Illinois
Fertilizer Conference Proceedings
January 26-28, 1998
| Home | 1998 Index | Search |
| Home | 1998 Index | Search |
The adoption of precision farming technology by producers has increased interest in managing inputs on a site-specific basis as opposed to a whole field approach. The use of yield monitors has demonstrated the magnitude of yield variability within a field. This variability may indicate different requirements for N fertilizer in different parts of a field. If so, N applications made on a site-specific basis should improve the efficiency of N fertilizer use.
The first step in developing site-specific N recommendations is to determine what factors can be used to adjust current N recommendations for a specific site. A yield monitor can be a valuable tool to evaluate N response within a field. Davis et al. (1996) examined the variability of corn grain yield and its relation to N response. Carr et al. (1991) suggested that fertilization by soil types may increase fertilizer profitability. Sawyer (1994) reviewed variable rate fertilization concepts, mostly with regard to P and K.
The objective of this study was to examine N responses as measured by yield monitors from field-scale N rate strips in farmer fields, and to use these responses along with soil type maps to see whether or not optimum N rates vary sufficiently within fields to support the use of site-specific N applications.
This trial was conducted at 20 site-years in Central Illinois from 1995 to 1997. Over the three years of this study, ten farmers participated, with trial locations in eight counties. Cooperators were selected based on their interest, experience in operating combine-mounted yield monitors, and availability of accurate N fertilizer application equipment. The list of cooperators and their locations is given in the following table:
Cooperator |
County |
|
Cooperator |
County |
Adams, John |
Logan |
Johnson, Mick |
Morgan |
|
Corzine, Leon |
Christian |
Reifsteck, John |
Champaign |
|
Dalenberg, Ken |
Piatt |
Sasse, Dean |
Dewitt |
|
Harford, Doug |
Grundy |
Sauder, Ken |
Tazewell |
|
Hunt, Andy |
Grundy |
Western, Kent |
Morgan |
Field-length strips with different N rates were established using
sidedressed N fertilizer (usually anhydrous ammonia) at each site. There were
two or three replications at each site. Yield data were collected by each participating
farmer, using his own yield monitor equipped with global positioning receivers
and differential correction systems to obtain field position. Each of the N rate
strips in the trial was assigned as a "load" on the yield monitor to
simplify identification of the strips.
The GIS programs ArcView 3.0a (ESRI, 97), Green Plan Analyst (Milby 1997), and Ag Link 3.1 (Agris, 97) were used to analyze the yield data. Yield data were filtered and outliers were removed. Then, data were extracted and analyzed from areas within each field representing different soil types and different yield levels. To avoid assigning low N rate strips as low-yielding areas, yield levels in low-N strips were corrected by adding to each data point in those strips the difference between the yield under optimum N rates and those under reduced N rates, calculated over the entire trial. Soil types were identified from digitized Soil Conservation Service maps. Responses to N were calculated both on the basis of all data from a particular yield level or soil type, and on the basis of areas selected visually within a particular yield level or soil type. This was done to avoid uncertainty in soil type boundaries, and to avoid the inclusion of yield monitor data that reflected yield monitor fluctuations rather than true yield level of that part of the field.
The quadratic-plateau model was used to develop a N response curve for each site, for the different soil types, and for areas of differing yield level. The economically optimum N rate was calculated from these curves based upon a N cost of $0.20/lb and a corn price of $2.60/bu.
Due to the extensive work required to analyze yield monitor data, results from 1997 are not yet available. Of the fourteen sites in this study in 1995-1996, eleven were responsive to N fertilizer. In order to present contrasting soil types, results from 1996 from two locations near one another—Harford and Hunt, both in Grundy County—but on quite different soil types will be shown and discussed.
The overall N responses (estimated from all yield data) were quite similar at the Harford and Hunt locations in 1996, with optimum N rates 28 lb/acre higher at the Harford site than at the Hunt site, but with optimum yields about 8 bu/acre higher at the Hunt site than at the Harford site (Figure 1). The Harford field, which consists of silt loam and silty clay loam soils, is considerably more uniform in soil type than the Hunt field, where soils range from silty clay loam to fine sandy loam (Figure 2). Yield ranges within the field were, however, similar for the two fields in 1996 (Figure 3).
Responses to N rate within defined areas of different soil types were similar
at the Harford location (Figure 4), perhaps as anticipated
due to the relatively high productivity ratings for all of these soil types.
At the Hunt location, the optimum N rate was considerably higher for the two
more productive soil types, while the least productive soil type (Ridgeville
fsl) produced the lowest yield but had the highest optimum N rate (Figure
5)
Our results (most of
which are not shown here) indicate that, across all locations, yield levels, and
soil types examined to date, there was not a strong correlation between yield
level and optimum N rate. While our study has not been extensive enough to support
conclusive statements about the usefulness and expected economic return from making
variable-rate, site-specific N applications, our data have not provided a rational
basis on which to make site-specific decisions on N rates. Much more extensive
data of this sort will need to be collected in order to assess economic benefits
from, and to develop guidelines for, the use of variable-rate, site-specific N
applications.
Figure 4. Yield responses by (A) soil type and by (B) yield level at the Harford site, 1996.
Figure 5. Yield responses by (A) soil type and by (B) yield level at the Hunt site, 1996.
1 P.D. Harrington is a Farm Manager at First Mid-Illinois Bank and Trust, Mattoon IL; E.D. Nafziger and R.G. Hoeft are Professors and K.S. Polo is a Graduate Research Assistant, Dept. of Crop Sciences, University of Illinois, Urbana IL.AGRIS. 1995. Applications Mapping. Ag Link 3.51.
