Yield Monitor Maps for P and K Fertilizer Rate Prescription Maps??
Developing a field’s variable rate fertilizer prescription map can be costly, including the time in taking plant tissue and/or soil samples, sample analysis costs, and later map development time. Soil sample analysis is particularly important to phosphorus (P), potassium (K), and soil acidity (pH) management. Soil sampling time may be in short supply when crop harvest is to be followed by the establishment of a succeeding crop. Soil test results are not always timely, further delaying prescription map development. Due to the expense, grid, or zone, sampling is often done only every 2 to 4 years, which raises the question of how much fertilizer is to be applied in other years.
Other precision technologies, especially yield maps, are being used to reduce the time crunch. Fertilizer prescription maps based on nutrient removal can be developed directly from a yield map by multiplying the yield by grain P or K concentration values taken from published tables. Intuitively, nutrients would be applied to replace nutrients removed by the previous crop. A random sample of the grain could be analyzed if values from published tables were thought inappropriate.
There are potential problems with this approach. Limiting factors other than nutrient deficiency, especially water stress (both too little and too much), can drive field yield patterns. Should this year’s insect/disease pressure or weed competition patterns drive fertilizer P and K application? If a low soil test P limits crop yield in one area, should that area then be fertilized according to the low P removal found with the low-yielding, P-deficient crop? The yield monitor map might improve fertilizer prescriptions, but how does it compare with other options?
We compared four approaches to generating field-scale fertilizer rate prescriptions: a) the “grid,” based on (expensive) grid soil sampling a field on a 180 x 200 ft grid (about 1 sample per 0.83 acres) and spatial analysis of the soil test results; b) the “composite,” based on a single average soil test value from all grid soil samples taken in the field; c) “yield map nutrient removal-tabular,” based on the field’s yield map, a single published grain P concentration value, and spatial analysis of the calculated nutrient removal values; and d) “yield map nutrient removal-local composite,” based on the field’s yield map, a single grain P concentration value determined on a composite grain sample taken from that field, and spatial analysis of calculated nutrient removal.
Two producer fields, designated 112 (51.4 ac) and 950 (43.4 ac), were chosen. The dominant soil in both fields was a well-drained Crider silt loam, but there were sizeable areas of only moderately well-drained soil (Lowell, Nicholson, or Tilsit silt loams). Field 112 had a history of chemical fertilizer applications, and 950 had a history of swine manure and fertilizer N applications. Corn yield was determined with a calibrated yield monitor. Grain and soil samples were taken at the same grid point, shown in Figure 1A for field 950. A digital elevation map was determined for each field (also shown in Figure 1A for field 950). Soil test P was determined by the Mehlich III extraction procedure at the University of Kentucky’s Division of Regulatory Services soil test laboratory. This lab also determined soil pH and organic matter on each soil sample. Grain tissue was analyzed for P by the University of Kentucky Plant and Soil Sciences Department’s Analytical Services Laboratory.
Maps were generated for crop yield/nutrient removal and soil test P. The tabular value used to calculate nutrient removal maps was 0.326 % P = 0.353 lb P2O5/bu. Table 1 shows the fertilizer rate prescription as related to P removal or soil test P values.
“Composite” soil test, grain yield, and grain tissue P information for the two fields are given in Table 2. On average, field 950 was higher in soil test P and organic matter, but soil pH values were similar. Grain yield was lower and more variable in 112 than in 950. For 950, grain P was close to the tabular value, and grain 112 was lower than the tabular value.
Figures 1a and 1b show sample point locations, elevations, and yields in 950. We generally found that lower elevation, soil erosion, and less than well-drained soil decreased corn yield in this moderately dry season. Considerable spatial variation in soil test P within 950 is shown in Figure 2a, but no fertilizer P would be recommended (Table 3) for the grid or composite soil test approaches because there were no areas with a soil test P value below 60 lb/acre. The nutrient removal-based fertilizer prescription map for 950 (Fig. 2b), using the yield map and the tabular grain P concentration, gave only two areas with rate prescription differences - due entirely to yield differences between these two areas (Fig. 1b). Comparing the four approaches to getting a fertilizer P prescription for 950, the nutrient removal-based fertilizer prescriptions always called for more fertilizer than the soil test based prescriptions for this field (Table 3). In fact, areas in the removal-based map calling for a greater fertilizer P rate (Fig. 2b) were often those areas with higher soil test P (Fig.2a).
The soil test P map for field 112 (not shown) also showed considerable spatial variation. Comparing prescription approaches for this field, fertilizer P is over-prescribed, relative to that recommended by “grid” sampling, by both nutrient removal approaches (Table 4). In 112, the greater difference between the grain P concentration value for grain taken from the field and the value taken from the table caused the removal-based fertilizer P prescriptions to differ. The “composite” soil analysis recommended a uniform rate of 30 lb P2O5 per acre for field 112.
Relative to grid soil sampling, the “composite” P rate prescription was appropriate for a third of the field, over-fertilized a third of the field, and under-fertilized a third of the field (Table 4).
We concluded that composite soil sampling was not always inferior to grid soil sampling in terms of the resulting fertilizer P or K prescriptions, especially when both approaches confirmed that no fertilizer was needed. In general, using yield-nutrient removal maps to derive fertilizer prescription maps resulted in greater prescribed P and K fertilizer rates than either soil test approach. We also observed that as the tabular grain P concentration value deviated from the field grain P concentration there was more of a difference in the nutrient removal-based fertilizer P prescription. Our results indicate that using yield monitor maps and grain P or K concentration information to develop variable rate fertilizer P and K rate prescription maps rests upon an assumption that was often not valid. We found P and K removal by the most recently harvested crop is not better related to the need for fertilizer P and K for the next crop than current soil test P and K values.
That said, our experience indicates that yield maps can be used, along with soil, topographic and other spatial information (satellite imagery) to divide a field into “management zones” that better capture crop production differences than simple square/rectangular grids. These zones, likely fewer, would then be soil sampled for nutrient management information.
Co-authors: Dr. John Grove, Professor of Agronomy/Soils Research and Extension, University of Kentucky, Dr. Eugenia Pena-Yewtukhiw, Assoc. Professor, Soil Physics and Management, Director, WVU Soil Testing Laboratory, West Virginia University, Morgantown, WV