Illinois Fertilizer
Conference Proceedings
January 24-26, 2005
| Home | 2005 Index | Search |
|
|
| Home | 2005 Index | Search |
S. A. Ebelhar, E. C. Varsa, G. K. Robertson, T. D. Wyciskalla, C. D. Hart, and J. Borrenpohl1
One of the biggest problems with utilizing soil tests for fertilizer recommendations under a variable rate application system is the uncertainty that the crops growing in that particular spot where the test is taken yields the same as the crops in other locations in the field. If a field averages 150 bu/acre corn, then maintenance fertilizer recommendations call for replacement of the P and K removed by the crop across the whole field. In reality, fields have a wide range in yield (for example, from 90 to 210 bu/acre). Using yield maps to correctly identify where the high and low yielding (plus everything in-between) areas are located would allow a more exact replacement of nutrients removed by a previous crop. Relying solely on soil tests for fertilizer recommendations tends to over-fertilize low-yielding areas and under-fertilize high yielding areas. Combining yield mapping and soil testing would reduce the amount of over-fertilizing of low yielding areas and under-fertilizing of high yielding areas. This may prove to be economically as well as environmentally friendly.
The objectives of this study were to
Field studies were started in 2001 at two locations in southern Illinois. The first location, Irvington, was a 48 acre field with a variable soil test K history and the second location, Belle Rive, was a 26 acre field with a variable soil test P and K history. The field at Irvington had such a high soil test P level that no P fertilizer was required, thus there was no attempt to variably apply P at this location. At both locations, soybeans were grown in 2001 and 2003, with corn grown in 2002 and 2004.
There were three fertilizer application treatments. Treatment 1 (VRT-1) consisted of a variable fertilizer application based on a soil test map for buildup but whole field average yield for maintenance. Treatment 2 (VRT-2) consisted of variable fertilizer application based both on a soil test map for buildup and normalized yield map for maintenance. Three years of previous crop history was normalized by expressing the yields as a percentage of the average yield for that crop that year. This normalized yield map was used as a productivity map to adjust yields higher or lower than the field average for the crop to be grown. For 2001 and 2003, we assumed a yield level of 35 and 45 bu/acre of soybeans for Belle Rive and Irvington, respectively. For 2002 and 2004, we assumed a yield level of 115 and 150 bu/acre of corn for Belle Rive and Irvington, respectively. Treatment 3 consisted of a uniform application or P and/or K based on a field-average soil test and buildup/maintenance applied uniformly for the whole field. Table 1 shows the range of fertilizer P (as 0-46-0) and K (as 0-0-60) spread across the two locations and the field average soil test P and K levels.
A randomized complete block design was used with field-length strips of each of the three treatments serving as plots. Plot widths were determined by the width of the fertilizer application equipment (60' for our study). There were six replications at Belle Rive and eight replications at Irvington. Harvest was performed with a combine equipped with a yield monitor and GPS. Yield maps were created using the yield monitor data. Corn (Mycogen 2E705) was planted on April 21 at Irvington and corn (Pioneer 34D72) was planted at Belle Rive on June 10, 2004.
Soil tests are being monitored yearly for changes in P and K levels, with samples taken in early spring each year. However, only the first year’s soil samples were used for determining the fertilizer recommendations and for the economic analysis.
The soil test K map and normalized yield map (Figure 1) were used to develop variable rate yield maps for VRT-1 and VRT-2 (Figure 2) for Irvington. Both treatments had a large acreage (15.4 acres out of 48) which received no fertilizer K in each of the four years. This is due to very high soil test K levels and our cut-off of K application when soil test levels exceeded 360 lb/acre. There was not a huge difference between the two VRT maps, an indication that yield variability across the field was not great. This means that the yields across the field were close to the field average yield (most were within 15% of mean).
