Illinois Fertilizer
Conference Proceedings
January 26-28, 2004
| Home | 2004 Index | Search |
|
|
| Home | 2004 Index | Search |
R.L. Mulvaney, S.A. Khan, J.J. Warren, L.C. Gonzini, T.J. Smith,
and R.G. Hoeft1
There is growing evidence from N-response studies that yield-based N recommendations are often inaccurate. This was the case, for example, with 75 N-response trials conducted in Illinois between 1990 and 1992, which included 55 sites where the recommended N rate exceeded the optimum N rate determined experimentally, often by more than 100 pounds per acre (Brown, 1996). More recently, Lory and Scharf (2003) concluded that maximum economic yield is a poor predictor of economically optimum N rate for corn, based on a lack of statistical significance in correlating these parameters for 298 N-response experiments in five Midwestern states (r = 0.04). The same conclusion has been reached previously by researchers in Wisconsin (Vanotti and Bundy, 1994; Bundy, 2000), Pennsylvania (Fox and Piekielek, 1995), Ontario (Kachanoski et al., 1996), and Iowa (Blackmer et al., 1997). This lack of correspondence between optimum and predicted N rates can be attributed to the fact that yield-based N recommendations do not take mineralizable soil N into account, with the result that under- or overfertilization is apt to occur in any given growing season.
Recent work at Illinois has implicated amino sugar-N as a specific fraction of soil organic N that reduces corn yield response to N fertilization (Mulvaney et al., 2001). Subsequent work has led to a simple soil test, the so-called Illinois soil N test (ISNT), that estimates amino sugar-N to detect sites where corn is unlikely to respond to N fertilization (Khan et al., 2001). This paper summarizes existing evidence from more than 100 N-response studies in six growing seasons, in which the effectiveness of the new test has been compared to the proven-yield method described in the Illinois Agronomy Handbook (2002), for detecting sites where corn is nonresponsive to N fertilization. Additional objectives of the work reported were to ascertain whether a quantitative relationship exists between N-test values and crop N requirement, and whether the ISNT is sensitive to within-field variability in soil productivity.
The soils used were profile (0–6, 6–12, or 12–24 inches) samples from 105 on-farm N-response trials that represented a wide range in soil type and management practices, each of which utilized a randomized complete block design with four replications. Fifty of these samples were from sites studied by Brown (1996) in 1990 to 1992, and an additional 55 N-response sites were sampled in 2001, 2002, or 2003. In all cases, soil sampling was done with a 1-inch (dia.) soil probe (five cores per sample) in late March or early April prior to planting corn. The soil samples were oven-dried at 40°C for at least two days. Before use, each sample was crushed to pass through a 2 mm screen and thoroughly homogenized.
The N test described by Khan et al. (2001) was performed as specified in a Technical Note cited in the References list. Two modifications were made to improve the uniformity of heating with the griddle employed (West Bend Model 76220): (i) replacement of the original temperature controller with an electronic unit, and (ii) rotation of jar positions after heating for 1.5 and 3 h. To ensure the validity of soil test data, all samples were analyzed in triplicate, and a check sample was included on each griddle. A single ISNT value was computed for 0–12 inches at each site, by averaging replicate data for 0–6 and 6–12 inch samples.
Mean values for the percentage response of corn to N fertilization were calculated as 100 ´ (optimum yield – check-plot yield)/check-plot yield, where optimum yield was determined by fitting N-rate and corresponding yield data to a quadratic plateau model by nonlinear regression. A corn:N price ratio of 0.08 was assumed in calculating the optimum N rate. Percentage values for N use efficiency were obtained as 100 ´ (optimum yield – check-plot yield)/optimum N, assuming one pound of N per bushel of grain. Nonresponsive sites were identified when the optimum yield did not differ significantly (P < 0.10) from the check-plot yield.
In Illinois, fertilizer N recommendations for corn are normally made according to a realistic yield goal, on the assumption that 1.2 pounds of N will be required, on average, per bushel of grain. Unfortunately, this approach does not account for mineralization of soil N, which usually supplies the majority of mineral N taken up during the growing season, even when the soil has been heavily fertilized with N (Bigeriego et al., 1979; Olson, 1980; Kitur et al., 1984; Torbert et al., 1993; Jokela and Randall, 1997; Omay et al., 1998). Not surprisingly, numerous investigations have indicated very little, if any, relationship between maximum economic yield and optimum N rate (Vanotti and Bundy, 1994; Fox and Piekielek, 1995; Kachanoski et al., 1996; Blackmer et al., 1997; Bundy, 2000; Lory and Scharf, 2003; Nafziger et al., 2003). Faced with this dilemma, scientists in Iowa and Wisconsin have abandoned yield goal as a basis for fertilizer N recommendations for corn.
