The ISNT was initially developed as a means to identify

Evaluation of the Illinois Soil Nitrogen Test in the North Central Region of the United States C. A. M. Laboski,* J. E. Sawyer, D. T. Walters, L. G. Bundy, R. G. Hoeft, G. W. Randall, and T. W. Andraski Nitrogen Management ABSTRACT Recently the Illinois soil nitrogen test (ISNT) was proposed as a means to identify fields where corn (Zea mays L.) will not respond to additional N fertilizer and which may also be used to predict the economic optimum N rate (EONR). Data from 96 corn N rate response trials across Iowa, Illinois, Michigan, Minnesota, Nebraska, and Wisconsin were compiled to evaluate the usefulness of the ISNT in identifying nonresponsive fields, predicting EONR, and estimating mineralizable N. At each trial site, multiple rates of fertilizer N were applied, including zero N and nonyield limiting rates. Corn was grown following several crops. The ISNT could not accurately predict nonresponsive sites, nor could it reliably estimate EONR. Subsetting the data based on soil drainage class and previous crop did not improve the predictive capability of the ISNT even though ISNT values were significantly different among previous crops and soil drainage classes. The ISNT was strongly correlated to soil organic matter (OM) and was apparently measuring a constant fraction of total soil N (TN). The lack of correlation between the ISNT and relative N uptake (check plot N uptake/n uptake at the maximum N rate) suggests that the ISNT is not measuring the readily mineralizable fraction of soil N. Based on results of this project, the ISNT is not suggested for use in adjusting N rate recommendations for corn in the North Central Region (Corn Belt) of the United States. The ISNT was initially developed as a means to identify fields where corn would not respond to the addition of N fertilizer (Khan et al., 2001). The ISNT was developed by Mulvaney and Khan (2001) as a simplified version of a diffusion technique that determines different forms of N in soil hydrolysates. Based on the analysis of archived soil samples collected from N rate trials, Mulvaney et al. (2001) found that N fertilizer response in corn was related to soil amino-sugar N. They also found that as amino-sugar N increased, corn N fertilizer response decreased to zero and remained nonresponsive to the application of N fertilizer above a threshold amino-sugar N value. The ISNT was subsequently shown to be strongly correlated to amino-sugar N (Khan et al., 2001). Favorable characteristics of the ISNT that could aid in the adoption of the test are that soil samples can be taken from the 0- to 15-cm soil at the same time as routine soil sampling (Khan et al., 2001) and samples can be taken in the fall or spring before applying N (Barker et al., 2006a; Hoeft et al., 2001). C.A.M. Laboski, L.G. Bundy, and T.W. Andraski, Dep. Soil Sci., Univ. of Wisconsin-Madison, 1525 Observatory Dr., Madison, WI 53706; J.E. Sawyer, Dep. of Agronomy, 2104 Agronomy Hall, Iowa State Univ., Ames, IA 50011; D.T. Walters, Dep. of Agronomy, Univ. of Nebraska, 261 Plant Sci. Building, Lincoln, NE 68583; R.G. Hoeft, Dep. of Crop Sci., Univ. of Illinois at Urbana-Champaign, 1102 S. Goodwin Ave., Urbana, IL 61801; G.W. Randall, Univ. of Minnesota, Southern Research and Outreach Center, 35838 120th St., Waseca, MN 56093. Received 23 Aug. 2007. *Corresponding author (laboski@wisc.edu). Published in Agron. J. 100:1070 1076 (2008). doi:10.2134/agronj2007.0285 Copyright 2008 by the American Society of Agronomy, 677 South Segoe Road, Madison, WI 53711. 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. Other researchers have since evaluated the ISNT in other geographic regions. The ISNT was shown to predict potentially mineralizable N (R 2 = 0.51) in a 24-wk laboratory incubation (Sharifi et al., 2007) using soils from New Brunswick, Quebec, Manitoba, Saskatchewan, and Maine. However, total soil N (TN) and total soil organic C along with NaHCO 3 extractable-n and NaOH direct distillation-n were more highly correlated (R 2 ranged from 0.60 0.74) with potentially mineralizable N than was the ISNT. Klapwyk and Ketterings (2006) compared the ability of the ISNT and pre-sidedress nitrate test (PSNT) to predict sidedress N response in corn silage production in New York. They confirmed that the PSNT is a good predictor of sidedress N response and found that ISNT alone was not. However, ISNT combined with OM concentration did provide an adequate predictor of the probability of sidedress N response. Klapwyk and Ketterings (2006) concluded that ISNT critical values above which no N response is expected should be adjusted for soil OM. In North Carolina, Williams et al. (2007b) found that the ISNT could predict economic optimum N rate (EONR), delta yield (ΔY, yield at the yield maximizing N rate zero N control yield), and N fertilizer