Utilizing either the VRT-1 or VRT-2 method would have resulted in 900+ lbs less total potash applied at the Irvington location (Table 2) in 2001 and 2003, and 575+ lbs less in 2002 and 2004. The economics of this cost savings will be determined and discussed at a later date. Overall, there were no differences in grain yields for 2004 (similar to 2001-2003, please see prior years reports) among the three fertilizer application treatments even though there were some differences in yield maps (Figure 3). Field average yields were 160+ bu/acre across the three treatments, which is twice the yields of 2002. Because this study was laid out in replicated strips, further evaluations of the data can be made by analysis of variance procedures. Differences among treatments were less when compared by this method (Table 3) and there was no significant treatment effects. Over the four-year period, the VRT-1 treatment has had the lowest relative yields, but not significantly different than the other treatments.
Belle Rive site
The Belle Rive location had soil test P and K levels which indicated a need for additional fertilizer. The soil test P and K maps and normalized yield map (Figure 4) were used to develop the variable rate yield maps for VRT-1 and VRT-2 (Figures 5 and 6). Because there was not a large acreage at this site which needed no fertilizer P or K, the amount of fertilizer used was nearly equal among the fertilizer treatments both years (Table 2). There was not a huge difference between the two VRT maps either for P or K, an indication that yield variability across the field was not great.
There were no significant yield differences associated with method of fertilizer application at this site in 2004, very similar to past years (Tables 2 and 3).
Soil test changes
Soil test P and K levels have been monitored at each location with yearly samples taken in the spring prior to application of fertilizer treatments. Soil test levels from these samples have been summarized by breaking the data into categories of soil test levels ranging from very low to very high (Figure 7) as outlined in the Illinois Agronomy Handbook. At Irvington, the VRT-1 and VRT-2 applications are doing a better job of building up the frequency of samples in the 260-360 range. At Belle Rive, where P levels were relatively low, all of the treatments decreased the frequency of samples < 20 lb P/acre, but none appear to be increasing the frequency in the 50-70 range, which suggests that we are not applying enough fertilizer to build up soil tests. Soil test K levels are predominantly in the 100-200 lb K/acre range for all of the treatments but levels are increasing in the 200-260 range, most favorable in the VRT-1 treatments, but levels do not appear to be increasing as expected, an indication that our K rates are too low or that the handbook rate of 4 lb K20/lb soil test K to build-up K levels is not adequate for this location.
Comparing the relative uniformity of soil tests across the field show some improvement of soil test K with the variable application treatments at Irvington (Figure 8), but the best improvement appears to be with soil test P at Belle Rive.
When soil test levels are above the critical level where no fertilizer is needed for a large enough acreage, less fertilizer may be required using a VRT method of application. In our study, we found a fertilizer savings of about 900 lbs per year for a 48 acre field of soybeans and a savings of 575 lbs for corn. Whether this is an economical cost savings has yet to be determined. At our other location, there was no difference in fertilizer application among the three treatments for either P or K. Yield differences were very small among the treatments.
Soil tests are not building up as expected, but there is some indication of better uniformity of soil test values with the variable rate applications..
Table 1. List of fertilizer P and K application ranges for Belle Rive and Irvington, 2001-03.
Table 3. Strip trial grain yields, 2001-2004.
Figure 1. Soil test K map and normalized yield map, Irvington.
Figure 2. Variable rate K application maps for Irvington, 2004.
Figure 3. maps, Irvington, 2003.
Figure 4. Soil test P and K, and normalized yield maps, Belle Rive.
Figure 5. VRT-1 and VRT-2 P application maps, Belle Rive, 2004.
Figure 6. VRT-1 and VRT-2 K application maps, Belle Rive, 2004.
Figure 7. Changes in soil test levels over years for Irvington and Belle Rive.
Figure 8. Changes in soil test compared to uniform rates of application
1 S. A. Ebelhar is an agronomist, Dept. of Crop Sciences, Univ. of Illinois; E. C. Varsa is a professor, Dept. of Plant, Soil and General Ag., So. Illinois Univ.; G. K. Robertson is a graduate student; T. D. Wyciskalla is a researcher, Dept. of Plant, Soil and Gen. Ag., So. IL Univ., C. D. Hart is a senior research specialist, Dept. of Crop Sciences, Univ. of Illinois; and J. Borrenpohl is Precision Ag. Coordinator., AgriPride FS.