Table 1 provides further reason to question the validity of yield-based N recommendations for corn production, particularly in a state such as Illinois, where approximately 1 million tons of N are applied annually, at a cost of more than $350 million. The data in Table 1 summarize results from 105 N-response studies conducted during six growing seasons at on-farm sites throughout Illinois. None of these sites was subject to drought during the growing season, yet remarkably, 52 were completely nonresponsive to N fertilization, representing almost exactly one-half of the total site-years studied. Some of the nonresponsive sites identified had been manured prior to the growing season studied, but the majority (54%) had not, and were cropped to continuous corn or corn after soybean.
On the basis of the proven-yield (PY) method described in the Illinois Agronomy Handbook (2002), only six of the 52 nonresponsive sites studied in our work would have been identified correctly for their complete lack of yield response to N fertilization. In each case, the N credit from manure (170–2,240 lb N acre-1) would have exceeded the N requirement estimated for the yield goal, although this sometimes occurred because of an additional N credit for corn following soybean (40 lb N acre-1). Of the 15 remaining manured sites, 13 were nonresponsive to N fertilization, but owing to a limited application of manure, none of these would have been predictable by the PY method. Likewise, this method was of no value whatsoever for detecting nonmanured sites where N fertilization was completely ineffective, even in cases where the maximal credit (100 lb N acre-1) was allowed for previous cropping to alfalfa.
Further examination of Table 1 suggests that the PY method was much more appropriate for the 53 responsive sites studied, in that N fertilization was invariably recommended. This assertion seems to be substantiated by the excellent agreement that was observed when experimentally determined values for optimum N (133 lb acre-1) and PY N recommendations (134 lb acre-1) were averaged for these 53 sites, although a considerable difference emerges when the same comparison was extended to all 105 sites, in which case the average N recommendation (126 lb acre-1) was almost twice the experimentally determined optimum N (67 lb acre-1).
A more critical analysis of PY N recommendations on a site-by-site basis reveals that most of these recommendations were subject to serious error, involving either under- or overfertilization. The magnitude of this error is summarized by Table 1, which shows that responsive as well as nonresponsive sites were affected, and that, on average, recommended N applications deviated from optimum N requirements by 33 to 123 lb N acre-1. Of the 105 site-years studied in our work, 70 were overfertilized, while 23 were underfertilized. Only for 12 site-years was the PY N recommendation accurate to within 10 lb acre-1, and in 6 of these cases a complete lack of N response was expected, owing to excessive application of manure prior to the growing season studied.
The recurring evidence of serious inaccuracy in N recommendations by the PY method has obvious economic implications for individual farmers, and also raises concern about environmental pollution. The only hope of improving N fertilizer recommendations in a humid region such as Illinois is to account for a soil’s capacity to supply plant-available N through mineralization. The ISNT was developed for precisely this purpose, and is designed to estimate a labile fraction of soil organic N, nominally referred to as amino sugar-N, rather than soil NO3– or total organic matter. Recent work by Haney et al. (2003) has confirmed that N determined by this test is highly correlated with soil microbial activity and net mineralization. As originated, the ISNT is employed to identify sites where corn responds very little, if at all, to N fertilization, and thereby to differentiate responsive from nonresponsive sites. The present paper permits a critical assessment of the performance of the ISNT, relative to PY N recommendations, in small-plot research evaluations that have been conducted to date.