response in corn for mineral soils only. In a followup study, the ability of the ISNT to predict EONR increased substantially when sites were separated into well or poorly drained classes (Williams et al., 2007a). Very poorly drained soils, which had the highest ISNT values with approximately average Abbreviations: ΔY, delta yield; EONR, economic optimum nitrogen rate; ISNT, Illinois soil nitrogen test; NFR, nitrogen fertilizer response; OM, soil organic matter; PPN, nitrate N concentration in the 0- to 30-cm soil at preplant sampling; PSN, nitrate N concentration in the 0- to 30-cm soil at pre-sidedress sampling; RU N, relative nitrogen uptake; RY, relative yield; TN, total soil nitrogen; U N0, nitrogen uptake with no nitrogen applied; Y EONR, yield at the economic optimum nitrogen rate; Y N0, yield with no nitrogen applied; Y YMNR, yield at the yield maximizing nitrogen rate; YMNR, yield maximizing nitrogen rate. 1070 Agronomy Journal Volume 100, Issue 4 2008

EONR, were excluded from the analysis because the authors felt that the high humic matter concentrations might be affecting the ISNT threshold levels. Because both studies (Williams et al., 2007a, 2007b) lacked nonresponsive sites, the validity of the ISNT critical level proposed by Khan et al. (2001) could not be evaluated. In contrast, separate studies in Iowa (Barker et al., 2006b) and Wisconsin (Osterhaus et al., 2008) found that the ISNT was not predictive of the EONR for corn and could not be used to separate responsive from nonresponsive sites. In fact, in these studies, using the ISNT critical threshold proposed by Khan et al. (2001) would result in many responsive sites being categorized as nonresponsive and many nonresponsive sites being declared responsive. Both of these studies found that the ISNT was measuring a constant fraction of total soil N (TN) and was not sensitive to mineralizable N. Additionally, Mulvaney et al. (2005) reported data from 102 N response studies conducted in Illinois in 1990 to 1992 and 2001 to 2003. They found that the ISNT had an overall failure rate of 20.6% and incorrectly identified 6.0% of the nonresponsive sites (nonresponsive sites were identified as being responsive, TYPE B failures, results in overapplication of N fertilizer) and 27.5% of the responsive sites (responsive sites were identified as nonresponsive, TYPE A failures, results in yield loss from underfertilization). Mulvaney et al. (2005) hypothesized that these failures may have resulted from a limitation to soil N mineralization or crop N utilization caused by moisture stress, weed competition, soil fertility limitation, quality/quantity of carbon inputs, or corn plant population. However, they did acknowledge that experimental data to substantiate these hypotheses was not collected. Incorrectly classifying responsive sites as nonresponsive could have a large negative economic impact to farmers as yield losses from under fertilization would have occurred, and would result in their losing confidence in the test. Fully understanding factors affecting ISNT performance is essential to providing farmers with criteria for successful use of the test. The objectives of this paper were to use data on corn yield response to applied N from response trials across the North Central Region of the United States to assess the effectiveness of the ISNT for: (i) predicting fields where corn will not respond to additional N fertilizer; (ii) estimating optimum N fertilizer application rates; and (iii) estimating mineralizable soil N. MATERIALS AND METHODS Data were compiled from 96 N rate studies in Iowa, Illinois, Michigan, Minnesota, Nebraska, and Wisconsin that were conducted from 2001 to 2004 as part of the regional CSREES NC-218 project, Assessing Nitrogen Mineralization and Other Diagnostic Criteria to Refine Nitrogen Rates for Crops and Minimize Losses. Results of some of these studies have been published or reported by Osterhaus et al. (2008), Barker et al. (2006b), and Laboski (2004). Summary information such as previous crop, manure history, soil texture, and drainage class for each site is provided in Table 1. Nitrogen rate trials consisted of either small plots or field strips where N fertilizer was applied at multiple rates, including a zero N rate and several rate increments up to a nonyield limiting rate (typically 202 375 kg N ha 1 depending on the rotation, environment, and soil series). Each rate was replicated four times. Nitrogen fertilizer was applied as anhydrous ammonia, urea ammonium nitrate, or urea at preplant, sidedress, or split (starter plus sidedress or preplant plus sidedress). All N applications were made to minimize potential N losses. Sites were either on experimental research farms or farmer s fields. Corn was grown using locally adapted cultural practices to optimize yield. One composite soil sample (20 cores per