Emerson D. Nafziger, Howard M. Brown, and Fabián G. Fernández 1
There has been a great deal of recent work on the response of corn yield to fertilizer N rates. Much of this work has been stimulated by the need to provide a stronger basis for determining optimal economic rates of N application, and by the need to minimize environmental consequences of corn production. Most results of such work have shown a large amount of variability in N response. Our own work, funded by FREC (Project Number 190) has been quantifying this variability, and these results have stimulated us to propose a new approach to using N rate data to formulate N use guidelines (Nafziger et al., 2005).
The variability in response to N fertilizer has been confirmed in numerous studies going back more than 50 years in Illinois. In the 1950s and 1960s, this variability was accommodated by recommendations that called for the use of N rates within a rather wide range (the 1971 Illinois Agronomy Handbook suggests using N rates “between 125 and 250 lb per acre” for corn following corn in Illinois). In the 1970s, this wide range of rates was narrowed by using the results of trials averaged over years to calculate a factor that related N rate to anticipated yield. This resulted in the “Yield Goal” or “Proven Yield” method (Hoeft and Peck, 2002), in which the recommendation was to apply “no more than” 1.2 lb of N for each bushel of expected yield for corn following corn, with credits given when corn follows a legume or when manure has been applied to the field. When N costs have been relatively low compared to the corn price, this recommendation system has suggested reasonable N rates.
A problem with the proven yield method when corn follows soybean is that our recent research has shown that the correlation between yield and the N rate needed to optimize yield is very weak. In other words, the use of expected yield to set N rate is not supported by research findings. This is not really surprising, because the proven yield method was developed by averaging research results over a number of trials, finding the optimum N rate and the yield at that N rate, then dividing the optimum N rate by the yield. This resulted in the factor of 1.1 to 1.2 (depending on the N:corn price ratio) used to set N rates. Much of this work was conducted on corn following corn, and there tends to be more correlation between yield and the N rate needed to reach that yield in continuous corn than where corn follows soybean. Hence the use of yield to suggest N rate may be somewhat more justifiable for corn following corn than for corn following soybean.
Working with agronomists in other Corn Belt states, we have developed a new approach to making N rate guidelines (Sawyer and Nafziger, 2005). This approach uses data from most of the recent N response work that has been conducted in Illinois, including both small-plot and on-farm results. This method involves calculating return to N based on data from each trial, then averaging these returns (at each N rate increment) over appropriate trials.
One of the main concerns in taking such an approach is whether or not the database is sufficiently large to support the outcome adequately. We undertook this work to enlarge the database on which N rate guidelines are based, using replicated, field-scale, N rate trials at numerous on-farm locations throughout Illinois, representing the diversity of soils and weather in the state. We also included sampling for the Illinois Soil Nitrogen Test (ISNT) in order to test the effectiveness of this test in describing/predicting N response in on-farm N rate trials.
Cooperators throughout Illinois were solicited to participate in this study through various publications, as well as in person. Criteria set out for participation were flexible, and included only that corn follow either corn or soybean, that N rates applied the previous fall or at the time of planting (for example, as starter or herbicide carrier) not exceed 40 or so lb N per acre, and that producers have the ability to apply N rates accurately. Partly because the request was made late (after funding was received), nearly all of the participants applied rates in the spring, mostly as sidedress after crop emergence. Approximately 60 cooperators participated.
During March and April, most fields where trials were scheduled to go were sampled for determination of the ISNT values. Samples were taken from 0 to 12” and from 12 to 24” deep using a truck-mounted probe. At most sites, nine samples were taken in the field within the general area where the trial was to be established. Sampling and analyses were done under contract by Cropsmith, Inc., a company located in Monticello, Illinois.
Nitrogen rates of 0, 50, 100, 150, and 200 lb N per acre (in addition to basal N rates applied uniformly across the field) were applied in a randomized complete-block design with three reps. Strip length and width were chosen by cooperators, with the general suggestion that harvest be done in only part of the width of strips in order to reduce border effects, and that plot length be no more than one-quarter mile long, in order to minimize total yield loss due to low N rates.
Data were taken either with a weigh wagon or a combine-mounted yield monitor. Yields were calculated for each strip on the basis of 15% moisture. Yields were averaged across reps, and curves were fit to the data using PROC NLIN of SAS, with the quadratic + plateau function chosen as the default. The quadratic portion of the curve was used to calculate the economically optimum N rate (EONR) and yield at that N rate. In order to simplify the presentation, we used the N cost:corn price ratio of 0.1 to calculate EONR for each site.
The ISNT values were averaged over the samples taken at each site, and the 0-12” average values were compared to EONR values over locations.
The return rate of data from these trials was high, with more than 90 percent of sites providing usable data. For reporting and consolidation, individual sites were separated into Northern, Central, and Southern locations, and in Central and Northern locations into corn following corn (CC) and corn following soybean (SC) sites. There were no CC sites in Southern Illinois in 2006.