Table 1 provides ample evidence that the ISNT far surpasses the PY method in identifying sites where N fertilization will be ineffective for increasing corn yield. Considering all 52 of the nonresponsive sites studied in our work, the ISNT was 90% effective, and unlike the PY method, the same effectiveness was achieved for the nonmanured and the manured sites. Four of the 5 failures observed occurred in 2003, and may be due, at least in part, to the stimulatory effect of excellent growing conditions on mineralization, which would have decreased the critical value for detection of nonresponsive sites by the ISNT. Spatial variability in soil N availability is also implicated, because the yield data, both within and among replicate plots, were erratic whenever the ISNT failed to detect a nonresponsive site, and test values also varied widely among plots. The latter finding is consistent with previous work showing substantial variation of the ISNT within a few feet (Boast et al., 2003), but more importantly, ISNT values tended to vary directly with check-plot yield and inversely with fertilizer N response. These relationships were eliminated when yield and soil test data were averaged for each site, with the result that the ISNT may not have accurately represented the area harvested for yield measurement. The same difficulty almost certainly vitiated a recent attempt to calibrate the ISNT by field-scale strip tests (Hoeft et al., 2003).
With respect to the 53 responsive sites studied in our work, Table 1 shows that the ISNT did not always predict when N fertilization increased corn yield, whereas no such error occurred with the PY method, as would have been expected because of the underlying assumption that all sites respond to N fertilization, except for cases involving extensive N credits. Critical examination of all information collected for the 10 sites where these failures occurred provides valuable insight concerning several likely factors. The most common occurrence was a marked decline in ISNT values with depth, which was particularly evident for samples taken between 12 and 24 inches. Yield and ISNT data often varied substantially among the replicate plots, and in cases where at least one of these plots was nonresponsive, the mean test value sometimes exceeded the critical range for 0–12 inches (225–240 ppm). Two of the 10 failures was probably due to excessive soil acidity, while another occurred where rye was plowed down after manuring, resulting in extensive immobilization that would have consumed the mineral N derived from the manure. Of the 10 responsive sites where the ISNT failed, five sites would have been overfertilized by the PY method (14–126 lb N acre-1), and five would have been underfertilized (15–44 lb acre-1).
As would be expected for any soil test that measures a mineralizable form of N, strong negative relationships have been observed between ISNT values and the percentage response to N fertilization, optimum N, and N use efficiency. This is illustrated by Figure 1, which was generated using data collected in 1990-1992, 2001, and 2002, from 79 of the site-years studied in our work. When all 105 site-years were included, the correlation coefficients were –0.51*** for yield response, –0.58*** for optimum N, and –0.48*** for N use efficiency, while a positive correlation was obtained in relating ISNT data to check-plot yield (r = 0.34*).
Figure 2 provides further evidence that the ISNT is sensitive to quantitative differences in soil N availability. For a nonmanured field under continuous corn for at least 34 yr, test values were substantially higher for a prairie soil (Drummer) than for a forest soil (Sabina). The spatial pattern in these values coincided with the differences observed in crop development and grain yield, suggesting an unprecedented potential for variable-rate N management. The same conclusion was reached in a recent study by Ruffo et al. (2003).
In 105 small-plot N-response evaluations to date, the ISNT identified 90% of the sites where corn did not respond to N fertilization, whereas < 6% of these sites were identified by the existing PY method, which often led to under- or overfertilization when evaluated on a site-by-site basis. Results to date have supplied remarkable evidence that the ISNT consistently detects spatial differences in crop N response. A very real opportunity now exists for site-specific N management, so as to improve the profitability of corn production and control the adverse environmental effects associated with N fertilization.
Appreciation is expressed to Cecilia Azpiroz Gutierriez, Edith Bermeo, and Esam El-Naggar for technical assistance in processing soil samples and collecting ISNT data.
1 R.L. Mulvaney is a professor, S.A. Khan is a research specialist, and T.J. Smith is a visiting research specialist, Department of Natural Resources and Environmental Sciences, University of Illinois. J.J. Warren and L.C. Gonzini are senior research specialists and R.G. Hoeft is a professor, Department of Crop Sciences, University of Illinois.
Bigeriego, M., R.D. Hauck, and R.A. Olson. 1979. Uptake, translocation and utilization of 15N-depleted fertilizer in irrigated corn. Soil Science Society of America Journal, 43:528–533.
Blackmer, A.M., R.D. Voss, and A.P. Mallarino. 1997. Nitrogen fertilizer recommendations for corn in Iowa. Pm-1714. Iowa State University, Ames, IA.
Boast, C.W., T.R. Ellsworth, T.J. Smith, R.L. Mulvaney, S.A. Khan, E.M. El-Naggar, and R.G. Hoeft. 2003. Spatial and temporal variability in the Illinois N test. In: Illinois Fertilizer Conference Proceedings 2003 (R.G. Hoeft, ed.). pp. 15–19.