sample) was collected per replication before planting at s of 0- to 15, 15- to 30, and/or 0- to 30-cm and in late spring before sidedress N application to a of 30 cm. However, samples were not collected from all sampling s at each site. Preplant soil samples were analyzed for nitrate N (Brown, 1998), TN (dry combustion), OM (loss on ignition or calculation from dry combustion C analysis), and ISNT (Khan et al., 2001). For all samples, the ISNT analysis was performed at the University of Illinois within a year after sample collection to eliminate any potential lab to lab variation in ISNT values. Soil samples taken in late spring were analyzed for nitrate N. Pertinent soil data for each site are provided in Table 2. At physiological maturity, final plant populations were counted and whole aboveground plant samples were collected and analyzed for total N. Plant samples were not collected at every site or in all plots at a given site. At sites where plant uptake was determined, relative N uptake (RU N ) was calculated as the N uptake in the zero N control treatments (U N0 ) divided by the N uptake in the highest N rate treatment. Corn grain yield and moisture were measured and grain yield was adjusted to 155 g kg 1 moisture content. Analysis of variance was performed to determine N rate treatment effects on grain yield. Corn yield response to applied N was fit to either linear, linear plateau, quadratic, quadratic plateau, or spherical models. The model with the best R 2 for each site was chosen by each respective researcher to represent the yield response. The yield maximizing N rate (YMNR) was determined for each site using the response model and is the N rate where yield was maximized. Delta yield (ΔY) was calculated as Y YMNR yield at zero N (Y 0N ). Relative yield (RY) was calculated as the Y 0N divided by Y YMNR. Nitrogen fertilizer response (NFR) was calculated as ΔY divided by Y 0N. The EONR and yield at the EONR (Y EONR ) were calculated at the first derivative for each response model using $0.14 kg 1 ($0.30 lb 1 ) N fertilizer and $0.0243 kg 1 ($3.00 bu 1 ) corn grain. The Y 0N, YMNR, yield at the YMNR (Y YMNR ), EONR, and Y EONR for each site are provided in Table 2. All statistical analyses were computed in SAS 9.1 (SAS Institute, 2002). For correlation and regression analyses, statistical significance was determined at a maximum probability level of 0.01, unless specifically stated otherwise, because the number of observations in the various analyses was large (n ranged from 49 96). RESULTS AND DISCUSSION Effectiveness of the Illinois Soil N Test for Predicting N Fertilizer Response and Rate Requirement The correlation between ISNT measured at the 0- to 15-cm and 0- to 30-cm soil s was very strong (r = 0.98, P < 0.001), substantiating the conclusions of Khan et al. (2001) Agronomy Journal Volume 100, Issue 4 2008 1071

Table 1. Sites, soil characteristics, tillage, and crop history, 2001 to 2004. Irrigated, Previous Soil Drainage tile crop# Site State Year texture class drained Tillage 1 yr 2 yr 1 IA 2001 cl p ND CP S C 2 IA 2001 l mw ND CP S C 3 IA 2001 sicl p ND CP S C 4 IA 2001 cl sp ND CP S C 5 IA 2001 sicl p ND CP S C 6 IA 2001 l mw ND MB S C 7 IA 2001 sicl sp ND CP S C 8 IA 2001 sicl w ND MB S C 9 IA 2001 sicl w ND CP S C 10 IA 2001 sicl w ND CP S C 11 IA 2001 sicl p ND CP S C 12 IA 2001 sicl mw ND CP S C 13 IA 2001 sicl sp ND NT S C 14 IA 2001 sicl p ND MB S C 15 IA 2002 sl w ND CP S C 16 IA 2002 sicl p ND CP S C 17 IA 2002 sicl w ND NT S C 18 IA 2002 sicl w ND ST S C 19 IA 2002 sicl w ND CP S C 20 IA 2002 sicl p ND CP S C 21 IA 2002 sil w ND NT S C 22 IA 2002 sil w ND NT S C 23 IA 2002 sicl w ND CP S C 24 IA 2002 sicl w ND CP S C 25 IA 2002 sicl sp ND NT S 26 IA 2003 l mw ND CP S C 27 IA 2003 sicl p ND CP S C 28 IA 2003 sicl w ND ST S C 29 IA 2003 sicl p ND CP S C 30 IA 2003 l mw ND CP S C 31 IA 2003 sil w ND CP S 32 IA 2003 sicl p ND NT S C 33 IA 2003 sil mw ND NT S C 34 IA 2003 sil w ND NT S C 35 IA 2003 sicl mw ND MB S C 36 IA 2003 sicl sp ND CP S C 37 IA 2003 sil p ND MB S C 38 IA 2003 sicl p ND CP S C 39 IA 2003 sicl w ND NT S C 40 IA 2003 sicl mw ND NT S C 41 IA 2003 sicl mw ND CP S C 42 IA 2003 sil w ND CP S C 43 IA 2003 sicl p ND CP S C 44 IL 2003 sil sp T CP S C 45 IL 2003 sil sp T NT S C 46 IL 2003 sicl p T CP S A 47 IL 2003 sil sp T CP S C 48 IL 2003 sil sp T CP C C 49 MI 2002 sl sp N CP S C 50 MI 2002 sicl p T MB S S 51 MI 2003 sl sp N CP S C 52 MI 2003 sicl p T MB DB W 53 MN 2002 sil w N MT S C (Continued next column.) and Barker et al. (2006a). This indicates that conclusions which may be drawn based on one sampling are applicable to the other sampling as well. Thus, statistical analysis of all soil test data presented here are based on the 0- to 15-cm soil because there are more sites with data for all analyses from that. The exceptions to this are the preplant and