D.W. Franzen 1
In 1961 a quietly radical soil sampling project was initiated by Drs. Sig Melsted and Ted Peck at the University of Illinois. A forty-acre field about 20 miles west of Champaign, Il, and another field near Urbana, were sampled in an 80-ft grid. The purpose of the study was to examine the variability of soil pH, P and K in actual farm fields, and to devise less aggressive and expensive sampling strategies to determine the central-tendency of those nutrients within fields. There are no records of the Urbana field ever being sampled again, however, the west field, called ‘Mansfield’ (the field is about 1 mile southwest of Mansfield, IL) or the ‘Warren Tract’ (the field is owned by the University of Illinois) was periodically sampled until 1982, then sampled annually from 1986 until 2002 under the direction of Dr. Ted Peck.
In 1982, Dr. Ted Peck began sampling another University of Illinois-owned forty-acre farm northwest of Thomasboro, IL in the same 80-ft grid as Mansfield. This field was sampled continuously from 1986 until 1999.
The following are soil sampling and grid-harvested crop yield dates for the two fields:
Mansfield- soil sampling dates- 1961, 1976, 1982, 1986, 1987, 1988, 1989, 1990, 1991, 1992, 1994, 1999.
Mansfield corn yield- 1991, 1993, 1999.
Mansfield soybean yield- 1992, 1994.
Thomasboro- soil sampling dates- 1982, 1986, 1987, 1988, June and October 1989, 1990, 1991, 1992, 1994, 1995, 1999, 2001.
Thomasboro corn yield- 1992, 1994, 1995, 1999.
Thomasboro soybean yield- 1996.
My involvement in the study began in 1989 as a graduate student working under the direction Dr. Peck. From 1989 through 1992 we worked together to accumulate soil, plant and yield data on the two farms. After I received my degree, the sampling became more periodic than annual. Dr. Peck began to have health problems related to a rare cancer, and became increasingly unable to continue the studies. A few months before he passed away, he asked if I would summarize his work and I agreed. I received all the data and with the assistance of the Illinois Fertilizer Research and Education Council funding committee, I have been able to summarize the data and will make it available in its entirety as a technical publication available in paper and on the web. The objective of this paper is to present a summary of the summary, and invite those interested in viewing or working with the data to access the technical publication for details.
All soil sampling was conducted using the same procedure in all years. The fields were flagged at each sampling location using a tape measure. The samples were 80-ft. apart, or 5-rods distant each other. The Mansfield site had 256 samples taken each year. The GPS coordinates of Mansfield are found in Figure 1.
The Thomasboro site was a US Air Force radar facility before being acquired by the University of Illinois. In the southeast, a small building was present, along with a V-shaped radar pole slightly to the northeast of the building. At Thomasboro, there was a cottonwood tree belt along the west border, and another tree belt between ranks 6 and 7 from the west side running almost the entire length of the field. The number of samples at Thomasboro varied due to the building site, but generally there were about 250 sampling points.
Each soil sample consisted of 5 soil cores 0-6 inches in depth. These cores were collected in a common bag, dried and ground prior to analysis for soil pH (1-1 water), P by the Bray P1 test, and K by ammonium acetate.
Corn yield in 1991 was determined by determining plant harvest population in each plot, then collecting 10 representative ears. In 1992, after considerable discussion about this method, a measured 20-foot of row from two interior rows in each plot were collected. Soybean yield was determined using a plot combine in 1992 using a long pole to separate the end of each plot area so that the combine did not extend farther into the next plot. From 1994 corn and soybean yields were determined using a plot combine.
Mapping for this publication was conducted using Surfer 8.0 for windows (Golden Software Co., Golden, CO). Parameters for kriged contour maps were determined using GS+ 5.0 for Windows, with the maps developed using these parameters within Surfer 8.0.
Elevations for use in terrain modeling were measured October, 2006 using a laser transit device with a stationary laser emitter. The readings are relative within each field independently.
Electrical conductivity measurements were conducted in November, 2006 using a Geonics EM-38 magnetic electrical conductivity sensor set for shallow EC measurement.
It would be impossible and impractical to present each soil/crop factor year of data into this proceeding. However, the following data are presented to provide support for the summarization of the data for this conference.
Lesson 1- Soil pH, P and K patterns were persistent unless obscured by large fertilizer and limestone application.
Between 1976 and 1991, limestone applications were made to parts of Mansfield. As Figure 3 shows, patterns in 1991 were very similar to those in 1976. Areas high in pH remained high, those with a tendency towards acidity retained that characteristic.
Soil P and K patterns also were persistent as shown in Figure 4. Thomasboro P patterns showed particularly low P in the northwest, southcentral areas. After large P application between 1982 and 1988, the P was allowed to decline due to crop removal. The final P map shows patterns similar to those in 1982, although the values of P had generally increased in the field due to earlier fertilization.