Brown, H.M. 1996. Evaluation of nitrogen availability indices. Ph.D. thesis. University of Illinois, Urbana-Champaign, IL.
Bundy, L.G. 2000. Nitrogen recommendations and optimum nitrogen rates: How do they compare? In: North Central Extension Industry Soil Fertility Conference Proceedings, St. Louis, MO. Vol. 16. Potash and Phosphate Institute, Brookings, SD. pp. 5–13.
Fehrenbacher, J.B., J.D. Alexander, I.J. Jansen, R.G. Darmody, R.A. Pope, and M.A. Flock. 1984. Soils of Illinois. Bull. 778. University of Illinois, Urbana-Champaign, IL.
Fox, R.H., and W.P. Piekielek. 1995. The relationship between corn grain yield goals and economic optimum nitrogen fertilizer rates. Agron. Ser. 136. Penn State University, University Park, PA.
Haney, R., L.R. Hossner, S.A. Senseman, and A.J. Franzluebbers. 2003. Relationships between soil microbial activity and amino sugar nitrogen. Agronomy Abstracts. American Society of Agronomy, Madison, WI.
Hoeft, R.G., R.L. Mulvaney, S.A. Khan, E.D. Nafziger, J.J. Warren, L.C. Gonzini, T.K. Lehman, and A. Gulso. 2003. Illinois N soil test: Temporal and spatial variation and prediction of N response. In: Illinois Fertilizer Conference Proceedings 2003 (R.G. Hoeft, ed.). pp. 9–14.
Illinois Agronomy Handbook, 23rd edn. 2002. University of Illinois, Urbana-Champaign, IL. pp. 101–116.
Jokela, W.E., and G.W. Randall. 1997. Fate of fertilizer nitrogen as affected by time and rate of application on corn. Soil Science Society of America Journal, 61:1695–1703.
Kachanoski, R.G., I. P. O’Halloran, D. Aspinall, and P. Von Bertoldi. 1996. Delta yield: Mapping fertilizer nitrogen requirement for crops. Better Crops with Plant Food, 80:20–23.
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 Science Society of America Journal, 65:1751–1760.
Kitur, B.K., M.S. Smith, R.L. Blevins, and W.W. Frye. 1984. Fate of 15N-depleted ammonium nitrate applied to no-tillage and conventional tillage corn. Agronomy Journal, 76:240–242.
Lory, J.A., and P.C. Scharf. 2003. Yield goal versus delta yield for predicting fertilizer nitrogen need in corn. Agronomy Journal, 95:994–999.
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 Science Society of America Journal, 65:1164–1172.
Nafziger, E.D., R.G. Hoeft, E. Adee, A.H. Anderson, R.E. Dunker, S.A. Ebelhar, L.E. Paul, and G.A. Raines. 2003. Assessing variability in corn response to N rate. In: Illinois Fertilizer Conference Proceedings 2003 (R.G. Hoeft, ed.). pp. 21–25.
Olson, K.R., and J.M. Lang. 2000. Optimum crop productivity ratings for Illinois soils. Bull. 811. University of Illinois, Urbana-Champaign, IL.
Olson, R.V. 1980. Fate of tagged nitrogen fertilizer applied to irrigated corn. Soil Science Society of America Journal, 44:514–517.
Omay, A.B., C.W. Rice, L.D. Maddux, and W.B. Gordon. 1998. Corn yield and nitrogen uptake in monoculture and in rotation with soybean. Soil Science Society of America Journal, 62:1596–1603.
Ruffo, M.L., G.A. Bollero, D.S. Bullock, and D.G. Bullock. 2003. Field scale variability of corn yield response functions to nitrogen fertilizer. Agronomy Abstracts. American Society of Agronomy, Madison, WI.
Technical Note 02-01, rev. e. 2003. 15N Analysis Service, University of Illinois, Urbana-Champaign, IL.
Torbert, H.A., R.G. Hoeft, R.M. Vanden Heuvel, R.L. Mulvaney, and S.E. Hollinger. 1993. Short-term excess water impact on corn yield and nitrogen recovery. Journal of Production Agriculture, 6:337–344.
Vanotti, M.B., and L.G. Bundy. 1994. An alternative rationale for corn nitrogen recommendations. Journal of Production Agriculture, 7:249–256.