pre-sidedress samples for nitrate N (PPN and PSN, respectively) which were collected from the 0- to 30-cm, and comparisons of ISNT for the 0- to 30-cm soil with RY. (Table 1, Continued.) Irrigated, Previous Soil Drainage tile crop# Site State Year texture class drained Tillage 1 yr 2 yr 54 NE 2002 sl w I RT C C 55 NE 2002 sl w I MT DB C 56 NE 2002 ls mw I NT S C 57 NE 2002 sil w I RT C C 58 NE 2002 sil w I RT C C 59 NE 2002 sil w I NT S C 60 NE 2002 sicl w I NT S C 61 NE 2002 sil w I RT C C 62 NE 2002 ls w I MT C C 63 NE 2002 sil mw I MT S C 64 NE 2002 sil w I MT DB C 65 NE 2002 sicl mw I NT S C 66 NE 2003 sl w I RT C C 67 NE 2003 sl w I MT DB C 68 NE 2003 sil w I NT S C 69 NE 2003 sil w I RT C C 70 NE 2003 sil w I NT S C 71 NE 2003 sicl w I NT S C 72 NE 2003 sil w I RT C C 73 NE 2003 ls w I MT C C 74 NE 2003 sicl mw I NT S C 75 NE 2003 sil mw I MT S C 76 NE 2003 sil w I MT DB C 77 NE 2004 sl w I MT DB C 78 NE 2004 sl w I MT C C 79 NE 2004 ls mw I MT S C 80 NE 2004 sil w I RT C C 81 NE 2004 sil mw I NT S C 82 NE 2004 sil w I RT C C 83 NE 2004 sicl w I NT S C 84 NE 2004 sicl w I MT S C 85 NE 2004 sicl mw I NT S C 86 NE 2004 sil sp I MT S C 87 NE 2004 ls w I MT C C 88 WI 2002 sil w N CP C C 89 WI 2002 sil w N CP A A 90 WI 2002 sil w N CP C A 91 WI 2002 sil w N NT S C 92 WI 2003 sil w N CP C C 93 WI 2003 sil w N NT S C 94 WI 2004 sil w N CP C C 95 WI 2004 sil w N NT S C 96 WI 2004 sil w N MB C C cl, clay loam; l, loam; ls, loamy sand; sl, sandy loam; sicl, silty clay loam; sil, silt loam. p, poorly drained; sp, somewhat-poorly drained; mw, moderately-well drained; w, well drained. I, irrigated; T, tile drained; N, not irrigated and not tile drained; ND, not irrigated and tile drainage not determined. CP, chisel plow; NT, no-till; MB, moldboard plow; MT, mulch tillage; RT, ridge tillage; ST, strip tillage. # Previous crop 1 and 2 yr before the study year. A, alfalfa [Medicago sativa L.]; C, corn; DB, dry bean; S, soybean; W, wheat. Manure applied 1 yr before study year. A correlation matrix was initially used to explore whether any soil test (ISNT, OM, TN) was related to various measures of yield (Y 0N, Y YMNR, Y EONR ), crop response to additional N fertilizer (RY, ΔY, NFR), and crop fertilizer N requirements (YMNR, EONR). Both ISNT and OM were significantly (P < 0.001) correlated to YMNR, Y YMNR, EONR, and Y EONR with r values ranging from 0.32 to 0.41 (Table 3) but were not significantly correlated to RY, ΔY, or NFR. Soil TN was not significantly correlated to any measure of yield, crop response, or fertilizer N requirement. Lory and Scharf (2003) reported 1072 Agronomy Journal Volume 100, Issue 4 2008

Table 2. Soil ph, organic matter (OM), total nitrogen (TN), Illinois soil nitrogen test (ISNT), preplant soil nitrate-nitrogen (PPN), pre-sidedress soil nitrate-nitrogen (PSN), corn grain yield without nitrogen (Y 0N ), yield maximizing nitrogen rate (YMNR), yield at the YMNR (Y YMNR ), economic optimum nitrogen rate (EONR), and yield at the EONR (Y EONR ) at the 96 sites. 0- to15-cm soil 0- to 30-cm soil Site OM TN ISNT ISNT PPN PSN Y 0N YMNR Y YMNR EONR Y EONR g kg 1 mg kg 1 kg ha 1 1 66 2.75 355 295 9,322 43 10,827 40 10,815 2 44 1.57 284 277 9,253 183 10,683 113 10,485 3 40 1.62 270 244 10,005 0 10,005 0 10,005 4 55 2.14 353 316 7,383 175 11,417 153 11,348 5 92 4.29 605 515 9,893 0 9,893 0 9,893 6 41 1.85 256 247 12,552 73 13,286 52 13,226 7 40 1.58 265 258 7,917 88 11,373 82 11,357 8 51 2.23 336 335 9,378 224 11,160 157 10,886 9 42 1.73 268 256 10,451 44 11,047 35 11,021 10 40 1.60 270 259 10,413 224 11,103 0 10,507 11 41 1.69 259 260 11,404 181 11,975 38 11,566 12 37 1.08 225 206 9,416 0 9,416 0 9,413 13 39 1.25 256 257 8,594 124 13,079 114 13,048 14 54 1.80 332 313 8,776 118 10,413 92 10,341 15 37 1.93 294 250 17 13,888 0 13,888 0 13,884 16 40 2.11 264 248 8 8,625 166 11,208 136 11,121 17 43 2.52 327 307 15 10,746 134 11,919 95 11,795 18 48 2.60 337 314 14 10,896 102 12,596 85 12,547 19 48 2.69 340 307 13 8,594 132 11,197 113 11,146 20 64 3.27 428 416 14 10,056 96 11,869 81 11,826 21 40 2.28 288 277 9 10,909 70 13,572 64 13,556 22 44 2.48 317 267 11 13,594 0 13,594 0 13,589 23 40 2.24 272 266 14 3,293 171 4,377 105 4,189 24 34 2.08 263 252 14 10,520 123 11,988 91 11,889 25 41 2.10 273 246 12 11,354 172 15,074 149 15,006 26 42 2.02 290 280 7,929 179 11,128 151 11,048 27 60 2.67 324 316 11,078 137 14,811 123 14,769 28 51 2.57 327 343 7,019 187 12,490 169 12,439 29 64 2.94 432 402 8,763 141 13,525 129 13,492 30 25 1.40 174 157 8,939 136 13,606 125 13,576 31 27 1.61 245 203 11,160 201 12,722 131 12,518 32 90 4.32 558 569 8,544 0 8,544 0 8,541 33 30 1.67 222 198 10,890 0 10,890 0 10,886 34 31 1.87 239 229 9,297 92 10,645 74 10,596 35 40 1.89 296 282 11,059 0 11,059 0 11,056 36 41 2.11 281 277 10,275 152 13,311 131 13,247 37 40 1.83 241 225 7,245 153 10,545 132 10,486 38 62 2.94 376 303 7,120 199 13,619 181 13,568 39 41 2.27 266 260 5,859 178 8,826 146 8,736 40 34 1.66 223 191 8,043 130 10,764 113 10,715 41 41 2.09 298 264 9,686 121 15,168 113 15,144 42 36 1.99 287 275 9,817 81 12,352 73 12,333 43 54 2.47 343 286 6,505 224 12,314 204 12,246 44 24 1.30 191 174 11 6 1,777 178 10,623 178 10,623 (Continued next column.) (Table 2, Continued.) 