Lesson 2- When P and K applications were discontinued, yields remained high and total P and K removal at both sites exceeded P and K inputs, while P and K levels remained high. The rate of P soil test draw-down was slower than fertilization buildup rates. The rate of K soil test draw-down was faster than fertilization buildup rates.
Figure 5 shows an example of differences in rate of K buildup compared to draw-down rate. The rate of draw-down is slower than buildup. During this period, crop yields remained high, as soil test K levels suggest.
At Mansfield, 5.4 lb/a P2O5 added resulted in a 1 lb/a increase in P1 test. It took 7.9 lb P2O5 draw-down to decrease the P1 test 1 lb/a. It took 6 lb/a K2O to increase the K test 1 lb/a, but only 1.7 lb/a K2O draw-down to decrease the K test 1 lb/a.
At Thomasboro, 5.8 lb P2O5 added resulted in a 1 lb/a increase in the P1 test. It took 15 lb/a P2O5 draw-down to decrease the P1 test 1 lb/a. It took 7.1 lb/a K2O to increase the K test 1 lb/a, but it took only 3 lb/a K2O draw-down to decrease the K test 1 lb/a.
Lesson 3- Nutrient difference maps did not generally represent historic nutrient patterns within two-three years of each sampling, however, long-term difference maps tended to be similar in pattern to soil/landscape.
Usually, two to three-year nutrient difference analysis did not show any particularly characteristic pattern in the field. However, often, particularly with pH, but also with P and K, nutrient change maps over a period of years showed characteristic patterns in the field initially seen with the nutrient levels of each nutrient. This indicates that patterns are initially seen in the field because there are intrinsic properties of the soils within the field to reveal those patterns over time. Figure 6 is an example of this property.
Lesson 4. Yield frequency mapping more consistently represented nutrient patterns and were more related to nutrient levels than single-year yield maps.
Yield frequency maps use data from multiple years of normalized yield, regardless of crop type, combined to create new data that reflects tendencies of yield from most consistently high yield to most consistently low yield. Figure 7 shows the yield frequency map for Thomasboro. It shows the effects of the tree line and also several characteristic zones that relate directly to nutrient levels.
Lesson 5- Simulated zone sampling using topography and soil EC were generally similar in pattern to dense (1 sample per acre) grid sampling for pH, P and K.
Mapping the elevation surfaces of both fields provided insight into crop yield and nutrient patterns. Patterns of elevation were similar to long-term patterns in pH and other nutrients. Layering EC with elevation provided additional relationships with nutrients (Figure 8).
Table 1 shows the correlation values (r) for Mansfield pH, P and K between the original data and meta-data for those points generated using a 220 ft or 330 ft grid kriged map. A 220 ft grid, with r values higher than 0.15 would be significantly correlated. Tables 2 and 3 show correlation developed similarly for electromagnetic sensor zones (EM), elevation (EL) zones and combination zones. P was particularly difficult at Mansfield due to heavy past fertilization and an old feedlot/building site in the northwest corner (see technical bulletin) unrelated to landscape or other field physical properties. However, the zone relationship with pH and K was similar with zone maps as the 220 ft grid relationships.
Lesson 6- Satellite imagery of July-growing crops showed similar patterns as nutrients, soil EC, yield frequency maps and topography.
Lesson 7- Small-scale nutrient sampling (8-foot grids) showed that attention needs to be paid to fertilizer application to avoid streaking. The small-scale sampling also shows that although zone and grid sampling maps suggest long transitions between nutrient levels, the transition can also be abrupt.
The data set of Mansfield consists of grid-sampled data dating to 1961. The data set of Thomasboro dates from 1982. These data are presented in a technical publication that is soon to be released. The lessons learned from this summary should help guide growers and consultants in their movement towards site-specific management of crop nutrients.
Funding to support the summary of these data, and also funding for a considerable portion of the modern data collection was made possible through support from the Illinois Fertilizer Research and Education Council. Thanks also to Agri Images, Maddock, ND for acquiring satellite imagery for the sites at no charge to the project.
Table 2. Comparisons of electro-magnetic and elevation zones with soil pH, P and K, Mansfield, 1961.
Table 3. Comparisons of electro-magnetic and elevation zones with soil pH, P and K, Mansfield, 1991.
Figure 1. GPS coordinates for Mansfield.
Figure 3. Mansfield pH, 1976 and 1991.
Figure 4. Thomasboro P, 1982 and 1991.
Figure 5. Differences in rate of K buildup and K draw-down due to crop removal.
Figure 6. K changes in Mansfield, 1988-1990.
Figure 7. Yield frequency map for Thomasboro.
Figure 8. Elevation surface of Mansfield.
Figure 9. Mansfield Landsat image, July 4, 1999.
Figure 10. Thomasboro Landsat image, July 16, 2006.
1 D.W. Franzen is an Extension Soil Specialist at North Dakota State University, Fargo, ND