The SC trials generally showed responses similar to those we have seen previously. In Northern Illinois, one of six SC sites showed little response to N, but N increased yields by an average of about 60 bushels per acre (Fig. 1). In Central Illinois, where there were 21 SC trials, the projected yield without N (most trials had some basal N, typically 20 to 30 lb) averaged 94 bu per acre, while the yield at the optimum N rate averaged 180 (Fig. 2). Only one trial showed a calculated optimum N rate greater than the highest N rate used, in which case the optimum was set at the highest N rate used. In Southern Illinois, N more than doubled average yield, from 83 bu per acre without N to 169 bu per acre at the optimum N rate (Fig. 3). Two of 10 trials in the South, both from the same area in Hamilton County, had optimum N rates higher than the highest N rate used.
Responses to N of corn following corn (CC) were unusually small in most trials in 2006. In Northern Illinois, N increased yield of six CC sites by an average of less than 40 bu per acre (Fig. 4). In Central Illinois, N increased yield by about 70 bu per acre (Fig. 5), less than the 86-bu yield increase from N in SC trials. One of the CC trials in Central Illinois showed almost no response to N, while another showed a decline as N rate increased (Fig. 5). The latter site was not used in calculating the overall response to N. Both of these non-responsive sites were in areas where dry weather affected the crop.
In order to see how results from these on-farm sites in 2006 differed from N response data (gathered through 2005) used to formulate current guidelines, the N rate required to produce the maximum return to N (MRTN) for these trials was compared to the MRTN values returned by the N rate calculator at the website http://extension.agron.iastate.edu/soilfertility/nrate.aspx. In Northern Illinois, the 2006 SC trials produced higher MRTN values than did the calculator, while the values based on six 2006 CC trials were more than 40% less than those returned by the calculator based on 23 previous trials (Table 1). This difference also existed in Central Illinois, but the MRTN for the 2006 CC trials was only about 18% less than for the previous 42 trials used by the calculator. In both Northern and Central Illinois, though, the MRTN for SC trials was considerably larger than the MRTN calculated from CC trials in this study. This is very likely due to the dry weather in 2005 and through the winter into 2006, resulting in much more than normal amounts of carryover N.
In Southern Illinois, rainfall was generally good in 2006 and yields were relatively high in most areas. There were no CC trials in Southern Illinois, but the MRTN based on the ten SC sites was 192 lb N per acre, or about 20% higher than the MRTN returned by the calculator based on 51 previous studies (Table 1).
There was a tendency for fields with lower ISNT values to require more N to optimize yields (Fig. 6). This may have been due in part to the weather effects discussed above, which resulted in CC trials showing unusually low responses to N, especially in Northern Illinois where ISNT values tend to be higher. Conversely, responses to N were unusually high in 2006 in Southern Illinois, where ISNT values tend to be low.
On-farm trials went very well in 2006, due mostly to excellent cooperators and those who worked with them. Credible data were received from nearly all locations. Trials where corn followed soybean (SC) produced more or less “typical” response data, with optimum N rates generally similar to those calculated from earlier trials. The N response was unusually low in most corn following corn (CC) trials, however, with average optimum N rates substantially less than those calculated from earlier studies, and also less than optimum N rates calculated for the 2006 SC trials. We think this was due mostly to larger than normal amounts of carryover N from the dry weather in 2005. Amounts of N needed to optimize yield tended to decrease as the ISNT values in the 0-12” soil depth increased. This reflected to some extent the high yields and high N rates needed in Southern Illinois, where ISNT values are low, and the lower than normal response to N, especially in the CC trials, where ISNT values tended to be higher.
Figure 1. N response for the 2006 on-farm, corn following soybean (SC) trials in Northern Illinois.
Figure 2. N response for the 2006 on-farm, corn following soybean (SC) trials in Central Illinois.
Figure 3. N response for the 2006 on-farm, corn following soybean (SC) trials in Southern Illinois.
Figure 4. N response for the 2006 on-farm, corn following corn (CC) trials in Northern Illinois.
Figure 5. N response for the 2006 on-farm, corn following corn (CC) trials in Central Illinois.
1Emerson Nafziger, Howard Brown, and Fabián G. Fernández are Professor, Growmark Agronomy Manager and Adjunct Associate Professor, and Assistant Professor, Dep. of Crop Sciences, University of Illinois.
Hoeft, R.G. and T.R. Peck. 2002. Soil testing and fertility. In: Illinois Agronomy Handbook, 23rd Edition. University of Illinois Extension, College of ACES, University of Illinois at Urbana-Champaign.
Nafziger, E.D., R.G. Hoeft, Eric Adee, R.E. Dunker, S.A. Ebelhar, and L.E. Paul. 2005. Assessing variability in corn response to N rate. pp. 12-18, Proc. Illinois Fertilizer Conference, Peoria, Illinois, January 24-26, 2005.
Sawyer, John E. and Emerson D. Nafziger. 2005. Regional approach to making nitrogen fertilizer rate decisions for corn. Proc. North Central Extension-Industry Soil Fertility Conference, November 16-17, 2005, Des Moines, IA.