0- to15-cm soil 0- to 30-cm soil Site OM TN ISNT ISNT PPN PSN Y 0N YMNR Y YMNR EONR Y EONR g kg 1 mg kg 1 kg ha 1 45 37 1.60 283 260 11 12 6987 110 12,576 103 12,558 46 39 1.50 287 280 16 6 8,458 129 13,799 120 13,773 47 24 1.20 196 180 9 9 8,234 195 13,879 177 13,827 48 37 1.60 332 306 17 13 8,030 102 9,162 78 9,092 49 26 1.78 189 166 4 5 6,183 139 9,578 123 9,532 50 33 1.20 232 211 6 9 6,639 122 10,014 122 10,014 51 32 1.60 241 219 5 7 5,821 164 9,677 143 9,615 52 33 1.70 216 218 7 7 7,556 167 10,770 142 10,700 53 40 2.15 294 278 3 5 7,631 137 10,027 116 9,968 54 15 113 97 1 5 11,282 321 14,634 321 14,634 55 19 165 133 13 8 15,244 0 15,244 0 15,244 56 9 83 69 2 3 9,092 172 13,103 157 13,060 57 26 185 168 7 13 10,969 178 12,872 142 12,766 58 28 193 168 8 7 6,988 193 14,651 184 14,623 59 34 243 243 5 27 9,745 114 11,448 98 11,402 60 34 223 200 6 6 13,318 209 15,478 166 15,350 61 22 120 95 6 13 8,677 242 14,805 223 14,749 62 14 101 87 4 10 13,814 206 15,030 123 14,771 63 29 216 202 12 11 11,716 129 16,150 121 16,128 64 21 90 80 10 140 12,093 143 13,375 108 13,270 65 30 190 154 4 13 6,605 125 11,365 118 11,347 66 13 107 85 12 3 13,440 365 16,828 280 16,571 67 17 95 84 4 8 11,911 194 15,016 170 14,945 68 19 142 131 6 14 10,599 151 15,849 142 15,824 69 26 188 164 7 4 8,646 106 16,802 103 16,794 70 32 239 228 5 14 12,658 190 14,323 142 14,179 71 28 264 250 6 22 8,866 135 14,651 128 14,632 72 23 134 116 10 11 8,790 238 14,289 217 14,229 73 14 101 88 3 5 8,809 170 15,641 162 15,617 74 23 219 209 4 8 7,905 99 13,561 96 13,551 75 34 217 192 11 13 11,672 123 16,486 117 16,467 76 16 108 104 16 10 11,522 258 14,264 208 14,117 77 27 107 103 12 10 12,181 91 12,915 66 12,838 78 12 100 81 5 4 5,638 194 12,006 183 11,972 79 11 118 94 3 8 9,983 237 12,236 183 12,074 80 23 187 172 5 13 10,492 293 16,734 265 16,653 81 27 212 209 10 2 12,137 104 14,849 96 14,826 82 27 187 179 7 10 10,511 170 16,002 160 15,972 83 35 246 243 3 7 9,713 135 15,182 128 15,163 84 30 220 203 5 11 11,446 75 13,798 70 13,784 85 32 217 206 4 7 10,731 150 15,731 141 15,705 86 27 207 176 9 15 10,335 145 15,633 137 15,611 87 18 109 95 9 5 10,687 161 14,003 145 13,957 88 22 1.24 181 136 5 25 3,638 122 7,589 122 7,589 89 24 178 125 11 26 8,593 0 8,593 0 8,593 90 26 193 139 10 23 4,453 101 7,841 93 7,816 91 27 1.16 165 120 5 21 3,261 173 8,803 157 8,758 92 44 2.32 304 292 17 1 7,150 235 11,497 209 11,394 93 36 1.98 258 230 9 1 6,021 211 13,141 193 13,089 94 45 2.32 283 254 18 8 8,593 155 11,831 133 11,770 95 46 2.18 279 261 20 6 8,655 135 11,959 119 11,914 96 35 1.77 202 199 4 6 6,021 211 12,858 211 12,863 Table 3. Correlation (r) matrix of soil test values, total N uptake in corn, and various measures of corn yield, crop response to applied N, and N fertilizer requirement. The number of observations is shown in parentheses. Variable U N0 RU N Y N0 YMNR Y YMNR EONR Y EONR ΔY RY NFR r ISNT 0.37* (51) 0.24 (49) 0.09 (96) 0.39** (96) 0.34** (96) 0.41** (96) 0.33** (96) 0.27 (96) 0.22 (96) 0.16 (96) OM 0.34 (51) 0.19 (49) 0.09 (96) 0.35** (96) 0.33** (96) 0.37** (96) 0.32** (96) 0.25 (96) 0.22 (96) 0.17 (0.96) TN 0.22 (15) 0.29 (13) 0.22 (60) 0.19 (60) 0.04 (60) 0.15 (60) 0.05 (60) 0.20 (60) 0.27 (60) 0.25 (60) * Signifi cant at the 0.01 probability level. ** Signifi cant at the 0.001 probability level. ΔY, delta yield; EONR, economic optimum nitrogen rate; ISNT, Illinois soil nitrogen test; OM, soil organic matter; NFR, nitrogen fertilizer response; RU N, relative nitrogen uptake; RY, relative yield; TN, total soil nitrogen; U N0, nitrogen uptake when no nitrogen fertilizer was applied; Y EONR yield at the economic optimum nitrogen rate; Y N0, yield when no nitrogen fertilizer was applied; YMNR, yield maximizing nitrogen rate; Y YMNR, yield at the yield maximizing nitrogen rate. a strong linear relationship (R 2 = 0.47) between EONR and ΔY and suggested that ΔY might be useful for developing N fertilizer recommendations. In the present study, ΔY was significantly correlated to EONR (r = 0.68). The ISNT was significantly and negatively correlated (r = 0.41, P < 0.001) to EONR but was not significantly correlated to ΔY (r = 0.27). Thus, the ISNT and ΔY apparently explain different types and amounts of variability in EONR between sites. Additionally, the relationship between ISNT and EONR in the present study is not nearly as strong as that found by Williams et al. (2007b) (R 2 = 0.90). The relationship between ISNT and EONR was further explored with linear regression analysis (Fig. 1, Table 3). Though statistically significant, the relationship between ISNT and EONR (R 2 = 0.17) is not strong enough to use the ISNT as a practical means to predict the amount of N needed for a crop as evidenced by a low R 2 and a wide range (often >100 kg N ha 1 ) in EONR for most ISNT values. Additionally, ISNT values above the critical level of 230 mg N kg 1 proposed by Khan et al. (2001) are associated with sites with both zero and high N requirements. Khan et al. (2001) used NFR to develop the ISNT critical level. In the present study, there was no cor- Agronomy Journal Volume 100, Issue 4 2008 1073

