institute. 1993. SAS User's Guide: Statistics. SAS Institute, Cary, NC.
Schmitt, M.A., and G.W. Randall. 1994. Developing a soil nitrogen test for improved recommendations for corn. Journal of Production Agriculture, 7:328-334.