Fig. 2. Evaluation of the Illinois soil nitrogen test (ISNT) critical level (230 mg kg 1 ) published by Khan et al. (2001) using the relationship between INST at the 0- to 30-cm and relative yield (RY). TYPE A failures = ISNT > 230 mg kg 1 and RY <90%. TYPE B failures = ISNT 230 mg kg 1 and RY 90%. Fig. 1. Relationship between Illinois soil nitrogen test (ISNT) at the 0- to 15-cm and economic optimum nitrogen rate (EONR) and nitrogen fertilizer response (NFR) for 96 sites. ***Statistically significant at the 0.001 probability level. Not statistically significant at the 0.05 probability level. relation between ISNT and NFR (Fig. 1, Table 3). Even though there were several nonresponsive sites in this data set, no ISNT critical level could be determined that would separate sites with large or small NFR. These findings contradict the work of Khan et al. (2001). Critical levels for soil tests are often determined using the relationship between RY and the soil test value. There is no relationship between RY and ISNT in the present study Table 4. Effect of previous crop and drainage class on mean economic optimum nitrogen rate (EONR), yield at EONR (Y EONR ), delta yield (ΔY), and Illinois soil nitrogen test (ISNT) values. Parameter n EONR Y EONR ΔY ISNT kg ha 1 mg kg 1 Previous crop Alfalfa 1 0 8,593 0 178 Soybean 70 105 b 12,180 3065 b 272 a Dry bean 6 116 b 13,519 1846 b 130 b Corn 19 177 a 13,360 4655 a 175 b P <0.01 0.08 <0.01 <0.01 Drainage class Well 51 130 12,734 3246 215 b Moderately well 17 100 12,865 3051 220 b Somewhat poorly 12 131 12,072 4374 255 b Poorly 16 96 11,447 2759 345 a P 0.15 0.22 0.24 <0.01 Probability level does not include previous crop alfalfa. Values within each column for each parameter followed by the same letter are not signifi cantly different at the 0.05 probability level. (Fig. 2, Table 3). Furthermore, the 230 mg kg 1 ISNT critical level published by Khan et al. (2001) was evaluated for its failure rate with regard to identifying responsive and nonresponsive sites on the basis of RY above and below the critical level. TYPE A failure rates were defined as sites with an ISNT value >230 mg kg 1 and a RY <90%. For all sites with an ISNT value >230 mg kg 1, the TYPE A failure rate was 76.1%. TYPE B failure rates were defined as sites with an ISNT 230 mg kg 1 and a RY 90%. For all sites with an ISNT value 230 mg kg 1, the TYPE B failure rate was 14.0%. For all 96 site-years in this study, there was an overall (TYPE A and TYPE B) failure rate of 43.8%. Mulvaney et al. (2005) found that when the ISNT incorrectly predicted responsive sites (ISNT classified them as nonresponsive), the previous crop was most often soybean [Glycine max L. Merr.]. This observation is also true for the present study. Mulvaney et al. (2005) furthered suggested that a higher ISNT critical level may be needed for corn grown after soybean. In an effort to quantify the effect that previous crop had on ISNT interpretation, ANOVA with Fisher s protected LSD was used to compare the mean EONR, Y ENOR, ΔY, and ISNT values for each previous crop (Table 4). There were no significant (P = 0.08) differences among mean Y EONR for any previous crop. Sites with corn as the previous crop had significantly (P < 0.01) greater EONR and ΔY compared to soybean and dry bean [Phaseolus vulgaris L.]. The average ISNT where soybean was the previous crop was significantly (P < 0.01) greater than the ISNT for corn and dry bean. The EONR decreased in the following order: corn > dry bean = soybean. However, ISNT increased in the following order: dry bean = corn < soybean. Linear regression was used to define the relationship between ISNT and EONR for each previous crop and for the data set as a whole (Table 5). Subsetting the data based on previous crop did not improve the relationship between ISNT and EONR (R 2 values remained low) compared to using the whole data set. Based on our evaluation, the ISNT could not be effectively calibrated for the determination of EONR for corn regardless of previous crop. 1074 Agronomy Journal Volume 100, Issue 4 2008

Table 5. Regression equations for the relationship between Illinois soil nitrogen test (ISNT) at the 0- to 15-cm soil and economic optimum nitrogen rate (EONR) for all data and for subsets of data based on previous crop and drainage class. Parameter n Equation Model P R 2 All data 96 y = 189 0.288x <0.01 0.17 Separated by previous crop Soybean 70 y = 165 0.218x <0.01 0.11 Dry bean 6 y = 161 0.350x 0.66 0.05 Corn 19 y = 237 0.345x 0.12 0.14 Separated by drainage class Well 51 y = 206 0.355x <0.01 0.15 Moderately well 17 y = 195 0.421x 0.07 0.20 Somewhat poorly 12 y = 199 0.269x 0.15 0.20 Poorly 16 y = 181 0.247x 0.09 0.19 x, ISNT (mg kg 1 ); y, EONR (kg N ha 1 ). In research on North Carolina soils, Williams et al. (2007a) found that the ability of the ISNT to predict EONR was improved when the data were separated into soil drainage classes. In the present study, drainage class had no significant effect on EONR, Y EONR, and ΔY; however, the ISNT was significantly (P < 0.01) greater for poorly drained soils compared to all other drainage classes (Table 4). Furthermore, as drainage class became more poorly drained, the ISNT increased, reflecting a greater OM content under more poorly drained conditions. Williams et al. (2007a) also found that more poorly drained soils had greater ISNT values. The relationship between ISNT and EONR was evaluated for the data set as a whole and when subsetted based on drainage class (Table 5). Subsetting the data resulted in marginal change in R 2 values (difference in R 2 of 0.02 to 0.03). Multiple regression was pursued as another means to evaluate the ability of ISNT to predict EONR. A stepwise regression procedure using maximum R 2 and C(p) statistics was used to evaluate multi-parameter models (Freund and Littell, 2000). The best regression models for predicting EONR included two parameters, ISNT and previous crop. Adding previous crop into the model improved R 2 from 0.17 to 0.26. Addition of any other parameter into the model increased R 2 by <0.01. Overall the two parameter model, though significant, does not predict EONR well enough to be useful for making N rate recommendations. Illinois Soil Nitrogen Test as a Measure of Mineralizable Soil Nitrogen Relative N uptake (RU N ) can be considered an in situ index of the relative amount of N that was available to the crop from all sources (net N mineralization, residual inorganic N, precipitation, or irrigation water) other than fertilizer N. Sites with low RU N had high NFR, although this relationship is distinctly not linear (Fig. 3). The RU N was also significantly (P < 0.001) correlated to the other measures of crop responsiveness (ΔY and RY; r = 0.85 and 0.91, respectively), but was not significantly correlated to N fertilizer requirement (EONR or YMNR; r = 0.13 and 0.29, respectively) (data not shown). The ISNT was not significantly correlated to RU N (r = 0.24) and was moderately correlated to U N0 (r = 0.37, P < 0.01) (Table 3, Fig. 4). The amount of net N mineralized and nitrified between the pre-sidedress and preplant soil sampling dates was also not significantly correlated to ISNT (r = 0.26) (data not shown). The negative correlations between ISNT and RU N, U N0, and early season net N mineralization indicate that the ISNT is apparently not measuring a fraction of mineralizable and crop available soil N. The ISNT was found to be strongly correlated to OM (r = 0.96, P < 0.001) across a wide range of OM levels (<10 to >90 g kg 1 ) at the sites studied. Using OM and ISNT data reported by Klapwyk and Ketterings (2006) for soils in New York, we determined that their OM and ISNT values were also strongly correlated (r = 0.95, P < 0.001). In the present study, ISNT is strongly correlated to TN (r = 0.90, P < 0.001) and seems to be measuring a relatively constant fraction of TN (Fig. 5). In work published by Khan et al. (2001) and Klapwyk and Ketterings (2005), the ISNT was also correlated to soil TN (Fig. 5), although those authors did not explore that relationship. The slopes of the linear regression for TN and ISNT for each data set in Fig. 5 are not significantly (P > 0.05) different from each other (data not shown). Based on the slope of the regression lines, the ISNT is measuring 14.7% of TN when all data sets are combined, with a range from 13.5 to 16.4% for the individual data sets. By comparison, the ISNT was 15.0, 13.5, and 12.6% of TN as reported by Barker et al. (2006b), Marriott and Wander (2006), and Osterhaus et al. (2008), respectively. Thus, the ISNT is apparently measuring a constant Fig. 3. Relationship between corn nitrogen fertilizer response (NFR) and relative nitrogen uptake (RU N ) for 49 sites. ***Statistically significant at the 0.001 probability level. Fig. 4. Relationship between relative nitrogen uptake (RU N ) and Illinois soil nitorgen test (ISNT) for 49 sites. Not statistically significant at the 0.05 probability level. Agronomy Journal Volume 100, Issue 4 2008 1075

Fig. 5. Relationship between Illinois soil nitrogen test (ISNT) and total soil nitrogen (TN) for 0- to 15-cm samples. NC-218 data are from the present study. Other data were published by Khan et al. (2001) and Klapwyk and Ketterings (2005). **Statistically significant at the 0.001 probability level. fraction of TN for a wide range of soils rather than the readily mineralizable fraction of soil organic N as would be required to assess the soil s contribution to the available N supply. CONCLUSIONS The ISNT was not a good predictor of ΔY, RY, NFR, Y EONR, or Y YMNR in the North Central Region and did not provide reliable estimates of EONR or YMNR. Additionally, the ISNT was unable to differentiate responsive from nonresponsive sites. The overall failure rate for the ISNT in this 96 site-year study was 43.8% with the majority of the test failures as TYPE A failures where sites predicted as nonresponsive were found to need substantial N fertilizer additions. Subsetting the data based on previous crop or drainage class did not improve the ability of the ISNT to predict EONR. Results of this work indicate that the ISNT is not a useful predictor of N fertilizer need because it is measuring a constant fraction of TN rather than a specific fraction of mineralizable N as evidenced by the lack of correlation between ISNT and RU N. Based on results of this multi-state project, the ISNT is not suggested for use in adjusting corn N rate applications in the North Central Region (Corn Belt) of the United States. ACKNOWLEDGMENTS The authors thank Mr. Bonner Karger for his assistance in creating the database file used to archive the compiled data. We would also like to thank everyone who worked on various aspects of data collection. REFERENCES Barker, D.W., J.E. Sawyer, and M.M. Al-Kaisi. 2006a. Assessment of the amino sugar-nitrogen test on Iowa soils: I. Evaluation of soil sampling and corn management practices. Agron. J. 98:1345 1351. Barker, D.W., J.E. Sawyer, M.M. Al-Kaisi, and J.P. Lundvall. 2006b. Assessment of the amino sugar-nitrogen test on Iowa soils: II. Field correlation and calibration. Agron. J. 98:1352 1358. Brown, J.R. 1998. Recommended chemical soil test procedures for the North Central Region. NC Regional Res. Publ. 221 (Revised). SB1001. Missouri Agric. Exp. Stn., Columbia. Freund, R.J., and R.C. Littell. 2000. SAS system for regression analysis. 3rd ed. SAS Inst., Cary, NC. Hoeft, R.G., R.L. Mulvaney, and S.A. Khan. 2001. The Illinois nitrogen soil test. In 2001 Proc. of the North Central Extension-Industry Soil Fertility Conf., Des Moines, IA. 14 15 Nov. 2001. Potash and Phosphate Inst., Brookings, SD. Khan, S.A., R.L. Mulvaney, and R.G. Hoeft. 2001. A simple soil test for detecting sites that are nonresponsive to nitrogen fertilization. Soil Sci. Soc. Am. J. 65:1751 1760. Klapwyk, J.H., and Q.M. Ketterings. 2005. Reducing analysis variability of the Illinois soil nitrogen test with enclosed griddles. Soil Sci. Soc. Am. J. 69:1129 1134. Klapwyk, J.H., and Q.M. Ketterings. 2006. Soil tests for predicting corn response to nitrogen fertilizer in New York. Agron. J. 98:675 681. Laboski, C.A.M. 2004. Michigan prospects for using the Illinois N soil test. In Proc. 2004 Wisconsin Fert., Aglime, and Pest Management Conf., Madison, WI. 20 22 Jan. 2004. Univ. of Wisconsin, Madison. Lory, J.A., and P.C. Scharf. 2003. Yield goal versus delta yield for predicting fertilizer nitrogen need in corn. Agron. J. 95:994 999. Marriott, E.E., and M.M. Wander. 2006. Total and labile soil organic matter in organic and conventional farming systems. Soil Sci. Soc. Am. J. 70:950 959. Mulvaney, R.L., and S.A. Khan. 2001. Diffusion methods to determine different forms of nitrogen in soil hydrolysates. Soil Sci. Soc. Am. J. 65:1284 1292. Mulvaney, R.L., S.A. Khan, and T.R. Ellsworth. 2005. Need for a soil-based approach in managing nitrogen fertilizers for profitable corn production. Soil Sci. Soc. Am. J. 70:172 182. Mulvaney, R.L., S.A. Khan, R.G. Hoeft, and H.M. Brown. 2001. A soil organic nitrogen fraction that reduces the need for nitrogen fertilization. Soil Sci. Soc. Am. J. 65:1164 1172. Osterhaus, J.T., L.G. Bundy, and T.W. Andraski. 2008. Evaluation of the Illinois soil nitrogen test in Wisconsin cropping systems. Soil Sci. Soc. Am. J. 72:143 150. SAS Institute. 2002. SAS 9.1. SAS Inst., Cary, NC. Sharifi, M., B.J. Zebarth, D.L. Burton, C.A. Grant, and J.M. Cooper. 2007. Evaluation of some indicies of potentially mineralizable nitrogen in soil. Soil Sci. Soc. Am. J. 71:1233 1239. Williams, J.D., C.R. Crozier, J.G. White, R.W. Heiniger, R.P. Sripada, and D.A. Crouse. 2007a. Illinois soil nitrogen test predicts Southeastern U.S. corn economic optimum nitrogen rates. Soil Sci. Soc. Am. J. 71:735 744. Williams, J.D., C.R. Crozier, J.G. White, R.P. Sripada, and D.A. Crouse. 2007b. Comparison of soil nitrogen tests for corn fertilizer recommendations in the humid southeastern USA. Soil Sci. Soc. Am. J. 71:171 180. 1076 Agronomy Journal Volume 100, Issue 4 2008